from monash.edu.au - Monash University Research Repository

Transcription

TESTING THE NEUROBIOLOGICAL MODEL OF EMOTION-ENHANCED MEMORY WITH
EMOTION ELICITED BY MUSIC
PhD Thesis
by
SHERILENE M. CARR
Bachelor of Arts (Honours)
MONASH UNIVERSITY
School of Psychology and Psychiatry
Melbourne, Australia
April, 2013
TABLE OF CONTENTS
ABSTRACT.................................................................................................................................................... vii
GENERAL DECLARATION ............................................................................................................................. ix
ACKNOWLEDGEMENTS ................................................................................................................................ x
LIST OF TABLES ............................................................................................................................................ xi
LIST OF FIGURES ......................................................................................................................................... xii
LIST OF ABBREVIATIONS ............................................................................................................................ xiv
1
INTRODUCTION................................................................................................................................... 1
1.1 Aims and hypotheses .................................................................................................................. 3
1.2 Thesis overview ........................................................................................................................... 4
2
LITERATURE REVIEW ........................................................................................................................... 6
2.1 The neurobiological model of emotion-enhanced memory ....................................................... 6
2.1.1
Neurobiological modulation of memory in humans .................................................... 10
2.1.2
Psychophysiological changes associated with emotion-enhanced memory ............... 13
2.2 Cognitive processing and facilitated memory .......................................................................... 15
2.3 Differentiating encoding from consolidation effects on memory ............................................ 19
2.4 Evidence of emotion elicitation ................................................................................................ 25
2.4.1
Subjective feeling and peripheral efference ................................................................ 27
2.4.2
Cognitive ...................................................................................................................... 27
2.4.3
Motivation and motor expression ............................................................................... 30
2.5 Music as an emotion stimulus .................................................................................................. 32
2.5.1
Subjective feeling component ..................................................................................... 35
2.5.2
Peripheral efference and motor expression components ........................................... 38
2.5.3
Cognitive component ................................................................................................... 43
2.6 Music effects on memory encoding and consolidation ............................................................ 49
2.7 Individual differences................................................................................................................ 53
2.8 Conclusion ................................................................................................................................. 53
ii
3
GENERAL METHOD ........................................................................................................................... 56
3.1 Participants ............................................................................................................................... 56
3.2 Emotion elicitation .................................................................................................................... 57
3.3 Material to be remembered (MTBR) ........................................................................................ 59
3.4 Apparatus .................................................................................................................................. 59
3.5 Emotion Measures .................................................................................................................... 61
3.5.1
Subjective feeling ......................................................................................................... 63
3.5.2
Peripheral efference component ................................................................................. 65
3.5.3
Cognitive component ................................................................................................... 66
3.6 Memory measures .................................................................................................................... 68
3.7 Measures of individual differences........................................................................................... 70
4
EXPERIMENT ONE: Emotional story and emotional music effects on memory ............................... 71
4.1 Introduction .............................................................................................................................. 71
4.2 Aims and hypotheses ................................................................................................................ 72
4.2.1
Non-music aims and hypotheses ................................................................................. 73
4.2.2
Music aims and hypotheses ......................................................................................... 73
4.3 Method ..................................................................................................................................... 74
4.3.1
Participants .................................................................................................................. 74
4.3.2
Emotion manipulation ................................................................................................. 74
4.3.3
Apparatus ..................................................................................................................... 77
4.3.4
Measures ...................................................................................................................... 78
4.3.5
Procedure ..................................................................................................................... 80
4.4 Results ....................................................................................................................................... 81
4.4.1
Continuous measurement of subjective emotion ....................................................... 81
4.4.2
Non-music manipulation .............................................................................................. 82
4.4.3
Emotional response ..................................................................................................... 86
4.4.4
Interim Discussion ........................................................................................................ 90
4.4.5
Music manipulation ..................................................................................................... 91
4.4.6
Emotional response ..................................................................................................... 95
4.4.7
Extraneous music variables .......................................................................................... 99
iii
4.4.8
Interim Discussion ...................................................................................................... 100
4.5 General Discussion .................................................................................................................. 101
5
4.5.1
Neurobiological modulation of memory.................................................................... 101
4.5.2
Music as a source of extrinsic emotional arousal ...................................................... 105
4.5.3
Limitations.................................................................................................................. 106
4.5.4
Conclusion .................................................................................................................. 107
EXPERIMENT TWO: Participant-selected music effects on emotional arousal and memory ......... 108
5.1 Introduction ............................................................................................................................ 108
5.2 Aims and hypotheses .............................................................................................................. 109
5.3 Method ................................................................................................................................... 110
5.3.1
Participants ................................................................................................................ 110
5.3.2
Emotion manipulation ............................................................................................... 110
5.3.3
Material to be remembered (MTBR) ......................................................................... 110
5.3.4
Apparatus ................................................................................................................... 113
5.3.5
Measures .................................................................................................................... 113
5.3.6
Participant music selection procedure ...................................................................... 115
5.3.7
Laboratory procedure ................................................................................................ 116
5.4 Results ..................................................................................................................................... 118
5.4.1
Data screening and manipulation checks .................................................................. 118
5.4.2
Emotional response ................................................................................................... 120
5.4.3
Music effects on memory .......................................................................................... 124
5.4.4
Mood-congruent memory ......................................................................................... 126
5.4.5
Individual differences ................................................................................................. 127
5.5 Discussion................................................................................................................................ 127
6
EXPERIMENT THREE: Post-learning music effects on long-term memory ..................................... 131
6.1 Introduction ............................................................................................................................ 131
6.2 Method ................................................................................................................................... 132
6.2.1
Participants ................................................................................................................ 132
6.2.2
Emotion manipulation ............................................................................................... 133
iv
6.2.3
Material to be remembered (MTBR) ......................................................................... 134
6.2.4
Apparatus ................................................................................................................... 134
6.2.5
Measures .................................................................................................................... 135
6.2.6
Procedure ................................................................................................................... 136
6.3 Results ..................................................................................................................................... 137
6.3.1
Data screening and treatment ................................................................................... 137
6.3.2
Emotional response ................................................................................................... 138
6.3.3
Memory ...................................................................................................................... 144
6.3.4
Individual differences ................................................................................................. 147
6.4 Discussion................................................................................................................................ 147
7
6.4.1
Emotion effect on memory ........................................................................................ 148
6.4.2
Emotion components of good and poor memory performance ............................... 149
6.4.3
Limitations.................................................................................................................. 151
6.4.4
Conclusion .................................................................................................................. 152
GENERAL DISCUSSION .................................................................................................................... 154
7.1 Emotion elicitation .................................................................................................................. 157
7.1.1
Effective control of the emotional response ............................................................. 160
7.1.2
Chills as a measure of music induced emotion .......................................................... 162
7.2 Neurobiological modulation of memory ................................................................................ 165
7.3 Free recall and recognition memory differences.................................................................... 167
7.4 Methodological considerations .............................................................................................. 168
7.5 Conclusions and future directions .......................................................................................... 171
8
REFERENCES .................................................................................................................................... 177
9
APPENDICES .................................................................................................................................... 195
A. Definitions of information categories across studies ................................................................. 196
B. Summary of aims and hypotheses.............................................................................................. 202
C. Digital copy of Experimental Stimuli........................................................................................... 206
D. Exp. 1 Linguistic Properties of the Narratives ............................................................................ 207
v
E. Exp. 1 Narrative Pilot Testing ...................................................................................................... 208
F. Exp. 2 Emotion variables Pearson’s Correlation Results............................................................. 210
G. Exp. 3 Music selection pilot ........................................................................................................ 212
H. Exp. 3 Equipment set-up ............................................................................................................ 214
I. Exp. 3 Correlation matrix for the memory and emotion measures ............................................ 215
J. Exp. 3 Good vs. poor memory performers .................................................................................. 216
K. Exp. 3 Good vs. Poor memory t-test results for each emotion measure and each group ......... 218
L. Supplementary investigation of individual differences .............................................................. 220
L.1
Method of analysis ..................................................................................................... 223
L.2
Results ........................................................................................................................ 224
L.3
Testing information processing speed mediation of arousal and memory ............... 229
L.4
Conclusion .................................................................................................................. 232
M. High versus low chills responder analysis ................................................................................. 233
vi
ABSTRACT
Extensive research has revealed that central and peripheral physiological mechanisms that act to assess
and respond to negative and arousing emotional events also act to consolidate memory for the event.
An area of research yet to be fully investigated is the effect of positive and arousing emotion on longterm memory. The paucity of research may be due to the difficulty in experimentally manipulating
positive and arousing emotions in the research laboratory. A source of emotional arousal yet to be fully
explored in this context is music. A number of studies now demonstrate that music a) elicits strong
subjective feelings of emotion, b) activates limbic structures involved in emotion processing, and
c) elicits physiological responses consistent with emotion. The aim of the current research project was
to determine whether positive emotional arousal elicited by music could facilitate long-term declarative
memory.
Three experiments were conducted with a total of 127 participants ranging in age from 18 to 50 years
(M = 28.10, SD = 10.05), 68% of which were female. Full exploration of the relationship between
emotion elicited by music and non-music stimuli and memory was made possible with the use of a wide
range of emotion measures and material to be remembered. The aims of Experiment 1 were to: a)
replicate the emotion-enhanced memory effect reported by previous researchers using a three-phase
slideshow paradigm; and b) determine whether the presentation of emotionally arousing background
music further enhanced memory for the slideshow. The results revealed that relative to non-music and
neutral music comparison conditions, the experimenter-selected music had no effect on emotional
arousal, or on long-term declarative memory. Experiment 2 was designed to establish the emotion
inducing properties of music with participant-selected positive and arousing music, and to determine
whether emotion elicited in this way influenced memory. The results confirmed that participantselected music elicited subjective and physiological arousal responses that were consistent with positive
and arousing emotions, and that this music facilitated the early stages of long-term memory.
Experiment 3 tested the memory modulating effect of positive and arousing music on memory
consolidation by presenting music soon after learning. Results revealed that after controlling for high
baseline arousal levels and memory ability, music modulated memory consolidation processes.
However, music tended to have an impairing effect on memory in this experiment.
vii
Conclusions drawn from this project were that music has the capacity to elicit positive emotion, and that
this music induced emotion has the potential to both enhance and impair long-term memory. The time
of music presentation, and participants’ choice in music utilized, may determine the direction of this
effect. These conclusions must, however, be accepted with caution given the low sample size for many
of the statistical analyses. The major contribution of this research project is the methodological rigor
applied to understanding the conditions under which emotion elicited by music influences memory.
Future research investigating the influence of music on cognitive functioning can thus proceed from a
stronger base of empirical evidence.
Under the Copyright Act 1968, this thesis must be used only under the normal conditions of scholarly fair
dealing. In particular no results or conclusions should be extracted from it, nor should it be copied or
closely paraphrased in whole or in part without the written consent of the author. Proper written
acknowledgement should be made for any assistance obtained from this thesis.
I certify that I have made all reasonable efforts to secure copyright permissions for third-party content
included in this thesis and have not knowingly added copyright content to my work without the owner's
permission.
viii
GENERAL DECLARATION
I declare that this thesis contains no material which has been accepted for the award of any other
degree or diploma in any university or other institution.
I affirm that to the best of my knowledge this thesis contains no material previously published or written
by another person, except where due reference is made in the text of the thesis.
…………………………………………………………………….
……………………………………..
Sherilene Carr
Date
ix
ACKNOWLEDGEMENTS
This thesis was made possible with the support of my supervisor, Nikki Rickard, the people who
volunteered their time to participate in these experiments, and to my family. Nikki’s mentoring has
been invaluable in teaching me the discipline of the experimental method and the essential value of
research rigour. It goes without saying that research volunteers are critical to furthering our
understanding of the human mind. My deepest gratitude goes to my family, who have provided me
with unwavering support in my quest to understand the power of music and how it influences our
thinking and behaviour.
x
LIST OF TABLES
Table 2.1
Post-encoding arousal effects on human memory ...................................................................22
Table 2.2
Component Process Model of emotion ....................................................................................26
Table 2.3
EEG activation in response to music .........................................................................................46
Table 3.1
Emotion elicitation across the three experiments ....................................................................58
Table 3.2
Apparatus utilised across experiments .....................................................................................60
Table 3.3
Emotion component measures utilised across experiments ....................................................62
Table 3.4
Memory measures and testing delay across experiments ........................................................68
Table 4.1
Exp. 1 IAPS images and narratives.............................................................................................75
Table 5.1
Exp. 2 Summary of emotion response results ........................................................................123
Table 5.2
Exp. 2 Multiple regression results ...........................................................................................126
Table 6.1
Exp. 3 Summary of subjective emotion response results .......................................................139
Table 6.2
Exp. 3 Summary of ANS response results ...............................................................................141
Table 6.3
Exp. 3 Summary of cortical activation results ........................................................................143
Table 6.4
Exp.3 Correlations between emotion component variables...................................................144
Table 6.5
Exp. 3 Memory results .............................................................................................................145
Table 7.1
Summary of variables tested and outcomes ..........................................................................156
Table 7.2
Predicted versus actual stimulus arousal and valence ............................................................158
Table 7.3
Summary of recommendations ..............................................................................................174
Table G1
Exp. 3 Music selections pilot tested .......................................................................................212
Table L1
Moderator matrices across experiments ...............................................................................225
xi
LIST OF FIGURES
Figure 2.1
Neurobiological model of amygdala mediated emotion-enhanced memory ............................ 8
Figure 2.2
Parallel cognitive and autonomic emotional responses regulated by the amygdala ...............16
Figure 2.3
Neural Network Model ..............................................................................................................17
Figure 2.4
Hypothesised information processing effects on encoding and consolidation ........................20
Figure 2.5
Solution to isolate information processing effects on memory ...............................................21
Figure 2.6
The circumplex model of emotion ............................................................................................37
Figure 2.7
Summary of studies measuring physiological responses to music ...........................................41
Figure 2.8
Music effects on memory reported in previoius studies ..........................................................51
Figure 3.1
Affect grid used in Experiments Two and Three .......................................................................64
Figure 3.2
Positive and negative face line drawings ..................................................................................67
Figure 4.1
Continuous self-report of emotion ...........................................................................................78
Figure 4.2
Number of correct 4AFC recognition memory test items per phase (no music) .....................82
Figure 4.3
Number of participants to recall each slideshow image (no music) ........................................83
Figure 4.4
4AFC mean percentage correct for each image (no music) .....................................................84
Figure 4.5
Images freely recalled as a function of image valence (no music) ...........................................85
Figure 4.6
4AFC recognition correct as a function of image valence (no music) ......................................86
Figure 4.7
Non-music mood valence and arousal ratings ..........................................................................87
Figure 4.8
Non-music continuous changes in physiological response .......................................................88
Figure 4.9
Non-music recognition memory (4AFC) narrowing ..................................................................89
Figure 4.10 Number of participants to recall each slideshow image (music) .............................................92
Figure 4.11 4AFC mean percentage correct for each image (music) ..........................................................93
Figure 4.12 Images freely recalled as a function of image valence (music) ................................................94
Figure 4.13 4AFC recognition correct as a function of image valence (music) ............................................94
Figure 4.14 Music mood valence and arousal ratings ..................................................................................96
Figure 4.15 Music continuous changes in physiological response ...............................................................97
Figure 4.16 Music recognition memory narrowing ......................................................................................99
Figure 4.17 Memory comparison with previous studies using three-story-phase slideshow ...................102
Figure 5.1
Exp. 2 IAPS image collage ........................................................................................................112
Figure 5.2
Stimuli and tasks used in Exp. 2...............................................................................................118
xii
Figure 5.3
Exp. 2 Subjective and physiological responses........................................................................121
Figure 5.4
Exp 2. Free recall .....................................................................................................................124
Figure 5.5
Exp 2. Valence free recall ........................................................................................................126
Figure 6.1
Exp 3. I-PANAS-SF adjective ratings.........................................................................................140
Figure 6.2
Exp 3. Music-Mix BVA change .................................................................................................142
Figure 6.3
Exp 3. Good and poor free recall performer RSP and Frontal Theta values ...........................146
Figure 6.4
Exp 3. Good and poor word recognition performer valence and IBI values ...........................147
Figure 6.5
Exp.3 Word list freel recall and recognition comparison with previous research ..................148
Figure 7.1
Emotion responses across experiments mapped onto two-dimensional emotion space ......159
Figure E1
Pilot study emotion statement agreement ratings .................................................................208
Figure E2
Pilot study 4AFC recognition memory .....................................................................................209
Figure L1
Influence of BIS/BAS sensitivity on the relationship between RT and free recall ...................227
Figure L2
Influence of music listening on the relationship between skin temp and free recall .............228
Figure L3
Influence of music enjoyment on the relationship between SCL/BVA and free recall ...........229
Figure L4
Testing RT mediation of the relationship between SCL and memory for low BAS scorers.....231
Figure L5
Testing RT mediation of the relationship between SCL and memory for high BAS scorers ...231
xiii
LIST OF ABBREVIATIONS
%HF HRV
°C
2DES
4AFC
ACC
ACh
ANS
BIS/BAS
BLA
BMEQ
BVA
CNS
EEG
EEM
EMG
EOG
EPI
FFT
fMRI
GABA
HR
HRV
IAPS
IBI
IPIP
LC
ln
LTM
LTP
MRA
ms
MTBR
NAc
NE
NTS
OsM
proportion of high frequency
components of heart rate
variability
degrees Celsius
two-dimensional emotion space
four-alternative forced choice
anterior cingulate cortex
acetylcholine
autonomic nervous system
behavioural inhibition and activation
system
basolateral amygdaloid complex
brief music experience questionnaire
blood volume amplitude
central nervous system
electroencephalography
emotion-enhanced memory
electromyography
electro-occulargram
epinephrine (adrenalin)
fast Fourier transformation
functional magnetic resonance imaging
PA Music
PET
PFC
PNS
PsM
Q1, Q2
Q3, Q4
RMS
RSA
RT
RTln
SCL
SCR
SEC
SNS
Temp
μS
y-aminobutyric acid
heart rate
heart rate variability
international affective picture system
interval between normal-to-normal beats
International Personality Item Pool
locus coeruleus
natural log
long-term memory
long-term potentiation
multiple regression analysis
milliseconds
material to be remembered
nucleus accumbens
norepineprhine (noradrenalin)
nucleus tract solitaris
other-selected music
xiv
positive and arousing music
positron emission tomography
prefrontal cortex
peripheral nervous system
Participant-selected music
quadrant 1, quadrant 2
quadrant 3, quadrant 4
root mean square
respiratory sinus arrhythmia
response time
response time natural log transformed
skin conductance level
skin conductance response
stimulus evaluation check
sympathetic nervous system
skin temperature
micro Siemens
1 INTRODUCTION
The phenomenon of enhanced long-term declarative memory for emotional events is well
documented. The memory is often vivid and encompasses the cause, context, subjective feelings,
actions, and consequences of the event. These memories can be decades old yet remain as vivid as if
they occurred relatively recently. Memory for emotionally powerful events, such as the 9/11 terrorist
attack, confirm that people recall such events in more detail (though not necessarily with greater
accuracy; see Neisser & Harsch, 1992) than less emotional events (Sharot, Martorella, Delgado, &
Phelps, 2007; Talarico & Rubin, 2003a; Wolters & Goudsmit, 2005a). Furthermore, the phenomenon is
sufficiently robust to be elicited by moderate emotional stimuli in research laboratories. Mechanisms
postulated to cause such robust memory effects include emotion induced changes in brain chemicals
that modulate permanent neural change in memory networks, and increased attention, rumination, and
integration of salient information within the current knowledge store. The capacity to manipulate
emotion-enhanced memory (EEM) in the research laboratory raises the possibility that the phenomenon
could be used as a memory aid in everyday situations. This is an exciting proposition given memory is
one of the most important functions of any living organism. The laboratory based research findings
could have innumerable practical applications, such as the development of programs to enhance
learning in schools and in the workplace, and to challenge memory decline in an increasingly aging
population.
One of the most empirically validated models of EEM, and thus the primary theoretical basis of
this thesis, is founded upon the neurobiology of emotion. The neurobiological model posits that
emotional stimuli activate the amygdala, a limbic structure involved in the generation of emotional
responses, which then modulates a chain of biochemical responses in the autonomic and central
nervous systems, leading to protein synthesis and dendritic growth in cortical structures that store longterm declarative memory. The principal focus of the neurobiological model is on the interaction
between emotion-induced changes in the adrenal arousal hormones epinephrine (EPI), norepinephrine
(NE), and glucocorticoids, and beta-adrenergic receptors within the basolateral amygdaloid complex
(BLA).
An alternative theory of memory formation that will be tested in this thesis is founded upon
Hebbian principles and is more ‘top-down’ in nature. It has been postulated that EEM is caused by the
interaction between primed goal related or schematic neural networks and incoming perceptual
1
CHAPTER 1. INTRODUCTION
___________________________________________________________________________________
information. The reduced firing thresholds in these primed neural networks would increase the
probability that ‘bottom-up’ stimulus driven neural impulses will activate the network. Continued firing
of the associated neurons during a post-encoding elaboration and rehearsal period would increase the
probability of dendritic connection formation. Arousal elicited by the emotion stimulus would result in
increased excitatory neurotransmitter activity (e.g., acetylcholine and glutamate) in the cortex that
underlies efficient synaptic transmission, thus facilitating the formation of permanent dendritic
connections (long-term potentiation). Research that simultaneously tests neurobiological and neural
network models of EEM is therefore crucial to advance understanding of this complex phenomenon.
The neurobiological model has provided significant insights into EEM in animals. Nevertheless,
the contributions of memory-facilitating processes that may be unique to humans, such as postencoding elaboration upon unusual or interesting experimental stimuli, potentially confounds previous
human research results. Furthermore, the stimuli used in human studies may be described as surprising
or attention-capturing, thus eliciting an orienting reflex, which is qualitatively different to emotion. An
orienting reflex is accompanied by an acute increase in central and peripheral activation, possibly
mediated by the amygdala, which quickly dissipates. It is thus possible that amygdala activation is
associated with facilitated memory for perceptually salient information in the absence of a true
emotional response. This research project will therefore measure multiple components of emotion to
determine whether facilitated memory can indeed be accounted for by emotion, as distinct from normal
fluctuations in physiological arousal levels associated with abrupt or unusual information.
An additional aspect of EEM yet to be fully explored is the effect of positive emotions on
memory. This may be due in part to the effectiveness of negative stimuli to elicit the intended amygdala
mediated arousal response. However, the choice of negative and arousing emotional stimuli limits our
understanding of the full range of emotion effects on memory. For instance, negative and arousing
emotional stimuli have been demonstrated to strengthen memory for central details of events while
memory for peripheral information tends to be impaired. In contrast, according to ‘broaden and build’
theory of information processing, memory for positive emotional events may be less detailed but cover
a broader range of information. This could mean that the elicitation of negative emotional states
facilitates memory accuracy for a narrow range of information, while positive emotional states could
benefit memory more generally. A greater understanding of variations in emotion valence effects on
memory is required before attempting to apply the phenomenon in everyday contexts.
2
___________________________________________________________________________________
An effective emotion stimulus yet to be fully utilised in EEM research is music. Music has a
number of attributes that make it a practical emotion-eliciting stimulus. Several studies now
demonstrate that music is capable of: a) eliciting strong subjective feelings of emotion; b) activating
brain regions involved in emotion processing; and c) eliciting physiological responses consistent with
emotional arousal. Furthermore, music can elicit a range of emotions, from positive and arousing to
negative and deactivated, thereby enabling comparison of differing emotion states on memory. An
important and unique attribute of music is that its emotional meaning does not rely on the material to
be remembered. It is thus possible to test whether an extrinsic source of emotional arousal, presented
before, during, or after unrelated material to be remembered, can facilitate memory. Furthermore, the
ease of application and non-invasive nature of music provides a more tolerable source of emotional
arousal than the aversive emotion treatments commonly used within this research field (e.g. gruesome
images, stressful psychological tests, or pharmacological interventions). Perhaps most importantly,
music is already extensively used to regulate emotions, thereby making it an ecologically valid,
increasingly portable, even enjoyable emotion source that could be applied in a variety of contexts to
facilitate memory. These unique properties of music could ensure compliance with memory
enhancement programs that may arise as a result of this line of research.
1.1 Aims and hypotheses
The primary aim of this research project was to determine whether positive emotion elicited by
music could be utilised to facilitate long-term memory (LTM) for unrelated information. To achieve this
aim, multiple components of emotion were measured to validate the emotion inducing properties of
music. The research commenced using a previously validated emotion-memory methodology (adapted
to incorporate experimenter-selected background music as an emotion stimulus) to re-test the
phenomenon. Further investigation of the emotion and memory-enhancing properties of music was
then conducted in a second experiment utilizing participant-selected music, with the aim of validating
music as an effective emotion stimulus. With the knowledge gained up to this point, the aim of the final
experiment was to verify the neurobiological model of EEM. Emotionally arousing positive music was
presented during the consolidation period (instead of prior to, or during, learning) with the aim of
facilitating memory for previously presented neutral information. In this way, emotion effects on
3
___________________________________________________________________________________
information processing and encoding could be isolated from physiological arousal effects on memory
consolidation, thus supporting the neurobiological model of EEM.
Hypotheses were developed to test both neurobiological and neural network theories of EEM.
In brief, it was hypothesised that: a) emotional music would elicit changes in multiple components of
emotion, thus confirming the validity of music as an emotion stimulus; b) emotion elicited by music
before, during, and after unrelated material to be remembered would facilitate memory relative to
active listening non-emotional control conditions; and c) that the memory enhancing effect could be
most parsimoniously accounted for by the neurobiological model of memory consolidation.
1.2 Thesis overview
In the following chapters, the Literature Review will provide an overview of the neurobiological
model of EEM. Particular attention will be paid to the interaction between BLA and brain stem
structures and the hypothalamus to modulate an autonomic emotion response, and the effects of
autonomic activation on the consolidation of declarative memory. The measurement of multiple
components of emotion to determine the effectiveness of the emotion stimulus will then be discussed.
From this review, it will be demonstrated that synchronised changes in two fundamental dimensions of
emotion, valence and arousal, may act as proxy for amygdaloid activity, and that concomitant memory
improvements relative to neutral controls would support the hypothesis that the amygdala modulates
emotional memory. Alternative explanations for facilitated memory that are founded upon cognitive
processing models will then be discussed. The emphasis here will be on the influence of emotional
stimuli on attention and increased cortical activation that facilitates information processing. An
argument will then be made for the use of music as a source of emotional arousal to investigate
emotion effects on long-term declarative memory. The value of music will be described in terms of its
activating effect on emotion structures in the brain (the limbic system, including the amygdala and
associated areas) and the autonomic nervous system (ANS); and its ability to elicit positive and negative
emotions of varying arousal intensity. Research investigating the effect of music on working memory
and short-term memory will then be reviewed with a focus on identifying the strengths and weaknesses
of music as an emotion source.
4
___________________________________________________________________________________
An overview of the methodology employed in this research project will be described in Chapter
3, while methodology unique to each experiment will be described in the relevant empirical chapters.
Hypotheses were tested with three experiments. The first experiment (Chapter 4) was designed to
replicate previous EEM research in humans and to extend the findings by testing whether emotionally
powerful background music could facilitate memory. The second experiment (Chapter 5) was designed
to fully exploit the emotional power of music by examining the effect of participant-selected enjoyed
music on emotion components and the early stages of LTM. The third experiment (Chapter 6) utilised
the knowledge gained from Experiment 1 and 2 to examine whether emotional arousal elicited by music
during the consolidation period could facilitate memory. Chapter 7 is dedicated to integration of the
results of the three experiments, their implications, and suggested pathways for future research.
5
2 LITERATURE REVIEW
Events that induce an emotional response tend to have privileged status in autobiographical
memory. This may be an evolutionary phenomenon in that survival and personal well-being are
optimized when details surrounding emotionally significant, and therefore by definition personally
salient events, are well remembered. The storage of information surrounding the emotion eliciting
event, whether it is the achievement of reaching personal goals, frustration in not achieving those goals,
or encounters with life threatening situations, enables the organism to make informed decisions should
similar events be encountered in the future. There has been widespread investigation of emotionenhanced memory. For example, memory for emotionally powerful events, such as the 9/11 terrorist
attack, confirm that people recall such events in more detail than less emotional events (Sharot et al.,
2007; Talarico & Rubin, 2003b; Wolters & Goudsmit, 2005b). The relevance of emotional memory for
eyewitness testimony also encouraged extensive research into the quality of this enhanced memory,
and demonstrated that central events may be recalled intensely at the cost of peripheral details (see
memory narrowing, the weapon-focus effect and the Easterbrook hypothesis reviewed by Christianson,
1992a, 1992b; Deffenbacher, Bornstein, Penrod, & McGorty, 2004; Kensinger, Garoff-Eaton, & Schacter,
2007b). There are a number of theories explaining emotion-enhanced memory. One theory that has
gained particular empirical support, and thus is the focus of investigation in this dissertation, is founded
upon the memory modulating effects of arousal hormones and neurotransmitters released during
emotional experiences (McGaugh, 2004).
2.1 The neurobiological model of emotion-enhanced memory
The neurobiological model of EEM involves the interaction between arousal hormones and
many brain regions and circuits, the activation of which is argued to be initiated by amygdala response
to emotional stimuli (Cahill & McGaugh, 1998; McGaugh, 2000; McGaugh, 2004). The amygdala is a
limbic structure composed of distinct clusters of neurons responsible for processing and responding to
emotional information (see Figure 2.1 for a diagram of the projections and functions of the major nuclei
of the amygdala). Emotional percepts are received by the thalamus, which projects both directly to the
dorsolateral amygdala and to sensory cortex where the information is processed at a more fine grained
level (LeDoux & Phelps, 2000). The strength of the stimulation determines whether the signal is further
6
CHAPTER 2. LITERATURE REVIEW
___________________________________________________________________________________
transmitted to the basal nucleus and/or the central nucleus. At this point, the stimulation must be
sufficiently high to overcome the strong inhibitory network within the amygdala (see LeDoux, 2007 for a
review of the structure and function of the amygdala). If sufficient stimulation occurs, central nucleus
projections to brain stem structures lead to adaptive responses such as generalized cortical arousal,
freezing, hypothalamic mediation of adrenal arousal hormone release, and changes in heart rate and
blood pressure (Sah, Faber, Lopez De Armentia, & Power, 2003). Basal nuclei projections to prefrontal
cortex (PFC), ventral striatum, and polymodal association cortex lead to monitoring and regulating
responses, the initiation of instrumental behaviours, such as running to safety, and drawing on existing
knowledge, such as object recognition and episodic memory, that aid decision making. Projections to
the hippocampus and entorhinal cortex initiate anatomical changes in cell structure necessary for the
consolidation of declarative memory, including long-term potentiation (LTP). LTP is a long-lasting
change in synaptic efficiency believed to underlie long-term declarative memory (Squire & Kandel,
1999).
7
___________________________________________________________________________________
Adapted from Le Doux (2007) and McGaugh (2000)
Figure 2.1 Inputs to and outputs from amygdala nuclei. Solid lines represent axon projections, dashed
lines represent humoral projections. Labels: dark blue, cortical regions; light blue, brain stem structures;
white, associated limbic structures; green, peripheral structures. Abbreviations: B, basal nucleus; Ce,
central nucleus; itc, intercalated cells; La, lateral nucleus; M, medial nucleus; NE, norepinephrine; DA,
dopamine; ACh, acetylcholine; 5HT, serotonin; and EPI, epinephrine. Note: Labels and projections have
been added to the original model presented by Le Doux according to the review of amygdala anatomy
and physiology by Sah, Faber, Lopez De Armentia and Power (2003); and label position is not intended
to be anatomically correct.
McGaugh’s neurobiological model (McGaugh, 2000; McGaugh & Roozendaal, 2002) focuses on
the effect of the adrenal release of EPI and glucocorticoids (e.g. cortisol) on β-adrenergic receptors
within the BLA. In the model, amygdala activation in response to an emotion stimulus leads to
activation of the sympathetic branch of the autonomic nervous system (ANS), resulting in physiological
changes such as increased cardiac activity, vasoconstriction, pupil dilation, and EPI and NE secretion
8
___________________________________________________________________________________
from the adrenal medulla. The adrenal secretion of EPI activates β-adrenergic receptors on the
ascending vagus nerve, which projects to the nucleus of the solitary tract (NTS) located in the brain
stem. The NTS in turn has NE projections to the locus coeruleus (LC), which is a major site of NE
projections throughout the brain, including the BLA. β-adrenergic receptor activation in the BLA
reinforces amygdala projections throughout the cortex, including to memory structures such as the
hippocampus and entorhinal cortex. These ‘waves’ of autonomic activity (Cahill & McGaugh, 1998)
provide sufficient activation for the gene transcription and protein synthesis required for cell growth in
the hippocampus and entorhinal cortex (see Figure 2.1) and the consolidation of the memory trace.
Additional waves of BLA stimulation arise from adrenal cortex release of glucocorticoids into the blood
stream. The glucocorticoids cross the blood brain barrier to further reinforce BLA activity and memory
consolidation.
The role of noradrenergic receptors within the amygdala in memory consolidation has been
demonstrated using inhibitory avoidance training and testing in many animal species (e.g. mice,
Campolongo et al., 2009; day old chicks, Gibbs & Summers, 2002; and rats, Gold & McGaugh, 1975). For
instance, rats are trained to lick from a water spout and then the behaviour is paired with a footshock.
The latency to drink from the spout after a 24 hour delay is then used as a measure of memory retention
(Gold & McGaugh, 1975). It has been revealed that rats with high footshock training have elevated
plasma epinephrine concentrations and facilitated memory for the task. In addition, rats infused with
epinephrine post-training show similar plasma epinephrine concentrations and memory performance as
those who received high footshock training. Finally, an arousal dose-response effect has been
demonstrated in that high footshock training combined with post-training infusion of epinephrine
impairs memory (reviewed in McGaugh, 1989). This arousal dose-response effect has been replicated
numerous times under different training paradigms with different species (Baldi & Bucherelli, 2005). The
critical memory modulating role of noradrenergic receptors in the BLA has been demonstrated by
impairing memory with infusions of β-adrenoceptor antagonists directly into the amygdala. Further
confirmation of the memory modulating role NE has been demonstrated by reversing the memory
impairing effect of β-adrenoceptor antagonists with administration of norepinephrine (McGaugh, 2004).
9
___________________________________________________________________________________
2.1.1 Neurobiological modulation of memory in humans
Support for the critical role of the amygdala and the β-adrenergic system in EEM in humans has
been gathered from imaging and neuroendocrine studies. Imaging studies have revealed a relationship
between amygdala activity during the encoding of emotional material and LTM. For example, Cahill et
al. (1996) exposed study participants to 12 emotionally arousing films and 12 emotionally neutral films
in two separate positron emission tomography (PET) sessions. They revealed that the emotionally
arousing films were rated as subjectively more emotional than the neutral films. Memory for the films
rated as emotional was also significantly higher than for the neutral films after a three week delay.
Finally, amygdala activity during film encoding was highly correlated with LTM. Other investigations into
the relationship between amygdala activity and memory for emotional material have yielded similar
findings (Cahill, Uncapher, Kilpatrick, Alkire, & Turner, 2004b; Canli, Zhao, Brewer, Gabrieli, & Cahill,
2000; Hamann, Ely, Grafton, & Kilts, 1999). Further evidence of the critical role of the amygdala in
mediating emotional memory has been revealed by studies of patients with amygdala damage caused
by trauma or disease, in which amygdala damage attenuates emotional memory (Adolphs, Tranel, &
Denburg, 2000; Hamann, Monarch, & Goldstein, 2002).
The memory modulating effect of adrenal hormones in humans was initially demonstrated in a
seminal paper by Cahill, Prins, Weber and McGaugh (1994). They revealed that long-term declarative
memory for emotional events could be attenuated with propranolol hydrochloride, a substance that
blocks the action of β-adrenergic receptors in the central and peripheral nervous systems. Their study
demonstrated that blocking an adrenal stress response to an emotionally arousing story attenuated
EEM compared to placebo controls, despite participants reporting an emotional response at the time.
The memory modulating effect of the noradrenergic system has been replicated numerous times
(reviewed by Chamberlain, Müller, Blackwell, Robbins, & Sahakian, 2006). For instance, using the same
experimental materials as those used by Cahill et al., O’Carroll et al. (1999b) revealed that yohimbine
hydrochloride, a noradrenergic agonist, facilitated LTM, and metoprolol, another β-adrenergic
antagonist, impaired LTM relative to placebo. Furthermore, memory improvements were associated
with increased heart rates during story viewing. Increased heart rate in yohimbine treated participants
was related to higher memory scores, and lowered heart rate in metoprolol treated participants was
related to lower memory scores. These results indicated that increases in heart rate, which is
modulated by noradrenergic receptors in the ANS, indexed noradrenergic modulation of EEM.
10
___________________________________________________________________________________
A limitation of the use of β-adrenergic receptor antagonist propranolol and metoprolol to
support the neurobiological model of EEM is that these drugs readily cross the blood-brain barrier. The
unique contribution of the release of noradrenergic neuromodulators in the peripheral nervous system
(PNS) on EEM has therefore yet to be unequivocally established in humans. This distinction is important
for the neurobiological model, which posits that the adrenal release of arousal hormones in the PNS and
subsequent vagal afferents is critical to the modulation of EEM (McGaugh & Roozendaal, 2002). If the βadrenergic receptor antagonists block NE transmission in both the central nervous system (CNS) and
PNS, then attenuated EEM could be attributed to the blockade of β-adrenergic receptors in the
amygdala and brain stem structures alone (e.g., the LC, see Figure 2.1), without the input of vagus nerve
projections initiated by adrenal release of NE and EPI. This ‘site of action’ problem was addressed by
van Stegeren, Everaerd, Cahill, McGaugh and Gooren (1998), who compared long-term memory
between groups who received propranolol, which readily crosses the blood brain barrier, and nadolol,
which does so to a lesser extent. Their study revealed that memory was attenuated with propranolol
relative to placebo but not with nadolol, indicating that peripheral β-adrenergic receptor activation was
not critical to EEM. This result raises the possibility that, contrary to the neurobiological animal model,
adrenal release of arousal hormones in humans is not critical to the consolidation of emotional
memories. That is, a life-preserving stress response (e.g., fight or flight) involving the release of adrenal
arousal hormones may be critical to EEM in animals, but not in humans. What is critical to EEM in
humans is activation of β-adrenergic receptors in the CNS, most likely in the BLA and LC. In other
words, amygdala mediated EEM in humans may be supported by more central than peripheral
mechanisms.
An additional limitation to studies designed to manipulate EEM in humans is the variability in
results. Many studies use a three phase slideshow of 12 still images accompanied by a one-sentence
voiced narrative per image. Long-term memory is compared: within-subjects who viewed neutral
phases (start and end) and an emotional phase (middle) of the slideshow; between-subjects who viewed
an emotional or neutral version of the slideshow; and between-subjects who were administered a βadrenergic agonist, antagonist, or placebo. Inconsistencies arise in the reliability of β-adrenergic
modulation of memory. For instance, attenuation of EEM with 40mg of propranolol in the van Stegeren
et al. (1998) was not replicated in a study conducted by O’Carroll, Drysdale, Cahill, Shajahan and
Ebmeier (1999a). The major difference between studies was sample size, with van Stegeren and
colleagues recruiting 75 participants compared to 36 recruited by O’Carroll and colleagues. With an
11
___________________________________________________________________________________
increase in propranolol dosage (40mg to 80mg), Maheu, Joober, Beaulieu and Lupien (2004)
demonstrated attenuation of EEM with a sample size of 42. The influence of sample size and dosage
level on memory indicates that either a reliable methodology has yet to be developed to test EEM in
humans, or that the underlying putative mechanism requires further clarification.
There has been more empirical success demonstrating the facilitatory effect of the adrenal
release of glucocorticoids on EEM in humans. Buchanan and Lovallo (2001) administered 20 mg of
hydrocortisone or a placebo to participants before they viewed emotionally arousing (positive and
negative) and neutral images. An incidental memory test1 conducted one week later revealed that
when cortisol treatment increased salivary cortisol levels (consistent with psychological stress), there
was greater recall of arousing images relative to the placebo controls. Arousal level was the key factor
underlying facilitated memory as images were similarly recalled regardless of whether they were
pleasant or aversive. A similar cortisol effect using a higher dose of hydrocortisone (30 mg) and similar
methodology was demonstrated by Kuhlmann and Wolf (2006). EEM may thus be accounted for by the
interaction between centrally mediated NE and peripherally mediated glucocorticoids in the BLA (see
review by Roozendaal, 2002). Support for the interacting effect of glucocorticoids and NE within the
human amygdala was revealed in a functional magnetic resonance imaging (fMRI) study conducted by
van Stegeren et al. (2007). In this study, participants with high endogenous cortisol levels showed
greater amygdala activity in response to emotionally arousing images compared to participants with low
cortisol levels. Furthermore, amygdala activity in high cortisol participants could be attenuated with
administration of a β-adrenergic receptor antagonist.
The study conducted by van Stegeren et al. (2007) supported the neurobiological model by
revealing an interaction between NE receptor activation and cortisol levels in the human amygdala.
Nevertheless, there was no evidence that the interaction facilitated memory. The failure to detect a
relationship between adrenal neuromodulation, amygdala activation, and memory may have been due
to study methodology. Memory was tested by presenting the same and matched foil images to
participants after a two week delay and requesting them to make ‘old’, ‘new’, or ‘not sure’ responses.
As humans have an astounding ability to remember visual scenes (Standing, 1973), discrimination
between old and new images may have been too easy, thus yielding a ceiling effect. Alternatively,
forced choice responses may have led to higher variability due to guessing (Van Stegeren et al., 2005),
1
Participants were not informed of the delayed memory test so as to minimise the confounding influence of
rehearsal or other memory strategies.
12
___________________________________________________________________________________
thereby reducing the power of inferential tests to detect differences between condition means. In a
separate study using a similar methodology, Segal and Cahill (2009) revealed a significant relationship
between endogenous cortisol levels and free recall memory. Image free recall is a more difficult method
of memory testing than recognition as it requires participants to recall details of the event in the
absence of cues. Reducing the margin for error by eliminating guessing (e.g., false positives or
negatives) may have thus increased the sensitivity of the test and improved detection of memory
differences. The inconsistent outcomes of these similar studies demonstrate the importance of
selecting a reliable measure of memory.
In summary, attenuation of EEM with β-adrenergic receptor antagonists that act both centrally
and peripherally, but not with an agent that is selective to the PNS, indicates that NE modulation of
memory in humans occurs in the CNS. Furthermore, low doses of β-adrenergic receptor antagonists,
low sample sizes, or insensitive methods of memory testing can result in the failure to detect EEM,
indicating that effects are mild and easily masked by random variation. Nevertheless, the interaction
between endogenous cortisol levels and NE receptor activation in the human amygdala supports
McGaugh’s neurobiological model of EEM.
2.1.2 Psychophysiological changes associated with emotion-enhanced memory
According to McGaugh’s model of EEM, activation of the amygdala is critical to enduring
memory of emotional information. The amygdala is activated when information or events are
significant enough to threaten or enhance the well-being of the organism. The amygdala has
projections to cortical and subcortical brain structures (illustrated in Figure 2.1) that regulate fright,
flight or fight responses. A method of determining whether an emotion has been elicited, and by
association that the amygdala has been activated, has therefore been to measure autonomic (e.g., heart
rate and skin conductance) and neuroendocrine (e.g., EPI and NE) changes then test whether these
changes are related to facilitated memory.
The ANS can be monitored with relatively non-invasive methods compared to the
neuroendocrine system, which requires saliva or blood sample assaying. Autonomic responses to
emotional stimuli can be determined by increased sympathetic nervous system (SNS) activity and
decreased parasympathetic activity. Within the context of EEM research, heart rate and perspiration
(determined by skin conductivity) have been most widely used to index SNS responses to emotion
13
___________________________________________________________________________________
stimuli. Monitoring SNS activation also enables measurement of the efficacy of β-adrenergic agents. As
sympathetic control of the peripheral nervous system is supported by post-ganglionic noradrenergic
receptors, receptor blockade results in decreased heart rate and blood pressure. Measuring these
indices of sympathetic response therefore confirms the efficacy of the pharmacological intervention.
Measuring tonic and phasic shifts in SNS activation enables evaluation of emotion (sustained increases
in arousal) relative to orienting reflex (instantaneous and rapidly inhibited) effects on memory (Bradley,
2009).
Studies that have measured autonomic indices of emotion response to stories, still images, and
films reveal that tonic increases in heart rate and/or skin conductivity are associated with enhanced
memory (Abercrombie, Chambers, Greischar, & Monticelli, 2008; Laney, Campbell, Heuer, & Reisberg,
2004; Vecchiato et al., 2010). Likewise, phasic physiological changes, such as heart rate deceleration
and skin conductance peaks at the onset of emotional images and words, are associated with enhanced
memory (Abercrombie et al., 2008; Buchanan, Etzel, Adolphs, & Tranel, 2006; Cahill & Alkire, 2003;
Palomba, Angrilli, & Mini, 1997). These findings demonstrate that increased tonic and phasic
sympathetic responses can act as ‘proxy’ measures of noradrenergic modulation of EEM (Abercrombie
et al., 2008).
In sum, testing of EEM in the research laboratory has demonstrated that emotional arousal
facilitates memory. Subjective and sympathetic nervous system responses consistent with an emotion
stress response have been elicited by emotionally arousing images, words and stories. These emotional
responses predict better recall of the emotional stimuli relative to neutral comparison stimuli. Evidence
of modulation of memory via emotional arousal has been most widely documented with subjective
reporting of emotional responses and concomitant tonic and phasic physiological changes. The
relationship between increased SNS activity and facilitated memory may therefore provide indirect
evidence of noradrenergic modulation of the BLA and memory consolidation, as proposed by McGaugh.
14
___________________________________________________________________________________
2.2 Cognitive processing and facilitated memory
The modulating influence of adrenal arousal hormones on the BLA and memory consolidation is
one of many possible mechanisms that underlie strengthened memory. For instance, activation of the
amygdala by the emotion stimulus also leads to efferent projections to frontal brain regions involved in
cognitive processes2, such as contextualising information within current goals or activities (Barbas,
Zikopoulos, & Timbie, 2011). An adaptive plan of action can then be programmed, and the event can be
integrated with stored information for future reference (refer to the depth of processing model of
learning proposed by Craik & Tulving, 1975). Amygdala regulation of cortical activation via
monoaminergic (serotonin, dopamine and norepinephrine) and acetylcholine (ACh) projections to
widespread cortical areas (see Figure 2.1) would support information processing efficiency. As such, the
amygdala may also mediate cognitive processes that establish a more robust memory trace. A model
illustrating the influence of such cognitive processes on memory is presented in Figure 2.2. The model
has been adapted from that presented by Cahill and McGaugh (1998, p. 295) to include a secondary
pathway (dotted lines) that represents cognitive emotional effects on immediate coping behaviour and
memory storage.
2
Cognitive processes are defined as mental processes, such as attention, pattern recognition, learning, and
memory, that operate on knowledge structures (Haberlandt, 1997)
15
___________________________________________________________________________________
Figure 2.2 Parallel cognitive and autonomic emotional responses regulated by the amygdala. External
or internal stimuli that are significant to the well-being of the organism activate the amygdala, which
then projects to widespread cortical structures that support an adaptive response. The frontal cortex is
one of the locations where information is interpreted and emotional responses are regulated. Parallel
projections from the amygdala to the hypothalamus and brain stem structures also initiate an
autonomic stress-hormone response. Pathways to frontal regions regulate the cognitive emotional
response, and pathways from brain stem regions regulate the autonomic stress-hormone response.
Both pathways have an immediate influence on coping behaviour, and both pathways can influence
memory storage. The cognitive emotional response influences memory storage via greater depth of
information processing, and the autonomic stress-hormone response influences memory storage via
neuromodulation of BLA projections to medial temporal lobe memory structures.
Memory binding could be explained by Associated Neural Network theory. Associationism, also
known as connectionism, is founded on the Hebbian principle that when two neurons are
simultaneously excited, there will be an increase in the strength of the connection between them (Hebb,
1949, cited in Baddeley, 2005). Several cognitive psychologists have contributed to the development of
the associated network theory of learning (e.g. McClelland & Rumelhart, 1985). In simple terms, the
model posits that information processing consists of input units, hidden units, and output units (see
Figure 2.3). Input units represent sensory information, output units represent motor output, and hidden
units represent all operations in between. Connections between units can be excitatory or inhibitory
and have varying strengths, thereby explaining such cognitive processes as learning (strengthened
16
___________________________________________________________________________________
connections) and habituation (inhibition of connections). A unit could represent any construct, for
example words, objects, memories, or moods. When a unit is activated, a mood state for instance, units
associated with that mood are excited to sub-threshold level. When the sub-threshold associated units
are further excited by additional input units, such as visual information congruent with the current
mood, the activation threshold is reached more efficiently. Associated neurons that have
simultaneously fired will therefore be more likely to form a connection. EEM could thus be accounted
for by active neural networks ‘firing and wiring’ in response to incoming congruent information.
Figure 2.3 Neural Network model consisting of input units (sensory neurons), hidden units (where
knowledge is stored), and output units (motor neurons). Sourced from Garson (2010).
Phenomena such as mood-congruent and schema consistent memory can be explained in terms
of associated neural network activation (see Bower, 1981, for a description of associated network
effects on memory). Support for the mood (or emotion3) effect on memory has been demonstrated by
facilitating memory for information that is congruent with specific moods induced before learning
(Bower, 1981; Eich & Forgas, 2003; Eich & Schooler, 2000; Kenealy, 1997). The mood-congruence effect
on memory is also well documented in depression research, where depressed individuals pay more
attention to, and remember, more negative environmental information than do non-depressed controls
(e.g. Bouhuys, Geerts, & Gordijn, 1999; Watkins, Vache, Verney, Muller, & Mathews, 1996).
3
Emotion and mood tend to be used interchangeably in this theory. Emotion, however, is distinct from mood in
that it usually occurs within a short time-frame and is accompanied by physiological, cognitive and behavioral
changes that facilitate flexible responses to environmental challenge. Mood is typically a more stable state
sustained over hours or longer.
17
___________________________________________________________________________________
Schemata could have a similar priming effect on memory. Schemata are stored concepts that
represent our knowledge of the world and can be used to help interpret ambiguous situations
(Baddeley, 1999). In doing so, attention will become selective to information that is consistent with the
active schema and lead to memory bias. The influence of active schemata on memory bias was
demonstrated by Boltz (2001). Music was chosen as the method of schema induction based on its
prolific use to enhance the movie viewing experience. Boltz argued that schema are elicited when a
movie viewer hears positive and uplifting music and subsequently expects, based on prior experience,
positive and uplifting story elements. Likewise, when they hear suspenseful music, they are likely to
expect negative story elements. Positive or negative music was therefore presented to participants
whilst they viewed ambiguous film clips. Memory testing after a one week delay revealed that negative
music led to recall of more negative than positive film details and positive music led to recall of more
positive than negative film details. This finding indicates that music provided a schematic framework for
interpreting ambiguous information which biased attention and memory. Mood-congruent memory
and schema bias are therefore examples of cognitive processing influences on LTM.
These cognitive processing theories may account for the memory effects observed in the
neurobiological research previously described. A typical experiment involves informing participants that
they will be viewing a slideshow with a story narrative and images that are pleasant, unpleasant, or
basically neutral. This information is fairly ambiguous and may therefore have elicited schemas of
pleasant, unpleasant, and neutral images. Incoming image information that was congruent with these
active schemas could have provided sufficient excitation for the associated neurons to reach firing
threshold and establish a memory trace. The observed difference in memory for the neutral relative to
emotional versions of these slideshows could be explained by the emotional narrative enhancing the
unpleasantness of the images, further increasing excitation of associated units and the strength of the
memory trace. It is therefore possible that facilitated memory for the ‘emotional’ story could be
attributed to schema bias.
A final cognitive processing influence on memory yet to be fully explored is post-encoding
rehearsal. Two of the basic principles of successful learning are repetition and rehearsal (Baddeley,
2005). As such, increasing the presentation frequency of the material to be remembered (MTBR)
increases the probability that it will be retained (see the classic nonsense syllable learning study
conducted by Ebbinghaus, cited in Baddeley, 2005). Likewise, rehearsing the MTBR increases the depth
to which it is integrated into the memory store and thus retained (Craik & Tulving, 1975). With regard to
18
___________________________________________________________________________________
EEM, the definition of an emotional event is that it is significant to the well-being of the individual.
Thus, an emotional event will more likely be thought about many times after the event has taken place
compared to a neutral event. According to the repetition principle, every time the event is retrieved the
memory would be strengthened. According to the rehearsal principle, the personal meaningfulness of
the event will ensure that the information is more elaborated upon and thus more deeply encoded than
a neutral event.
Investigators of EEM have attempted to control these potential confounds of the
neurobiological model by withholding information regarding memory testing from participants. The
intention has been to minimise practice or rehearsal of the MTBR. This type of control is appropriate for
experiments investigating memory processes underlying neutral material (e.g., word lists, nonsense
syllables, or languages). However, the very nature of emotional stimuli increases the probability that it
would be rehearsed relative to neutral comparison material. Thus, memory for the emotional event
could be facilitated by post-encoding cognitive processes. A solution to this confound is difficult to find.
It would be counterproductive to ask participants to refrain from thinking about the emotional material
as this would more than likely increase post-encoding pondering. Post-encoding elaboration of
emotional material is therefore a serious confounding variable of neurobiological explanations of EEM,
thus warranting careful consideration when interpreting memory effects.
2.3 Differentiating encoding from consolidation effects on memory
Neurobiological modulation of the consolidation of memory has been supported primarily with
animal studies. Memory has been facilitated by administering neuromodulators systemically or directly
into the brains of animals during the consolidation period (Gibbs & Summers, 2002; Gold, Hankins,
Edwards, & McGaugh, 1975; McGaugh et al., 1993). There is empirical support for the same
neurobiological mechanism to underlie memory consolidation in humans (e.g., Buchanan & Lovallo,
2001; Kuhlmann & Wolf, 2006). However, as described in the previous section, there are caveats to this
assumption. First, emotion effects on cortical activation and strengthened encoding have not been
excluded as causes of facilitated LTM, and second, the memory-strengthening effect of post-encoding
elaboration of meaningful or interesting information has not been adequately controlled. Figure 2.4
illustrates how paradigms that present the emotion stimulus during encoding of the MTBR are subject to
increased circulating arousal hormones and more efficient information processing (attention, appraisal,
19
___________________________________________________________________________________
integration of information according to goals or active schemas, and post-encoding elaboration) during
the encoding and consolidation period. The effect of emotion stimuli (Figure 2.4A) relative to neutral
stimuli (Figure 2.4B) on memory may therefore be caused by factors other than adrenal modulation of
memory consolidation.
Figure 2.4 Task paradigm illustrating the confounding influence of more efficient information processing
on encoding and consolidation. Grey rectangles represent matched MTBR, solid red lines represent
expected circulating arousal hormones, and dashed green lines represent expected level of information
processing. (A) Blue arrow represents the emotion stimulus, and (B) white arrow represents the neutral
stimulus.
This confound of the arousal hypothesis has been recognised by a number of researchers (e.g.,
Cahill & Alkire, 2003; Cahill, Gorski, & Le, 2003; Preuss & Wolf, 2009). Consequently the methodological
approach used in animal studies of administering the emotion or arousal stimulus after encoding, during
the consolidation period, has been employed in humans. Figure 2.5 illustrates how presenting the
emotion stimulus post-encoding is hypothesised to strengthen the consolidation of previously presented
neutral material. In this paradigm, cognitive processes that influence learning are directed towards the
post-learning emotion stimulus and not the previously presented neutral MTBR. Thus, when comparing
emotional (Figure 2.5A) and neutral conditions (Figure 2.5B), strengthened memory caused by cognitive
processes directed towards the emotion stimulus can be more readily isolated from neurobiological
modulation of memory for the previously presented MTBR.
20
___________________________________________________________________________________
Figure 2.5 Solution to isolate the effects of emotion on memory
A breadth of research studies has tested whether post-encoding emotional arousal facilitates
the consolidation of memory in humans. Within this paradigm, several methods of arousal elicitation
have been used and memory has been tested for a variety of information types. In general, there is
consistent evidence that post-encoding emotional arousal facilitates LTM for the target MTBR (e. g.
Anderson, Wais, & Gabrieli, 2006; Judde & Rickard, 2010; Nielson & Bryant, 2005; Nielson & Powless,
2007; Nielson, Yee, & Erickson, 2005; Soetens, Casaer, D'Hooge, & Hueting, 1995). A summary of
studies investigating post-learning arousal effects on memory is presented in Table 2.1.
21
____________________________________________________________________________________________________________________
Table 2.1
Post-encoding Arousal Effects on Human Memory (chronological order)
Authors
Year
MTBR
Soetens et al.
1995
10 lists of 20
words
(1 per second)
240 words,
120 target
and 120
distractor
(4x60 word
lists)
60 words
Method of arousal
elicitation
EXP 1 & 2 oral
amphetamine (1 hr
delayed action)
Controls
DVs
IV s
Results
placebo
free recall
memory after each list (FR)/after
all lists(FFR)/1 hr delay (1HD)/ 24
hour delay (1DD)
Oral amphetamine after learning too late to have effect
on memory (1 hr delay of action). Sig effect of
amphetamine on acquisition up to 1 hour post
administration.
EXP 3 & 4
intramuscular
amphetamine
immediately after
word list learning
EXP 5 intramuscular
amphetamine
immediately after
word list learning
placebo
free recall
memory FR/FFR 20 min after
drug/1DD
FFR not sig, 1DD sig.
placebo
recognition memory
memory immediate/1 day/1
week
No difference between placebo and amphetamine at
immediate, trend at 1 day, sig effect at 1 week.
oral cortisone
(cortisol) 1 hr delayed
action
placebo
memory recall and recognition;
subjective stress ratings
cortisol 1hr before
acquisition/immediately after
acquisition/1hr before retrieval;
memory immediate/delayed
(24hour)
Sig memory impairing effect of cortisol on free retrieval
but not recognition. No effect on acquisition or
consolidation. No effect of cortisol on subjective stress
ratings.
Memory didn't differ between yohimbine and placebo.
Yohimbine responders had higher mean memory scores
than yohimbine non-responders. MHPG change
predicted memory (linear regression), irrespective of
yohimbine or placebo (but stronger effect for placebo).
Correlation between yohimbine and memory not
significant (n=14). Low numbers for regression analysis,
could be random variation.
EPI sig more recall for primacy slides than placebo. EPI
80 recalled more primacy slides than saline. No
differences between conditions for recency slides.
Faster HR for primacy than recency slides. Greater SCR
for primacy than recency slides.
De Quervain,
Roozendaal, Nitsch,
McGaugh, and Hock
2000
Southwick et al.
2002
12 slide
arousing story
(Cahill et al
1994)
intravenous
yohimbine 5 mins
after slide.
placebo
LTM (7 days); MHPG
MHPG: -30/15/+20/+60/+120min
Cahill and Alkire
2003
intravenous injection
of epinephrine
saline
HR; slide SC peak less slide
baseline: first three averaged
(primacy)/last three averaged
(recency); EPI; free recall
epinephrine dose: saline/40/80;
primacy/recency; HR 1/2/3min
Cahill et al.
2003
21 IAPS
images low to
moderate
arousal pos,
neg and
neutral
valence
21 IAPS
images low to
moderate
arousal pos,
neg and
neutral
valence
cold pressor
warm water
salivary cortisol; pulse
plethysmograph (HR ruse);
subjective slide arousal;
subjective slide valence; % of
total (all participants) free
recall of slides (written); % of
total slide details
cortisol t0 before slide
viewing/t10 after water;
arousal/neutral slides based on
arousal median split
22
CPS after presentation of images enhanced memory for
self-rated arousing images but not self-rated neutral
images. No gender effect.
____________________________________________________________________________________________________________________
Authors cont.
Year
MTBR
Method of arousal
elicitation
$1 surprize reward or
praise
Controls
DVs
IV s
Results
Nielson and Bryant
2005
30 neutral
words (110
distractors)
nothing
free recall, recognition (error
corrected)
reward, intrinsic
(praise)/extrinsic (surprize
reward)/nothing/posted
(aware they would get a
reward); memory,
immediate/delay (1 week)
Free recall and recognition significantly greater for extrinsic
and posted groups than control.
Nielson et al.
2005
30 neutral
words (110
distractors)
arousing video (oral
surgery)
neutral video
(teeth
brushing)
free recall, recognition (error
corrected); heart rate;
subjective affect; skin
conductance
memory, 30min/24hr;
physiol change, stimulus
and post stimulus
HR sig increase for arousal video compared to neutral. No
SCL change. Mood sig more negative for arousal group.
Memory for all three tests (30min, 24hr recall and
recognition, controlling for initial differences in immediate
free recall using ANCOVA), better for arousal group.
Anderson et al.
2006
108 male and
female
neutral
photos, 108
pictures of
houses
IAPS 72 negative (eg
mutilation), 72
positive (eg erotica)
IAPS 72 neutral
error corrected recognition of
test items and modulators
(correct hits less proportion of
false alarms) for face and
house stimuli
modulator delay short
(4s)/long (9s); subjective
arousal Q1/Q2/Q3/Q4
Modulator memory: pos and neg modulator images sig more
arousing than neut. Memory greater for pos and neg scenes
- degree of arousal rather than valence predicted memory.
Quadratic arousal effect on recognition of modulator
images. Test item memory: short delays (4s) between test
and modulator enhanced memory, but not longer delays
(9s). No gender differences.
Nielson and Powless
2007
30 neutral
words (110
distractors)
positive (comedy) or
negative (dental
procedure) video
no postlearning
stimulus
immediate free recall; delayed
corrected recognition (1 week);
subjective arousal; subjective
mood
time of video, none
(control)/0/10/30/45min;
good/poor learners
Immediate free recall used as covariate for recognition
memory analysis. Subjective arousal and mood sig change
from baseline for all time groups and pos and neg videos
(more for negative). Sig delayed recognition memory effect
for all arousal delays except 45mins. Memory analysed by
good/poor learners - no interaction with group, pattern of
effect the same for good and poor learners irrespective of
arousal conditions
Smeets et al.
2008b
simple motor
actions
cold pressor
arm immersed
in warm water
reality monitoring (24 hr
delay); reality monitoring is
discrimination between
internally vs externally
generated memories; GC
change from baseline;
recognition % correct; correct
performed; perform errors; Pr
discrimination index; Br bias
index
performing simple motor
acts/imagining simple
motor acts; male/female;
cortisol t0/t10/t20;
low/high cortisol
responders
Faster GC response in females than males, but otherwise no
gender effect. Stress group higher memory than non-stress
group on all measures except response bias. No gender
effects for memory. No difference between low and high
cortisol responders. No correlation between GC and memory
in the stress group.
Smeets, Otgaar,
Candel, and Wolf
2008a
DeeseRoedigerMcDermott
word list
learning
paradigm.
Neutral and
emotional
versions.
Cold pressor before
encoding (1); during
consolidation (after
encoding) (2); before
retrieval (3).
not stressed (4)
arm immersed
in warm water
Word list recognition (24
hours); GCs; sAA (measure of
sympathetic activity)
23
Enhanced correct present (recognition) memory for stress
during consolidation group. Stronger memory for emotional
words. Effect related to cortisol and sAA. No effect of stress
on false present (errors). Sig correlations between memory
for neutral and emotional words and cort and sAA for the
consolidation group but not for the encoding or no stress
groups.
____________________________________________________________________________________________________________________
Authors cont.
Year
MTBR
Method of arousal
elicitation
Controls
DVs
IV s
Results
Liu, Graham, and
Zorawski
2008
25 neg, 25
post and 25
neutral IAPS
images (plus
45 matched
foils for
recognition
testing)
pleasant (the tonight
show, comedy) or
aversive (oral surgery)
video
neutral video
(toothbrushing)
incidental free recall memory
(1 week); mood adjective
check list (UWIST)
condition: pos/neg/neutral
film; images:
pos/neg/neutral; gender;
memory type: gist/detail;
mood arousal after image
viewing/after manipulation
Free recall: neg and pos images better recalled than neutral.
IAPS arousal rating correlated with recall. No memory
differences between males and females at any level. More
images recalled by those in pos and neg arousing video
conditions compared to neutral condition and in these
conditions, more positive and negative vs neutral images
recalled. Slight advantage for neg images to be better
recalled (not sig). Not many sig effects for recognition, no
group effects.
Nielson and Lorber
2009
60 ANEW
words, 15 for
each 2DES
quadrant,
randomly
presented (80
foil words for
recognition
test)
Jingleheimer Junction
(comedy video)
presented 10 minutes
after MTBR
documentary
video
words recognised, corrected
for guessing; subjective mood
valence and arousal; emotion
regulation questionnaire (ERQ)
suppression and reappraisal
sub scales; arousal
predisposition (APS)
reappraisal, suppression
and arousal (APS) median
split hi/lo
Word arousal ratings: hi APS scorers rated words more
arousing than low APS scorers; high suppression had higher
arousal ratings for neg high arousal words and lower ratings
for pos low arousal words; hi reappraisal had lower ratings
of neg high arousal words and higher ratings of pos low
arousal words. Memory: low arousal neg less retained than
other quads; low arousal pos better retained than high
arousal neg; post-learning arousal group better memory; no
group x primacy/recency interaction (no interaction with
serial position of words). Individ diffs; high APS in arousal
group had better memory; high reappraisers had less benefit
of arousal induction than low reappraisers.
Preuss and Wolf
2009
Trier Social Stress Test
(TSST): Video taped
oral presentation and
an arithmetic test
before a reserved
panel (2 people)
lasting 15 mins.
oral
presentation
and arithmetic
task but not in
front of a panel
or video taped.
Picture and narrative
immediate free recall; picture
and narrative delayed free
recall % retained (24 hours);
picture and narrative delayed
recognition (24 hours);
gist/detail memory; PANAS
(before stimuli, after stress,
before 24hr recall test);
subjective valence of stimuli;
subjective arousal of stimuli;
sAA; GCs
gist/detail; stimuli
positive/neutral/negative;
treatment arousal/neutral;
male/female
Immediate free recall: negative images most, then positive,
then neutral; delayed free recall: neutral most recall v
negative and positive (possibly due to male results);
recognition no effect of stress, sig effect of valence, negative
more remembered than positive or neutral; gist vs detail no
interactions with stress, main effect of gist (more
remembered than detail); cortisol correlated with memory
for neutral material for all participants and trend for
negative (controlling for group with partial correlations)
Judde and Rickard
2010
Neutral (5),
positive
arousing (5)
and negative
arousing (5)
mostly IAPS
pictures
accompanied
by a one
sentence
narrative
unique to
each picture
(Buchanan et
al)
30 words
(110
distracters)
Pos music (Beethoven
S No 6, 3rd mvt),
negative music
(Mussorgsky Night on
bare mountain)
matched on arousal,
familiarity and liking
nothing (filler
tasks)
PANAS, music familiarity,
liking; subjective valence and
arousal; recognition memory
(7 days); BIS/BAS; immediate
free recall;
time of arousal: 0/20/45min
Sig 'time' memory effect for both music excerpts. 20 min
delay higher memory than 45 min delay. Difference between
0min and 20min not reported. No music valence main
effect. No music valence by time interaction. Positive music
rated as neutral (1 sample t test compared to neutral score
of 4). Negative music rated more arousing than neutral.
Negative music not different to neutral score of 4, positive
music rated as more positive than neutral. Both similar on
familiarity and liking. BAS drive sig effect on memory at
20min, but not 0 and 45 min.
24
___________________________________________________________________________________
It has also been revealed that the arousal inducing properties of the encoded material can
interact with post-learning arousal. For instance, Lieu et al. (2008) demonstrated that participants who
viewed an arousing video (positive or negative) post-learning recalled more arousing than neutral
images after a one-week delay. Similarly, Smeets et al.(2008a) revealed that a cold-pressor stress test
after word list learning had a greater facilitatory effect on memory for emotional than neutral words
when tested 24 hours later. This effect is consistent with the neurobiological model of ‘waves of
activation’ discussed in Section 2.1 (see also the literature on memory tagging, e.g., Reymann & Frey,
2007; Richter-Levin & Akirav, 2003). Memory bias towards arousing material may therefore be
explained in terms of the establishment of a memory trace that survives long enough to be modulated
by the post-learning emotion treatment. In sum, these results indicate that memory strength is
modulated by emotional arousal elicited during the consolidation period, and that the influence of
cognitive processes on encoding is less useful in explaining facilitated memory.
2.4 Evidence of emotion elicitation
The emotion stimuli used in many studies of EEM have had high sensory impact in that they
could elicit acute reactions of shock or surprise. For instance, Heuer and colleagues (Burke, Heuer, &
Reisberg, 1992; Heuer & Reisberg, 1990) and Cahill and colleagues (Cahill & McGaugh, 1995; Cahill et al.,
1994) used images of exposed viscera and/or surgically reattached legs, Buchannan and colleagues used
profanities or sexually explicit words (Buchanan et al., 2006), Kensinger and colleagues used images of
snakes (Kensinger et al., 2007b), and Adolphs and colleagues used images of dead bodies (Adolphs,
Denburg, & Tranel, 2001). The acute reactions to these stimuli may be considered emotional in nature.
They may also be considered orienting reflexes that are devoid of the full breadth of central and
peripheral responses that characterise emotion. Orienting reflexes (indexed as phasic autonomic
changes) have been associated with enhanced memory (Abercrombie et al., 2008). Orienting has also
been proposed as an explanation for the relationship between decreased heart rate and facilitated LTM
revealed by Burke et al. (1992) and Heuer and Reisberg (1990). It is therefore possible that facilitated
memory in the aforementioned studies was elicited by a stimulus driven orienting reflex that elicited
neuromodulation of memory.
Emotions, however, are more complex than the orienting reflex. Therefore, to consider the
paradigm of facilitated memory discussed thus far as emotional in nature, it should be established that a
25
___________________________________________________________________________________
true emotion has been elicited. According to Scherer’s Component Process Model of emotion (2001), an
emotion consists of the synchronisation of a number of cognitive and physical components that produce
the most adaptive, situation specific, and well-being enhancing response. The process consists of the
synchronisation of what Scherer argues to be five distinct emotion functions: appraisal or evaluation,
subjective feelings, action tendencies, physiological responses to support action tendencies, and
expressive behaviour (see Table 2.2 for a description of the various components and functions of the
Component Process Model). The range of emotions is therefore not readily described by discrete
labels, but by activation of multiple subjective, behavioural, cognitive, and physiological responses that
support an adaptive response in various situations. For instance, a fast approaching ominous object
would cause an organism to cease current activity, evaluate the object, subjectively feel a sense of
danger, grimace and hunch or prepare to run, and undergo physiological changes such as increased
secretion of arousal hormones, increased heart rate, perspiration and pupil dilation, to support the
action. This synchronisation of multiple emotion components would traditionally be labelled as a fear
response.
Table 2.2
Component Process Model: the relationship between the functions and components of emotion and the
subsystems that subserve them
Emotion function
Emotion component
Organismic subsystem (and
major substrata)
Evaluation of objects and events
Cognitive
Information processing (CNS)
System regulation
Peripheral efference
Support (CNS, NES, ANS)
Preparation and direction of
Motivational
Executive (CNS)
action
Communication of reaction and
Motor expression
Action (SNS)
behavioural intention
Monitoring of internal state and Subjective feeling
Monitor (CNS)
organism-environment
interaction
CNS: central nervous system; NES: neuroendocrine system; ANS: autonomic nervous system; SNS:
somatic nervous system. The organismic subsystems are theoretically postulated functional units or
networks (Scherer, 2001, p. 93)
26
___________________________________________________________________________________
2.4.1 Subjective feeling and peripheral efference
Emotion-enhanced memory research reported to date at the minimum has measured emotion
responses in terms of the subjective feeling component. According to Scherer’s model, subjective
feelings of emotion reflect the conscious experience of changes in all emotion components by the
emotion stimulus (Scherer, 2004). Subjective feelings have been measured in terms of ‘emotionality’
(e.g. Cahill & McGaugh, 1995; Cahill et al., 1994; Heuer & Reisberg, 1990; O'Carroll et al., 1999b; van
Stegeren et al., 2007), or degree of subjective arousal levels and positive or negative valence (e.g.
Buchanan, Denburg, Tranel, & Adolphs, 2001; Buchanan et al., 2006; de Quervain et al., 2007). A
number of studies have also measured the peripheral efference component (e.g. Abercrombie et al.,
2008; Adolphs & Damasio, 2001; Buchanan et al., 2006; Burke et al., 1992; Cahill & Alkire, 2003; Heuer &
Reisberg, 1990). According to the component process model (Scherer, 2001, 2004), the peripheral
efference component would be activated by a stimulus that (1) disturbed ongoing homeostatic
regulation and smooth behavioural coordination, and (2) prepared the organism for an appropriate
adaptive response. Peripheral efference would thus be indexed by increased SNS activity (skin
conductance responses, increased heart rate, vasoconstriction, increased skin conductivity, pupil
constriction, increased respiratory rate, and decreased salivation), increased CNS activity (EEG alpha
desynchronisation), and neuroendocrine response (increased secretion of cortisol and epinephrine).
Nevertheless, only a handful of studies have reported the synchronisation of two or more
emotion components to confirm the emotion manipulation. For instance, facilitated memory for the
emotion stimulus employed by Burke et al. (1992) and Heuer and Reisberg (1990) can be more readily
attributed to emotion as subjective emotionality was correlated with skin conductance responses for
the emotion stimulus, but not for the neutral stimulus. Likewise, Adolphs et al. (2001) revealed that
aversive stimuli were rated as more negative and arousing than neutral stimuli, elicited higher and more
frequent skin conductance responses, and the that skin conductance responses were correlated with
subjective arousal ratings and memory.
2.4.2 Cognitive
Scherer proposed that the function of the cognitive component of emotion was to appraise a
situation for its significance to the well-being and goals of the organism, and to drive coordinated
changes in the subjective, peripheral efference, motor expression, and motivational components of
27
___________________________________________________________________________________
emotion. Detection of changes in CNS activation with imaging techniques, particularly in emotion
regulating structures such as the amygdala, anterior cingulate gyrus, parahippocampal gyrus, caudal
orbitomedial prefrontal cortex, and anterior insula, can index the increased information processing
required to appraise the emotion stimulus.
CNS activation can also be determined by electroencephalography (EEG), in which global
changes in surface cortical oscillations are recorded. The various frequencies and locations of these
oscillations signify generalised information processing states. The slower oscillations (delta, 1 – 4 Hz;
and theta, 4 – 8 Hz frequency range) signify drowsiness, sleep, meditation, or neurological pathology
(reviewed by Pizzagalli, 2007). When theta oscillations are present in the frontal midline, they signify
mental effort and sustained attention and are postulated to originate from the anterior cingulate cortex
(ACC). The presence of frontal midline theta may therefore signify emotion processing (Sammler,
Grigutsch, Fritz, & Koelsch, 2007). The alpha rhythm (8 – 13 Hz) has been argued to be the normal
resting rhythm of healthy awake and relaxed adults, with decreases in alpha rhythm associated with
increased brain activity. The alpha rhythm in posterior regions has been associated with increased
attention and vigilance, while asymmetry in frontal alpha has been associated with trait differences in
motivation to withdraw from or approach emotionally salient objects or situations (Davidson, 1992), and
with the direct effect of approach (joy, anger) and withdrawal (fear, sadness) emotions on frontal
activation (reviewed by Coan & Allen, 2004). The higher frequency bands (beta, 13 – 30 Hz and gamma,
36 – 44 Hz) are associated with the cognitive processes of attention, perceptual processing, and cortical
integration and synchronisation (Pizzagalli, 2007). Changes in CNS activity in response to emotional
stimuli could thus act as an index of the cognitive component of emotion.
Cognitive processes that underlie the appraisal of an emotion stimulus could also be determined
by stimulus evaluation checks (SECs). SECs are evaluations that determine “the major families of
emotional states” (Scherer, 2001, p. 94) and include appraisal of the stimulus relevance, its causal
implications, the degree of coping potential required, and the internal (self) and external (social)
normative significance of the emotional situation. The relevance of the stimulus is appraised in terms of
its novelty, intrinsic pleasantness, and goal relevance. Information that is novel (i.e. sudden and
intense), intrinsically pleasant or unpleasant (e.g., chocolate or spiders), and relevant to the goals or
needs of the organism, will attract greater attention focus and information processing. Attention focus
can be determined by the scope of memory accuracy (e.g. the Easterbrook hypothesis, reviewed by
Christianson, 1992a; the Goal Relevance hypothesis proposed by Levine & Edelstein, 2009; and
28
___________________________________________________________________________________
'memory narrowing', reviewed by Reisberg, 2006), thus operationalizing one aspect of the cognitive
component of emotion.
The use of intrinsically unpleasant emotional stimuli has been particularly successful in
demonstrating cognitive processing effects on memory (reviewed by Levine & Edelstein, 2009; Reisberg
& Heuer, 2004). Images of death and mutilation typically elicit greater memory accuracy for details
relating to the central emotive content than for background or peripheral details. For instance, Adolphs,
Denburg and Tranel (2001) revealed verbal and visual recall of the central features of aversive images,
such as the location of a dead body, were enhanced while memory for peripheral details, such as the
colour of the ground or the orientation of the body, were impaired. Similarly, Kensinger, Garoff-Eaton
and Schacter (2007b) presented negative items (e.g. snakes) on neutral backgrounds (e.g. near a river)
and found memory for negative items was enhanced while memory for neutral background details was
impaired. This effect was not evident for neutral items (e.g. chipmunks) presented on neutral
backgrounds. As these emotional stimuli were purposely designed to elicit an immediate and aversive
response, they probably elicited stimulus driven attention capture (Reisberg, 2006), possibly mediated
by the amygdala (Buchanan & Adolphs, 2004; Kensinger, 2009), leading to a narrowing of information
encoding and retention.
Intrinsically pleasant emotional stimuli tend to be less activating (Taylor, 1991) and therefore
may not have the same attention capturing impact as unpleasant stimuli. Nevertheless, stimuli that
elicit positive emotion tend to imbue information processing strategies that are distinct from negative
emotion. For instance, taking from theories of positive affect on cognitive processing (e.g., Ashby, Isen,
& Turken, 1999), positive affective states increase cognitive flexibility and creativity, elicit information
processing strategies that use heuristics to decrease decision making time and effort, and broaden the
scope of information processed (reviewd in full by Taylor, 1991). Appraisal of positive emotional stimuli
could thus be determined by decision making speed and a broad and inclusive memory scope.
Evaluating the type and scope of information retained may therefore assist in determining whether the
cognitive component of emotion was activated by the emotion stimulus.
Interestingly, memory narrowing is a phenomenon well documented in laboratory experiments,
yet it appears that it is not as common in naturally occurring emotional events (with the exception of
the witnessing of traumatic events such as violent crime, reviewed by Christianson, 1992b). Laney,
Heuer and Reisberg (2003) demonstrated this point by revealing that descriptions of past emotional
events did not always include reference to visually explicit information. They found that of the
29
___________________________________________________________________________________
memories described, on average 75% did not have a distinct visual focus. An example given was
notification by telephone of the death of a family member. These ‘non-visual’ memories were described
by the authors as thematic in nature.
In a follow up study, Laney et al. (2004) elicited an emotional response in study participants with
thematically induced emotion, defined as involvement and empathy with an unfolding event in the
absence of visual emotional information, to test the effect on memory. Emotion was elicited by a
narrative of a woman being attacked when out on her first date. A neutral comparison story told of a
woman who went on an uneventful date. Both versions of the story were accompanied by neutral
images. An emotion response to the attack story was confirmed by significantly higher ratings of
subjective anger and a trend for increased heart rate compared to the neutral story. Of interest, the
authors revealed that for both groups, both gist and central information was better recalled than
peripheral information, indicating that generalised memory narrowing occurred even in the absence of
attention capturing visual information. In terms of the neurobiological model of EEM, the failure to
detect differences in memory narrowing, and to elicit heart rate increases greater than chance may have
reflected a weak emotional response. Facilitated memory for the emotional story may therefore have
been caused by cognitive processes. For instance, the emotional story may have been more cohesive
than the neutral story, thus making it more memorable (Levine & Edelstein, 2009). Further
investigation of memory narrowing for thematic emotional events, which may be more consistent with
typical emotional experiences, is therefore warranted.
This section has identified that the cognitive process of appraising emotional information may
influence the type of information attended to and retained. Differences in the scope of information
retention for emotional compared to neutral stimuli may therefore be a useful index of the cognitive
component of emotion proposed by Scherer (2001).
2.4.3 Motivation and motor expression
The remaining components of emotion that have yet to be fully explored in neurobiological
paradigms of EEM are directional (motivational component), and communicative (motor expression
component). The motivational component reflects the preparation of specific behaviours or action
tendencies to cope with the emotion elicitor. The most readily defined action tendency is the
approach/withdraw dichotomy, in which appetitive stimuli elicit approach behaviour and aversive
stimuli elicit withdraw behaviour (Cacioppo & Gardner, 1999). Other action tendencies include moving
30
___________________________________________________________________________________
over (dominance), moving under (submission), brief gratuitous contacts (playful exuberance), and
unfocussed attentional stance (receptive openness) (Frijda, 2009).
The motivational component of emotion could be determined by controlling the moderating
influence of trait approach/avoidance motivation on attention scope. For instance, Gable and HarmonJones (2008) argued that if an individual was predisposed to approach appetising or rewarding stimuli,
measured with Carver and White’s (1994) behavioural inhibition and activation system (BIS/BAS) scales,
then they would have greater attention narrowing for appetising and rewarding images than individuals
who were not approach motivated. Their hypothesis was supported with regression analyses, which
revealed that after viewing approach motivating images, attention narrowing was predicted by
individuals with higher BAS scores but not with lower BAS scores. Attention scope, and by extension
memory scope, may thus provide useful information about the motivation component of emotion.
The final emotion component, motor expression, reflects the communication of emotion and
adaptive responding. Emotion can be communicated by facial expression, vocal expression, gestures, or
posture. The adaptive response may be to clench the teeth for biting in defence or attack, or to protrude
the tongue to eliminate odorous food (disgust). Considerable research on the facial expression of
emotion has identified facial muscle activity that corresponds with specific emotions. For instance,
emotions that elicit feelings of pleasure correspond with eye and mouth opening, nostril dilation, and
upturned corners of the lips, while emotions that are unpleasant correspond with lowered brows, eye
and nostril closing and upper lip raising (Scherer, 2001). Measurement of facial muscle activity thus
provides an additional index of complex emotional responding.
Testing the relationship between Scherer’s emotion components and memory would further our
understanding of the parameters of the memory effect. Detection of a relationship between facilitated
memory and activation of the peripheral efference component, but not the cognitive component of
emotion, would support an orienting reflex explanation of facilitated memory. That is, the attention
capturing stimulus would elicit an interruption to homeostasis that supported information processing at
a perceptual level, but not to great detail at an appraisal and evaluation level. Conversely, detection of a
relationship between facilitated memory and activation of the cognitive component of emotion, but not
the peripheral efference component, would support a non-neurobiological explanation of facilitated
memory; rehearsal or elaboration for instance. Detection of a relationship between facilitated memory
and the synchronisation of multiple components of emotion, ideally subjective feeling, peripheral
efference, and cognitive would further support the neurobiological model of EEM. When considering
31
___________________________________________________________________________________
application of these research findings outside the research laboratory, knowing whether memory is
modulated by an orienting reflex, by rehearsal or elaboration, or by emotional arousal would assist with
the development of appropriate memory treatments.
2.5 Music as an emotion stimulus
For an emotion to occur, the organism must feel that an event or object is relevant to its current
needs, goals, values, or general well-being (Frijda & Scherer, 2010). It can be difficult to select ‘relevant’
emotional stimuli that can elicit these feelings in the experimental laboratory. Attempts have been
made using stimuli such as highly evocative images, images combined with tragic stories, or films; or by
psycho-social means, such as receiving an unexpected reward or preparation for public speaking. These
methods have been effective, as demonstrated by subjective reports and physiological correlates of
emotion. Nevertheless, the strength, reliability, and external validity of these methods remain
questionable (Eich, Ng, Macaulay, Percy, & Grebneva, 2007). For instance, it is likely that emotion is
only weakly elicited due to the low level of relevance of the stimuli. Furthermore, due to the nature of
the stimuli (e.g., evocative images) or the complexity of delivering the emotion stimulus (e.g., surprising
someone with a gift), it would be impractical to use these methods to manipulate emotion in applied
settings (e.g., in school settings, or aged-care facilities) to enhance memory.
Music as a source of emotional arousal has yet to be fully explored in neurobiological
investigations of EEM. This is surprising given that music has been described as a powerful source of
emotional experiences (Gabrielsson, 2001) and is used widely to regulate emotion (Juslin & Laukka,
2004; Juslin, Liljestrom, Vastfjall, Barradas, & Silva, 2008; North, Hargreaves, & Hargreaves, 2004). The
emotional function of music may have even contributed to the evolution of language (Brown, 2000). For
instance, words provide referential meaning in language while the prosody of language, the variation in
musical qualities such as pitch, rhythm, tone, melodic contour, and stress provides emotional
information (Fernald, 1993; Trainor, Austin, & Desjardins, 2000). Music can portray a wide range of
emotions, most notably exemplified in opera, and elicit strong feelings of emotion in listeners.
Gabrielsson and Lindstrom-Wik (2003) demonstrated this point when they asked 900 participants to
describe their experience of music. Factor analysis of participant responses revealed a number of
emotion components, such as physical reactions, changes in behaviour, feelings of emotion, and
existential and transcendental feelings. These results are meaningful as they provide empirical support
32
___________________________________________________________________________________
for the common notion that music can elicit powerful responses in listeners. Additionally, categories of
responses felt by these research participants were consistent with components of emotion (Scherer,
2010).
One potential argument against the use of music to manipulate EEM is that it is not capable of
eliciting adaptive emotions such as fear, anger, disgust, sadness, or happiness (see Koelsch, Siebel, &
Fritz, 2010 for a review of the arguments for and against music as a source of real emotions). This is an
important point when considering music as a source of emotional arousal to manipulate EEM. Music
cognitivists such as Kivy (1990) and Konecni (2008) argue that music in itself may not have the capacity
to elicit adaptive emotions, yet it may elicit a sense of emotion by way of learned and culture specific
music features that portray emotions (e.g. in Western culture minor mode is perceived as sad and major
mode is perceived as happy). They do not seek to minimize the subjective experience of emotion that
can be instilled by music; their argument is that the experience is not derived from an adaptive response
that supports survival, but rather an aesthetic appreciation of the music (aesthetic emotion) or musicinduced cognitive associations. As such, music induced aesthetic emotions may not initiate the full
range of physiological and behavioural responses that would be expected from adaptive emotions.
There are, however, a number of music theorists who propose that music emotion does have
adaptive function (Ball, 2010; Huron, 2003; Levitin, 2008; Perlovsky, 2010). Two such functions are
pleasure derived from music listening, and social facilitation. Pleasure is felt when goals have been
achieved, when exploratory behaviour results in reward, when one receives an unexpected gift, and so
on. The pleasure state then imbues an information processing style that is more open to new ideas,
elaborative, creative, broad and inclusive, flexible, intuitive, and uses less cognitively demanding
heuristics to make decisions (see Ashby et al., 1999; Taylor, 1991, for reviews of empirical studies testing
positive affective states on information processing). Information processing in this way reflects an
outward looking view of the world, an exploration of new possibilities, and a broadening of horizons
(see Fredrickson’s (1998) ‘broaden and build’ theory of positive affect and Cacioppo & Gardner’s (1997)
‘positivity offset’ hypothesis), all of which could be considered adaptive. Pleasurable music may also
allow people to feel positive and to maintain self-esteem and preservation of self-concept (the 'mood
repair hypothesis' proposed by Isen (1984) and reviewed by Taylor, 1991). In contrast to the assertion
that pleasure from music has no adaptive value (e.g., Pinker, 1997), it may be speculated that music can
have a direct influence on adaptive cognitive processes and can elicit mood states that support
emotional well-being, exploration, and hence survival.
33
___________________________________________________________________________________
On another level, music may have functions that facilitate social affiliation through coordinated
movement, bonding (mediated by the release of oxytocin, Huron, 2003), and relief of the anxieties
associated with living in social groups (Benzon, 2001; see also Thompson, 2009 for a review of the
proposed adaptive functions of music). Perlovsky (2010) maintains that the purpose of music is to
bridge the gap between knowledge of the world and the emotional self, thereby enabling individuals to
maintain their sense of self and their meaning for existence in an ever changing social environment. This
gap was created with the advent of language, which anatomically separated emotional vocalisation
centred in ancient brain structures from communication of “specific and complicated knowledge”, (p. 7),
rewired to temporal cortical structures. The effect of music is therefore argued to not only elicit
pleasure in the individual, with its subsequent facilitatory effects on cognitive processing, but also have
a cohesive effect on the functioning of society (see also Koelsch et al., 2010). One of the primary
outcomes of music as a social cohesive may therefore be emotion regulation, in which individuals have
access to a means of anxiety relief, to share emotions with others, and a safe medium for selfexploration.
A recently developed theoretical framework describes several psychological mechanisms by
which emotion can be evoked by music (Juslin, Liljestrom, Vastfjall, & Lundqvist, 2010; Juslin & Vastfjall,
2008). The authors have named the theoretical framework BRECVEM, with each letter corresponding to
one of the following mechanisms: (1) Brain stem reflexes, where sudden changes in auditory sensations
(e.g. loudness, tempo, and dissonance) directly activate brain stem structures like the reticular system in
a reflexive way to produce changes in central and peripheral nervous systems; (2) Rhythmic
entrainment, in which strong rhythmical patterns in music interact with internal body rhythms and
become synchronised (e.g. respiratory or heart rate increases or decreases to match the music tempo),
with the proprioceptive feedback leading to a feeling of emotion in the listener; (3) Evaluative
conditioning, where certain music types that are repeatedly presented with positive or negative stimuli,
consciously or unconsciously, become associated; (4) emotional Contagion, in which characteristics of
emotion that can be expressed by music (e.g. minor mode and slow tempo generally expresses sadness,
major mode and fast tempo generally express happiness, and musical instruments can mimic emotional
vocal expressions) become internalised by the listener; (5) Visual imagery, by which certain music
characteristics (e.g. increasing sound frequency elicits feelings of increasing height) can evoke mental
images that enhance the music’s emotionality; (6) Episodic memory, in which music is associated with a
previously experienced emotional event (e.g. the ‘their playing our song’ phenomenon) and when
34
___________________________________________________________________________________
heard, acts as an emotional cue4; and (7) Musical expectancy, related to knowledge of musical syntax,
and surprise when that syntax is violated. The BRECVEM theoretical framework can be applied to
empirical research to elucidate possible mechanisms that caused an emotional response to music (e.g.
Blumstein, Bryant, & Kaye, 2012; Olsen & Stevens, 2013; Vuoskoski & Eerola, 2011).
Given the continuing controversy over whether music elicits adaptive emotions, it is imperative
that efforts are made to verify music as an emotion eliciting stimulus. To meet the emotion criteria,
music must elicit synchronised changes in the emotion components proposed in Scherer’s Component
Process Model. There are now numerous studies that demonstrate music has the capacity to elicit
responses that are consistent with many of these emotion components.
2.5.1 Subjective feeling component
Asking participants to listen to music then reflect upon and report their feelings is the most
commonly used method to measure emotional responses to music. The results of studies measuring
subjective emotional responses to music across a wide range of participants, listening contexts, and
music genres has provided strong support for the hypothesis that music elicits emotions. However,
there is some concern that participants often report the emotion expressed in the music, rather than
the emotion they feel. Explicit instructions to participants to report the emotion they feel and not the
emotion expressed by the music may address this potential confound (see Schubert, 2007). Subjective
responses to music have been measured in a variety of ways, ranging from simple 5 point Likert scales
presented after music listening to measure discrete emotional responses, to more sophisticated
methods that measure fine grained continuous responses to music in two-dimensional emotion space
(Nagel, Kopiez, Grewe, & Altenmuller, 2005; Schubert, 1999).
Empirical evidence of subjective emotional responses to music in naturalistic settings has been
provided by numerous researchers. For instance, Sloboda (1991) revealed that of a sample of 83 16-70
year old musicians and non-musicians, 85-90% reported shivers, laughter, a lump in the throat, and
tears when listening to music. Sloboda and O’Neill (2001) asked eight non-musicians aged between 1840 years to report their music listening habits every two hours over a period of two weeks. They
4
Note that the episodic memory mechanism differs from the evaluative conditioning mechanism in terms of
autobiographical memory (former) and conscious or unconscious paired associations (latter).
35
___________________________________________________________________________________
revealed that there was a 44% chance that music would be listened to in a 2 hour period (purposeful
music listening occurred only 2% of the time), and that music was experienced during personal
maintenance, personal travel, and active leisure. This music listening resulted in participants becoming
more positive, more alert, and more focused on the present. Juslin and Laukka (2004) explored
expression, perception, and induction of emotion in the context of ordinary listeners’ interactions with
music in everyday life. In their study, 141 participants from the general population described basic
emotions such as anger, sadness, joy and fear as being expressed by music more often than complex
emotions such as jealousy and shame, and that emotion was induced about 50% of the time. Emotions
felt in response to music were commonly positive, such as happy, relaxed or moved. Their study
revealed that music not only expressed emotions, but also evoked emotions and was used to change
mood, match mood, and vent emotions. In addition, the most reported motives for listening to music
were to express, release and influence emotions, to relax, and for enjoyment, fun, and pleasure. All of
these responses point to a specific and active use of music to manipulate emotion (see also Saarikallio,
2011; vanGoethem & Sloboda, 2011).
In a study comparing music-induced emotion to non-musical emotion, Juslin et al. (2008) asked
32 college students to answer questions relating to music use at random times during the day
(participants were notified by a personal electronic device). Responses revealed that music listening
evoked feelings of happiness, elation, nostalgia and longing most frequently. These responses to music
were in contrast to anger, irritation, boredom, indifference, anxiety, and fear reported most frequently
for non-music emotions. Participants reported that their motive for listening to music was to relax, to
evoke calmness or contentment, and to influence feelings of sadness or melancholy. With the aim of
creating and validating a specific rating system for musical emotions, Zentner et al. (2008) surveyed a
total of 1393 participant’s emotional responses to music over four separate studies. In the first two
studies (n = 354), they revealed that techno and Latin American music was rated as activating and
classical and jazz evoked feelings of amazement and peacefulness. These results indicate that
purposeful music listening can influence emotion.
Studies of subjective emotional responses to music in naturalistic settings demonstrate that,
according to a wide range of music listeners in different listening contexts, music elicits feelings of
emotion. A general characteristic of music is that it tends to elicit positive and activating affective
states, irrespective of the emotion expressed by the music. Furthermore, some musical genres can be
more activating than others. Subjective emotional responses to music are also well documented in
36
___________________________________________________________________________________
experimental laboratory contexts using both dimensional and discrete methods of emotion
measurement. Dimensional models of emotion classify subjective feelings of emotion along the two
primary axes of arousal and valence. A third dimension resembling emotional intensity has also been
proposed. However, as there is a lack of consistency in defining this third dimension and measurement
of three dimensions increases the complexity of the task, dimensional models are mostly presented in
two dimensions. The circumplex model developed by Russell (1980) and illustrated in Figure 2.6 is a
widely cited dimensional model of emotion. The model has been validated by numerous researchers
using a broad range of stimuli, including music (Bigand, Vieillard, Madurell, Marozeau, & Dacquet, 2005;
Nagel et al., 2005; North & Hargreaves, 1997; Nyklicek, Thayer, & Van Doornen, 1997; Schubert, 1999),
and a variety of response measures (reviewed by Posner, Russell, & Peterson, 2005).
Q2
Q1
Q3
Q4
Figure 2.6 The circumplex model of emotion (Russell, 1980, p. 1164). The horizontal axis reflects
affective valence, from negative to positive. The vertical axis reflects arousal, from passive to active.
Q1 = Quadrant 1, Q2 = Quadrant 2, Q3 = Quadrant 3, and Q4 = Quadrant 4 (Schubert, 1999).
37
___________________________________________________________________________________
According to the circumplex model, emotions can be broadly categorised into four quadrants;
positive and aroused or activated emotions such as joy and happiness would occupy Quadrant 1 (Q1,
top right), negative and aroused or activated emotions such as fear would occupy Quadrant 2 (Q2, top
left), negative and calm or deactivated emotions such as sadness would occupy Quadrant 3 (Q3, bottom
left), and positive and calm or deactivated emotions such as contentment would occupy Quadrant 4
(Q4, bottom right). In a study testing the validity of a method of continuously measuring emotional
responses to music, Schubert (1999) confirmed that participants spent longer in Q1 when reporting their
response to positive and active music (Slavonic Dance, composed by Dvorkac) and more time in Q2 and
Q3 when reporting their response to negative low arousal music (Morning, composed by Grieg). Similar
effects of music on arousal and valence ratings have been reported by Gomez and Danuser (2004),
Grewe, Nagel, Kopiez and Altenmuller (2007), and Witvliet and Vrana (2007). In sum, it can be
concluded that music elicits subjective feelings of emotion.
2.5.2 Peripheral efference and motor expression components
Activation of the peripheral efference and motor expression components of emotion have been
widely investigated. Many of these studies focus on facial expression and ANS activity. In the primate,
facial expressions communicate emotion to others (Ekman, 1992; Ekman, Levenson, & Friesen, 1983).
Activation of facial muscles such as the corrugator supercilii, located on the upper central brow (frown),
has been demonstrated to predict negative emotions, and the zygomaticus major, located on the cheek
(smile), predicts positive emotions (see reviews by Bradley & Lang, 2000; Cacioppo, Berntson, Larsen,
Poehlmann, & Ito, 2000). Meta-analyses of ANS activity studies (cardiac function, respiration,
temperature, electrodermal activity, facial muscle activity, and movement) conducted by Cacioppo et al.
(2000) revealed reliable differentiation of emotion could be obtained when emotion was categorised in
terms of valence, for example how positive or negative it was. Their meta-analyses revealed increased
cardiac function in response to negative compared to positive stimuli was a reliable predictor of
emotion. A series of studies reported by Bradley and Lang (2000) revealed that skin conductance
reactivity (or skin conductance response, SCR) was higher for negative and positive images and words
compared to neutral comparisons, and that skin conductance reactivity was linearly related to
subjective reports of arousal. There was, however, no difference in reactivity between positive and
38
___________________________________________________________________________________
negative stimuli, indicating the skin conductance response is a reliable measure of arousal, but does not
differentiate the valence of the emotion.
Cacioppo et al. (2000) and Bradley and Lang (2000) have proposed that physiological responses
to emotional stimuli can be best explained in terms of the motivation to approach (positive) or avoid
(negative) the stimulus, and the degree of urgency to which the motivational action is required
(arousal). For instance, both positive and negative emotion stimuli elicit action tendencies (approach or
withdraw) that require ANS support. Therefore, emotional stimuli that requires action increases
physiological arousal levels, which can be mapped onto the arousal dimension of the circumplex model.
Further, negative stimuli are more likely to pose immediate threat to survival or wellbeing, thus leading
to greater ANS response than positive stimuli (Taylor, 1991). The degree of arousal could therefore
differentiate positive and negative emotions, with higher levels of arousal more likely to index ‘negative
and aroused’ emotions in Q2 of the two-dimensional space. Previous research has indicated that the
valence of low arousal emotions, such as sadness and contentment, are difficult to differentiate based
on their physiological correlates. Therefore the location of emotions that reduce physiological activity
on the valence dimension (bottom half of the two-dimensional emotion space) is difficult to predict.
The emotion categorisation framework described above will be used to summarise the findings
from a selection of studies investigating music effects on emotion and the peripheral nervous system
(Figure 2.7); see Hodges (2010) for a full review of studies investigating physiological responses to music
dating back to the beginning of the 20th century. Each of the studies presented in Figure 2.7 has been
assigned a number that corresponds to the author’s study. As the method of comparison between
conditions across studies varies a great deal, for instance, reported comparisons included those that
were between music that varied in valence, arousal, both valence and arousal, or to non-music listening
conditions (such as viewing a musician without audio compared to hearing the musician without vision,
Chapados & Levitin, 2008), the results summarised in Figure 2.7 have been marked with ‘Q1’, Q2’, ‘Q3’
or ‘Q4’ to clarify the results being compared. Therefore, a study number presented in Q1 of the
emotion space with the superscripts Q2 and Q3 would represent a study that compared music from Q1
(positive and arousing) with music from Q2 (negative and arousing) and Q3 (negative and deactivating).
The two graphs at the bottom of each quadrant represent the predicted5 (left graph) and actual (right
graph) percentage of studies reporting increases, no change, and decreases in SNS response. The Q1
5
Predictions were based on subjective reports cited in music-emotion studies reviewed in section 2.5.1 and nonmusic studies of physiological emotion response reviewed in the current section.
39
___________________________________________________________________________________
prediction graph reflects prediction of the highest proportion of studies reporting SNS increases and the
least reporting no change; Q2 reflects the prediction of increased SNS responses, however with a
greater proportion of studies reporting no change (due to lower arousal levels elicited by positive
emotions); and Q3 and Q4 reflect the prediction that in most instances (approximately 66%), low
arousal music will reduce SNS levels and a large proportion (approximately 33%) will report not change.
A few points should be noted when interpreting the contents of Figure 2.7. Some studies used
multi-modal sources of emotional arousal, such as music varying in emotional valence and arousal whilst
participants viewed films or images (Baumgartner, Esslen, & Jancke, 2006a; Ellis & Simons, 2005).
Therefore, as physiological effects of music were of interest, only music main effects or conditions
containing music and no other stimulus have been summarised. The chills response – the experience of
goose bumps or thrills when listening to music – has been included as a physiological response due to
research conducted by Eckart Altenmuller and colleagues (Grewe, Kopiez, & Altenmuller, 2009; Grewe,
Nagel, Kopiez, & Altenmuller, 2005; Nagel, Kopiez, Grewe, & Altenmuller, 2008) demonstrating a strong
relationship between chills, emotional responses to music, and skin conductance responses. Facial
activity was not considered an SNS response therefore EMG results were excluded from the SNS graphs
in each quadrant. Finally, participant-selected music was treated as positive and arousing, even if the
music itself expressed negative emotion.
40
___________________________________________________________________________________
Q2 Negative Activated
Increase
No Change
Decrease
Q4
Chills
Q1
10
+v
Q2,Q4
10
13
HRR
RR
2 13
Q3 Q3 Q3
1 3 5
SCL
SCR
Temp
5 8
Q1
1
Q2,Q4,Q3 Q3
2
Q1,Q4,Q3
Q4,Q3
2
13
Q1,Q3 Q1 Q4,Q3
1
5 9
Q1,Q4
10
Q3
13
Q1 Q1 Q1
1 3 5
Q1,Q4,Q3
9
Q4
10
Q3
1
Q3 Q3
Q1,Q4
1 5 10
BL Q3 Q4
3 5 8
Q1,Q4
Q3 Q4
Q1
Q3
13
Q1
9
7 13
-a Q3 BL Q2,Q3
4 5 6 9
NM
11
Q1
12
Q3 Q3 -a
Q1
Q1
5 10
+v
2
Q1 Q1 Q4
1 5 8
25
+
0
0
%
-
42
+
Expected
Increase
Chills
EMGf
EMGs
HR
HRR
RR
SCL
SCR
Temp
0
Q3 BL
1 3 4 5 6
Q4,Q3
9
Q2,Q3
Q2,Q4
NM
5
10
11
-v Q3
2 7
Q2 Q2,Q3
1 5
60
21
-
+
Actual
0
2
7
Q2,Q1,Q4
Q2,Q4
2
13
Q1,Q3 Q1 Q2,Q3
1
7 9
Q2
13
Q2,Q3
9
Q1,Q4
2 13
Increase
Q1
1
Q1
7
66
%
33
%
0
+
0
Expected
-
27
+
35
0
11
48
%
-
+
BL
6
5
2
Q2
8
-
Actual
13
0
-
Q2
10
13-v
Q1
13
+a Q2
4 8
-v
Q2,Q3
2 13
+a Q1
4 9
Q1
Q2
8 10
Q2
10
-v
48
40
Actual
10
Q2,Q1,Q3
2
Q2,Q1,Q3
Q2,Q3
2
13
Q2,Q1,Q3
Q2,Q1
9
10
Q4
13
BL Q2Q3
6 9
Q2,Q1 Q2,Q1 Q2,Q1
1
Q2,Q1 Q2
1
5
1Q2,Q3
7Q3
Q4 Positive Deactivated
No Change
Decrease
Q1 Q1
Q2
2-v 13Q2,Q3
NM
Expected
13 7
BL Q2,Q1 Q1
3 5
9
1
3
Q1
9
Q2,Q1 Q1
5
7
+v
2
Q1
5
Q2,Q4
1 10
0
Q2,Q1,Q4 Q1
13
Q3
40
%
Q3 Negative Deactivated
No Change
Decrease
+v
+v
36
13-v
2
7
Q2,Q4,Q3
2
Q1,Q3 BL Q2 Q3
1
3 5 7
Q4
Q2,Q4
9 10
BL
Q4
12 13
Q2 Q2 Q2 Q1
1 3 5 9
NM
11
Q2,Q4,Q3
75
%
Decrease
NM
11
Q1,Q4,Q3
EMGf
EMGs
HR
+v
10
Q1 Positive Activated
Increase
No Change
Q1
Q2,Q1
10
66
%
33
%
0
+
0
Expected
-
40
50
0
-
10
+
Actual
Figure 2.7 Summary of studies measuring physiological responses to music. Horizontal axis represents valence
(negative to positive), vertical axis represents activation (activated to deactivated). The left graph at the bottom of
each quadrant represents predicted proportion (%) of studies reporting SNS increase (+), no change (0), and
decrease (-). The right graph represents actual proportion. Measures: EMGf electromyogram frown (corrugator
supercilii), EMGs electromyogram smile (zygomaticus major), HR heart rate, HRR heart rate response, RR
respiratory rate, SCL skin conductance level, SCR skin conductance response, Temp finger temperature.
Comparison Key: BL Baseline, NM neutral music, Q2 negative active, Q1 positive active, Q4 positive calm, Q3
negative calm, -a low arousal, +a high arousal, -v negative valence, +v positive valence. Study authors: (1)
Baumgartner, Esslen, and Jancke (2006a), (2) Ellis and Simons (2005), (3) Etzel, Johnsen, Dickerson, Tranel, and
Adolphs (2006), (4) Iwanaga and Moroki (1999), (5) Krumhansl (1997), (6) Lingham and Theorell (2009), (7)
Lundqvist, Carlsson, Hilmersson, and Juslin (2009), (8) Nater, Abbruzzese, Krebs, and Ehlert (2006), (9) Nyklicek, et
al. (1997), (10) Rickard (2004), (11) Salimpoor, Benovoy, Longo, Cooperstock, and Zatorre (2009), (12) Sammler,
Grigutsch, Fritz, and Koelsch (2007), (13) Witvliet and Vrana (2007).
41
___________________________________________________________________________________
The consistency between expected and actual patterns of physiological responses to music in
the studies summarised in Figure 2.7 makes it clear that music has the capacity to increase physiological
arousal, although many studies also reported no change. It can be seen that there are more studies
reporting increased SNS activity than decreased activity in response to arousing music (both negative
and positive). Conversely, there are more studies reporting decreased SNS activity than increased
activity in response to deactivating or calming music (negative or positive). These trends are consistent
with those predicted for physiological responses to emotional stimuli. Interestingly, the most consistent
increases in physiological arousal were elicited by positive and arousing music, which was in contrast to
the prediction that negative and arousing music would elicit the highest physiological responses. This
trend is consistent with the subjective reports of positive and activated feelings elicited by music
(reviewed in Section 2.5.1). Nevertheless, there remain a substantial number of studies that report no
music effect. The null results that are common in these studies may be due to a number of confounding
factors. For instance, pleasure elicited from participants’ own music selections results in greater
physiological responses than experimenter-selected arousing music (Rickard, 2004). Music preference,
liking or familiarity also influences physiological activity, with highly familiar music polarising
physiological responses (Witvliet & Vrana, 2007). Physiological responses also appear to depend on
gender, with females responding with greater activation to negative and arousing music (Nater et al.,
2006). These and many other confounding variables result in highly variable data, thus weakening the
capacity of statistical tests to detect differences between conditions. However, overall there is evidence
to support the hypothesis that emotionally arousing music activates the ANS, thus validating the use of
music to manipulate emotion and facilitate memory.
Studies of music effects on the neuroendocrine system are less prevalent than those on
subjective or physiological responses. With the small number of studies conducted in this area comes
even wider variability in the results. Of the studies conducted, there have been reports of music
decreasing, increasing, or having no effect on neuroendocrine activity. Cortisol levels were reported to
decrease after listening to music that was meditative (Mockel et al., 1994), relaxing (Khalfa, Dalla Bella,
Roy, Peretz, & Lupien, 2003; Yamamoto, Naga, & Shimizu, 2007), or sung by a choir (Kreutz, Bongard,
Rohrmann, Hodapp, & Grebe, 2004), and activating music has been reported to increase cortisol levels
(Gerra et al., 1998; Vanderark & Ely, 1992; 1993, Music Major group). However, both relaxing and
activating music has also been reported to have no effect on cortisol levels (Gerra et al., 1998; Knight &
42
___________________________________________________________________________________
Rickard, 2001; Mockel et al., 1994; Nater et al., 2006; Rickard, 2004; Stefano, Zhu, Cadet, Salamon, &
Mantione, 2004; Vanderark & Ely, 1993, Biology Major group).
The effects of music on plasma concentrations of NE, EPI, and ACTH have been investigated to a
lesser extent, and the findings are similarly inconsistent. Gerra, et al. (1998) exposed participants to 30
minutes of positive relaxing classical music (60-100 bpm) and techno music (130-200 bpm), separated by
three days. They revealed that techno music significantly increased plasma NE and ACTH whereas
classical music had no effect. Hirokawa and Ohira (2003) reported a trend for uplifting music to increase
NE levels. Yet Mockel et al. (1994) reported no effect of classical music on ACTH, NE or EPI, and
Vanderark and Ely (1992, 1993) reported no effect of classical music on NE. There is therefore weak
evidence for music to modulate the endocrine system. The inconsistency of the music effects
demonstrates that various additional factors may need to be controlled to reduce variability. For
instance, music effects have been found to be moderated by features of the music (e.g. moderate
compared to fast tempo, Gerra et al., 1998), individual differences in musical instrument training
(Vanderark & Ely, 1992, 1993), personality traits such as novelty-sensation seeking (Gerra et al., 1998),
and gender (Nater et al., 2006). The stringent and intrusive methodology required for collection of
neuroendocrine assays, combined with unreliable results, limits the practicality of measuring
neuroendocrine changes to index emotional responses to music.
2.5.3 Cognitive component
Increased central nervous system activation, particularly in the limbic system and orbitofrontal
cortex, is an index of the increased cognitive processing required to appraise emotional stimuli. Using
music as the emotion elicitor, researchers investigating the neural correlates of emotion have reported
activation of a number of limbic structures involved in emotion processing (see review by Koelsch et al.,
2010). In brief, various types of music (computer generated, naturalistic, consonant, dissonant,
experimenter-selected, and participant-selected) modulate activity in the orbitofrontal cortex,
amygdala, hippocampus, parahippocampal gyrus, anterior cingulate gyrus, insula, and ventral striatum.
Activation of these brain regions is consistent with cognitive appraisal of emotional information. In a
PET study, Blood and Zatorre (2001) scanned musician’s brains while they listened to their own
selections of intensely pleasurable music or another participants music. They revealed that the intensity
of chills elicited by music, argued to be a reliable marker of pleasurable response to music, was
43
___________________________________________________________________________________
correlated with increased blood flow to ventral striatum, midbrain, thalamus, anterior cingulate,
supplementary motor area, and cerebellum. Many of these structures, particularly a cluster of neurons
in the ventral striatum called the nucleus accumbens (NAc), form part of the reward system (Spanagel &
Weiss, 1999), a system inherently related to seeking behaviour and the experience of positive emotions
upon goal attainment (Panksepp, 1998). Additionally, chills intensity was correlated with decreased
blood flow to the amygdala, hippocampus, and ventral medial prefrontal cortex. The authors
interpreted decreased activity in the amygdala and hippocampus as inhibition of negative emotion.
Thus, pleasurable music was argued to not only activate the reward system, but also inhibit structures
involved in negative emotion. Similar brain regions have been activated when participants passively
listen to unfamiliar pleasant and arousing music (Brown, Martinez, & Parsons, 2004; Menon & Levitin,
2005), and self-selected intensely pleasurable music (Salimpoor, Benovoy, Larcher, Dagher, & Zatorre,
2011).
In a different approach to understanding the neural correlates of emotion, Baumgartner et al.
(2006b) combined emotional background music with emotional images taken from the International
Affective Picture System (IAPS). Using fMRI, they revealed that relative to viewing a fixation point,
viewing negative emotional images activated the right dorsolateral prefrontal cortex, indicating that the
cognitive process of evaluation had occurred. Additional activity in limbic and paralimbic structures was
elicited by negative emotional images combined with negative emotional music (fearful and sad classical
music). Specifically, the amygdala was activated by combining negative images with negative music, but
not by negative images presented alone. To further investigate whether the amygdala responds to
positive in contrast to negative auditory stimuli, Koelsch et al. (2006) presented participants with both
joyful dance tunes and digitally manipulated dissonant versions of the same tunes. Their results were
consistent with those reported by Blood and Zatorre (2001) in that positive and pleasant music
decreased activity in the amygdala. In addition, Koeslch et al. (2006) revealed that the dissonant and
unpleasant music comparison increased amygdala activity. In sum, these studies demonstrate that
positive music can both activate brain structures involved in reward and positive emotion, and inhibit
those that are involved in negative emotion. Similarly, negative music (when combined with pictures) or
music that is unpleasant can activate brain structures more commonly involved in negative emotion.
The extant literature has demonstrated that increases in frontal midline theta, generalised
decreases in alpha, and frontal asymmetry are correlated with emotion processing. The studies
summarised in Table 2.3 demonstrate that music can also elicit such changes. Positive and arousing
44
___________________________________________________________________________________
music has been shown to increase frontal midline theta relative to unpleasant or negative low arousal
comparisons in two studies (Lin, Duann, Chen, & Jung, 2010; Sammler et al., 2007), and to increased left
frontal activation (indexed by reduced alpha power) relative to negative music or sounds in an
additional four studies (Altenmuller, Schurmann, Lim, & Parlitz, 2002; Flores-Gutierrez et al., 2007; Kim,
Yoon, Kim, Jho, & Lee, 2003; Schmidt & Trainor, 2001). Nevertheless, it should be noted that two
studies did not detect any frontal asymmetry in response to positive or negative music (Iwaki, Hayashi,
& Hori, 1997; Sammler et al., 2007). Music has also been reported to reduce alpha power relative to
resting states (Iwaki et al., 1997). Interestingly, a number of the studies summarised in Table 2.3
measured a broad range of frequency bands and revealed that the alpha band was the most sensitive to
the music manipulations. Additionally, females were reported to be more responsive to emotional
music than males (Altenmuller et al., 2002). In contrast to these reports of brain activity in response to
music that are consistent with emotional response, Baumgartner et al. (2006a) revealed that music
alone relative to pictures alone or music combined with pictures had the least effect on brain activity. It
is not clear from this report however whether the music nonetheless increased brain activity relative to
baseline levels.
45
____________________________________________________________________________________________________________________
Table 2.3
EEG activation in response to music
Author/s
Age
(years)
12 - 15
Gender
Baumgartner
, Esslen, and
Jancke
(2006a)
Mean
26.1,
SD 5.3
FloresGutierrez et
al. (2007)
Iwaki,
Hayashi, and
Hori (1997)
Altenmuller,
Schurmann,
Lim, and
Parlitz (2002)
Handed
-ness
Right
Music
Training
Yes
22 F
Right
?
Mean
25.0,
SD 3.05
11 M
8F
Right
No
21 – 28
5M
5F
?
No
8M
8F
Music stimuli
15 s excerpts
40 jazz
40 rock-pop
40 classical
Half negative,
half positive,
matched on
arousal.
3 x 70 s excerpts;
fear, sadness,
happiness.
Classical genre.
Presented alone
or combined with
images.
3 full pieces,
pleasant and
happy (1 soft, 1
vigorous), and 1
fearful.
2 x 3 min 40 s
excerpts;
Fear, calm.
Classical genre.
46
Non-music
stimuli
40 x 15 s
environmental
sounds
48 IAPS and
other images;
fear, sadness,
happiness.
Presented alone
or combined
with music.
30 s white noise
alternating with
30 s music.
1 min resting
baseline
Electrodes
Frequency
Results
32, 10/20
system.
dc – 30 Hz
Positive music, ↑ L front
activation. More for females.
Negative music, bilateral
activation.
30, 10/20
system
collapsed to
4 clusters;
anterior L
and R,
posterior L
and R.
19, 10/20
system
8 – 13 Hz
(alpha)
Music alone least activating
relative to pictures alone and
music and pictures combined.
3 – 8 Hz,
8 – 10 Hz,
11 – 14 Hz,
14 – 25 Hz
Positive music ↑ L activation,
negative music bilateral
activation.
12
7.6 – 15.4
Hz (alpha1,
alpha2,
alpha3, and
beta).
↑ alpha-2 amplitude (cortical
activation) at the start of both
music excerpts relative to rest.
Arousing but not calming music
↑ frontal coherence relative to
rest. Calming music had similar
coherence values to rest.
____________________________________________________________________________________________________________________
Author/s cont.
Particip
ant age
(years)
Mean
24.6,
SD 2.5
Gender
Handed
-ness
Music
Training
Music stimuli
14 M
10 F
Right
No
30 s excerpts,
movie
soundtracks,
positive arouse,
negative arouse,
positive calm,
negative calm.
Sammler et
al.(2007)
20 – 30
8M
10 F
Right
No
10 x 1 min
consonant joyful
music tunes.
10 x 1 min
dissonant
manipulations of
same tunes.
6 x 30 s baseline
silence periods
63, 10/20
system
Schmidt and
Hanslmayr
(2009)
Mean
22,
SD 2
8M
8F
13 Right
3 Left
?
3 x 2 min classical
or rock music
excerpts; positive,
negative, neutral.
2 min resting
baseline
61,
equidistant
whole head
montage.
Schmidt and
Trainor (2001)
18 - 34
29 M
30 F
Right
?
4 x 60 s classical
music excerpts;
intenseunpleasant,
intense-pleasant,
calm-pleasant,
calm-unpleasant.
Lin et al. (2010)
47
Non-music
stimuli
Electrodes
Frequency
Results
32 channel
10/20
system.
1 – 3 Hz
(delta)
4 – 7 Hz
(theta)
8 – 13 Hz
(alpha)
14 – 30 Hz
(beta)
31 – 50 Hz
(gamma)
Individual
alpha
frequency
defined
theta, lower
alpha 1,
lower alpha
2, upper
alpha, beta.
8 – 12 Hz
(alpha)
4 – 8 Hz
(theta)
12 – 20 Hz
(beta1)
Positive valence and high
arousal music related to ↑
midline theta.
F3, F4, P3,
P4, 10/20
system
8 – 13 Hz
(alpha)
↑ frontal midline theta relative
to baseline for music with high
pleasant ratings. No change
from baseline for music with
high unpleasant ratings.
No frontal asymmetry in the
alpha bands detected.
Alpha asymmetry; ↑ left
baseline activity = ↑ positive
ratings and ↑ enjoyment for all
music compared to ↑ right
baseline activity. Negative
music enjoyed most by left
active. No effect for the other
frequency bands.
Positive music ↑ left activity,
negative music ↑ right activity.
Positive music ↑ frontal
activity relative to negative
music.
___________________________________________________________________________________
The research on music effects on the neural correlates of emotion generally supports the claim
that music is capable of stimulating the cognitive component of emotion, indexed by increased cortical
activity in emotion processing brain regions. Music has therefore been found to activate brain
structures involved in emotion processing and elicit valence specific changes in electrical brain activity
that is consistent with emotional responding.
This section has reviewed the effects of music on multiple components of emotion; subjective
reports of emotion feeling, cognitive processes, and changes in the peripheral nervous system, somatic
nervous system, and neuroendocrine system. In general, music has been subjectively reported as
playing an important role in emotion regulation and for the elicitation of positive and activated mood
states. Consistent with the subjective reports, music elicits brain activity (cognitive component) that is
correlated with emotion processing in general, and positive emotion in response to rewarding stimuli.
Additionally, negative and unpleasant music elicits activation of the amygdala, which is an important
structure involved in modulating EEM. Physiological responses to music provide further evidence that
music has the capacity to modulate the physical, ‘action tendency’ component of emotion.
Nevertheless, the high variability in these responses indicates that music effects can be inconsistent and
warrant strict control of moderating variables. Such variables include gender, musical features, music
familiarity, and musical instrument training.
A unique advantage of music, however, is that it can act as both an intrinsic and an extraneous
source of emotional arousal. In terms of an intrinsic source of arousal, music can be the source of
emotion and the target for memory testing, much like emotional images, words or stories. In this
context, memory for emotionally arousing music could be compared to memory for neutral music
(arousal intrinsic to the MTBR). The unique quality of music is that it can also be used as an extraneous
source of arousal, applied in a similar way to other exogenous arousal treatments (e.g., administering
cortisol or exposure to psychological stress) before, during or after presentation of unrelated MTBR. In
terms of external validity, music as a source of extraneous arousal would clearly be more acceptable
than existing psychological or pharmacological interventions designed to manipulate EEM.
48
___________________________________________________________________________________
2.6 Music effects on memory encoding and consolidation
Anecdotal evidence of music’s effect on memory includes the ‘they’re playing our song’
phenomenon, whereby certain music tracks can cue vivid recall of autobiographical memory (e.g.,
Janata, Tomic, & Rakowski, 2007). The emotional nature of the music may have therefore activated the
neurobiological memory consolidation mechanism, thus strengthening memory for the event.
Unfortunately, this proposition is difficult to test retrospectively as there is no means of measuring
neurobiological activity during the encoding of the event. Nevertheless, the phenomenon is powerful
and supports music as an emotional stimulus to facilitate LTM.
Empirical evidence of the effect of music on short-term and/or working memory is more
forthcoming. The background to many of these studies has been to investigate the effect of background
auditory stimuli on phonological processing (e.g. Boyle & Coltheart, 1996; Salame & Baddeley, 1989) and
learning (e.g. de Groot, 2006; Jancke & Sandmann, 2010), or as a means of inducing mood to test moodcongruent, mood-dependent, or context-dependent memory. In the phonological processing domain,
various properties of the music and the task have been manipulated to increase or decrease the
demands on information processing. The main findings are that when information processing demands
are increased by increasing the tempo, loudness, complexity or familiarity of background music, or by
including lyrics; memory performance declines (Cassidy & MacDonald, 2007; Iwanaga & Ito, 2002;
Nittono, 1997; Salame & Baddeley, 1989; Sousou, 1997; Woo Ee & Kanachi, 2005). These impairing
effects contrast with facilitatory effects of soft or simple instrumental background music (Hallam, Price,
& Katsarou, 2002; Kiger, 1989).
Studies have revealed that the effects of background music also depend on individual
differences in extraversion and preference for background music while studying. Furnham and Allass
(1999) revealed a classic cross-over interaction between music that differed in complexity (high and low)
and participants that scored high or low on the extraversion scale of the Eysenck Personality
Questionnaire. In the presence of complex background music, participants high on the extraversion
scale performed better on an object memory task than their low scoring counterparts, and low scorers
on the extraversion scale performed better in the presence of simple background music. Extraversion,
introversion, and preference for background music while studying (related to extraversion) were also
reported to moderate memory by Cassidy and MacDonald (2007) and Crawford and Strapp (1994). The
results of these studies demonstrate that background music that increases demand on limited cognitive
49
___________________________________________________________________________________
processing capacity can impair short-term memory performance if the individual is overly sensitive to
arousing stimuli (one of the primary characteristics of introverts, Strelau & Eysenck, 1987), while simple
low demand background music can facilitate memory performance. The latter conclusion should be
treated with caution however, given the low number of studies reporting facilitatory effects of music
relative to those that report no effects or memory impairment. Of note, moderating variables identified
in these studies include music familiarity, music complexity, and task complexity, but not gender (Hallam
et al., 2002; Kiger, 1989). With the exception of gender, these memory moderating variables are
consistent with the emotion moderating variables identified in Section 2.4.
Mood-state or mood-dependent memory is the phenomenon of facilitated LTM for an event
when current or induced mood at recall is congruent with mood during encoding. The mood state at
recall is posited to activate a network of associated memory nodes, thereby making the memory trace
more accessible to retrieval (Eich & Forgas, 2003). In this context, music has been used to induce mood
before encoding and then again before recall to test mood-dependency, or to influence the affective
tone of encoded information and then assess memory performance when the same music, mood
incongruent music, or no music is presented during recall. Empirical work demonstrates the
effectiveness of music for this purpose (Balch, Bowman, & Mohler, 1992; Balch & Lewis, 1996; Balch,
Myers, & Papotto, 1999; Boltz, Schulkind, & Kantra, 1991; Boltz, 2001; de l'Etoile, 2002; Tesoriero &
Rickard, 2011; Thaut & de l'Etoile, 1993), thus validating the use of music to manipulate mood.
In general, background music impairs or has no effect on short-term or working memory if
individual differences in arousal responsiveness are not controlled, but can act as a cue for the retrieval
of LTM. Studies that have specifically investigated the effects of music before, during, or after encoding
on long-term declarative memory have also generally failed to detect long-term facilitatory effects of
music, relative to silence (Boltz, 2001; de Groot, 2006; de l'Etoile, 2002; Thaut & de l'Etoile, 1993). The
studies conducted by de Groot, Boltz, and Thaut and de l’Etoile, presented music in the background
while participants either viewed short films (Boltz), learnt foreign language words (de Groot), or created
antonyms for words (Thaut & de l’Etoile). Although comparison of the various music conditions tested
in these studies failed to reveal statistically significant facilitatory effects of background music, trends in
the Thaut and de l’Etoile data suggest that music may have had some beneficial effect. More promising
music effects on LTM were revealed by Judde and Rickard (2010), who presented music during the
consolidation period. Figure 2.8 presents the mean percentage of total items recalled for each of the
above cited studies, with the exception of de Groot, which presents the percentage of items retained
50
___________________________________________________________________________________
from learning (week one) to retrieval (week two). Note that the means for Boltz, and Judde and Rickard
are approximations as values for these specific comparisons were not reported.
70
60
Percent
50
40
Music
30
Silence
20
10
0
Boltz
de Groot
de l'Etoile
Judde &
Rickard
Thaut &
de l'Etoile
Figure 2.8 Mean percentage of item recall between participants exposed to music and to silence
Similar trends for facilitatory effects of music relative to silence are apparent in the results
reported by Judde and Rickard (2010), and de l’Etoile (2002). Judde and Rickard presented arousing
music (based on pilot testing) immediately after word list learning or at a delay of twenty or forty five
minutes. They revealed a significant increase in the number of words recalled after a one-week delay
when positive or negative music was presented at the 20 minute interval relative to the 45 minute
interval (which was comparable to a control silence group). Further analysis of the moderating effect of
arousal responsiveness, measured with the BIS/BAS scale, revealed a significant facilitatory effect of
music at the twenty minute interval over silence for participants high on the BAS drive subscale.
Analysis of the subjective ratings of mood in response to the music revealed that the arousing and
positive music (Beethoven’s Symphony No. 6, 3rd mvt.) elicited positive mood valence without being
arousing, and arousing and negative music (Mussorgsky’s Night on Bald Mountain) was arousing without
eliciting negative mood. Based on these subjective ratings, it may be surmised that both positive nonarousing music and valence-neutral arousing music modulated LTM. Alternatively, as suggested by the
study authors, the subjective ratings may not have reflected changes in physiological arousal. As such,
replication with measures of physiological arousal is required. Importantly, the significant effect of post-
51
___________________________________________________________________________________
encoding music on LTM is consistent with arousal modulation of memory consolidation. Additionally,
the stronger memory effects for participants more responsive to arousing stimuli (high BAS drive
scorers) adds further support to an arousal-based explanation of facilitated memory in this group.
Turning to the effects of music presented before encoding on LTM, de l’Etoile (2002) induced
positive mood by asking participants to listen to the first five minutes of Mozart’s Clarinet Concerto in A,
Opus 107 before writing antonyms for a list of 40 words. Despite the trend in the data, free recall of the
antonyms the next day failed to reveal an advantage of music over silence. Closer scrutiny of each of
these studies indicates that arousal may not have been fully considered as a memory modulator (with
the exception of Judde & Rickard, 2010), despite convincing evidence of the modulatory effect of
arousal on the consolidation of memory (reviewed in Section 2.1). The failure to facilitate memory may
therefore be due to the use of pleasant but not particularly arousing music (e. g. part of J. S. Bach’s
Fourth Brandenburg Concerto and Mozart’s Clarinet Concerto in A), relative to the subjectively reported
arousing music used by Judde and Rickard (Beethoven’s Symphony No. 6, 3rd mvt. and Mussorgsky’s,
Night on Bald Mountain), or by not controlling for arousal differences between concurrently presented
music excerpts. For instance, Boltz (2001) presented participants with three ambiguous film clips
combined with either positive or negative music with various levels of arousal, or no music. Pre-testing
confirmed that the positive and negative music selections were subjectively perceived as positive and
negative. However, the arousal properties of the music were not controlled. Forced choice, old or new
recognition memory testing of film objects after a one week delay revealed that positive items were
more readily recalled for films presented with positive music, and that negative items were more readily
recalled for films presented with negative music. Nevertheless, music had no additive effect on memory
relative to no music. Thus, music influenced the encoding of information but did not increase the total
number of items retained. Furthermore, three different music excerpts combined with ambiguous films
were repeatedly presented to participants. It is therefore possible that arousal elicited by any one of
the music tracks influenced encoding and consolidation of all films, thus decreasing mean memory
differences between conditions after the one-week delay.
The investigation of the effect of music on short-term and LTM has demonstrated that music
can influence mood and modulate subjective experience of arousal. Nevertheless, it has yet to be
clearly demonstrated that the mood and arousal changes elicited by music are powerful enough to
modulate LTM. The findings that individuals who respond positively to arousing music, such as those
who score highly on extraversion or BAS scales, also have higher memory scores indicates that music
52
___________________________________________________________________________________
induced arousal may indeed be modulating performance. At time of writing, there have been no reports
of studies that monitor concomitant ANS responses to music to confirm definitively that music
facilitated memory is modulated by emotional arousal.
2.7 Individual differences
As has been demonstrated in this review, emotion and memory are complex constructs
comprised of many interacting variables. The use of music as an emotion stimulus adds further
complexity given the wide range of responses it can elicit in individuals. Variability that arises from
testing such complex phenomena can be reduced to some degree by controlling known moderating
variables. For instance, emotion effects on memory have been observed after controlling for individual
differences in arousal responsiveness (Bloise & Johnson, 2007; Judde & Rickard, 2010; Nielson & Lorber,
2009). Gender differences have also been demonstrated to moderate memory performance (e.g. Bloise
& Johnson, 2007; Cahill & van Stegeren, 2003). As individual differences in music experience can both
influence emotional response to music (e.g. Steinbeis, Koelsch, & Sloboda, 2006) and have a direct
influence on verbal memory (e.g. Chin & Rickard, 2010), it would seem likely that they would have a
moderating influence on the relationship between emotion and memory. Likewise, familiar and/or
enjoyed music can influence arousal, which can then influence memory. The familiarity and enjoyment
of music, including music that is disliked or not enjoyed may also be distracting, thus impairing memory
encoding. The control of individual differences is thus warranted when testing the complex relationship
between emotion, music, and memory.
2.8 Conclusion
Emotion-enhanced memory is a well-documented phenomenon that is amenable to
experimental manipulation. The empirical studies reviewed in this chapter have provided insight into
the mechanisms that underlie various aspects of memory, from perception of and attention to the
target material, to the consolidation of the material into long-term declarative memory. The principal
finding in both animal and human studies is that activation of the BLA is critical to EEM. The BLA is
53
___________________________________________________________________________________
sensitive to emotional stimuli and regulates responding by projecting to a number of brain structures
that support action. The BLA also projects to brain structures that regulate gene transcription and
synaptic growth required for permanent memory storage. Level of activation is an important property
of this mechanism. Under normal circumstances, the amygdala is strongly inhibited. As such, the
amygdala either requires highly activating stimuli to reach the threshold required for memory
consolidation, or reinforced activation from feedback loops from the central and peripheral nervous
systems. McGaugh and others have convincingly demonstrated that arousal hormones released from
the adrenal medulla activate brain stem structures that in turn have noradrenergic projections to the
BLA, and hormones released from the adrenal cortex cross the blood-brain barrier to activate the BLA
directly. Concomitant autonomic activation provides the organism with the physiological support to act
in response to the arousal stimulus. The detection of a positive relationship between ANS activity and
LTM would thus support the neurobiological model of EEM.
One of the limitations of research investigating arousal-modulated memory is the type of
stimulus used. Human studies typically include highly arousing, often negative images or stories to elicit
the required arousal response. As a consequence, there is a paucity of research testing the effects of
positive emotional arousal on memory. An argument against comparing positive and negative emotions
has been that arousal is the critical memory-modulating factor, therefore similar memory effects should
apply to both positive and negative arousing stimuli. Nevertheless, the use of negative emotional
stimuli to facilitate LTM may be limited in ecological settings as compliance rates and motivation to
continue treatment may be compromised. The research field could therefore be extended by
considering the lasting effect of both negative and positive emotions on memory.
A source of positive emotional arousal yet to be comprehensively exploited in this context is
music. The effect of music on subjective experience of emotion, brain activity, and physiological
responses consistent with emotion has been validated in a number of studies. Furthermore, there are
promising trends for music to effectively facilitate the consolidation of LTM, although at this point in the
absence of evidence of ANS activation. These findings indicate that music could be a valid means of
experimentally manipulating EEM. Furthermore, the flexibility of music delivery means that emotion
elicited by music could be applied in almost any context (e.g. via portable music players) and for any
MTBR (e.g. played before, during, or after encoding). Music applied in this way may therefore provide
the link between facilitating memory in the research laboratory to facilitating memory in everyday
settings, including clinical, educational, and occupational applications.
54
___________________________________________________________________________________
In sum, the source of facilitated memory in humans (i.e. orienting reflex rather than emotional
response; or cognitive processes, such as encoding efficiency and post-encoding elaboration rather than
adrenal neuromodulation) has yet to be unequivocally demonstrated. There is also a paucity of research
investigating the effects of arousal elicited by positive emotions on memory consolidation. The
controlled use of music as the emotion stimulus may allow for these gaps in the knowledge to be filled.
There are many properties of music that make it appealing as an emotion stimulus: it is widely used in
the community to regulate emotions; it can be physiologically activating or calming; it can elicit feelings
of both positive and negative emotions; and most importantly, it can be used as an extraneous source of
emotion applied before, during, or after the MTBR, thus lending flexibility to the type of information the
emotion treatment can be applied to. Music may therefore be a novel method of facilitating memory
for a diverse range of populations, in varying ecological contexts, and for a wide range of information.
55
3 GENERAL METHOD
Three controlled experiments were conducted to test the hypotheses raised in this thesis. An
overview of the methods used in each experiment and method details that were consistent across
experiments will be provided in this section. Elaboration of methods specific to each experiment will be
provided in the relevant empirical chapters.
3.1 Participants
One hundred and twenty seven individuals participated in this research, 68% of whom were
female. Participant age ranged from 18 to 50 years with a mean of 28.10 (SD = 10.05). The majority of
participants were university students or staff members recruited through university newsletters or
posters. Thirty eight percent were recruited through the university’s first year psychology research pool
in exchange for course credit. The greater ratio of females reflects the cohort typically encountered in
psychological research. Criteria for participation were: an age between 17 and 50 years; normal or
corrected to normal vision and hearing; and an absence of blood pressure or mood stabilising
medication.
Age was set at a maximum of fifty years to minimise differences in memory quality due to agerelated encoding strategies. A number of differences in memory associated with increasing age have
been observed, including positive bias (reviewed by Mather, 2004), faster decay (Charles, Mather, &
Carstensen, 2003), and enhanced general (as distinct from visual) emotional memory (Kensinger, GaroffEaton, & Schacter, 2007a). Given the high variability in emotion responding and memory performance
inherent in this research field, a maximum of 50 years was set for the current experiments to clearly
discount age related memory biases as an explanation for EEM.
Individuals taking blood pressure and/or mood stabilising medications were excluded from
participation as these substances could attenuate EEM. Blood pressure medication includes βadrenergic blocking agents. As β-adrenergic receptor activation and attenuation in the amygdala has
been demonstrated to modulate memory in animals and humans (see Section 2.1), participants taking
these medications may be immune to the experimental manipulation of EEM (e.g. see the seminal paper
by Cahill et al., 1994). Mood stabilising medications, such as a class of drug called Benzodiazepines,
have anxiolytic properties by binding to GABA receptors and depressing the central nervous system.
56
CHAPTER 3. GENERAL METHOD
___________________________________________________________________________________
The action of these drugs may therefore attenuate the intended experimental manipulation of
increasing CNS activity with emotionally arousing stimuli.
Participants were asked to refrain from taking caffeinated beverages before the experiments in
an attempt to minimise extraneous sources of arousal (caffeine is a CNS stimulant). Adherence to this
rule, however, was not strict given the adverse effect caffeine withdrawal can have on concentration
and mood. All testing occurred between the hours of 9am and 5pm. Participants of multiple session
experiments were tested at the same time of day for each session. It is possible that endogenous
arousal levels varied at different times of the day as cortisol, which is an arousal hormone that can
modulate memory, generally peaks in the morning (Touitou & Haus, 2000). Nevertheless, controlling
this variable by conducting all testing within a narrow timeframe (e.g. in the morning) would have been
overly restrictive. Likewise, the stage of the female menstrual was not controlled despite the possible
influence on endogenous arousal levels.
Potential participants were provided with a document containing an overview of the purpose
and possible benefits of the study, participant exclusion criteria, and what the study involved. Within the
document, participants were informed they could withdraw from the study at any time, that arrival at
the research laboratory implied informed consent to participate, and that their data could not be
identified (only aggregate data would be reported). All participants thus gave their informed consent
before commencement of the experiments. The experiments were approved by the university’s
Standing Committee on Ethics in Research Involving Humans.
3.2 Emotion elicitation
Emotion was manipulated with a negative story with or without emotionally powerful and
negative background music in Experiment 1, and emotional music alone in Experiments Two and Three.
To delineate the unique effect of music induced emotional arousal on memory, particular emphasis was
placed on the inclusion of valid auditory control conditions. The emotion manipulations and control
conditions for each experiment are presented in Table 3.1.
57
___________________________________________________________________________________
Table 3.1
Emotion elicitation across the three experiments
Emotion
Controls
Experiment 1
Experiment 2
Experiment 3
Participant-selected
positive
Experimenter-selected
positive
Non-music
negative
Negative narrative
Music
Low arousal
Random music mix
Non-music
Neutral narrative
Other participant’s
music self-rated as
neutral
Radio interview
Music
Radio interview
All music excluded lyrics to reduce the influence of language processing on encoding and recall.
All experimenter-selected excerpts were chosen based on studies reporting significant self-report and
physiological increases in arousal. The primary criterion for participant selections was that their music
be intensely enjoyed (emotion treatment) while at the same time being relatively neutral for other
participants (others’ neutral music control condition). The emotion manipulation occurred during
learning in Experiment 1, before learning in Experiment 2, and after learning in Experiment 3. The
emotional music was intended to elicit negative arousal in Experiment 1, and positive arousal in
Experiments 2 and 3. The only emotional stimulus that was not musical in nature was in Experiment 1.
As this experiment was a partial replication of the methodology employed by Cahill and colleagues, a
non-music emotional narrative was presented to participants. The emotion manipulation for each
experiment will be described in full in the empirical chapters.
Adequate music control conditions are imperative for making reliable conclusions about the
unique effects of music. For example, it has been claimed that music listening before performing
cognitive tasks facilitates performance (e.g. Rauscher, Shaw, & Ky, 1995). However, more rigorous
testing of the effect using appropriate non-music listening controls, such as listening to a narrated story
rather than sitting in silence, eliminates the music effect (Nantais & Schellenberg, 1999a). It would
therefore be equally valid to conclude that the experimental effect was due to reduced arousal in the
control condition (boredom, de-activation) compared to normal arousal in the treatment condition
58
___________________________________________________________________________________
(monitoring). To control for this possible confound, any arousal manipulation should include active
listening non-music controls rather than passive sitting in silence.
Control conditions in Experiment 1 included a neutral narrative matched to an emotional
narrative on word count, word frequency and word imagability, and arousing and negative background
music was compared to pleasant and non-arousing music. In Experiments Two and Three, the control
conditions included active listening to an Australian radio interview (similar to the non-music control
method employed by Schellenberg & Hallam, 2005). Experiment 3 included the additional control
condition of a random mix of the music used in the music arousal condition (similar to the music control
method employed by Menon & Levitin, 2005). All controls were thus active in nature, in that they
engaged attention and information processing. Differences between conditions could therefore be
more readily attributed to the emotion manipulation.
3.3 Material to be remembered (MTBR)
In all three experiments, the target MTBR was visual in nature. IAPS images were used in
Experiments One and Two, with arousal, valence, and the semantic categories of images controlled.
Images that were deemed overly attention capturing, such as erotica and mutilation, were not included.
The complexity of image memory allowed for the detection of cognitive processes, such as moodcongruent memory and valence bias (Eich & Forgas, 2003). The MTBR in Experiment 3 was simplified to
the same target and distractor word lists employed by Nielson and colleagues (Nielson & Bryant, 2005;
Nielson & Powless, 2007; Nielson et al., 2005), and Judde and Rickard (2010), as the influence of
cognitive processes on memory were of less relevance. Full details of the MTBR are provided in the
relevant empirical chapters.
3.4 Apparatus
Visual stimuli were presented to all participants on a high resolution 19” LCD color monitor.
Auditory stimuli were presented through Sennheiser closed headphones via a Kenwood midi hi-fi system
with volume adjusted to a comfortable level for each participant. A range of apparatus to support
software and hardware stimulus presentation and data collection methods were employed across the
59
___________________________________________________________________________________
three experiments, a summary of which is presented in Table 3.2. Paper and pencil data collection
methods will be described in the next section.
Table 3.2
Apparatus utilised across experiments to present stimuli and collect emotion and memory data
Experiment 1
Stimulus
presentation
Experiment 2
EMuJoy *
E-Prime
Microsoft PowerPoint
Microsoft PowerPoint
Experiment 3
E-Prime
ProShow Gold *
Data
collection
Self-report
Physiological
EMuJoy *
On-line survey *
Bioview
Reaction
time
Recognition
memory
Bioview
Nexus-10
E-Prime
E-Prime
E-Prime
E-Prime
* Described in Chapter 4
The timing of the procedures for Experiments 1 and 2 were controlled using similar Microsoft
PowerPoint slideshows to cue each step. Stimuli were presented with ProShow Gold in Experiment 1,
PowerPoint in Experiment 2, and E-Prime in Experiment 3 depending on the requirements of each
experiment. Physiological responses were collected with Bioview V2.11 Series IV (Zencor, 1998), Nexus10 and BioTrace+ (MindMedia, 2004), and reaction times and forced choice ‘old’ or ‘new’ recognition
memory data were collected with E-Prime 2® (Schneider, Eschman, & Zuccolotto, 2002). Bioview
contains four recording channels, each digitised at an output rate of 2 Hz. Heart pulse was detected by a
light-dependent resistor attached to the left ear lobe. A clip anchoring the HR lead to the participant’s
clothing reduced movement artefact. Activity of the left corrugator supercilii (frown) muscle was
detected by taping disposable 3M Red Dot Ag/AgCl electrodes above the middle and inner most point of
the left brow. The reference electrode was taped to the centre of the forehead near the hairline. Skin
conductance changes were detected by two electrodes attached by Velcro straps to the second and
60
___________________________________________________________________________________
third distal phalanges of the non-dominant hand. Skin temperature was detected by a sensor placed
under the Velcro strap on the second distal phalange.
Physiological data in Experiment 3 were collected with a Nexus-10 wireless data acquisition
system and BioTrace software. The Nexus-10 contains 10 recording channels, four of which acquired
electrophysiological signals through built-in DC amplifiers, referenced to a ground electrode. These
channels had a sample rate of 2048 Hz, thus allowing alpha and theta frequency recording. A further
four channels with a sample rate of 128 Hz suitable for recording ANS activity were used. Electrode
cables were carbon coated with active shielding to reduce movement artefact and mains interference.
The EEG data were acquired via two dual channel cables, each providing four standard snap-on
connectors for two bi-polar inputs. Cardiac activity was recorded from a photoplethysmograph sensor
clipped to the distal phalanx of the third finger of the non-dominant hand, and skin conductivity was
recorded from electrodes attached to the distal phalanx of the second and fourth fingers. Respiratory
rate was recorded from an expansion sensor placed over clothing and around the diaphragm. All
electrodes were Ag-AgCl.
E-Prime software was used to present the stimuli in Experiment 3, and record reaction time and
recognition memory data in Experiments Two and Three. E-Prime enabled the precise timing of
stimulus presentation in the auditory and visual modalities, the precise measurement of reaction times
to such stimuli, and for the recording of categorical decisions. The procedure for Experiment 3 was
simplified by using E-Prime to replace PowerPoint as the sole method of stimuli presentation.
3.5 Emotion Measures
As Scherer’s multiple component model of emotion was utilised as the theoretical framework of
emotion in this project, emotion measurement was structured in terms of the cognitive component,
which relates to the evaluation of objects and events; peripheral efference, relating to physiological
responses to support action; motivation, related to preparation and direction of action (approach or
withdraw); motor expression, related to communication of emotion; and subjective feeling, related to
monitoring of internal state. Cognitive, self-report and peripheral efference components were
measured in all three experiments, but not motor expression and motivation due to resource and time
limits. Details pertaining to the motor expression and motivational components of emotion will be
61
___________________________________________________________________________________
described in full in the relevant empirical chapters. Table 3.3 summarises how the emotion components
were measured in each experiment.
The measurement of more than one emotion component enabled the testing of synchronisation
across emotion components, which is rare in EEM research. Such testing would provide further
confirmation that the experimental stimuli elicited emotion enhanced memory and not orienting reflex
enhanced memory. Furthermore, the multi-component approach is novel in that it does not rely solely
on self-report to confirm the elicitation of an emotional response. One of the limitations of self-report
of emotion is that emotion can occur without conscious awareness and is therefore not available for
self-report (Davis & Thaut, 1989; Rickard, 2004).
Table 3.3
Emotion component measures utilised across the three experiments
Emotion component
Experiment 1
Experiment 2
Experiment 3
Subjective feeling
EMuJoy*
Arousal
Valence
Affect Grids
Affect Grids
IBI
SCL
SCR Max
TEMP
Chills
IBI
IBIsd
SCL
SCR Min
TEMP
%HF HRV
BVA
Chills
IBI
RSP
SCR Min
Motivational
EEG alpha asymmetry
Motor expression
Facial EMG
Cognitive
Memory narrowing
Mood-congruent
memory
Memory narrowing
Valence judgement
reaction time
EEG (alpha and theta)
Note. %HF HRV = percentage of high frequency heart rate variability; BVA = blood volume amplitude;
EMG = electromyography; IBI = inter-beat interval mean; IBIsd = inter-beat interval standard deviation;
RSP = respiratory rate; SCL = skin conductance level; SCR Max = maximum skin conductance response
amplitude; SCR Min = frequency of skin conductance responses in a minute; TEMP = skin temperature.
* Described in Chapter 4.
62
___________________________________________________________________________________
3.5.1 Subjective feeling
The two-dimensional emotion space (2DES) approach to emotion report (Russell, 1980) was
preferred over discrete or other models due to its sensitivity to the two factors of primary interest in
this dissertation; arousal and valence. The dimensional approach has been criticised for oversimplifying
a complex construct (Scherer, 2009), and participants may prefer using categorical emotion scales when
rating their responses to music (Zentner et al., 2008). Nevertheless, the overwhelming advantage of
scales representing 2DES is that they are quick and intuitive (Scherer, 2004; Schubert, 1999) and not
dependent on verbal capacity, thereby allowing for a greater number of reports without overly taxing
participants. Furthermore, high interrater agreement of music-induced emotion using 2DES was
revealed by Zentner, et al. (2008). Cronbach’s alphas were greater than 0.95 for emotions that fit within
the negative and aroused, positive and aroused, and pleasant and calm quadrants. High interrater
agreement demonstrates that 2DES is adequate for capturing these emotions. Inadequacy in the 2DES
approach was found for music-induced emotions that fit within the negative-deactivated quadrant
(Cronbach’s alpha less than 0.95). This is not surprising given that music generally elicits positive and
activated affective states (reviewed in Section 2.5.1). The use of 2DES to capture emotional responses
to music may therefore be considered reliable for most emotions except those that are negative and
deactivating, for example sadness. As manipulation of high arousal and/or positive emotions was the
primary focus in this project, the poorer reliability of the low arousal negative quadrant was deemed
inconsequential.
Two-dimensional emotion space was presented to participants in Experiment 1 as two separate
11 point scales representing mood, ranging from ‘negative’ (-5), to ‘positive’ (5); and arousal, ranging
from ‘calm’ (1), to ‘aroused’ (11). Grid format of the two dimensions was presented to participants in
Experiment 2 and 3 (Figure 3.1). Arousal was presented on the vertical axis and ranged from ‘calm’ (-5)
to aroused (5). Valence was presented on the horizontal axis and ranged from negative (-5) to positive
(5). Zero on both axes represented a neutral affective state (neither calm nor aroused, and neither
positive nor negative).
63
___________________________________________________________________________________
2DES GRID
5
4
3
AROUSAL
2
1
0
-1
-2
-3
-4
-5
-5
-4
-3
-2
-1
0
1
2
3
4
5
VALENCE
very aroused/energised and very negative (anger, fear, anxiety)
very aroused/energised and very positive (joy, happiness, excitement)
very deactivated/sleepy and very negative (melancholy, sadness)
very deactivated/sleepy and very positive (meditative, contented)
Figure 3.1 Affect grid used in Experiments Two and Three. The cross indicates a rating of moderately
positive mood valence (2) and moderate arousal (3). The location of the cross within 2DES indicates the
emotion felt would be one of joy or happiness. The face icons act as visual cues for the type of emotion
represented in each of the four quadrants. The icon descriptions were presented at the bottom of the
page for each affect grid.
The reporting of emotional responses to music can vary depending on whether the instructions
are to report the actual (or felt) emotions induced by music, or to evaluate the emotion intended by the
music creators/performers (often referred to as expressed emotion, Schubert, 2007). There is therefore
some concern that participants who are not aware of this distinction may report the emotion expressed
in the music, rather than the emotion they feel. To address this potential confound, participants were
explicitly informed of the distinction, and instructed to report how the music made them feel.
64
___________________________________________________________________________________
3.5.2 Peripheral efference component
Sweating is primarily a thermoregulatory function innervated by the SNS (Low, 2004) that is
detected by electrophysiological measurement of skin conductivity. Skin conductivity was measured in
all three experiments. Tonic skin conductance levels (SCL) were derived from the mean, in micro
Siemens (μS), of the total listening duration in Experiment 1 and 2. Mean skin conductance levels were
not recorded in Experiment 3 due to inherent drift typically observed in skin conductance levels
confounding interpretation of differences between conditions with differing durations. Skin
conductance responses (SCR) provide an index of attention to emotionally salient or novel stimuli
(Bradley, 2009; Dawson, Schell, & Filion, 2007). There were two skin conductance response (SCR)
measures. In Experiment 1, SCR Max reflected the maximum amplitude of the SCR within the
experimental manipulation (all conditions of the same duration). To account for the differences in
duration of the music excerpts in Experiments 2 and 3, the number of SCRs per minute was calculated.
Values were derived from peaks greater than .05 μS for successive 10 second epochs (Duncko, Cornwell,
Cui, Merikangas, & Grillon, 2007). Skin conductance was recorded from sensors attached to the distal
phalanx of the second and fourth fingers of the non-dominant hand.
Grewe, Kopiez, and Altenmuller (2009) have demonstrated that the subjective experience of
chills (or thrills, goose pumps, shivers, etc.) are correlated with increased skin conductance levels. Thus,
the number of chills experienced during music (and control) listening conditions was captured
immediately after presentation in Experiment 2 and 3 by self-report. Chills were not recorded in
Experiment 1, in which the music was played during presentation of the MTBR, to minimise encoding
interference.
Increased sympathetic control of cardiac function is reflected by increased heart rhythm and
vasoconstriction (Hamill & Shapiro, 2004), decreased blood volume amplitude (BVA; MindMedia, 2004),
and decreased parasympathetic control of heart rhythm (Berntson, Quigley, & Lozano, 2007).
Decreased parasympathetic control of heart rhythm can be inferred from decreases in the proportion of
high frequency components of heart rate variability (%HF HRV; Berntson et al., 2007). Heart rhythm,
indexed as the interval, in milliseconds (ms) between normal-to-normal beats (IBI), was monitored in all
three experiments. Limitations to the Experiment 1 and 2 electrophysiological recording equipment
precluded measurement of BVA and %HF HRV. The standard deviation of IBI was therefore used as a
65
___________________________________________________________________________________
rudimentary measure of heart rhythm variability. The unit of measure for BVA was the root mean
square (RMS) of peak-to-peak micro Volts (µV pk-pk).
In Experiment 3, IBI and BVA were recorded from a photoplethysmograph sensor attached to
the distal phalanx of the third finger of the non-dominant hand at a sample rate of 128 Hz. Manual
checks for ectopic beats were conducted and the corresponding data were removed. Automatic IBI
artefact rejection criteria were set at intervals between beats greater than 1500 ms (less than 40 beats
per minute) and less than 500 ms (greater than 120 beats per minute), or beats greater than 25% of the
preceding interval. All removed beats were replaced with interpolated values using the formula:
IBI [n] = (IBI [n-1] + IBI) / 2
The heart rate variability (%HF HRV) measure was derived from the artifact corrected IBI
recordings and was calculated with the BioTrace+ software as the high frequency proportion of very low
frequency power (0.0033 to 0.04 Hz) plus low frequency power (0.04 to 0.15 Hz) plus high frequency
power (0.15 to 0.4 Hz).
Decreased skin temperature reflects sympathetically mediated vasoconstriction in the distal
extremities (Kihara, Sugenoya, & Low, 2004). Vasoconstriction in the extremities, e.g. hands and feet,
redirects blood to deep veins and vital organs. Additionally, blood flow in the extremities is a major
source of heat loss, thus vasoconstriction preserves core temperature. Skin temperature in Experiment
1 and 2 was measured in degrees Celsius (°C) from the distal phalanx of the second finger of the nondominant hand. Due to equipment malfunction, skin temperature was not recorded in Experiment 3.
RSP in Experiment 3 was recorded at a sample rate of 32 Hz from a sensor placed over clothing and
around the diaphragm.
3.5.3 Cognitive component
Memory narrowing, the time taken to judge whether objects are positive or negative, and the
degree of cortical alpha and theta desynchronisation can index cognitive processing of emotional
stimuli. Increased memory for specific types of information, such as information central to an emotional
event, reflects appraisal processes. Decreases in the time taken to make judgements, and
66
___________________________________________________________________________________
desynchronisation of alpha and theta brain oscillations, reflects increased cortical activation and
concomitant increases in information processing efficiency.
Judgement speed was measured in Experiment 2 with a modification of Bouhuys, Bloem, and
Groothuis’s (1995) face valence judgement reaction time task. Participants were instructed to respond
as quickly and accurately as possible to line drawing of positive and negative faces (illustrated in Figure
3.2). Activation of the cognitive component of emotion was implied by decreases in valence judgement
response times.
Figure 3.2 Positive and negative line drawings of faces used for the face valence judgement task.
Images were modified versions of those created by Bouhuys et al. (1995). The faces differed in terms of
mouth shape (turned up or down) and eyebrow position (high or low), such that in the above, figures
represent two variations of positive faces, and two variations of negative faces.
Cortical activation was determined by desynchronisation of oscillations in the alpha and theta
frequency bands. EEG was employed in Experiment 3 to record alpha and theta rhythms from three
scalp locations; the left and right frontal poles and a central parietal site. Electrodes were placed
according to the international 10/20 system in the corresponding scalp locations; Fp1, Fp2, and Pz
(Jasper, 1958). Placement of electrodes at these sites allowed for determination of frontal alpha
asymmetry (the difference between Fp1 and Fp2), frontal relative to posterior theta desynchronisation
(the mean of Fp1 and Fp2 relative to Pz), and generalised cortical activation (alpha and theta
desynchronisation at all sites).
67
___________________________________________________________________________________
Eye movement was detected with electrooculargram (EOG), collected from the external canthus
of the left eye, for off-line detection and removal of lateral eye movement artefact (vertical eye
movement was not monitored as eyes were closed for all periods of interest). Electrode offset was
monitored continuously throughout data acquisition to ensure it remained within an acceptable input
range. Artefact free data, band pass filtered at 1-64 Hz, were then extracted using a Hamming window
of 2 s width with 75% overlap for fast Fourier Transformation (FFT). Frequency power (µV pk-pk2) in the
theta (4.0-8.0 Hz) and alpha (8.0-12.0 Hz) range was derived from the FFT. To normalize the distribution
of the data, all power values were natural log (ln) transformed.
3.6 Memory measures
The approach to memory measurement varied across the three experiments (see Table 3.4). An
overview of the memory measures will be described in this section, with full details provided in the
relevant empirical chapter.
Table 3.4
Memory measures and testing delay across the three experiments
Memory measures
Experiment 1
Experiment 2
Experiment 3
Free recall
1 week delayed free and
cued recall of 9 images
5-60 minute delayed
written recall of 120
images
Immediate and 1 week
delayed written recall
of 30 words
Recognition
1 week delayed 72 item 4
alternative forced choice
written questions
30 minute delayed 240
item old/new forced
choice
1 week delayed 120
item old/new forced
choice
As Experiment 1 was a partial replication of Cahill and colleague’s studies (Cahill & McGaugh,
1995; Cahill et al., 1994), a similar approach to the measurement of memory was taken. To reduce the
likelihood of rehearsal, participants were not informed that the study was investigating memory.
Instead, they were told that the study aim was to determine the effect of music on physiology. To
ensure memory consolidation was being tested, a delay of one week elapsed before memory testing
occurred. At this time free recall was tested first, then recognition memory in the form of a four
68
___________________________________________________________________________________
alternative forced choice memory test. The MTBR and the memory test were purposefully designed to
differentiate visual from thematic details, and central from peripheral details. Long-term memory could
therefore be compared between conditions that varied in the emotion manipulation, and within each
condition to determine whether the quality of memory varied with emotional valence.
Free recall and recognition memory was tested in Experiment 2 after an average delay of 30
minutes. The purpose of memory testing was to determine whether emotional music facilitated the
early stages of LTM, and whether emotional valence influenced the quality of information encoded. In
contrast to Experiment 1, participants were informed that the purpose of the study was to investigate
memory. Knowledge of the memory test thus ensured participants fully attended to the MTBR.
Knowledge of the experimental aim was not deemed a concern due to the absence of any substantial
delay between the task and the memory test. Furthermore, the repeated measures nature of the task
and the inter-condition distracter tasks (described in Chapter 5) meant that participants had little
opportunity for rehearsal. Memory testing involved free recall of the MTBR, followed by a forced
choice, ‘old’ or ‘new’ recognition memory test.
Free recall memory for material presented in Experiment 3 was tested immediately and then
again after a delay of one week. Participants were informed on their first visit that their memory would
be tested. To prevent rehearsal, they were not informed of the delayed memory test. An index of
retention that controlled for individual differences in memory strategies and ability was obtained by
calculating the percentage of words retained between immediate and delayed testing. Recognition
memory was tested after a one week delay with a forced choice, ‘old’ or ‘new’ recognition memory test.
Free recall and recognition memory testing are sensitive to different aspects of memory. Free
recall is more demanding than recognition testing as the entire memory is reconstructed in the absence
of cues. The scope of memory freely recalled is therefore not confined to the limits of the cues provided
in the test. Information that was salient to the individual when the event was encoded would therefore
be detectable with free recall. Such influences could include mood-congruent learning or use of
schemata to interpret ambiguous situations (reviewed in Section 2.2). One of the limitations of free
recall testing, however, is the increased possibility of floor effects. As incidental memory is tested in
most cases, the event to be remembered may not be purposefully encoded, leading to the loss of many
details. The nature of recognition memory testing is to provide a cue and then ask the respondent to
decide whether the cue represented a previously presented item or not. The cue may thus activate a
69
___________________________________________________________________________________
memory trace previously unavailable for recall, making recognition memory a more sensitive measure of
weakly encoded information.
Memory quality, which can be influenced by cognitive processes associated with mood valence
and currently active schemas or goals, was measured in Experiments 1 and 2. As the emotion
manipulation occurred after the MTBR in Experiment 3, the influence of cognitive processes on memory
encoding were not tested. The operationalisation of memory quality differed between Experiment 1
and 2. In Experiment 1, the memory test was designed to detect memory narrowing in response to a
negative emotional stimulus (described in more detail in Chapter 4). In Experiment 2, the MTBR was
designed to detect broad memory scope in response to positive stimuli (described in more detail in
Chapter 5).
3.7 Measures of individual differences
A selection of measures of individual difference was included across experiments. Arousal
sensitivity was measured in Experiment 1 with the Extraversion scale of the International Personality
Item Pool (IPIP) Big 5 Personality Questionnaire (Goldberg et al., 2006), and with the BIS/BAS scales in
Experiments 2 and 3. Gender was recorded from a demographic survey in each experiment. Music
experience was measured in Experiment 1 with the Brief Music Experience Questionnaire (BMEQ;
Werner, Swope, & Heide, 2006), and in Experiment 2 and 3 with music performance and nonperformance based questions developed by Chin and Rickard (2010). Music distraction, enjoyment, and
familiarity, was recorded from Likert type scales across experiments. Full details of all measures of
individual differences are provided in Appendix L.
70
4 EXPERIMENT ONE: Emotional story and emotional
music effects on memory
4.1 Introduction
Successful manipulation of LTM with music as a source of extrinsic emotional arousal would
improve the generalisability of EEM. The purpose of the first experiment was thus to test whether
emotionally arousing background music could facilitate memory for unrelated and non-arousing
visual information. For continuity, the study methodology was similar to an established
methodology used in human EEM researchers (Burke et al., 1992; Cahill et al., 1994; Christianson,
1984; Heuer & Reisberg, 1990; described in Section 2.1.1). The aim was to replicate the EEM effect
with a refined methodology that: measured multiple components of emotion and their relationship
with memory; excluded visually aversive, thus attention capturing images; added emotionally
powerful background music as an extrinsic source of emotional arousal; measured cognitive
processing influences on memory; and controlled linguistic anomalies that may influence memory.
Multiple components of emotion (Scherer, 2001; reviewed in Section 2.4) were measured to
confirm that an emotion had indeed been elicited as intended. Facilitated LTM that was associated
with synchronised changes in multiple emotion components would thus support the emotional
arousal hypothesis of EEM. Of particular interest were autonomic changes suggesting noradrenergic
modulation of memory. For instance, an increase in cardiac activity indicates that the noradrenergic
system has been activated, and may also index acetylcholine innervation of the adrenal medulla and
secretion of NE and EPI into the blood stream. Localised circulating NE innervates the vagus nerve,
with subsequent activation of brain stem structures involved in noradrenergic modulation of BLA
dependent memory (McGaugh, 2000; reviewed in Section 2.1). Detection of a relationship between
cardiac activity and enhanced LTM would thus provide indirect support for neurobiological
modulation of memory in humans.
Visually aversive images were removed from the methodology to reduce or eliminate an
amygdala mediated orienting reflex explanation for enhanced memory (see Section 2.1). The
distinction between orienting and emotion is important when considering the generalisation of this
research, as orienting reflexes are easier to elicit than emotions. For instance, emotions, in the
adaptive sense, are more complex and thus more difficult to elicit reliably than an orienting reflex.
An emotion response can depend on factors such as the goals/needs of the individual and the
context in which the emotional stimulus is encountered, which are more difficult to control. In
71
CHAPTER 4. EXPERIMENT ONE
________________________________________________________________________________
contrast, an orienting reflex can be reliably elicited by a stimulus as simple as a sudden and loud
noise. Memory enhancement programs could then simply intersperse the MTBR with novel stimuli
to enhance memory. It is therefore important to test whether EEM is eliminated with the removal
of novel stimuli, such as traumatic images.
As visually aversive information was removed from the stimuli, emotion was elicited mostly
thematically with a negative narrative (e. g., Laney et al., 2004; reviewed in Section 2.2). Memory
narrowing was measured to determine whether emotional appraisal processes were active during
the encoding of the MTBR. However, defining what information is central and peripheral for the
purpose of detecting memory narrowing has been made difficult by ambiguity in the research field.
For instance, Bower (1992) described central information as the central idea; Adolphs et al. (2001);
Buchanan, Karafin, and Adolphs (2003); Burke et al. (1992); and Laney et al. (2004), described it as
gist; Christianson, Loftus, Hoffman, and Loftus (1991) described it as the topic of relevance; and
Kensinger et al. (2007b) as the central visual item (see Appendix A for a more complete list of
information category definitions). Memory test questions were thus created based on predetermined categorisation of central and peripheral image details by independent judges.
The addition of emotional background music to the methodology enabled testing of extrinsic
emotion effects on LTM. Music, unlike alternative sources of extrinsic arousal (e.g., exercise, glucose
administration, arousal hormone administration), is an emotion stimulus that can be accessed by a
wide range of populations. Music can thus be effortlessly applied in a number of learning
environments, increasing the generalizability of the EEM phenomenon (see section 2.5).
To determine whether non-neurobiological processes explained EEM, the methodology was
adapted to control the emotional valence of the MTBR. In this way, the influence on memory of
active cognitive processes, such as lowered firing thresholds in networks associated with current
mood (e.g. mood-congruent memory, reviewed in Section 2.2) could be measured. Further
modifications were made to the verbal narratives of the slideshows to control linguistic anomalies
that may influence memory, (e.g., story cohesiveness, word distinctiveness, and word count).
The overall aims in Experiment 1 were to replicate previous research by facilitating memory
with non-music emotional stimuli, and to extend previous research by facilitating memory with
emotional background music.
72
________________________________________________________________________________
4.2.1 Non-music aims and hypotheses
Aim 1 was to determine whether the newly developed materials replicated the intrinsic
emotion-memory effects reported by previous researchers. It was hypothesised that (1a) emotion
elicited by a negative narrative in Phase 2 of a three-phase slideshow would facilitate one week
delayed free recall and recognition of Phase 2 slideshow images relative to Phase 1 and Phase 3, and
(1b) that the increase in Phase 2 memory would account for greater total memory scores for the
negative slideshow relative to the neutral slideshow. Aim 2 was to determine whether any observed
EEM elicited by the narrative could be accounted for by emotional arousal or by mood-congruence
effects on encoding. It was hypothesised that (2a) for emotional arousal to be the underlying
mechanism, any increase in memory caused by the Phase 2 negative narrative would be associated
with reports of increased subjective arousal, increased skin conductivity, increased skin conductance
responses, decreased skin temperature, increased heart rate, increased frown muscle activity, and
increased memory narrowing. For mood-congruence to be responsible for facilitated memory, it was
hypothesised that (2b) greater memory for the negative slideshow would be attributed to more
negative image details retained relative to the neutral slideshow.
4.2.2 Music aims and hypotheses
Aim 3 was to determine whether extrinsic arousal elicited by emotionally powerful
background music facilitated memory for a neutral slideshow. It was hypothesised that (3a) emotion
elicited by arousing music played continuously in the background of a neutral story would facilitate
one-week delayed free recall and recognition of all slideshow images relative to the neutral music
and no music conditions. Aim 4 was to determine whether any observed EEM elicited by music was
best explained by emotion or mood-congruence effects on encoding. It was hypothesised that (4a)
for emotion to be the underlying mechanism, any increase in memory caused by the emotional
music would be associated with reports of increased subjective arousal, increased skin conductivity,
increased skin conductance responses, decreased skin temperature, increased heart rate, increased
frown muscle activity, and increased memory narrowing. For mood-congruence to be responsible
for facilitated memory, it was hypothesised that (4b) greater memory for the slideshow with
emotional background music would be attributed to more negative image details retained relative to
the no-music or neutral music conditions. A summary of aims and hypotheses is presented in
Appendix B.
73
________________________________________________________________________________
4.3 Method
4.3.1 Participants
Forty seven females and 22 males (N = 69) with a mean age of 25.01 (SD = 8.40, range 17 to
48 years) participated. Participants were largely first year psychology students who received course
credit for their participation. University staff and students and individuals recruited from the
general population made up the remainder of the pool. Each participant was randomly allocated to
one of the four experimental conditions in a between-subjects experimental design.
4.3.2 Emotion manipulation
Stimuli were four, three minute slideshows consisting of 12 IAPS images (Lang, Bradley, &
Cuthbert, 2005), the details of which were the target MTBR (a copy of the slideshows can be found
on the data disc in Appendix C). Images were presented for 15 seconds each and accompanied by a
three phase narrative. The narrative in Phase 1 and Phase 3 was identical in the neutral and
emotional versions of the slideshows. The emotion manipulation occurred in Phase 2, in which a
neutral or negative version of the story was presented. The negative narrative presented in Phase 2
thus formed the thematically elicited emotion manipulation that was intrinsic to the MTBR. Extrinsic
emotion was manipulated with negative and arousing background music and a neutral music
control. To control emotional narrative effects on memory, only the neutral narrative slideshow
condition was presented with background music. Even though the narrative of the music conditions
was neutral across the three story phases, three phase analyses were conducted to be consistent
with the non-music intrinsic (3 phase) emotion comparisons. The neutral music control condition
was included to account for memory differences caused by background sound. Image sequence was
the same across the four slideshows (illustrated in Table 4.1). Story phases were separated by an
abstract image of neutral valence and arousal presented for 15 seconds, thereby allowing clear
demarcation between each phase for physiological and memory comparisons.
74
________________________________________________________________________________
Table 4.1
IAPS Images by Phase with Accompanying Narrative and IAPS Details
Phase
Image
Narrative
No.
One
Two
Buffer Slide
7161
2.98
4.98
A man used to work at a busy port as a shipping
container administrator.
7036
3.32
4.88
He often went to a restaurant for lunch before
it burnt down.
9471
4.48
3.16
His wife is the primary carer of their two school
aged children.
2598
3.73
7.19
Buffer Slide
7179
2.88
5.06
2722
3.52
3.47
5220
3.91
7.01
7242
3.83
5.28
7160
3.07
5.02
Some time later a woman finds the daughter’s
jumper near her country house.
5779
3.57
7.33
The jumper is labeled with the family details so
the woman contacts the mother.
2516
3.50
4.90
After a long delay they meet while the mother
is visiting the cemetery.
9001
3.47
3.10
He takes his children to visit his brother who is
in jail for fine evasion.
He is in jail for beating his wife and plots his
revenge on her.
On the way they go walking in a park and the
father drops the daughter’s jumper.
When he gets out the father collects his
children and takes them into dense bush.
When they return home the mother wants to
wash the jumper but can’t find it.
When they don’t return home the mother
becomes frantic with concern for her children.
Buffer Slide
Three
IAPS details
Arousal Valence
Note. Each slide was presented for 15 seconds, each phase for one minute. Emotional narrative is in
italics. IAPS normative ratings range from 1 to 9 (Lang et al., 2005).
75
________________________________________________________________________________
The slideshow images were selected based on psychometrically validated ratings of neutral
arousal (4 on a scale ranging from 1 – 9), thereby reducing potential orienting reflexes to arousing
images. Each phase of the slideshows contained three images that were negative, positive, or
neutral and that depicted people, buildings, or scenes; the order of which was counterbalanced.
Image valence varied across phases to allow testing of emotion-congruent memory bias. The
semantic content of the images varied to maintain interest. Counterbalancing across each phase
ensured memory for image valence or semantic content was not confounded by primacy or recency
effects.
The narratives were created to match the presentation order of the selected images and
were recorded by a male professional actor who spoke in a clear and neutral voice. Slideshows of
the images combined with the narratives were then created, with each sentence timed to
commence two seconds after image onset. To control for memorability differences due to
differences in linguistic properties between the two narratives, care was taken to ensure that each
sentence was matched on word count, word frequency, imaginability, and concreteness (full details
described in Appendix D). Pilot testing of the narratives with two groups of undergraduate students
(neutral narrative, n = 22; negative narrative, n = 23) revealed that the negative narrative elicited
agreement with the statement that the story was sad and disagreement with the statement that the
story was happy. These ratings were relative to the neutral story, in which there was general
disagreement with various emotion statements. Students in both conditions disagreed with the
statement that the story was complex. There was, however, a small (non-significant) difference in
how understandable the stories were, with agreement that the neutral story was easily understood,
but neither agreement nor disagreement that the negative story was easily understood. The
emotional and neutral stories had similar ratings of emotional intensity, indicating that even though
the emotional story was perceived as emotional, it may have elicited relatively mild emotional
responses (see Appendix E for a detailed description of the narrative pilot test results).
Extrinsic emotional arousal was manipulated with an excerpt from Holst’s Mars, the Bringer
of War played in the background throughout the slideshow. Previous studies indicate this music
elicits subjective and physiological reactions consistent with negative and aroused emotion
(Baumgartner et al., 2006a; Krumhansl, 1997). An orchestral recording of Gymnopedies No. 1,
composed by Erik Satie, was used as a neutral background music control. The orchestral recording
of Gymnopedies was chosen over a single instrument recitation (piano or guitar) to minimize
differences in responses across conditions due to sound complexity. Background music was phased
in and out at the beginning and end of the slideshows and played at a volume that allowed the
narrative to be clearly heard. Volume was set to a comfortable level by each participant.
76
________________________________________________________________________________
4.3.3 Apparatus
A Microsoft PowerPoint presentation was used to display instructions and practice trials to
participants. The experimental procedure was presented via a purpose-built website. The slideshow
stimuli were created using ProShow Gold (PhotodexCorporation, 2004) allowing for the 15s
presentation of each slide accompanied by the neutral or emotional narrative, and the neutral or
emotional music. Slideshow files were saved to MPEG-1 format and embedded in the on-line
procedure. Physiological reactions were measured with the Bioview physiological data acquisition
system (described in full in Chapter 3). Changes in subjective emotion during slideshow presentation
were continuously measured with EMuJoy software (Nagel, Kopiez, Grewe, & Altenmuller, 2007;
http://musicweb.hmt-hannover.de/emujoy).
EMuJoy was developed to measure ongoing changes in subjective emotional response to
visual and/or auditory stimuli. Due to the dynamic nature of music and listener response as they
unfold over time, continuous as compared to static, post-listening measurement of emotional
response to music, was considered more ecologically valid (Schubert, 1999). EMuJoy was
programmed to initiate the onset and timing of visual and auditory stimuli. Emotional responses
were recorded by tracking the location of cursor movements within the two dimensions of valence
(horizontal axis, negative to positive) and arousal (vertical axis, calming to arousing). The two
dimensions were presented as light grey vertical and horizontal lines that overlay images presented
on a computer monitor. The emotions represented by each quadrant were described to participants
and they were instructed to move a cursor around the screen to reflect their emotions as they occur.
Visual feedback of emotion ratings was provided to participants by a worm shaped cursor, the face
of which changed to reflect the emotion represented by its location in the two dimensional emotion
space. Acquired data were time and X:Y coordinates at a sampling rate of 2 Hz, transmitted in real
time via the intranet to a remote computer. A visual representation of the cursor trace for one
participant is presented in Figure 4.1.
77
________________________________________________________________________________
A
B
1
1
0.5
0.5
0
-1
-0.5
0
0
0.5
1
-1
-0.5
0
0.5
1
-0.5
-0.5
-1
-1
Figure 4.1 Visual representation of continuous self-report of emotion for a participant viewing the
(A) negative narrative slideshow, and (B) neutral slideshow. Horizontal axis represents valence,
negative (-1) to positive (+1). Vertical axis represents arousal, calm (-1) to aroused (1).
The visual component of all tasks and stimuli were presented on a computer monitor and
the auditory component of the slideshows was presented through closed headphones. Participants
were seated approximately 1.5 m from the computer monitor (height and distance adjusted
according to preference) in a firm armchair. Three computers were used; the first for presentation
of digital media to participants, the second to collect physiological data, and the third to operate the
EMuJoy software. The second and third computers were located in a room adjacent to the testing
room to minimise disturbance to the participant during testing. EMuJoy data were collected from
the positioning of a wireless optical mouse and were transmitted to the experimenter computer via
the university intranet.
4.3.4 Measures
Memory
After a one-week delay, memory was measured by first asking participants to freely recall as
many visual details of the slideshow as they could. Slide details were recorded with a voice recorder
and later transcribed. One score was awarded for the description of a unique recognizable element
within an image that could not be attributable to the other images. An element was defined as the
description of an object, a color, a relevant association, or a visual perspective. Due to low levels of
ambiguous ratings by judges of element classification in previous studies (Cahill & McGaugh, 1995;
78
________________________________________________________________________________
van Stegeren et al., 2002), element classification was determined by one judge (SC) blind to the
experimental conditions. Recall of elements relating to the narratives was not scored.
A 72 item four-alternative forced-choice (4AFC) memory test contained eight questions per
image, four questions relating to central details and four questions relating to peripheral details
(random presentation order). To eliminate differences in recall due to differences in the narratives,
questions related to visual details of the images only. One question from slide one was answered
correctly by all participants. Otherwise, there was no evidence of ceiling or floor effects. Central
details were defined as information that immediately stood out or attracted attention. For example,
the second image in the story was of a burnt building and a central detail question was: ‘The building
had several windows that … a) were being painted; b) displayed items for sale; c) were being
dismantled; or d) had broken glass’. Peripheral details were defined as information that would only
become obvious after detailed visual scrutiny. For example, for the same image, a peripheral detail
question was: ‘The sheet of corrugated iron in the image was located … a) on the ground to the right
of the image; b) on the ground to the left of the image; c) hanging from the top of the building; or d)
hanging from one of the windows’. Categorization of the central peripheral distinction was
confirmed by six independent judges. There was an average consensus of 64% for the central item
distinction, and 68% for the peripheral item distinction. Memory scores were converted to percent
correct (4AFC % correct).
Emotional response
Subjective mood and arousal was recorded before and after the slideshow on two, eleven
point sliding scales; valence, ranging from negative (-5), to positive (5); and arousal, calm (-5), to
aroused (5). Scales were centered on neither negative nor positive, and arousing nor calming (0).
Continuous subjective ratings of emotional responses were recorded using the two axes presented
in EMuJoy; negative (-1) to positive (1), and arousing (1) to calming (-1). The central point on both
scales represented neither negative nor positive and neither arousing nor calming (0).
Psychophysiological measures were skin conductance level (SCL), maximum amplitude of
skin conductance response in each slide (SCR Max), skin temperature (TEMP), corrugator supercilii
electromyographic response (EMG), and inter-beat interval (IBI). IBI is the inverse of heart rate,
with high IBIs indicating low heart rate (beats per minute). See the Chapter 3 for a full description of
the Bioview software and measurement parameters. Mean activity for each 15 second slide was
calculated and then subtracted from the buffer slide introduced at the beginning of the slideshow,
yielding a moving change-by-slide index for each measure.
79
________________________________________________________________________________
Cognitive processing of emotional information was indexed by memory narrowing, a process
by which salient emotional information is attended to and retained more effectively than peripheral
information. An index of memory narrowing was created by subtracting the proportion of
peripheral 4AFC items correct from the proportion of central 4AFC items correct6. Positive values
therefore reflected memory for more central than peripheral image details and negative values
reflected memory for more peripheral than central image details. Free recall elements were not
categorised due to the time required to classify the responses and to ensure the classifications were
unambiguous and consistently agreed upon.
To determine whether music distraction influenced memory, music distraction, familiarity
and enjoyment was measured (enjoyed and familiar music can also be distracting). Music distraction
was rated on a five point scale ranging from none of the time (1) to 100% of the time (100) in 25 unit
intervals. Music enjoyment and familiarity was rated on two separate five point scales ranging from
disliked it very much, not familiar at all (1) to liked it very much, very familiar (5), respectively (the
scales also included a there was no music (0) option). Individual difference measures of
musicianship, music commitment, and extraversion are described in Appendix L.
4.3.5 Procedure
Participants were tested individually in two sessions held one week apart. Testing occurred
in a sound attenuated room with low lighting and participants were seated alone during
presentation of the experimental conditions. In the first session, participants were informed that
the purpose of the experiment was to investigate the emotional and physiological effects of
slideshows that may be unpleasant or basically neutral. Participants were then connected to the
physiological recording equipment and completed a training module on how to rate their emotional
response on EMuJoy. Instructions were given on how to initiate the on-line experimental procedure,
in which the slideshow would be presented. Participants were asked to attend to the slideshow as
they would a television show or film, and to signal the experimenter when the slideshow had
finished. The experimenter then left the room.
The experiment commenced when participants initiated the automated on-line preslideshow questionnaire by mouse clicking on the on-screen ‘start’ button. The on-line procedure
collected information on gender, mood, and arousal, followed by a two minute settling period.
During this time, participants read an on-line excerpt from Oliver Sacks’ The Power of Music (Sacks,
6
4AFC recognition memory test items were created based on independent judgements of what constituted
central and peripheral image details.
80
________________________________________________________________________________
2006). The slideshow was then presented. At the end of the slideshow, participants signaled the
experimenter, who re-entered the room to initiate the automated on-line post-slideshow
questionnaire. The post-slideshow questionnaire recorded mood and arousal ratings, music
distraction, familiarity, and enjoyment, and responses to the Brief Music Experience Questionnaire
(BMEQ) and Big 5 personality questionnaire (refer to Appendix L for supplementary analyses of
music experience and personality effects on memory). Once completed, the physiological recording
equipment was disconnected. Confirmation of the appointment for the second session, in which
participants were told would be similar to the one just completed, was confirmed before
participants left.
In the second session, participants were informed that the true aim of the study was to test
whether emotional stories and music had an effect on LTM. They were then asked to recall the
slideshow presented one week previously. All participants were naïve to this aim and agreed to the
memory tests. First, free recall of the slideshow was tested. Participants were encouraged to
remember as many details of the slides as possible. They were then given a cue for each of the nine
slides (e.g., slide one was a ship docked at a wharf) and open ended questions relating to the slide
were asked (e.g., what angle was the photo taken from?). After recall was exhausted, recognition
memory was tested using the 72 item 4AFC memory test. Participants were given 30 minutes to
complete the test and were asked to answer all questions, even if guessing. At the completion of
memory testing participants were debriefed.
4.4 Results
4.4.1 Continuous measurement of subjective emotion
During the course of data collection, it became evident that the task of attending to
slideshow content while at the same time continuously monitoring and reporting feelings of emotion
was difficult for participants to perform. A preliminary memory data check of the conditions without
music7 (neutral narrative n = 5, emotional narrative n = 7) revealed that Phase 2 of the emotional
slideshow was impairing memory relative to Phase 2 of the neutral slideshow (illustrated in Figure
4.2). The observed pattern of memory performance was the inverse of those revealed in the
narrative pilot test (see Appendix E). Ad-hoc interviews with participants suggested that the
competing task demands of continuously monitoring and recording emotional responses whilst also
attending to the content of the slideshow placed too much demand on cognitive resources. The
7
There were insufficient data in the remaining conditions for interpretation.
81
________________________________________________________________________________
decision was therefore made to discontinue continuous measurement of emotion and the
procedure was modified accordingly.
Recognition
4AFC Items Correct
16
Neutral Narrative
Negative Narrative
14
12
10
8
6
1
2
3
Phase
Figure 4.2 Mean number of correct 4AFC recognition memory test items for each phase of the
slideshow. Error bars represent standard error.
The detrimental effect of the continuous emotion reporting task on LTM was subsequently
confirmed by an independent samples t-test using data collected from a further 18 participants (no
EMuJoy reporting). Participants who continuously reported their emotional responses to the
slideshow recalled less image details (M = 34.16, SD = 6.04) than those who did not use EMuJoy (M =
39.55, SD = 3.50), the difference of which was significant, t (28) = -3.10, p < .01, d = 1.13. Due to the
change in task and the subsequent effect on the main variable of interest, LTM, all further analyses
were performed only on participants who did not continuously report emotion. Resulting sample
size was N = 36 (nine per condition).
4.4.2 Non-music manipulation
Intrinsic emotion effect on memory
Inspection of the number of slides freely recalled by each participant (presented in Figure
4.3) indicated that the image of the father in a jail cell (Slide 4, Phase 2) was more memorable for
the emotional narrative group than for the neutral narrative group. However, analysis of the mean
number of slides recalled in each story phase (within-subjects) for each narrative (between-subjects)
82
________________________________________________________________________________
with a mixed ANOVA8 revealed that free recall in each phase did not differ according to story
narrative (phase x narrative interaction not significant, F (2,32) = 0.07, p = .93). The main effect of
story phase was also not significant (F (2,32) = 0.02, p = .98), as was the main effect of narrative (F
(1,16) = 0.00, p = 1.00).
Free Recall
9
Number of participants
8
7
6
5
4
Neutral narrative
3
Negative narrative
2
1
0
1
2
Phase 1
3
4
5
6
7
Phase 2
8
9
Phase 3
Figure 4.3 Total number of participants in the neutral and emotional groups that recalled each
slideshow image.
Inspection of the recognition memory data presented in Figure 4.4 indicated that the image
of the family (Slide 3, Phase 1), the father in a jail cell (Slide 4, Phase 2), and the country cottage
(Slide 7, Phase 3) were more accurately recognised by participants in the emotional narrative group
than those in the neutral group. As with the free recall data, analysis of the recognition memory
scores revealed that memory for each phase did not differ according to story narrative (phase x
narrative interaction not significant, F (2,32) = 0.14, p = .86). However, the main effect of story
phase was significant (F (2,32) = 6.76, p < .01, ηp2 = .30) with post-hoc tests revealing that more
image details were recognised in Phase 1 (M = 62.50, SD = 11.52) than Phase 2 (M = 51.85, SD = 8.48,
p < .05, d = 1.08), and Phase 3 (M = 50.46, SD = 9.24, p < .01, d = 1.16) while the difference between
Phase 2 and Phase 3 was not significant (p = 1.0). There was also a significant main effect of story
narrative, with recognition memory accuracy significantly higher for the emotional narrative group
8
Despite the apparent influence of one slide on free recall, a phase by narrative mixed ANOVA was used to
maintain consistency with previous researchers using this three phase methodology.
83
________________________________________________________________________________
(M = 57.87, SD = 2.94) relative to the neutral narrative group (M = 52.00, SD = 4.71), F (1,16) = 10.01,
p < .01, ηp2 = .38. Despite the apparent emotion-memory effect in Slides 3, 4, and 7; between
subjects t-tests revealed that the differences failed to reach statistical significance (Slide 3, t(16) =
1.21, p = .24; Slide 4, t(16) = 2.06, p = .06; and Slide 7, t(16) = 1.73, p = .10).
Recognition
85
4AFC % Correct
75
65
55
Neutral narrative
Negative narrative
45
35
25
1
2
3
4
Phase 1
5
6
Phase 2
7
8
9
Phase 3
Figure 4.4 4AFC mean percentage correct for each image in the neutral and emotional conditions.
Mood-congruence effect on memory
Inspection of the mean number of images freely recalled presented in Figure 4.5 indicated
that more negative images were recalled when accompanied by a negative relative to a neutral
narrative. The apparent difference between conditions was tested with a narrative (2) by image
valence (neutral, positive, and negative; within-subjects) mixed ANOVA, which failed to detect a
main effect of valence (F (2,32) = 0.10, p = .91) or narrative (F (1,16) = 0.01, p = .91), or an
interaction between image valence and narrative (F (2,32) = 1.25, p = .30).
84
________________________________________________________________________________
Free Recall
Number of images
3
Neutral Narrative
Negative Narrative
2
1
0
Neutral
Positive
Negative
Image valence
Figure 4.5 Mean number images freely recalled as a function of image valence for the slideshows
with a neutral or negative narrative. Error bars represent standard error.
The only trend in the data to be consistent between free recall and recognition memory was
the tendency for more negative image details to be recognised by participants in the emotional
narrative group, illustrated in Figure 4.6. Positively valenced images also appeared to be more
accurately recognised. A mixed ANOVA revealed that differences in mean recognition memory
scores between neutral, positive, and negative images were at the predetermined chance level, F
(2,32) = 3.21, p = .05, ηp2 = .17. The interaction between image valence and condition was not
significant (F (2,32) = 1.25, p = .30). The observed difference between narrative groups was tested
with between subjects t-tests. The tests confirmed that for the negative narrative condition,
memory for the negative images was significantly higher than for the neutral narrative condition, t
(16) = 2.16, p < .05, d = 1.02, whereas the difference between conditions for positive images was not
significant t (16) = 2.00, p = .07, d = 0.94.
85
________________________________________________________________________________
Recognition
*
4AFC % Correct
70
Neutral Narrative
Negative Narrative
60
50
40
0
Neutral
Positive
Negative
Image Valence
Figure 4.6 Mean proportion of correct 4AFC recognition memory test items as a function of slide
valence for the neutral narrative slideshow and negative narrative slideshow. Error bars represent
standard error.
Pearson’s correlations were conducted to test the relationship between image valence
recognition scores and post-encoding mood valence scores. All three correlations failed to reach
statistical significance (neutral images, r = -.13, n = 18, p = .60, R2 = .02; positive images, r = -.13, n =
18, p = .06, R2 = .02; and negative images, r = -.44, n = 18, p = .07, R2 = .19).
4.4.3 Emotional response
Subjective feelings
Inspection of group means indicated that mood valence ratings from before to after both
slideshows became more negative, while arousal ratings remained relatively stable (illustrated in
Figure 4.7). Narrative (2) by time (before vs. after, within-subjects) mixed ANOVAs confirmed that
the reductions in mood valence ratings were significant for both slideshows, F (1,16) = 56.50, p <
.001, ηp2 = .78, and that mood valence ratings were more positive for the neutral slideshow, F (1,16)
= 5.20, p < .05, ηp2 = .24. The near significant interaction between narrative and time, F (1,16) = 4.11,
p = .06, ηp2 = .20, was due to greater mood valence decreases for the negative narrative (M = -4.44,
SD = 2.07) relative to the neutral narrative slideshow (M = -2.55, SD = 1.88).
The ANOVA testing arousal changes confirmed that subjective reports of arousal did not vary
from before to after the slideshows, F (1,16) = 0.51, p = .49. There was a main effect of narrative,
with arousal ratings being higher both before and after the emotional slideshow, F (1,16) = 7.58, p <
86
________________________________________________________________________________
.05, ηp2 = .32. The interaction between time and narrative was not significant, F (1,16) = 0.00, p =
1.00.
Figure 4.7 Group mean of mood valence (V, horizontal axis) and arousal (A, vertical axis) ratings for
the: (a) neutral narrative and (b) negative narrative slideshows. Arrows denote direction of change.
The possible confounding effect of baseline subjective arousal levels between groups was
further explored with ANCOVA. Memory data were re-analysed after partialling out the influence of
pre-slideshow arousal levels. Assumptions of normality, homogeneity of regression slopes, linearity,
and homogeneity of variance were met. A separate ANCOVA for the free recall and recognition
memory data revealed that pre-slideshow arousal levels were not related to memory (free recall, F
(1,15) = 0.21, p = .65; recognition, F (1,15) = 0.12, p = .73), and that memory difference between
conditions remained non-significant for free recall, F (1,15) = 0.03, p = .95; and significant for
recognition, F (1,15) = 8.35, p < .05, ηp2 = .36.
Peripheral efference and motor expression
Data were lost for four participants in the negative narrative group (case 1, all measures;
case 2, SCR Max; case 3, EMG; and case 4, IBI) and three participants in the neutral narrative group
(cases 1 and 2, all measures; case 3, SCL, SCR Max, and EMG). IBI data were lost due to high artefact,
and SCL, SCR Max, and EMG data were lost due to poor sensor connectivity to the corrugator
supercilii muscle (loose connection interfering with resistance of the SCL sensor). High baseline SCL
for one participant in the emotional narrative condition was normalised by replacing each SCL value
with the group mean plus two standard deviations. In this way, SCL values for this participant would
remain high whilst satisfying the assumption of parametric tests of being within a normal
distribution. There were moderate violations of the assumption of normality, particularly for the
87
________________________________________________________________________________
negative narrative condition. No further action was taken to normalise the distributions as ANOVA is
deemed to be robust to violation of this assumption (Howell, 2002).
Inspection of the means for each of the physiological measures (illustrated in Figure 4.8)
revealed a possible autonomic response. For the emotional narrative, there was attenuation of the
downward trend in SCL at slide five (Figure 4.8a), and of skin temperature increases in the final
phase of the slideshow (Figure 4.8d). Otherwise, apart from greater variability in the SCR Max and
IBI data for participants in the emotional condition, there was little differentiation between the two
(a)
(b)
1
1.0
SCR (max. S) Change
from Buffer 1
SCL (S) Change from
Buffer 1
groups.
0
-1
-2
Neutral Narrative
Negative Narrative
0.5
0.0
-0.5
-1.0
-1.5
-3
1
2
3
B
4
5
6
B
7
8
1
9
2
3
B
(c)
5
6
B
7
(d)
Temperature (C) Change
from Buffer 1
60
IBI (ms) Change from
Buffer 1
4
8
9
Slide (phase)
Slide (phase)
40
20
0
-20
Neutral Narrative
Negative Narrative
0.6
0.4
0.2
0.0
-0.2
-40
-0.4
1
2
3
B
4
5
6
B
7
8
9
1
2
3
B
4
5
6
B
7
8
9
Slide (phase)
Slide (phase)
(e)
Neutral Narrative
Negative Narrative
EMG (V) Change from
Buffer 1
1.0
0.5
0.0
-0.5
-1.0
1
2
3
B
4
5
6
B
7
8
9
Slide (phase)
Figure 4.8 Continuous changes in: (a) SCL, (b) SCR Max, (c) IBI, (d) Temp, and (e) EMG from the start
to the end of each of the neutral and negative slideshows. The mean of the first buffer slide was
subtracted from the mean of each subsequent slide. Numbered slides were accompanied by a
neutral narrative and B represents inter-phase buffer slides. Error bars represent standard errors.
88
________________________________________________________________________________
Data were analysed with a separate narrative (2) by slide (11, within-subjects) mixed ANOVA
for each of the physiological measures. The SCL slide main effect was the only effect to reach
statistical significance (F (4.25, 50.99) = 16.33, p < .001, ηp2 = .57, Huynh-Feldt ε = .42, p < .001).
Post-hoc tests revealed that the effect was due to progressive SCL decreases from the first to the last
slide in both conditions9. The apparent interaction between narrative type and slide for SCL and
TEMP failed to reach statistical significance (SCL, F (4.25, 50.99) = 1.41, p = .18; TEMP, F (1.97, 27.56)
= 0.44, p = .92, Huynh-Feldt ε = .20, p < .001).
Cognitive appraisal
Inspection of mean differences between groups (illustrated in Figure 4.9) indicated that
memory narrowing in the emotional phase (2) of the story did not occur. Interestingly, the
difference between central and peripheral detail memory was similar between the two groups in
Phase 1 and Phase 2 of the slideshow (more central than peripheral details recognised). However, a
change occurred in the final third of the neutral slideshow, where memory was relatively similar for
central and peripheral details (scores approaching zero).
Recognition
Central less Peripheral
4AFC % Correct
30
*
Neutral Narrative
Negative Narrative
20
10
0
-10
1
2
3
Phase
Figure 4.9 Recognition memory (4AFC) proportion of total central detail questions correct less
proportion of peripheral detail questions correct for each story phase (horizontal axis). Error bars
represent standard error. * p < .05.
9
The main effect of slide reflected downward drift usually observed in SCL data (Dawson et al., 2007). The
data were not corrected as it was assumed that the drift would be present for both groups, therefore any
differences between groups would be relative to comparable levels of SCL drift.
89
________________________________________________________________________________
A narrative (2) by phase (3) mixed ANOVA revealed a trend for the interaction between
narrative and phase to be significant, F (2,32) = 2.90, p = .07, ηp2 = .15. Post-hoc independent
samples t-tests comparing the two groups at each level of phase revealed that the trend was due to
a significant difference between groups at Phase 3 (t(16) = 2.20, p < .05, d = 1.07) that was not
present in Phase 1 (t(16) = 0.52, p = .61) or Phase 2 (t(16) = 0.57, p = .58).
4.4.4 Interim Discussion
A slideshow designed to elicit intrinsic emotion thematically with an emotional story
narrative, was more memorable than a closely matched control. On average, participants who
viewed a slideshow with a negative narrative recognized 11% more image details than those who
viewed a similar slideshow with a neutral narrative. It was predicted that memory would be most
accurate for Phase 2 of the story which contained the negative emotion manipulation. However,
memory performance was facilitated for all three phases of the slideshow. The failure to detect an
interaction between narrative and story phase indicated that facilitated memory was not specific to
details surrounding the Phase 2 emotion manipulation.
The failure to detect an interaction between narrative and story phase may be attributed to
the absence of visually aversive images in Phase 2 of the slideshows. The memory pattern revealed
in this study reflects engagement with the story as a whole, and thus more effective encoding of
story details. For instance, the emotional component of the story may have elicited post-encoding
elaboration upon the image of the mother and children, which was presented before the emotion
manipulation, and more detailed scrutiny of the cottage, which was presented after the emotion
manipulation, in a search for clues to the fate of key story characters.
There was some evidence of a mood-congruence effect on memory. The near significant
interaction between narrative and pre and post-slideshow mood valence, combined with greater
memory for negative images for those who viewed the emotional story, indicated that facilitated
memory could be accounted for by negative mood elicited by the emotional narrative increasing
encoding efficiency of negative information (a mood-congruence effect). Partial support for this
apparent relationship was provided by a weak correlation between post-encoding mood and
memory for negative images. Facilitated memory for the negative slideshow may therefore also be
partially explained by mood effects on associated neural networks and strengthened memory for
mood-congruent information.
90
________________________________________________________________________________
Investigation of the multiple components of emotion measured failed to detect change in
subjective arousal, peripheral efference (SCL, SCR, IBI, and Temp) or motor expression (facial EMG).
There was, however, evidence that the emotional slideshow decreased mood valence (see above)
and increased cognitive appraisal (indexed by memory narrowing). There were differences between
groups in the last phase of the slideshow, in which memory narrowing continued for the emotion
group, and broadened for the neutral group. The pattern of memory scope across story phases
indicated a general tendency to attend to and encode more central than peripheral visual details,
irrespective of the negative narrative emotion manipulation in Phase 2. The broadening of memory
in Phase 3 of the neutral slideshow reflected a return to ‘normal’ attention focus, thus diminished
appraisal. There was therefore some support for activation of the cognitive component of emotion
by the emotional narrative, albeit in Phase 3 of the slideshow, and not Phase 2 as predicted.
The results of this EEM manipulation indicate that facilitated memory can be accounted for
by increased cognitive processing in response to the emotional slideshow, indexed by story
engagement, mood-congruent memory, and appraisal processes. The failure to detect autonomic or
subjective change in arousal levels precludes interpretation of the memory effect in terms of
neurobiological modulation of memory. It is proposed that increases in negative mood valence
elicited by the emotional slideshow lowered firing thresholds in associated neural networks, leading
to more efficient encoding of mood-congruent negative images, while increased appraisal of the
emotional story led to greater integration of the information within memory networks. Postencoding elaboration of the emotional story relative to the neutral story would have prevented
decay of the newly connected networks.
4.4.5 Music manipulation
Extrinsic emotion effect on memory
The effect of extrinsic music-induced emotional arousal on memory for unrelated material
was tested. The effect of an emotionally powerful music selection on emotion components and
memory was compared to neutral background music or no music. Analysis of the data according to
story phase was retained, even though there was no emotion manipulation in Phase 2, for
consistency with the previous studies. Inspection of the number of participants that recalled each
image (illustrated in Figure 4.10) indicated that the images of the father in jail (Slide 4), the block of
apartments (Slide 6) and the cemetery (Slide 9) were recalled by all nine participants who viewed the
slideshow accompanied by emotional background music. By comparison, the maximum number of
participants to recall various slides in the neutral music and no music conditions was seven.
91
________________________________________________________________________________
However, a phase (3) by music (3) mixed ANOVA failed to detect differences in free recall memory
between music conditions, F (2,24) = 0.69, p = .51, at any level of phase (phase by music interaction
F (4,48) = 0.97, p = .43). There were also no differences in slide free recall between phases, F (2,48)
= 2.05, p = .14.
Free Recall
9
Number of particiapnts
8
7
6
5
No music
4
Neutral music
3
Emotional music
2
1
0
1
2
Phase 1
3
4
5
6
7
Phase 2
8
9
Phase 3
Figure 4.10 Total number of participants in the no music, neutral music, and emotional music
conditions that recalled each slideshow image.
Inspection of the recognition scores presented in Figure 4.11 indicated that memory was
generally more accurate across all slides for the neutral and emotional music conditions. This trend
was surprising given that the neutral-music condition was expected to elicit the same memory effect
as the no-music condition. Differences between conditions at each level of phase were tested with a
mixed ANOVA, which revealed that the differences in recognition scores between music conditions
was not significant, F (2,24) = 1.96, p = .16, at any level of phase (phase by music interaction F (4,48)
= 0.79, p = .54). There were also no differences in recognition scores between phases, F (2,48) =
1.44, p = .25. The trend for more accurate recognition memory in the music conditions was tested
with planned comparisons, which revealed that the difference between the neutral music and no
music conditions approached significance, p = .06, d = 1.09. The difference between the emotional
music and no music conditions was not significant, p = .41, d = 0.42.
92
________________________________________________________________________________
Recognition
90
4AFC % Correct
80
70
60
None
50
Neutral
40
Emotional
30
20
1
2
Phase 1
3
4
5
6
7
Phase 2
8
9
Phase 3
Figure 4.11 4AFC mean percentage correct for each image in the neutral and emotional conditions.
Mood-congruence effect on memory
Inspection of the mean number images freely recalled (Figure 4.12) indicated that emotional
background music elicited greater neutral and negative image recall than neutral music or no music.
The apparent image valence memory effect was tested with a music (3) by image valence (neutral,
positive, and negative; within-subjects) mixed ANOVA. The ANOVA failed to detect an image valence
effect (F (2,48) = 1.82, p = .26) or an interaction between image valence and music condition (F
(4,48) = 1.40, p = .25).
93
________________________________________________________________________________
Free Recall
Number of images
3
No music
Neutral music
Emotional music
2
1
0
Neutral
Positive
Negative
Image valence
Figure 4.12 Mean number images freely recalled as a function of image valence for the neutral
narrative slideshow accompanied by neutral or emotional background music, or no music. Error
bars represent standard error.
There was no effect of music condition on recognition memory (Figure 4.13). A mixed
ANOVA revealed that the mean difference in recognition memory scores for neutral, positive, and
negative images was not significant, F (2,48) = 1.48, p = .24, nor was the interaction between image
valence and condition (F (4,48) = 0.84, p = .52).
Recognition
No music
Neutral music
Emotional music
4AFC % Correct
70
60
50
40
0
Neutral
Positive
Negative
Image valence
Figure 4.13 Mean proportion of correct 4AFC recognition memory test items as a function of image
valence for the neutral narrative slideshow accompanied by neutral or emotional background music,
or no music. Error bars represent standard error.
94
________________________________________________________________________________
Pearson’s correlations were conducted to test the relationship between image valence
recognition scores for all participants and post-encoding mood valence. All three correlations failed
to reach statistical significance (neutral images, r = .17, n = 27, p = .39, R2 = .03; positive images, r =
.16, n = 27, p = .42, R2 = .03; and negative images, r = .07, n = 27, p = .73, R2 < .01).
Subjective feelings
Emotion responses were explored to determine whether the failure to facilitate memory
with music was attributed to the failure to manipulate emotion. For the subjective feeling
component of emotion, mood valence ratings from before to after slideshows with no music, neutral
music, and emotional music became more negative. Subjective arousal decreased slightly after
viewing the slideshow accompanied by neutral background music, and increased after viewing
slideshows accompanied by emotional music or no music (illustrated in Figure 4.14). Music (3) by
time (before vs. after, within-subjects) mixed ANOVAs confirmed that the decreases in mood valence
across all music conditions was significant, F (1,24) = 32.41, p < .001, ηp2 = .57. However, mood
valence change did not vary according to the music manipulation, F (2,24) = 0.06, p = .94, and the
interaction between time and music manipulation was not significant, F (2,24) = 0.55, p = .58).
Reports of subjective arousal did not vary from before to after the slideshow, F (1,24) = 0.03, p = .85,
nor did they differ between music conditions, F (2,24) = 1.63, p = .22, and the interaction between
time and music condition was not significant, F (2,24) = 0.56, p = .58.
95
________________________________________________________________________________
Figure 4.14 Group mean of the mood valence (V, horizontal axis) and arousal (A, vertical axis)
ratings for: (a) no music, (b) neutral background music, and (c) emotional background music. Arrows
denote the direction of the change.
Physiological and motor expression response
Data were lost for three participants in the no-music group (cases 1 and 2, all measures; case
3, SCL, SCR Max, and EMG), IBI for one participant in the emotional music group was excluded due to
high artefact, and high EMG values for another participant were normalised. There was no missing
data for the neutral music group. Inspection of the means for each of the physiological measures
(illustrated in Figure 4.15) indicated that there was very little difference between conditions for SCL,
SCR Max, IBI, and TEMP. There was attenuation of skin temperature increases for both music
conditions in the final phase of the slideshow (Figure 4.15d), and the EMG data revealed a possible
negative valence response to the emotional music, indexed by higher frown muscle activity (Figure
4.15e).
96
________________________________________________________________________________
(a)
SCL
(b)
None
Neutral
Emotional
1.0
SCR (max. S) Change
from Buffer 1
SCL (S) Change from
Buffer 1
1
SCRmax
0
-1
-2
-3
0.5
0.0
-0.5
-1.0
1
2
3
B
4
5
6
B
7
8
9
1
2
3
B
4
Slide
(c)
IBI
(d)
Temperature (C) Change
from Buffer 1
IBI (ms) Change from
Buffer 1
60
40
20
0
-20
-40
1
2
3
B
4
5
6
B
7
B
7
8
9
8
None
Neutral
Emotional
Temperature
0.4
0.2
0.0
-0.2
-0.4
1
9
2
3
B
4
5
6
B
7
8
9
Slide
None
Neutral
Emotional
EMG
1.0
EMG (V) Change from
Buffer 1
6
0.6
Slide
(e)
5
Slide
**
**
*
0.5
0.0
-0.5
-1.0
1
2
3
B
4
5
6
B
7
8
9
Slide
Figure 4.15 Continuous changes in SCL (a), SCR Max (b), IBI (c), Temp (d), and EMG (e) from the start
to the end of each of the three slideshows. The mean of the first buffer slide was subtracted from
the mean of each subsequent slide. Numbered slides were accompanied by a neutral narrative and
B represents a buffer slide. Error bars represent standard errors.
* p < .05. ** p < .01.
Data were analysed with narrative (2) by slide (11, within-subjects) mixed ANOVAs for each
of the physiological measures. The ANOVAs confirmed that there were no significant differences
between conditions for any measure. There were slide main effects for SCL (F (1.35, 28.49) = 15.58,
97
________________________________________________________________________________
p < .001, ηp2 = .43, Huynh-Feldt ε = .14, p < .001), SCR Max (F (3.44, 72.35) = 5.10, p < .001, ηp2 = .19,
Huynh-Feldt ε = .34, p < .001), and TEMP (F (2.01, 44.36) = 3.33, p < .001, ηp2 = .13, Huynh-Feldt ε =
.20, p < .001). Post-hoc tests revealed that the SCL main effect was due to progressively greater SCL
decreases from the first slide to the last slide across all three conditions. However, pairwise
comparisons failed to detect significant SCR Max or TEMP differences between slides. Slide main
effects were not significant for the IBI and EMG data.
The interaction between slide and condition was significant only for the EMG measure (F
(4,42) = 5.04, p < .01, ηp2 = .32, Huynh-Feldt ε = 1.0, p = .47). The EMG interaction was decomposed
with one-way between subjects ANOVAs and post-hoc pairwise comparisons (Bonferroni). The
ANOVAs revealed that differences in EMG activity between conditions was significant for the last
three slides only, (Slide 7, F (2,21) = 8.17, p < .01, ηp2 = .44; Slide 8, F (2,21) = 6.08, p < .01, ηp2 = .37;
and Slide 9, F (2,21) = 4.61, p < .05, ηp2 = .30), with the differences being due to increased EMG
activity for the emotional music and decreased EMG for the neutral music (Slide 7, p < .01; Slide 8, p
< .01; and Slide 9 p < .05). EMG differences between the non-music control and the neutral and
emotional music conditions were not significant.
Cognitive appraisal
Inspection of mean differences between groups (illustrated in Figure 4.16) indicated that for
all three conditions, more central than peripheral image details were recognised in the first two
phases of the story. In the final third of the story, memory was relatively similar for central and
peripheral details (scores approaching zero). A music (3) by phase (3, within subjects) mixed ANOVA
revealed that the only effect to reach statistical significance was the main effect of phase, F (2,48) =
17.38, p < .001, ηp2 = .42. Post-hoc pairwise comparisons revealed that the main effect was due to
more central than peripheral details recognised in Phase 1 and Phase 2 of the story than Phase 3
(Phase 1 vs. Phase 3, p < .01, d = 0.95; Phase 2 vs. Phase 3, p < .001, d = 1.38). The difference
between Phase 1 and Phase 2 was not significant (p = .29).
98
________________________________________________________________________________
Recognition
No Music
Neutral Music
Emotional Music
Central less Peripheral
4AFC % Correct
30
20
10
0
-10
1
2
3
Phase
Figure 4.16 Recognition memory (4AFC) proportion of total central detail questions correct less
proportion of peripheral detail questions correct for each story phase (horizontal axis). Error bars
represent standard error.
4.4.7 Extraneous music variables
Music distraction, familiarity, and enjoyment were recorded to test whether they influenced
memory. A between subjects t-test revealed that music distraction ratings were similar between the
neutral and emotional music conditions (neutral music M = 38.89, SD = 28.26, emotional music M =
50.00, SD = 21.65, t(16) = .93, p = .36). Similarly, music familiarity ratings were similar between the
two music conditions (neutral music M = 2.89, SD = 1.53, emotional music M = 3.66, SD = 1.00, t(16)
= 1.27, p = .22). There were, however, significantly higher enjoyment ratings for the neutral relative
to the emotional music (neutral music M = 4.55, SD = 0.53, emotional music M = 3.44, SD = 1.01,
t(16) = 2.91, p < .05, d = 1.44). Pearson’s correlations were conducted to determine whether these
music variables were associated with free recall and recognition memory. The results revealed that
for the 18 participants who viewed the slideshow with background music, none of the extraneous
music variables were correlated with memory scores (Free recall: music distraction r = -.33, n = 18, p
= .23; music enjoyment r = -.22, n = 18, p = .24; and music familiarity r = -.08, n = 18, p = .74.
Recognition: music distraction r = .13, n = 18, p = .62; music enjoyment r = .29, n = 18, p = .24; and
music familiarity r = .11, n = 18, p = .66).
99
________________________________________________________________________________
The analysis of individual differences failed to detect any moderators of the relationship
between emotion and memory. Refer to Appendix L for a complete report of the individual
differences analyses.
4.4.8 Interim Discussion
The results for Aim 3 of this study revealed that emotionally powerful music was ineffective
as an extrinsic source of emotion to facilitate memory for neutral material. Unexpectedly, neutral
music selected as a non-arousing music control had a mild facilitatory effect on memory for all
phases of the slideshow. On average, participants who viewed the neutral slideshow containing
neutral background music recognized 12% more image details than those who viewed the same
slideshow without music. Mood congruent memory could not explain the memory effect as memory
for all images, regardless of their emotional valence, was higher for the neutral music condition.
Investigation of the mechanisms that may explain the neutral relative to no-music memory effect
revealed that the only emotion measure to differ between the two conditions was subjective
arousal, for which there were slight decreases for the neutral music condition relative to slight
increases for no music (both non-significant).
Of the remaining measures of emotion, there were no differences in autonomic response
between conditions. The only physiological measure to differ was facial muscle activity, which
increased towards the end of the slideshow presented with emotional music. This may have been
caused by an emotional response. However, it was more likely that, in the absence of any other
emotion component activity, that the observed increase in frowning was due to some non-specific
characteristic such as the increasing volume and intensity of the music towards the end of the
slideshow. Alternatively, increased frowning may have been caused by the incongruent nature of
the negative background music and the neutral story narrative. For the cognitive component of
emotion, there was no memory narrowing differences between conditions. These results indicate
that the music manipulation had no effect on the peripheral efference or cognitive components of
emotion. Checks of music distraction, familiarity, and enjoyment between the neutral and
emotional music conditions revealed that the neutral music was more enjoyed than the emotional
music. Nevertheless, correlational analysis failed to reveal a relationship between music enjoyment
and memory.
The mild facilitatory effect of neutral music on memory may be attributed to the anxiolytic
effect of the music, and/or to increased engagement with a slideshow containing music that was
100
________________________________________________________________________________
congruent with the neutral nature of the story. Reductions in arousal in the neutral music condition
may have been due to the calming influence of the music, thereby improving engagement with and
attention to all slideshow elements. The congruence of the neutral background music and neutral
story may have increased the cohesiveness of the slideshow, thus making it more memorable than
the same story without background music. Combined with the failure to detect any evidence of
emotion, these results indicate that cognitive processes that underlie normal memory formation
most parsimoniously account for the neutral music effect on memory.
4.5 General Discussion
A slideshow containing a negative emotional story that was intrinsic to the MTBR facilitated
LTM. Participants who viewed the emotional slideshow recognized more image details than those
who viewed a closely matched neutral control. The effect of the experimental manipulation thus
confirmed that memory could be facilitated with a modified version of the three-story-phase
experimental paradigm used in previous human EEM research. There were, however, a number of
divergences in how the experimental manipulation influenced memory. First, memory
enhancement across all phases of the emotional slideshow indicated that the effect was not
narrowed to details surrounding the negative narrative in Phase 2. Second, the failure to detect
increases in autonomic or subjective arousal excluded the neurobiological account of EEM. Third,
moderate mood-congruence effects on memory and memory narrowing in Phase 3 of the emotional
story indicated that cognitive processes accounted for facilitated memory. When testing the effect
of extrinsic emotional arousal on memory, there was no evidence that negative music influenced
emotion or memory. The mild facilitatory effect of neutral background music on memory was most
parsimoniously accounted for by cognitive processes that underlie normal memory formation.
4.5.1 Neurobiological modulation of memory
Comparison of the results reported by Cahill et al. (1994) (Figure 4.17a) and Cahill and
McGaugh (1995) (Figure 4.17b) to those of Aim 1 of the current study (Figure 4.17c) indicate that
mean recognition memory scores were similar (Cahill et al. mean scores ranged from approximately
47% to 69% correct, and the current study scores range from 47% to 66% correct). There were,
however, some clear differences. More details from Phase 2 of both 1994 neutral and emotional
101
________________________________________________________________________________
slideshows were recognized relative to the neutral phases (P1 and P3; see Figure 4.17a, solid lines).
Furthermore, when taking into account generally higher memory accuracy across all three phases of
the emotional story, there appeared to be only slightly higher memory accuracy for the Phase 2
emotional story relative to the neutral story. There is a more distinct Phase 2 emotional story effect
in the 1995 study (Figure 4.17b). When comparing the 1994 and 1995 studies with the current
study, there is no consistency in the memory pattern across phases. In the current study, Phase 1 of
both the neutral and emotional slideshows were more memorable, with somewhat more details
retained in Phase 3 of the emotional slideshow relative to the neutral slideshow. The only
consistency observed across the three studies was therefore higher mean memory scores for the
emotional slideshows.
Figure 4.17 Mean proportion of correct recognition memory scores across story phases from (a)
Cahill et al. (1994), (b) Cahill and McGaugh (1995), and (c) the current study. A/BB and N/BB =
arousal and neutral story (consecutively) viewed after administration of a beta blocker. Note that
the 1994 and 1995 study values are approximates obtained from the published figures.
The inconsistency in results across studies using a similar experimental paradigm suggests
that extraneous factors may have influenced memory. Taking a placebo or beta blocker before
viewing the slideshow in the 1994 study, and participant expectation in the current study (discussed
further in the next section) may have elevated baseline arousal levels. Elevated arousal levels in the
1994 study may have increased the probability of the image of a boy’s amputated and reattached
legs eliciting an orienting reflex in both versions of the slideshow. The orienting reflex is related to
memory and is potentially mediated by the amygdala (reviewed in Section 2.1), thus explaining
attenuation of memory for the emotional slideshow with a drug that blocks amygdala activation. The
slightly higher memory scores for Phase 2 of the emotional story may have been caused by the
102
________________________________________________________________________________
negative narrative intensifying the orienting reflex. Increased arousal elicited by the experimental
procedure combined with an orienting reflex may thus explain facilitated memory in the 1994 study.
As there was no pharmacological intervention in the 1995 study, general arousal levels may have
been lower and an orienting reflex to the aversive image in the neutral story may not have been
elicited, thus explaining similar memory performance across all three phases of the neutral story.
It is proposed that an orienting reflex can affect memory via a similar amygdala mediated
mechanism as that proposed by McGaugh (e.g. 2000). As reviewed in Section 2.1, and illustrated in
Figure 2.1, the amygdala is a highly inhibited brain structure that receives inputs from the sensory
thalamus and sensory cortex. Activation of the amygdala by strong sensory input leads to orienting
and evaluation via interaction with widespread cortical regions. If the stimulus has little relevance to
the organism, cortical feedback will be weak and the amygdala returns to the inhibited state. If the
stimulus is relevant, for example has survival value, there will be greater cortical feedback, with the
strength of the feedback determining the strength of further amygdala activation. For strong
cortical feedback, the central nucleus of the amygdala mediates autonomic activation and hormone
secretion to enable the organism to respond to the emotion elicitor (approach, retreat, attack or
defend). The central nucleus of the amygdala also projects to brain stem structures that modulate
cortical arousal, including the nucleus of the solitary tract and locus coeruleus. These same brain
stem structures have been identified by McGaugh and colleagues to modulate the BLA, which in turn
modulates the consolidation of hippocampus dependent memory.
The failure to elicit a phase effect with thematically induced emotion indicates that an
orienting reflex is necessary for EEM. However, even though the orienting reflex may be part of the
emotion response, it is questionable whether it could be considered an emotion in Scherer’s terms
(e.g., eliciting multiple mental and physical components that support adaptation to the environment
to enhance well-being). Perhaps orienting-enhanced memory is a more accurate description of the
memory effect observed in previous studies. If so, the research could be extended by specifically
testing orienting-reflex effects on LTM. In terms of ecological validity, this line of research may lead
to a less complex, relative to emotion elicitation, means of facilitating memory in learning
environments. In an educational context, for example, an orienting reflex could be elicited in
students by random exposure to unexpected and novel information, such as unexpected classroom
visitors, halts to normal activity to view humorous or aversive YouTube videos, to listen to eclectic
music, or from the blast of a loud horn. The amygdala mediated orienting reflex may be sufficient to
modulate adrenal-BLA-hippocampal interactions and LTM formation. However, the influence of
such unexpected experiences on anxiety levels and subjective wellbeing would need to be carefully
considered, as would the requirement that stimuli were constantly novel and surprising. Factors
103
________________________________________________________________________________
such as orienting reflex effects on memory scope and retrograde and anterograde memory would
also be interesting avenues of investigation.
In the 1995 study, Phase 2 of the emotional slideshow was clearly more memorable than
Phase 2 of the neutral slideshow, thus facilitated memory could be attributed to differences
between the emotional and neutral narration. Nevertheless, in the absence of clear evidence of an
emotional10 response, memory could have been facilitated by greater engagement with the more
interesting emotional story. Greater story engagement would have led to stronger encoding of the
story details and to greater levels of post encoding elaboration. It is therefore not clear whether
facilitated memory could be attributed to emotion induced arousal hormones modulating memory
consolidation, to general elevated arousal levels intensifying an amygdala mediated orienting reflex
and stronger encoding of the story, or to greater engagement with the emotional story and
strengthened encoding and post-encoding elaboration.
In the current study, multiple components of emotion were measured, revealing that the
emotional story was not arousing. There was, however, evidence that the emotional story elicited a
negatively valent low arousal emotion. Mood ratings were more negative for participants who
viewed the negative slideshow and there were greater levels of memory narrowing, an index of
cognitive appraisal elicited by negative mood states. Improved memory for the negative narrative
may therefore have been caused by mood-congruence effects on network activation. Furthermore,
given that the most memorable images were those that were associated with the context, cause,
effect, and consequences of the actions of the story characters (the family, the father, and the
country house), participants may have attended to and more deeply processed the information
presented in these images than those in the neutral slideshow group. The emotionality of the story
may have also made it more interesting, thus increasing post-encoding elaboration and
strengthening of the memory trace. This interpretation could be applied to the Cahill and
McGaugh’s 1995 (Exp. 2) free recall results to explain why details relating to a central story character
in Phase 1 of their emotional slideshow, which was identical to their neutral slideshow, were more
accurately recalled.
The general conclusion drawn from these observations is that the modification to the threestory-phase experimental paradigm has demonstrated that complex processes are involved in
memory formation. The failure to elicit increased autonomic activity by the more memorable
10
Participants were asked to rate how emotional they found the story on a single 11 point self-report scale
ranging from 0 (not emotional) to 10 (highly emotional). The use of this type of subjective measure could have
led to participant reactivity, in that the wording of the question led them to rate their response as emotional,
not otherwise. Other non-emotional responses, such as interest or engagement, may have yielded similar
scores using this type of scale.
104
________________________________________________________________________________
emotional story excludes the adrenal-arousal hypothesis of memory consolidation as a causal
mechanism. Facilitated memory for the emotional story was therefore most parsimoniously
accounted for by cognitive processes that underlie encoding strength and post-encoding
elaboration.
4.5.2 Music as a source of extrinsic emotional arousal
Background music was added to the three-story-phase experimental paradigm to test the
effect of extrinsic emotional arousal on memory. Despite previous research reporting the emotional
properties of the selected music selection (Holst, the Bringer of War), the music failed to elicit an
emotional response, and failed to facilitate memory. The music selection was therefore not an
effective emotion stimulus to test the neurobiological model of EEM. It is speculated that the music
was ineffective because of the context in which it was presented. In the current experiment,
participants were asked to focus on the slideshow as if it were a television drama or movie.
Attention would thus have been distributed over multiple modes of perceptual information, rather
than specifically on the music. The level of attention to the music stimulus would therefore have
varied from previous studies, in which music was the sole source of focus, and the current study, in
which music was one component of a complex set of information. Giving participants the
opportunity to focus entirely on the music may have changed the emotional response. For instance,
dedicated listening to Mars, the Bringer of War may have evoked a number of BRECVEM
mechanisms (reviewed in Section 2.5). The sudden changes in loudness and tempo and the
dissonant aspects of this piece may have evoked brain stem reflexes; long excerpts with a definite
rhythm may have elicited rhythmic entrainment; the military style of the piece may have elicited
evaluative conditioning; the mode in which it was played and the instruments used may have elicited
emotional contagion; the music excerpt may have been associated with previous experiences of
negative events, therefore cuing episodic memory; sudden changes in the expected progression of
the excerpt may have violated musical expectancy. The context in which the music was presented
may therefore have diminished the influence of these mechanisms on the emotional response.
An alternative explanation for the failure to elicit EEM with music is that the emotion needs
to be intrinsically related to the MTBR (e.g., Cahill & Alkire, 2003). It may be argued that music
becomes intrinsically related to all manner of concurrently presented information, a classic example
being movie soundtracks (Cohen, 2001). Not surprisingly, the proposed mechanism for this music
affect infusion (Forgas, 1995) is cognitive in nature, in that knowledge of the emotional meaning of
105
________________________________________________________________________________
certain genres or characteristics of music infuse concomitantly presented information to give it more
emotional meaning (for empirical examples see Grady, Hongwanishkul, Keightley, & Lee, 2007; Tan,
Spackman, & Bezdek, 2007). Nevertheless, the degree of noradrenergic arousal elicited by
emotional background music, even if, by association, it does become intrinsically related to the
MTBR, may simply be insufficient to modulate amygdala mediated memory consolidation. It is
therefore possible that any attempt to facilitate memory for neutral material with an extrinsic and
concurrent source of emotion will fail. Fortunately, emotion can also be elicited after the MTBR to
test neurobiological accounts of EEM (reviewed in Section 2.3).
4.5.3 Limitations
An unforseen limitation to this study was that an expectation of an arousing emotional
event may have inadvertently been elicited in participants prior to exposure to the stimuli. This
expectation could have been elicited by phrases in the research explanatory statement, such as “It is
possible that some participants may find some of the stories distressing” and “should you become
distressed or upset while partaking in this study, you are free to discontinue”. The warnings were
included in the description of the study to ensure compliance with ethical requirements. The
expectation of a potentially distressing story may have heightened baseline arousal levels, which
would have had a confounding effect on the physiological data, and led to disappointment or
annoyance after viewing the slideshows, which in reality were rather mild. Unfulfilled participant
expectation would explain the significant decrease in subjective mood ratings across all conditions.
Higher memory scores for the first phase of the slideshows, in which arousal levels would have been
elevated, supports this interpretation. A possible resolution to this confound would be to remove
the negative narrative from the MTBR and elicit emotion solely with music. However, in the current
study at least, experimenter-selected music was found to be an ineffective source of emotional
arousal. To ensure an arousal response occurs, participant-selected arousing music could be used,
thus improving the probability of an emotion response. Any potential distracting effects of
participant-selected arousing music could be eliminated by presenting the music before learning.
Limiting memory testing to visual details may have also limited detection of the full scope of
memory (images and narrative). This decision was made to eliminate interpretation of memory
effects in terms of differences in the linguistic properties of the two narratives. Furthermore, testing
was limited to visual elements of the slideshow to eliminate ambiguity in the classifications of
central and peripheral narrative details. The inclusion of a narrative in the MTBR, when memory is
tested for visual elements only, thus brings into question the requirement for a narrative at all.
106
________________________________________________________________________________
Perhaps a slideshow of images combined with a story narrative is more consistent with everyday
events, hence the widespread use of the slideshow paradigm in previous research. Nevertheless,
the complexity of the slideshow stimulus restricts clear interpretation of emotion effects on
memory, particularly for central versus peripheral narrative elements. Simplifying the MTBR by
testing memory for images only would improve interpretation of emotional valence effects on
attention (e.g. attention narrowing) and memory.
A final consideration is the nature of the non-emotion comparison condition. In this and
many previous experiments, memory performance has been compared between emotional stimuli
that by nature are interesting, engaging, or unusual; or neutral stimuli, which tends to be bland,
boring, or disengaging. It would therefore be equally valid to interpret the results of the
experimental manipulation as attenuated memory for ‘boring’ stimuli (see Schellenberg, 2012, for
methodological limitations that could account for observed differences in cognitive performance).
Greater control of the properties of the comparison condition would reduce the influence of this
potentially confounding variable. One method of control would be to use a repeated measures
design. Participants would thus act as their own control and memory differences could be more
confidently attributed to the emotion manipulation.
4.5.4 Conclusion
The results of this study have demonstrated that an emotional story facilitated LTM,
whereas music intended to be emotionally powerful had little effect on emotion or memory, and
music selected to be emotionally neutral had a facilitatory effect on memory. As there was no
evidence that the music and non-music emotional stimuli elicited a noradrenergic modulated
autonomic response, the neurobiological model of EEM was not tested. Cognitive processes such as
mood-congruence, associated neural network activation, and post-encoding elaboration were thus
the most plausible explanations for facilitated memory. The next experiment in the current research
project was designed to shift the music-emotion manipulation to before the MTBR, thus allowing
participants to fully attend to and emotionally engage with the music stimuli.
107
5 EXPERIMENT TWO: Participant-selected music effects
on emotional arousal and memory
5.1 Introduction
Researchers interested in the effect of music on cognitive tasks (Husain, Forde Thompson, &
Schellenberg, 2002; Nantais & Schellenberg, 1999a; Rauscher, Shaw, & Ky, 1993; Schellenberg & Hallam,
2005; Thompson, Schellenberg, & Husain, 2001) have demonstrated a robust effect of music on spatialtemporal reasoning when music is presented up to 15 minutes prior to the task. In these studies, the
most effective music was enjoyed, preferred, and/or liked. As these positive feelings are related to
optimum arousal (Hargreaves & North, 2010), arousal elicited by enjoyed music has been posited to
explain facilitated performance within this paradigm (Jones, West, & Estell, 2006). Presenting emotional
music before the MTBR, during which time participants can devote their full attention to music listening,
may thus be an effective means of eliciting a noradrenergic modulated emotional arousal response that
facilitates memory. Emotion elicited by music prior to the MTBR could increase circulating arousal
hormones, leading to neurobiological modulation of LTM for information presentation in close temporal
proximity.
The emotion-inducing properties of music have been effectively demonstrated with an
experimental protocol designed by Salimpoor, Benovoy, Longo, Cooperstock, & Zatorre (2009). In this
paradigm, participant selected intensely pleasurable music acts as an emotion stimulus for one
participant, and a control stimulus for another participant that rated the same music as neutral. Arousal
and affective valence differences between two participants who heard the same music could therefore
be more confidently attributed to the emotion eliciting properties of the music. The use of this
experimental paradigm both increases the probability of a music-induced emotion response (being
participant-selected, not experimenter-selected music), and accounts for structural elements of the
music excerpts that may in themselves elicit feelings of emotion (e.g., tempo-arousal, mode-valence).
Salimpoor et al. (2009) also demonstrated that chills experienced while listening to music was a
reliable index of music induced autonomic arousal (reviewed in Section 2.4.1). As such, the chills
response could act as a simple measure of the peripheral efference component of emotion elicited by
music. The cognitive component of emotion could be measured with a speeded response task, with
108
CHAPTER 5. EXPERIMENT TWO
___________________________________________________________________________________
faster response times (RT in ms) indexing higher cortical arousal levels (Surwillo, 1963), and music
induced mood valence effects on information processing, such as mood-congruent memory (see Section
2.2). The combination of these measures of emotion components would further clarify the mechanism,
neurobiological or cognitive processing in nature, that underlies music-emotion effects on memory.
The use of enjoyed music selected by participants also allows for the testing of positive emotion
effects on attention and memory. It has previously been theorised that positive emotions imbue a
broadening of attention focus (reviewed in Section 2.4.2). At time of writing, there were no published
studies reporting the influence of positive emotion elicited by music on the scope of attention and
memory. The current research is therefore novel in that it aims to clarify whether positive emotion
elicited by music broadens the scope of information attended to and retained.
The primary aim of this experiment was to determine whether participant-selected enjoyed
music elicited multiple components of emotion and facilitated memory. Confirmation of music effects
on the various components of emotion and on memory would thus provide the rationale for further
testing of music as a source of extrinsic emotional arousal to facilitate the consolidation of LTM. It was
hypothesised (1) that relative to non-music and neutral music controls, participant-selected music (PsM)
would elicit multiple components of emotion, indexed by more positive and higher self-reported arousal
ratings (subjective component); greater chills frequency, higher skin conductance levels and responses,
faster heart rate, reduced heart rate variability, and reduced skin temperature (peripheral efference
component); and decreased judgement response times and increased attention scope (cognitive
component). Support for these hypotheses would confirm that PsM was a valid source of emotional
arousal.
It was further hypothesised that relative to non-music and neutral music controls, PsM would
elicit: (2) the highest memory scores; (3) if emotional arousal was the causal mechanism, then any
observed facilitated memory effect would be predicted by increases in autonomic activity; and (4) if
mood-congruence was responsible for facilitated memory, then any observed facilitated memory effect
would be explained by greater memory for images congruent with self-reported mood. See Appendix B
for a summary of the aims and hypotheses.
109
___________________________________________________________________________________
5.3 Method
5.3.1 Participants
Fifteen males and 22 females aged between 18-50 years (M = 35.03, SD = 9.76) were recruited
through University newsletters and posters or by convenience sampling. Participation was voluntary
with no reward or incentive provided. Sixty two per cent had played a musical instrument (48% of which
had played within the last year) and 92% had completed tertiary education.
Participant-selected music was used as the emotion manipulation as it is a reliable means of
inducing arousal (Blood & Zatorre, 2001; Rickard, 2004; Salimpoor et al., 2009). PsM also enabled
testing of positive emotion effects on memory. To control for the effect that different music features
may have on arousal (i.e., tempo is associated with arousal, Husain et al., 2002; Peretz, Gagnon, &
Bouchard, 1998a; Schubert, 2004) and mood (i.e., mode is associated with mood valence, Husain et al.,
2002; Peretz et al., 1998a), participants were assigned two music tracks from a pool of other
participants’ selections (other-selected music, OsM) to act as music controls. These music tracks were
selected in a pre-experiment phase of the study, the selection criteria of which are detailed in Section
5.3.6. A five minute excerpt from a national radio science show was used as the non-music active
listening control (Control). Replacement of a silence control with an active listening non-music control
reduces interpretation of effects in terms of reduced arousal caused by inactivity or boredom while
sitting in silence (refer to Schellenberg, 2012, for a review).
A data disc containing a sample of participant selected music and the non-music control can be
found in Appendix C. The radio excerpt was selected to engage attention without being overly
stimulating or boring. Differences between the control and music could therefore be more readily
attributed to the unique effect of music. To maximise music emotion effects, participants were left
alone and instructed to listen intensely to the music or control selections with their eyes closed.
5.3.3 Material to be remembered (MTBR)
The memory task was simplified by presenting participants with visual information only (a series
of IAPS image collages) after the emotion manipulation. The task required participants to attend to the
110
___________________________________________________________________________________
image collages with the knowledge that memory of the images would be tested at the end of the
session. The IAPS image collages allowed testing of cognitive processing effects on memory, such as
mood effects on attention locus (memory narrowing), and mood-congruent memory bias. The image
collages also enabled testing of attention locus by analysing the mean number of image details recalled
per image. High detail per image scores would thereby reflect memory narrowing. This method
eliminated the requirement for a priori categorisation of central and peripheral detail. Varying the
valence of the images in each collage allowed the testing of mood-congruent bias towards neutral,
positive, and negative images within the collages.
To reduce the risk of a ceiling effect (Standing, 1973), 120 images of moderate arousal (based on
normative values provided by Lang et al., 2005) were used, thereby minimizing image-induced arousal
as a potential confound. Images were reduced to 3.13cm by 4.17cm and collated to create five separate
six wide by four high 24 image collages (see Figure 5.1). Collages comprised an equal number of
positive, negative and neutral images from the semantic categories of animals, buildings, objects,
people, and scenes. The location of each image in the collage was randomized.
111
___________________________________________________________________________________
Figure 5.1 One of the five 6 x 4 IAPS image collages used in the experimental procedure.
The image collages were complex visual stimuli designed to test detailed memory and attention
locus. Images were sized to allow recognition of general content (e.g., a bird, a woman, or a car)
without difficulty, and to enable focussed analysis of the details within each image. In this way, a broad
locus of attention would be characterised by recall of few details for many of the 24 images within the
collage. In contrast, a narrow locus of attention would be characterised by recall of many details for
fewer images in the collage. Approximately two minutes were allowed for collage viewing. To minimize
test anxiety, no official timer was used, however most participants indicated that they had viewed each
collage for long enough within the two minute time-frame.
112
___________________________________________________________________________________
5.3.4 Apparatus
Instructions and stimuli were presented using Microsoft PowerPoint software on a 19”
adjustable screen. Participants sat in a firm armchair with the screen adjusted to a distance of 100 cm
and centred at eye level. The five music selections (including the radio excerpt) were digitized and
presented through closed headphones at a consistent and comfortable volume. Response time and
recognition memory test responses were recorded with E-Prime, and continuous physiological
responses were captured with a Bioview data acquisition system. Full apparatus details are provided in
the Method chapter.
5.3.5 Measures
Emotional response
To confirm an emotion response occurred, subjective dimensions of emotion (arousal and
valence), were recorded immediately after each listening condition on two dimensional affect grids. The
arousal axis was anchored by -5 (very deactivated) to 5 (very activated) and valence axis was anchored
by -5 (very negative) to 5 (very positive), both of which were expected to differ between conditions. The
frequency of chills (thrills or goose bumps) was also recorded. A second affect grid was completed at
the end of the task set to confirm that subjective affective states had returned to comparable baseline
levels across conditions. That is, it was expected that they would not differ between conditions. The
peripheral efference component of emotion was indexed by changes in skin conductance level (SCL),
skin conductance responses per minute (SCR Min), mean inter-beat interval (IBIm), the standard
deviation of IBI (IBIsd), and skin temperature (TEMP) during the music/control listening part of the
procedure (see the Method chapter for a full description of the Bioview software and measurement
parameters). Physiological data submitted for analysis included the period from 5 seconds after
stimulus onset to 5 seconds before stimulus offset.
The cognitive component of emotion was indexed by face valence judgement RT, controlled
with E-Prime, and attention locus. For the RT task, participants were instructed to decide as quickly and
accurately as possible whether a line drawing of a face was negative (pressing the number 1 on the
number keypad with their index finger) or positive (pressing the number 2 on the number keypad with
113
___________________________________________________________________________________
their middle finger). Before commencement of the experimental procedure, a RT task to judge whether
on-screen arrows were pointing left or right was conducted to control for psychomotor response
differences between the index and middle fingers. This task enabled the subtraction of RT differences
between the index and middle finger caused by psychomotor response bias. All judgement RTs greater
than 2 s were removed, as were incorrect responses. Skewed face judgement RT data were then
normalised with a natural log transform to create the psychomotor speed and skew corrected RT
variable (RTln). Attention locus was calculated separately for each collage by dividing the number of
details freely recalled by the number images recalled, thus giving a mean detail per image value. Higher
values indexed greater memory narrowing (more details per image recalled) relative to lower values
(minimum of 1).
To test what influence participation in the experiment had on subjective feelings of positive and
negative affect, the short form of the International Positive and Negative Affect Schedule (I-PANAS-SF)
developed by Thompson (2007) was administered before commencement of the experimental
procedure and again on completion. The I-PANAS-SF is comprised of five positive and five negative
affect adjectives (e.g., inspired, determined, ashamed, and nervous). Participants were instructed to
rate their level of agreement with the adjectives at the current time on a scale ranging from 1 (never) to
5 (always). Summation of the five positive adjective scores yields a positive affect index ranging from 5
to 25, and summation of the five negative adjective scores yields a negative affect index ranging from 5
to 25. Individual difference measures of music enjoyment, hours per day spent purposefully listening to
music (a measure of music engagement), gender, and BIS/BAS are described in Appendix L.
Memory
Participants were informed at the commencement of the procedure that their memory for the
collage images would be tested, thus ensuring full attention to the task. Memory for the five image
collages was tested after presentation of all five experimental blocks. The delay between encoding and
memory testing ranged from 10 to 90 minutes, depending on individual differences in task completion
and rest times. A difficult distracter task presented after each collage and the multi-task nature of the
procedure inhibited rehearsal and maintenance of the images in short-term memory, thus ensuring that
the memory data represented the early stages of LTM consolidation (McGaugh, 2000; Squire & Kandel,
1999). Free recall of the image collages was tested first. Participants were given A4 response sheets
containing 50 blank boxes, five boxes per sheet, and asked to write a detailed description of as many
114
___________________________________________________________________________________
images as they could recall; one image per box. Extra blank sheets were provided should participants
recall more than 50 of the 120 images. Participants were asked to provide enough detail to be able to
distinguish images from each other, and informed that the order of image recall was not important, nor
was correctness of spelling. Each detail correctly recalled received a score of one. Recall responses
were scored by two independent judges with high inter-rater reliability (r = .96).
Recognition memory for the collage images was tested with a forced choice old/new response
task. Each image was randomly presented using E-Prime presentation software. For each collage, there
was a foil matched on valence, arousal, and semantic category. Participants were instructed that they
were to respond ‘yes’ or ‘no’ with assigned keystrokes if they did, or did not recognise the image. To
allow a rest period, images were presented in two blocks of 120.
5.3.6 Participant music selection procedure
Part One: Participant music selection
After agreeing to participate in the study, participants were instructed to select two enjoyed
music tracks in preparation for the experiment. Tracks were chosen based on the criteria that they were
intensely and consistently enjoyed, preferably ‘unusual’ or ‘eclectic’ (to reduce the emotional impact on
other participants), between two and five minutes in duration, contained no lyrics unless in an
unfamiliar language (to avoid linguistic interference or mnemonic cues), preferably induced chills,
shivers or goose bumps, were not associated with significant past events, and were of similar
energy/activation levels. Each track was rated on the subjective valence, arousal, and chills scales
described above. A description of the difference between emotions intended by the music
composer/performer and emotions felt by the listener was also given by the researcher, and
participants were instructed to report the emotion they felt (see Salimpoor et al., 2009, for a discussion
of the importance of measuring felt rather than percieved emotion). Participants also rated how much
they enjoyed their music selections on a scale ranging from 1, not enjoyed at all, to 7, enjoyed very
much. The middle point was 4, neither enjoyed nor not enjoyed. Participants provided the time-frame
of their most enjoyed one-minute excerpt of each track to allow compilation of one-minute excerpts for
Part Two of the music selection procedure. Participants recorded their music selections onto a compact
disc (CD) and returned them by post to the experimenter along with the subjective rating scales, a
115
___________________________________________________________________________________
demographic questionnaire, and the BIS/BAS scales prior to the laboratory session. (All materials and
postage costs were provided for by the university.) This part of the procedure took from one day to
approximately two weeks to complete. Participants that took longer than two weeks to respond were
excluded from the remainder of the study.
Part Two: Control music selection
On return of music selections from groups of 10 participants, one-minute excerpts of each track
were burnt to CD. CDs comprising 18 one-minute tracks (20 tracks from 10 group members less the
respondents own selections) were then posted back to participants for rating. In this part of the
procedure, participants rated the 18 one-minute tracks on the scales of valence, arousal, enjoyment,
chills frequency, and familiarity. The familiarity rating scale was 7 point anchored by 1, not familiar at all
and 7, very familiar, with 4 representing ‘somewhat familiar’. Rating scales were then returned to the
experimenter. Completion of this part of the procedure took from 1 day to two weeks.
For each participant, two music control tracks were chosen based on the criteria of enjoyment
ratings of around ‘4’ (neither enjoyed nor not enjoyed). In this way, music enjoyment and music
indifference was the main differentiator of music condition. This method allowed all music selections
within the pool to act as either participant selected enjoyed music (PsM) or as an others’ selected music
(OsM) control (Salimpoor et al., 2009). The mean OsM1 enjoyment rating across the 37 participants was
3.76 (SD = 1.06), and the OsM2 mean was 3.70 (SD = 1.07).
5.3.7 Laboratory procedure
Emotion manipulation and memory testing occurred in the one testing session as emotion
appears to facilitate memory in a similar way regardless of time of test (albeit with generalised lower
memory performance with delays; Burke et al., 1992, Exp 1). This enabled truncation of the lengthy
procedure which involved participant directed music selection, control music selection, experimental
emotion manipulation, and memory testing.
On arrival at the laboratory, participants were given an overview of the experimental tasks,
connected to the physiological recording equipment, and given an opportunity to ask questions (a
timeline for the procedure is illustrated in Figure 5.2). Instructions for affect rating (with emphasis on
116
___________________________________________________________________________________
rating current feelings), chills, and enjoyment were given. Participants then completed practice trials for
the left/right arrow RT task and the face valence judgement task. The experiment proper was then
initiated, with five task-sets containing music or control listening, followed by affect, chills, and
enjoyment ratings, the face-valence RT task, image collage viewing, a semantic fluency task, and a
second affect rating. After viewing each image collage, participants were given 60 seconds to name as
many items as they could from one of the five categories of; things to wear, things that get you from
place to place, animals, things in the kitchen, and fruits and vegetables (semantic fluency task).
Responses were voice recorded and scored as a percentage of the total items described from each
category. The semantic fluency task acted as a filler to prevent rehearsal of the previously presented
collage and to normalise affect before commencement of the next condition. A second affect rating was
recorded after the semantic fluency task to confirm that valence and arousal had returned to
comparable levels across conditions.
The order of the five experimental conditions was presented so that the two PsM tracks were
presented together as a block, and the two OsM tracks were presented together as a block. The order
of the PsM block, OsM block, and non-music control was quasi-randomised so that the control, OsM,
and PsM were presented equally in the first and second position order, thus ensuring a primacy effect
could not account for facilitated memory. Completion of the five task-sets took approximately 60
minutes, after which the physiological recording equipment was removed and participants were given
the opportunity to take a short break (approximately 5 min). Participants were then instructed they had
20 minutes to write down as many details from the five image collages as possible. To reduce test
anxiety, participants were assured that they were not expected to recall all 120 images. They were,
however, encouraged to do their best. Response sheets were numbered to 50 to indicate the expected
number of images to be recalled. The procedure for recognition memory testing was then explained to
participants and they were given the opportunity to practice, after which recognition memory was
tested (approximately 10 min). Time to complete the full procedure was between 90 and 120 minutes.
The average delay between learning the collages and memory testing was 30 minutes.
117
___________________________________________________________________________________
Figure 5.2 Stimuli and tasks used in the experimental procedure. PsM = participant selected music, OsM
= others’ selected music, Con = non-music control. Music or control listening (a) was followed by post
music affective, chills, and enjoyment ratings (b), a reaction time task (c), 4 x 6 IAPS image collage
viewing (d), a semantic fluency filler task to prevent rehearsal (e), and post task affective rating (f). The
stimuli and tasks were presented five times, followed by a short break, then memory testing. The
numbers preceding each letter denote condition number (1 and 2 of 5).
5.4 Results
5.4.1 Data screening and manipulation checks
Data were screened for outliers and missing values. If data for any of the variables were missing
from one of the five conditions, they were replaced with the participant’s grand mean for that variable.
Outliers with values M ± 2SD were treated similarly. Due to equipment malfunction, all physiological
118
___________________________________________________________________________________
data were lost for two participants. IBI data were lost for five additional participants due to high
movement artefact. Final sample size was n = 35 for SCL and SCR Min, n = 32 for IBIm and IBIsd, and n =
37 for TEMP. Valence, arousal, and enjoyment data were negatively skewed for PsM, reflecting effects
expected for highly enjoyed music. SCR Min data were negatively skewed for all conditions, reflecting
low response for approximately 50% of participants (less than one skin conductance peak greater than
10 ohms per minute). TEMP and IBI data for the control condition were not normally distributed. As
ANOVA remains a robust test when the assumption of normality is violated and sample sizes are equal
(Howell, 2002), ANOVAs were used to test differences between conditions for all variables.
Analysis of the subjective music enjoyment ratings revealed that both PsM music tracks were
more enjoyed (PsM1 M = 6.59, SD = 0.80; PsM2 M = 6.32, SD = 1.03) than the OsM tracks (OsM1 M =
4.97, SD = 1.36; OsM2 M = 4.46, SD = 1.34) and the non-music control (M = 4.35, SD = 1.23). However, it
should be noted that mean OsM enjoyment ratings had increased by approximately 1 scale point from
the Part 2 control selection procedure (OsM1 M = 3.76, SD = 1.06; OsM2 M = 3.70, SD = 1.07) to the
laboratory procedure. Some enjoyment of music previously rated as neutral was therefore experienced
by participants. Nevertheless, a one-way repeated measures ANOVA revealed that mean enjoyment
ratings varied significantly between own and others’ music excerpts, and to the non-music control, F
(4,144) = 30.28, p < .001, ηp2 = .46. Post-hoc Bonferroni adjusted pairwise comparisons revealed that
the differences between PsM and OsM, and PsM and Control were significant (p < .001 for all
comparisons), and that there were no differences in enjoyment ratings between PsM1 and PsM2 (p =
1.00), between OsM1 and OsM2 (p = 1.00), or between OsM1 and OsM2 and the non-music control (p =
.55 and p = 1.00, respectively). These results confirmed that PsM had comparable enjoyment effects,
and that OsM and the non-music control had comparable enjoyment effects.
Analysis of subjective valence and arousal ratings after completing the semantic fluency
distracter task (Figure 5.2, 1f) was conducted to test whether valence and arousal levels had returned to
a relative baseline levels prior to the next music listening task (Figure 5.2, 2a). A one-way repeated
measures ANOVA on the mood valence ratings confirmed that there was no music condition effect on
subjective mood after completing the semantic fluency distracter task, F (2,70) = 1.31, p = .27.
Significant effects of music condition where, however, revealed for post-semantic fluency task arousal
ratings, F (2,70) = 3.39, p < .05, ηp2 = .09. Post-hoc tests revealed significantly higher arousal ratings for
the PsM conditions (M = 2.07, SD = 1.52) relative to the OsM conditions (M = 1.37, SD = 1.49), p = .01, d
= 0.46. No other comparisons were significant. These results confirmed that mood valence returned to
119
___________________________________________________________________________________
comparable baseline levels at the completion of each task-set prior to commencing the next music
listening task. Subjective arousal levels, however, were elevated in the task-sets that involved listening
to PsM, indicating a possible influence of arousal elicited by PsM conditions on the following OsM or
control conditions. To test whether condition order had any systematic effects on memory, a one-way
between subjects ANOVA was conducted on the collage free recall data. The four levels of condition
order were 2 x PsM – Control – 2 x OsM, 2 x OsM – Control – 2 x PsM, Control – 2 x PsM – 2 x OsM, and
Control – 2 x OsM – 2 x PsM. The results confirmed that memory scores were not affected by condition
order, F (3,33) = 0.50, p = .68.
As enjoyment ratings were similar for the two participant selected music tracks, and for the two
(relatively) neutral control music tracks, and condition order did not influence memory performance,
the averages of the two PsM and the two OsM tracks were calculated for all variables. Doing so
reduced variability in the data, thus improving the parametric quality of the distributions and the ability
to detect meaningful differences between conditions.
Analysis of the I-PANAS-SF data with a time (before vs. after) by affect (positive vs. negative) by
condition order (4) mixed ANOVA revealed that participants generally felt more positive (M = 14.40, SD
= 3.89) than negative (M = 6.45, SD = 1.29) at the commencement of the experiment, F (1,31) = 166.30,
p < .001, ηp2 = .84. No other main effects or interactions were significant. These results confirmed that
the more positive affective states at the commencement of the experimental procedure remained
constant and were thus not influenced by the experimental tasks.
The self-report results indicated that relative to the non-music control, PsM elicited greater
subjective feelings of positive mood valence and arousal (illustrated in Figure 5.3a and b), greater
physiological activity, indexed by more chills responses (Figure 5.3c), higher skin conductance levels
(Figure 5.3d), lower skin temperatures (Figure 5.3e), and more efficient cognitive processing, indexed by
shorter RTs (Figure 5.3f). PsM also elicited more positive mood valence ratings and chills than OsM.
However, the data revealed that OsM also had an activating effect relative to the control, indexed by
higher levels of subjective arousal, chills frequency, SCL, and shorter RTs. There were no clear trends in
120
___________________________________________________________________________________
the SCR Min, IBIm, IBIsd, and attention scope data (see Table 5.1 for means and standard deviations for
all measures).
(b)
5
4
3
2
1
0
-1
-2
-3
-4
-5
***
***
Con
OsM
Post-music arousal
Post-music valence
(a)
PsM
*
OsM
PsM
(d)
***
4
**
30
***
**
25
*
3
SCL (S)
Chills frequency
***
Con
(c)
2
20
15
10
1
5
0
0
Con
OsM
Con
PsM
(e)
OsM
PsM
(f)
6.8
36
p = .06
Response time (ms ln)
Skin temperature ( C)
5
4
3
2
1
0
-1
-2
-3
-4
-5
34
32
30
28
26
Con
OsM
**
6.6
6.4
6.2
6.0
PsM
Con
OsM
PsM
Figure 5.3 Mean (a) post-music listening subjective mood valence and (b) arousal, and music induced (c)
chills frequency, (d) skin conductance level (SCL), (d) skin temperature (TEMP), and response time for
the non-music control, OsM, PsM conditions. Error bars represent standard errors. * p < .05, ** p <
.01, *** p < .001.
121
___________________________________________________________________________________
A series of one-way within-subjects ANOVAs confirmed that there were significant differences
between conditions for all of the emotion measures except SCR Min, IBIm, IBIsd, and attention locus
(ANOVA results are presented in Table 5.1). Post-hoc decomposition of the significant ANOVAs revealed
that mood valence ratings for PsM were more positive than for the control (p < .001) and for OsM (p <
.001), while similar for the control and OsM (p = 1.00). PsM elicited significantly greater subjective
arousal responses (p < .001), chills (p < .001), SCLs (p < .01), a trend to significance for reduced skin
temperature (p = .06), and decreased RTs (p = .01) relative to the non-music control. However, OsM
also elicited significantly greater subjective arousal ratings (p < .05), chills (p < .01), and SCL (p < .05)
than the control, and differences between OsM and PsM were not significant for subjective arousal (p =
.09), SCL (p = .12), skin temperature (p = .17), and RT (p = .08).
122
___________________________________________________________________________________
Table 5.1
Mean and Standard Deviation (SD), ANOVA F Statistic, Degrees of Freedom (df), and Effect Size (ηp2) for
the Non-music Control, Others’ Music (PsM) and Participants’ Music (PsM) for each Emotion Measure
Omnibus ANOVA
Control
Mean (SD)
OsM Mean
(SD)
PsM Mean
(SD)
F
(df)
ηp2
1.11
(1.78)
1.11
(1.56)
2.62
(1.88)
13.47
(2,72)
.27
***
-0.59
(2.31)
0.62
(1.89)
1.81
(2.46)
12.35
(2,72)
.26
***
0.00
(0.00)
0.63
(1.12)
2.75
(2.17)
45.34†
(1.47,52.79)
.56
***
SCL
17.51
(7.59)
19.92
(7.68)
20.90
(8.21)
11.81†
(1.53, 52.02)
.26
***
SCR Min
1.10
(0.68)
0.98
(0.66)
1.12
(0.61)
1.49
(2,68)
.04
IBIm
819.33
(103.30)
818.33
(108.94)
816.77
(99.75)
0.10†
(1.76, 51.14)
.01
IBIsd
51.55
(17.96)
46.96
(17.82)
49.52
(16.65)
0.57
(2,58)
.02
TEMP
31.70
(4.11)
31.84
(3.75)
30.51
(5.06)
3.88†
(1.39,47.38)
.10
*
4.75
(2.05)
4.78
(1.48)
4.73
(1.72)
0.02†
(1.52,54.88)
.00
6.45
(0.24)
6.41
(0.23)
6.36
(0.24)
6.60
(2,72)
.15
**
Subjective report
Valence
Arousal
Physiological
Chills a
Cognitive
Attention scope
Race RT (ln)
Note. a No chills were reported for the Control condition. †Huynh-Feldt correction when the assumption
of sphericity was violated. * p < .05, ** p < .01, *** p < .001.
123
___________________________________________________________________________________
5.4.3 Music effects on memory
Trends in the free recall data illustrated in Figure 5.4 indicated that PsM facilitated memory
relative to the non-music control. The highest recall scores were for details of images presented after
listening to self-selected enjoyed music. The higher recall scores for OsM relative to the control
indicated that there was also a facilitatory effect of others’ music selections, even though they had
minimal effects on enjoyment. The music conditions had minimal effect on recognition of collage
images.
A one-way between subjects ANOVA confirmed that the observed free recall differences were
statistically significant, F (2,72) = 3.28, p <.05, ηp2 = .08. Post-hoc tests revealed that the difference
between the control (M = 20.54, SD = 12.31) and PsM (M = 26.03, SD = 13.68) approached significance (p
= .05, d = 0.42), while difference between the control and OsM (M = 24.40, SD = 12.52) was not
significant (p = .23), as was the difference between OsM and PsM, p = 1.00). The recognition memory
ANOVA revealed that differences between music conditions were not significant, F (2,70) = 0.91, p =.41.
All further memory analyses were therefore conducted on the free recall data only.
Free Recall
*
Details recalled
30
20
10
0
Control
OsM
PsM
Figure 5.4 Mean number of correct details freely recalled from image collages presented after the nonmusic control, OsM, and PsM. Error bars represent standard error. * p = .05.
124
___________________________________________________________________________________
Observation of the trends in the emotion measures presented in Figure 5.3 and free recall
memory scores presented in Figure 5.4 indicated that memory scores increased with activation of the
emotion components of subjective feeling, peripheral efference, and cognitive. Pearson’s correlations
were conducted to test whether the relationship between the various indices of emotion and free recall
in each condition were significant. The results revealed that increasing SCL was significantly correlated
with increasing free recall scores for all three conditions (control, r = .37, n = 35, p < .05, R2 = .14; OsM, r
= .42, n = 35, p < .05, R2 = .18; and PsM, r = .43, n = 35, p < .05, R2 = .18), as was decreasing IBIsd in the
PsM condition (r = -.40, n = 30, p < .05, R2 = .16). There was a trend for the correlations between
decreasing SCR Min and RT and increasing free recall scores to be significant (SCR Min r = -.33, n = 35, p
= .05, R2 = .11; RT r = -.29, n = 37, p = .08, R2 = .09). A correlation matrix for each condition is presented
in Appendix F.
These results indicated that increased tonic arousal levels (indexed by increased SCL and
decreased IBIsd, a measure of parasympathetic control of heart rate), increased cortical activation
(indexed by faster RT), and decreased phasic arousal responses (indexed by decreased SCR) were related
to increased memory performance. To further explore the extent to which these indices of arousal
uniquely predicted memory, a standard multiple regression analysis (MRA, enter method) was
performed for each condition. Free recall of collage details was the outcome variable, and SCL, and
IBIsd, and RT were the predictor variables. SCR Min data were skewed and therefore excluded from the
analysis. It should be noted however, that a sample size of 29 for a regression analysis with three
predictor variables was small. Results should therefore be interpreted with caution. Otherwise, all
assumptions of MRA were met.
The MRA for the control condition revealed that SCL, IBIsd and RT accounted for 18% (adjusted
R2 = .09) of free recall variance, yet failed to reach statistical significance, F (3,29) = 1.95, p = .15.
Unstandardised (B) and standardised (β) regression coefficients, and squared semi-partial correlations
(sr2) for SCL, IBIsd, and RT across the three regression models are presented in Table 5.2. For the OsM
condition, SCL, IBIsd and RT accounted for 37% (adjusted R2 = .30) of free recall variance and was a
reliable predictive model of free recall scores, F (3,29) = 5.18, p < .01. SCL, IBIsd and RT in response to
PsM accounted for most variance in memory scores (44%, adjusted R2 = .38), F (3,31) = 7.70, p < .01.
125
___________________________________________________________________________________
Table 5.2
Unstandardized (B) and Standardized (β) Regression Coefficients and Squared Semi-partial Correlations
(sr2) for Regression Models Predicting Free Recall Memory from SCL, IBIsd, and RT During a Non-music
Control, Others’ Music Selections (OsM), or Participants’ Music Selections (PsM)
Control
Variable
SCL
OsM
sr2
B
β
0.61*
.40
13%
B
0.52*
β
PsM
sr2
B
β
sr2
.34
13%
0.60*
.37
12%
-0.37**
-.45
20%
-.28
7%
IBIsd
-0.11
-.12
3%
-0.21
-.10
10%
RT
-5.20
-.10
2%
-19.72*
-.40
13%
-16.15
Note. * < .05. ** p < .01
5.4.4 Mood-congruent memory
Mood effects on memory were tested by comparing memory differences for neutral, positive,
and negative images across conditions. Inspection of mean free recall scores (illustrated in Figure 5.5)
indicated that negative images were more memorable than neutral images in the music conditions, and
positive images were more memorable than neutral images in the control and PsM conditions.
Interestingly, recall of positive images was similar to neutral images in the OsM condition.
Details Recalled
12
Neutral Images
Positive Images
Negative Images
10
8
6
4
2
0
Control
OsM
PsM
Condition
Figure 5.5 Mean number of correct details freely recalled from neutral, positive, and, negative images in
the control, OsM and PsM conditions. Error bars represent standard error.
126
___________________________________________________________________________________
The image detail recall data were analysed with an image valence (3) by music condition (3)
repeated measures ANOVA. Results revealed significant main effects of valence, F (1.79, 64.47) = 6.91, p
< .01, ηp2 = .16 (Huynh-Feldt adjusted degrees of freedom) and condition, F (2,72) = 5.06, p < .01, ηp2 =
.12. However, the interaction between valence and condition was not significant, F (4,144) = 1.54, p =
.19. Post-hoc tests revealed that the valence effect was due to significantly more positive image details
recalled (M = 8.02, SD = 3.66) than neutral image details (M = 6.43, SD = 3.45), p < .05, d = 0.45.
Similarly, more negative image details were recalled (M = 8.90, SD = 4.88) than neutral image details, p <
.01, d = 0.59. Recall differences between positive and negative images were not significant, p = .82.
A similar analysis conducted on the recognition memory data revealed a consistent image
valence effect, F (1.69, 60.98) = 4.20, p < .05, ηp2 = .10 (Huynh-Feldt adjusted degrees of freedom).
However, post-hoc tests revealed that the only comparison to reach statistical significance was the
difference between positive (M = 2.90, SD = 1.90) and neutral image recognition (M = 2.50, SD = 1.85), p
< .05, d = 0.42. Recognition memory differences between negative (M = 2.74, SD = 1.68) and neutral
images, and negative and positive images were not significant (p = .09 and p = 1.00 respectively).
Differences between music conditions were not significant (F (1.95, 70.37) = 1.14, p = .32), nor was the
interaction between image valence and music condition F (4.00, 144.00) = 0.24, p = .91).
5.4.5 Individual differences
Hours of purposeful music listening and BAS sensitivity moderated the relationship between skin
temperature and response times (respectively) on memory for control condition images. Furthermore,
enjoyment of others’ music moderated the relationship between SCL and memory. These results
indicated that individual differences in music engagement and arousal sensitivity influenced the
emotion-memory relationship. See Appendix L for a full report of the individual differences results.
5.5 Discussion
The results of this study supported the hypothesis that participant-selected enjoyed music
elicited changes in several components of emotion. Relative to a non-music control, PsM elicited
significantly higher subjective mood valence and arousal ratings, higher frequency of chills and higher
skin conductance levels, a trend for significantly decreased skin temperature, and significantly shorter
RTs. These responses reflect synchronised activation of the subjective feeling, peripheral efference, and
127
___________________________________________________________________________________
cognitive components of emotion. When considering emotional responses to participants’ own music
selections relative to other participants’ music selections, PsM elicited significantly higher subjective
mood valence ratings and higher frequency of chills. However, the differences in subjective arousal, skin
conductance levels, skin temperature, and RTs were not significant. These results revealed that others’
music and own music elicited arousal responses (top half of two-dimensional emotion space), but that
only personally selected music elicited both increased arousal responses, and more importantly, greater
positive valence responses (Q1 of 2DES). OsM was thus activating while PsM was emotional,
determined by increases in both the arousal and valence dimensions of emotion. All three conditions
were not differentiated by skin conductance responses, mean heart rhythm, heart rhythm variability, or
attention focus.
The hypothesis that PsM would elicit greater early LTM scores than the non-music control was
supported, as was the hypothesis that increasing levels of emotional arousal would be associated with
increasing memory scores. Memory scores were higher in the PsM condition than the non-music
control, and memory scores were positively correlated with skin conductance levels, and negatively
correlated with skin conductance response, heart rhythm variability, and weakly correlated with RT. A
multiple regression analysis for each condition was conducted to discern which of these variables,
excluding skin conductance response due to skew in the data, best accounted for the arousal-memory
relationship. The results revealed that increased central and peripheral nervous system arousal,
indexed by reaction time and skin conductance levels, reliably predicted memory for the OsM condition,
and increased peripheral nervous system arousal, indexed by skin conductance levels and heart rhythm
variability, reliably predicted memory for the PsM condition. The regression model for the non-music
control was not significant. It should be observed however, that skin conductance levels accounted for
similar levels of variability in recall scores in all three conditions, indicating that increased arousal,
regardless of the experimental manipulation, was associated with increased memory. This result is
consistent with an arousal dose-response effect on performance of a wide variety of cognitive tasks.
There was weak evidence that mood-congruence influenced memory in the OsM condition.
Although mood valence ratings were relatively neutral after listening to OsM, listening to others’ choices
of eclectic music may have elicited a mildly negative response and less positive image details attended
to and retained. However, as the interaction between music stimulus and image valence was not
significant, this observation remains speculative. Facilitated memory for positive and negative images
relative to neutral images was consistent with the well documented bias towards remembering
128
___________________________________________________________________________________
emotional stimuli. The failure to detect differences in attention focus, one of the two measures of
cognitive appraisal used in this study, may have been due to the insensitivity of the measure. In
previous studies that test the influence of emotion on attention scope, response time to decide the
global or local features of abstracted information, such as letters or shapes, has been used as the index
of attention bias (e.g., Gable & Harmon-Jones, 2008b). In the current experiment, the
operationalization of attention bias was intended to be more direct, in that attention to a narrow or
broad range of information presented in the image collages would be reflected in free recall memory of
the collages. This method of measuring attention scope was either ineffective, or, given the clear
emotion inducing properties of PsM music, the proposal that emotion can influence attention scope is
not valid. Further testing of music induced changes in the locus of attention are therefore required.
The finding that for the OsM condition, shorter RTs predicted early LTM indicates that cortical
activation elicited by others’ music was the mechanism underlying memory retention. The source of the
cortical activation may be attributed to brain stem reflexes evoked by the music (reviewed in Section
2.5). When participants first heard others’ music selections, they were in their own environment and
evaluating the music according to a broad range of criteria. Their general arousal levels may therefore
have been lower, and their attention to the music may have been divided. The combination of these
factors may have diminished the influence of the dynamics of the music (e.g. changes in loudness and
tempo etc.) on emotion. When participants heard others’ music in the research laboratory, their
baseline arousal levels may have been elevated due to the experimental context, and they were paying
full attention to the music. Paying full attention to others’ music with elevated arousal levels may
therefore have been sufficient to evoke brain stem reflexes that were absent on initial exposure.
Interestingly, for PsM there was no relationship between RT and memory. Instead, decreases in
heart rhythm variability predicted higher memory scores. The combination of increased SCL and shorter
RTs predicting memory in the OsM model may reflect a cortically mediated ‘response readiness’ elicited
by others’ music that improved information processing and memory. In the PsM model, the
combination of increased SCL and decreased heart rhythm variability, which is an index of increased
noradrenergic control of heart rhythm, predicted memory. PsM may therefore have elicited an
amygdala response, leading to increased noradrenergic control of heart rhythm (and by association
sympathetic control of the adrenal gland), increased circulating arousal hormones, and β-adrengeric
receptor activation in the BLA, leading to synaptic plasticity in the hippocampus and LTM formation. It
could thus be surmised that the memory gains elicited by PsM were permanent. This proposition,
129
___________________________________________________________________________________
however, needs to be tested further using more direct measures of amgydala activation and peripheral
noradrenergic activation.
In conclusion, the results of this study demonstrate that self-selected enjoyed music elicited
responses consistent with positive emotions. Furthermore, self-selected enjoyed music presented
before encoding was the most effective method of facilitating memory for visual details, possibly due to
the music effect on noradrenergic modulation of heart rhythm and memory consolidation. The
successful elicitation of multiple components of emotion with music, combined with concurrent
facilitation of early stage LTM, provides the rationale for further exploring the effects of positive
emotionally powerful music on the consolidation of LTM.
130
6 EXPERIMENT THREE: Post-learning music effects on
long-term memory
6.1 Introduction
Experiment 2 confirmed that participant-selected music elicited multiple components of
emotion, and that noradrenergic modulation of HRV elicited by this music predicted the early stages of
LTM. Others’ music selections increased physiological arousal, possibly had a mildly negative effect on
mood, and facilitated memory, although all to a lesser extent than that revealed for participant selected
music. These emotional arousal effects on memory may have been caused by arousal modulation of the
early stages of LTM consolidation. The memory effect may also have been caused by arousal effects on
information processing speed and encoding efficiency. The current experiment was therefore designed
to isolate encoding from consolidation by presenting the arousal treatment after learning. In this way,
any facilitation of memory could be more confidently attributed to emotional arousal effects on
memory consolidation, rather than on encoding.
Post-learning emotion treatments have been successfully applied in studies testing EEM in
humans (reviewed in Section 2.3). Presenting the emotion treatment after learning isolates emotion
effects on cognitive processes that contribute to memory formation (e.g., mood-congruent memory,
more efficient information processing at encoding, and post-encoding elaboration) from emotion
induced arousal effects on memory consolidation. The emotion treatment has variously included postencoding cold-pressor tests or psychological stress tests to increase cortisol levels, oral cortisol
treatments, aversive or humorous videos, unexpected rewards, or positive and negatively valent
arousing music (see Table 2.1). One of the most consistently tested methodologies within this research
paradigm is that designed by Nielson and colleagues (Nielson & Bryant, 2005; Nielson & Powless, 2007;
Nielson et al., 2005), in which participants view a wordlist, are then exposed to the emotion treatment,
and memory for the wordlist is tested immediately and again after a delay. The use of a wordlist as the
MTBR has the advantage of simplicity. Nevertheless, this simplicity limits the detection of cognitive
processes that underlie memory formation (Eich & Forgas, 2003). The wordlist learning paradigm is
therefore most suited to testing neurobiological modulation of memory, in which the effect of postencoding arousal treatments on consolidation of the memory trace are of primary interest.
131
CHAPTER 6. EXPERIMENT THREE
___________________________________________________________________________________
The participant selected music in Experiment 2 was clearly more effective as an emotional
arousal treatment than the experimenter-selected music in Experiment 1. However, the complicated
and time consuming music selection procedure proved onerous. An alternative method of music
selection that maintains sufficient control of extraneous confounding variables uses pilot testing. Music
is pre-selected based on the outcomes of previous research, and individuals are requested to provide
judgements of the emotionality of the music. Although potentially not as emotionally powerful as
participant-selected music, this method has the advantage of simplifying the music selection procedure
while maintaining good experimental control.
Specific patterns of cortical activity, measured with EEG, have been proposed to index emotions
(reviewed in Section 2.4.2). The direct measurement of cortical activation with EEG would thus increase
confidence in the detection of the cognitive component of emotion. Synchronised changes in the CNS
and PNS that were associated with facilitated memory would also index neurobiological modulation of
memory.
The aim of this third and final experiment was to determine whether arousal elicited by positive
and arousing experimenter-selected music, presented after learning, strengthened the consolidation of
LTM. It was hypothesized that relative to music and non-music control conditions: (1) PA music would
activate the emotion components of subjective feeling (indexed by increased subjective valence and
arousal ratings), peripheral efference (indexed by increased autonomic activity), and cognitive (indexed
by increased EEG activity) components of emotion; and (2) post-learning emotional arousal elicited by
PA music would facilitate long-term (one week) word list memory.
6.2 Method
6.2.1 Participants
Forty five healthy volunteers were recruited through University notices or convenience sampling
and gave their informed consent to participate. As incentive, all participants were offered the chance to
win one of two $50 iTunes music vouchers. First year psychology students also received course credit
for their participation. Thirteen males and 32 females aged between 18 and 47 years (M = 25.53, SD =
8.81) were randomly allocated to one of the three experimental conditions. Three participants were left
hand dominant, 5 ambidextrous, and 37 right hand dominant, as determined by the Edinburgh
132
___________________________________________________________________________________
Handedness survey (Oldfield, 1971). All participants were current or graduate university students, 36
had musical instrument training, and 23 had played within the previous 12 months.
Three types of sound stimuli were used; positive and arousing music (PA Music), a music control
comprised of mixed excerpts of the PA music (Music Mix), and a digital recording of outdoor street
sounds interspersed with a recording of science show radio interview (Traffic Mix). A data disk
containing the sound stimuli can be found in Appendix C. The music mix was used as a music control,
and the traffic mix was used as an active listening non-music control. Given the high variability in
individual music preferences, three PA Music tracks were selected to maximize the probability of
eliciting positive arousal. Life in Technicolor by Coldplay (2’33”), Also Sprach Zarathustra (1’40”)
composed by Richard Strauss, and O Fortuna (2’48”) composed by Carl Orff were selected based on
arousal and positive valence ratings from 17 pilot study participants (refer to Appendix G for a summary
of music selection criteria and results). The three music tracks were presented in the order described
above, with a silence interval of 5 s separating each track. The order of music presentation was held
constant to allow subsequent analyses of real-time music structure changes on continuous
psychophysiological responses (not reported in this dissertation). The duration of the PA Music stimulus
was 7 minutes and 5 seconds.
Music Mix was created by overlaying one music track on another with sound editing software
(CD Architect, Sony). The music-based control was thus the same music as the PA Music. Also Sprach
Zarathustra was overlaid on Life in Technicolor; at 1’45” Also Sprach Zarathustra faded out and O
Fortuna faded in; at 2’30” Life in Technicolor faded out and Also Sprach Zarathustra faded in. To avoid
startling participants, care was taken to ensure a smooth transition between music tracks. The resulting
stimulus contained the structural elements of all three music tracks yet was a somewhat cacophonous
and unpredictable listening experience. It was thus expected that the music mix would be an interesting
auditory stimulus that did not elicit feelings of positive emotion, increase arousal, or be considered
musical. The duration of the music mix was 4 minutes and twelve seconds.
Traffic Mix was created using sound editing software to fade a science show radio interview
(described in Chapter 5, Experiment 2) in and out of a digital recording of street sounds. The street
sounds were comprised mostly of vehicles stopping and starting with occasional bird calls and
133
___________________________________________________________________________________
pedestrian sounds. Sections of the radio interview were faded down to silence at random intervals
throughout the recording, before returning to a volume no greater than the traffic sounds. The purpose
of fading the radio interview in and out of a general auditory threshold was to ensure participants
remained attentive to the stimulus. Total duration of the traffic mix was 5 minutes and 10 seconds. The
CD Architect software was used to adjust volume to similar levels throughout all sound stimuli, thus
minimising differences in arousal caused by differences in the loudness of the stimuli (Schubert, 2004).
Each sound stimulus was saved as wave (‘wav’) files for presentation via stimulus presentation software.
Music Mix and Traffic Mix duration was not matched to PA Music (7 min 5 s) to reduce eliciting negative
affect caused by overexposure to these unusual stimuli. The difference in stimulus duration was
considered more acceptable than reducing PA Music duration and failing to elicit positive and aroused
affective states, or increasing Music Mix and Traffic Mix duration and risk eliciting negative affect.
6.2.3 Material to be remembered (MTBR)
A word list previously used by Nielson et al (Nielson & Bryant, 2005; Nielson & Powless, 2007;
Nielson et al., 2005), and Judde and Rickard (2010) was used as the target MTBR. This word list was
chosen to allow for the comparison of results between the current and prior studies. The word list was
comprised of thirty highly imaginable nouns of moderate valence and arousal, for example caterpillar,
child, lobster, and ship. Each word (coloured white) was presented sequentially in the centre of a blue
computer monitor in random order. Participants controlled the rate of word presentation. Word onset
was initiated by a keypad press, at which time a white fixation cross appeared in the centre of the
screen for five seconds, followed by the word.
6.2.4 Apparatus
Participants were seated in a firm armchair located one meter from a computer monitor. A
room divider was placed between the armchair and the rest of the room (see Appendix H for images of
equipment set-up). All stimuli and instructions were presented on screen using E-Prime. EEG and
autonomic nervous system data were acquired continuously and simultaneously using Nexus-10. Realtime psychophysiological data were continuously monitored by the experimenter on a second computer
monitor located behind the room divider. Sound stimuli were delivered via closed headphones with
134
___________________________________________________________________________________
volume adjusted to a comfortable level for each participant. Full details of the apparatus are provided
in the Method chapter.
6.2.5 Measures
Emotional response
As with Experiment 2, the subjective component of emotion was measured using a twodimensional affect grid capturing subjective arousal (-5 to 5) and valence (-5 to 5) after each listening
condition. At the same time, the frequency of chills was recorded, and the liking and familiarity of the
stimulus was measured on Likert style scales ranging from 1 (extremely disliked/not at all familiar) to 7
(extremely liked/very familiar). The peripheral efference component of emotion was indexed by
changes in mean heart rhythm (inter-beat intervals, or IBI) and the proportion of high frequency
components of heart rate variability (%HF HRV), blood volume amplitude (BVA), respiratory rate (RSP),
and spontaneous skin conductance responses per minute (SCR min) for each participant. Physiological
data submitted for analysis included the average of the period from 5 seconds after stimulus onset to 5
seconds before stimulus offset. The cognitive component of emotion was indexed by changes in alpha
and theta activity in two frontal locations (Fp1 and Fp2) and one parietal location (Pz). A full description
of these measures is provided in the Method chapter. To control for basal differences in physiological
levels, change from a two minute baseline was calculated for all physiological measures. Changes in
subjective feelings of affect from before to after the experiment were measured with the I-PANAS-SF
(described in the method section of Experiment 2).
Memory
Memory for the word list was tested immediately after presentation, and again after a delay of
one week. At both times, participants were instructed to write down as many words from the list as
they could recall. To control for individual differences in memory ability, particularly strategies used to
encode the words, the percentage of words retained from immediate testing to one-week testing was
calculated. Word recognition was also tested by asking participants to select the 30 target words from
110 matched randomly interspersed distracter words. Participants were instructed to respond by key
135
___________________________________________________________________________________
press with ‘yes’ or ‘no’ to the question “Do you recognize this word from last week?”. Recognition
scores were corrected for guessing using the formula reported by Nielson and Bryant (2005).
6.2.6 Procedure
Week One
Upon arrival at the research laboratory, participants were informed that the aim of the study
was to determine the effect of music on brain activity, physiology, and memory. They were informed
that in session one, EEG and physiological sensors would be used to measure their responses to various
stimuli and a word list memory task, with similar activities occurring in session two one-week later. As
such, unlike Experiment 1, participants were partly informed of the purpose of the study. Information
relating to the LTM test in week two was withheld. The EEG and peripheral (skin conductance, heart
rate, respiratory rate) recording sensors were connected to participants and they were seated in front of
the computer monitor. Before commencement of the tasks, participants were informed that a two
minute baseline was to be recorded, during which time they were to relax with their eyes closed,
followed by the word list memory task, a listening condition, and then report of subjective responses to
the listening condition. The sequence of these tasks and instructions were presented on the computer
monitor and participants were asked to follow the instructions accordingly.
The I-PANAS-SF was completed, after which headphones were set in place and the two minute
baseline period was initiated. A low volume tone at the completion of the baseline period queued
participants to open their eyes. At this time, participants were then given the keyboard commands and
the instructions for the word list learning procedure. The procedure was initiated and paced by
participants. On completion of the word list presentation, participants were given a sheet of paper with
a table containing 30 blank cells and asked to write down as many of the 30 words as they could recall
without concern for spelling errors. The recall task was not time limited and participants finished the
task in two minutes to 10 minutes (M = 4.32). Free recall memory testing at this point thus satisfied the
expectation of memory testing and prevented rehearsal during the one-week delay period.
After completion of the first recall test, participants listened to one of the three sound
conditions. Condition order was randomised. Participants were instructed that it was important that
they listen with their eyes closed and that they pay attention to the drums (if PA Music or Music Mix) or
136
___________________________________________________________________________________
the radio interview (if Traffic Mix), thus ensuring equal attention was paid to each condition and
minimising further elaboration of the previously presented words. Subjective ratings were completed at
the end of the listening period. Participants were then given demographic questionnaires to complete
before their arrival at the second session in one week’s time.
Week Two
After collection of completed take-home questionnaires, participants were informed that the
session involved further testing of memory for the words presented in the previous week. They were
also informed that their subjective and physiological responses to two more sound stimuli (the second
and third of the three experimental conditions, randomly presented) would be recorded.11 At this time,
participants were asked whether they were aware that their memory for the same material would be
tested. As per session one, participants were connected to the peripheral physiological recording
sensors. After set-up, word free recall was tested using the same procedure as in session one, followed
by recognition memory testing. Participants were informed that their task was to indicate which of the
140 words presented on the computer monitor (same font and colours as session one) were present in
the word list they learnt in the previous week. Instructions were given to respond with a ‘1’ for ‘yes, I
recognize the word from last week’, or ‘2’ for ‘no, I do not recognize the word’. Onset of each new
word was initiated by key press and word presentation was randomized for each participant.
Participants then listened to the remaining two sound conditions and subjective ratings were recorded.
6.3 Results
6.3.1 Data screening and treatment
All data for one participant in the Traffic Mix group was removed due to knowledge of the oneweek delayed memory test. The remaining data were screened before analysing the effects of the
experimental conditions on emotion and memory. For the Traffic Mix group, all autonomic data were
removed for one participant due to high movement artefact, technical failure led to the loss of
recognition memory data for a second participant, and theta and alpha data were not collected from
11
Physiological responses to the remaining two conditions was recorded to satisfy participant’s expectation of
music exposure in this study.
137
___________________________________________________________________________________
site Pz for a third participant due to poor scalp access at that location. Technical failure led to the loss of
all physiological data for one participant in the PA Music group. Occurrence of movement artefact did
not differ between the three conditions (Traffic Mix = 1.83%, Music Mix = 0.95%, PA Music = 1.09%).
Eye movement was marked where signals were above or below 50 µV pk-pk and then removed from all
channels. More data were removed from the Traffic Mix condition due to EOG artefact (8.45%) than the
Music Mix (1.48%), and PA Music (2.27%) conditions. It was unclear why Traffic Mix should elicit greater
eye movement than the music conditions, perhaps keeping one’s eyes closed during music listening was
easier than for the no-music control. Regardless, as the Traffic Mix was longer, greater data loss did not
interfere with overall interpretation. A test of the effect of the experimental conditions on brain
hemispheric laterality at the Fp sites (right handed participants only12, N=34), failed to detect EEG
differences between the three conditions. EEG data from the two frontal sites, Fp1 and Fp2, were
therefore collapsed.
Subjective feelings
The self-report emotion measures indicated that subjective feelings of emotion were elicited by
PA Music relative to Traffic Mix. Surprisingly, similar emotion responses were observed for the Music
Mix. There were reports of higher subjective arousal and chills frequency for both music conditions
relative to Traffic Mix. There were also reports of more positive mood valence and liking of both music
conditions. Means and standard deviations for each condition and ANOVA results are presented in
Table 6.1.
12
The laterality test was conducted on right handed participants only in order to control for the effect of hand
preference on hemispheric dominance.
138
___________________________________________________________________________________
Table 6.1
Self-report Means (M) and Standard Deviations (SD), ANOVA F Statistics, Degrees of Freedom (df), and
Effect Sizes (ηp2) for Traffic Mix, Music Mix, and PA Music
Traffic Mix
n = 14
Music Mix
n = 15
PA Music
n = 15
Omnibus ANOVA
ηp2
M
SD
M
SD
M
SD
F
(df)
Arousal
0.85
1.96
2.60
2.16
2.53
1.62
3.77
(2,41)
.16
*
Chills a
0.00
0.00
1.20
1.21
1.90
1.67
9.17
(2,42)
.31
**
Valence
-0.78
2.33
2.20
2.30
1.37
1.71
7.52
(2,41)
.27
**
Liking
3.57
1.22
4.93
1.39
5.10
1.34
5.78
(2,41)
.22
**
Variable
Note. a No chills were reported for the Traffic Mix condition. * p < .05. ** p < .01.
The ANOVAs confirmed that the differences between conditions were significant for all of the
self-report measures. Post-hoc Bonferroni adjusted pair-wise comparisons revealed that relative to
Traffic Mix, PA Music elicited more chills (p < .001, d = 2.27), more positive mood valence ratings (p <
.05, d = 1.16), and was more liked (p < .05, d = 0.98). However, the higher arousal ratings for PA Music
relative to Traffic Mix failed to reach statistical significance (p = .07). Mean differences between Traffic
Mix and Music Mix were also significant for all of the self-report measures except arousal (p = .06; chills,
p < .05, d = 1.98; valence, p < .01, d = 1.36; and liking, p < .05, d = 0.86), while differences between Music
Mix and PA Music were not significant. Analysis of the I-PANAS-SF data with a time (before vs. after) by
affect (positive vs. negative) by condition (3) mixed ANOVA revealed a significant interaction between all
three factors, F (2,36) = 3.57, p = .04, ηp2 = .17. Post-hoc within-subjects t-tests comparing mean
adjective ratings before to after the experimental procedure revealed that the interaction was due to
decreases in positive adjective ratings after listening to the traffic mix (t (11) = 2.78, p < .05), and
increases in positive adjective ratings after listening to PA music (t (12) = -2.92, p < .05). No such pattern
of effect was evident for negative adjective ratings (illustrated in Figure 6.1). Higher ratings for positive
139
___________________________________________________________________________________
than negative adjectives both before and after the experiment were also significant (F (1,36) = 161.58, p
< .001, ηp2 = .82).
Mean adjective rating
20
Before
After
*
*
15
10
5
Traffic Mix
Music Mix
PA Music
Traffic Mix
Positive Affect
Music Mix
PA Music
Negative Affect
Figure 6.1 Mean I-PANAS-SF adjective ratings for each condition before and after the experiment. Error
bars represent standard error.
* p < .05.
Autonomic response
Inspection of the autonomic nervous system means (presented in Table 6.2) indicated that
autonomic activity changed from baseline for all conditions. The experimental conditions reduced IBIs
(increased heart rhythm), increased the proportion of high frequency HRV, reduced BVA
(vasoconstriction), increased respiratory rate, and either reduced or increased SCRs. Differences
between the conditions were apparent in the degree of change from baseline, with Traffic Mix eliciting
the greatest IBI reductions and increases in high frequency HRV, and the Music Mix and PA Music
conditions eliciting the greatest decreases in BVA and increases in respiration rate and SCRs.
140
___________________________________________________________________________________
Table 6.2
Autonomic Nervous System Change from Baseline Mean (M) and Standard Deviation (SD) for Traffic Mix,
Music Mix, and PA Music
Variable
Traffic Mix
n = 13†
M
SD
Music Mix
n = 15
M
SD
PA Music
n = 14
M
SD
IBI
-26.96
35.11
-10.75
45.72
-11.41
28.80
%HF HRV
10.95
19.53
5.26
21.66
4.31
18.68
BVA
0.05
5.70
-4.41*
3.04
-2.92*
5.01
RSP
2.26
3.95
4.31
3.67
4.18
3.23
SCR Min
-0.35
1.19
0.59
1.75
0.43
1.25
Note. † n = 13 in the Traffic Mix group for all measures except SCR Min (n = 14). * Difference from
Traffic Mix, p < .05.
Mixed ANOVAs with time (baseline vs. condition) as the within-subjects measure and condition
(3) as the between-subjects measure revealed that the change from baseline was significant for IBI (F
(1,39) = 7.98, p < .01, ηp2 = .17), %HF HRV (F (1,39) = 4.87, p < .05, ηp2 = .11), BVA (F (1,39) = 11.35, p <
.01, ηp2 = .22), and RSP (F (1,39) = 40.97, p < .001, ηp2 = .51), but not for SCR Min (F (1,40) = 1.13, p = .29,
ηp2 = .03). Mean differences between conditions were not significant for all measures. The interaction
between time and condition was not significant for all measures except BVA, F (2,39) = 3.27, p < .05, ηp2
= .14 (illustrated in Figure 6.2). Post-hoc within-subjects t-tests revealed that the interaction was due to
significant BVA decreases from baseline for the Music Mix (t (14) = 5.62, p < .001), and PA music (t (13) =
2.18, p < .05) that did not occur for the Traffic Mix (t (12) = -.03, p = .97).
141
Blood Volume Amplitude (RMS)
___________________________________________________________________________________
20
Traffic Mix
Music Mix
PA Music
15
10
5
Baseline
Post-encoding Treatment
Time
Figure 6.2 Significant change in mean BVA from the two minute baseline to the post-encoding treatment
period for Music Mix (p < .05) and PA music (p < .05). Error bars represent standard error.
Cortical activation
Inspection of the cortical activation means (presented in Table 6.3) indicated that cortical
activity generally increased from baseline for all conditions, indexed by reduced theta and alpha
oscillations in both frontal and parietal sites. Differences between the conditions were most apparent in
the theta bandwidth, with the greatest descynchronisation occurring in the frontal sites for PA Music,
then Music Mix. There was relatively little change from baseline for Traffic Mix. The pattern of
desynchronisation was different for the parietal site, with the greatest and relatively equal theta
decreases in the Traffic Mix and PA Music conditions relative to little change for the Music Mix. Cortical
activation could not be differentiated by group in the alpha bandwidth.
142
___________________________________________________________________________________
Table 6.3
Cortical Activation Change from Baseline Mean (M) and Standard Deviation (SD) for Traffic Mix, Music
Mix, and PA Music
Variable
Traffic Mix
n = 14
M
SD
Music Mix
n = 15
M
SD
PA Music
n = 14
M
SD
Theta
Frontal
-0.08
0.31
-0.32
0.11
-0.48
0.20
Parietal
-0.22†
0.39
-0.12
0.15
-0.20
0.26
Frontal
-0.10
0.21
-0.08
0.22
-0.14
0.30
Parietal
-0.15†
0.34
-0.14
0.23
-0.18
0.26
Alpha
Note. All EEG data were log transformed (ln). † n = 13.
Mixed ANOVAs with time (baseline vs. condition) and EEG site (frontal vs. parietal) as the withinsubjects measures and condition (3) as the between-subjects measure revealed that the change from
baseline was significant for both frequency bands (theta, F (1,39) = 13.65, p < .01, ηp2 = .26; alpha, F
(1,39) = 14.05, p < .01, ηp2 = .26). There was also greater alpha desychronisation at the frontal sites (F
(1,39) = 48.45, p < .001, ηp2 = .55). No other main effects or interactions were significant.
Pearson’s correlations were conducted to test the strength of the relationships between
variables within each emotion component (refer to Appendix I for a full correlation matrix). To simplify
interpretation of the relationship between emotion component variables, mean baseline arousal levels
were first subtracted from mean experimental arousal levels to yield a change index. Variables
measuring the subjective component were correlated (+ve valence and liking), as were variables
measuring the peripheral efference component (-ve %HF HRV and SCR Min) and the cognitive
component (+ve frontal and parietal alpha, +ve frontal and parietal theta, +ve frontal alpha and theta,
+ve parietal alpha and theta, and +ve frontal alpha and parietal theta). Significant correlations between
emotion components were also revealed and are displayed in Table 6.4. The significant correlations
between components indicated synchronisation of multiple components of emotion. Nevertheless, due
143
___________________________________________________________________________________
to limitations in sample size, it was not possible to test whether the synchronisation differed according
to experimental condition.
Table 6.4
Significant Correlations Between Variables Within Each Emotion Component
Emotion component
Subjective feeling
Cognitive
Subjective arousal and RSP
r = .35, n = 42, p < .05
Subjective arousal and
frontal theta
r = .32, n = 43, p < .05
IBI and parietal alpha
r = .55, N = 41, p < .01
IBI and parietal theta
r = .41, n = 41, p < .01
SCR Min and frontal theta
r = .35, n = 43, p < .05
6.3.3 Memory
Inspection of mean memory scores between conditions (presented in Table 6.5) indicated that
PA Music had the greatest effect on the proportion of words retained (% of own words retained was
calculated for each participant to account for individual differences in baseline memory ability) while the
Music Mix had the greatest effect on word recognition. However, one-way between-subjects ANOVAs
failed to reach statistical significance (% of words retained, F (2,41) = 0.92, p = .41; words recognised,
error corrected, F (2,40) = 0.28, p = .76).
144
___________________________________________________________________________________
Table 6.5
Mean (M) and Standard Deviation (SD) for Words Recalled at Week 1, and Words Recalled, Percentage
of Own Words Retained, and Percentage of Words Recognised at Week 2 for the Traffic Mix, Music Mix,
and PA Music Conditions
Traffic Mix
n = 14
Music Mix
n = 15
PA Music
n = 15
M
SD
M
SD
M
SD
Week 1 words recalled
17.14
4.20
18.80
4.46
19.00
5.31
Week 2 words recalled
8.86
4.93
10.13
4.56
11.67
6.34
Percentage of own
words retained
49.37
16.82
52.58
18.24
58.35
19.11
Percentage of words
recognised (error
corrected)
61.61†
14.72
65.43
12.03
62.77
14.96
Note. † n = 13.
Pearson’s correlations were conducted to test the significance of the relationship between each
of the emotion measures and memory. The correlation was significant for subjective reports of
emotional valence and words recognised (r = .31, n = 43, p < .05). No other emotion measures
correlated with the memory measures (refer to Appendix I for a correlation matrix of memory and
emotion variables).
To further explore whether the experimental conditions moderated the relationship between
emotion components and memory, the three groups were split according to good memory performers
(above the median) or poor memory performers (below the median)13. To improve interpretation of
experimental treatment effects on emotion, only synchronised emotion component variables were
analysed (see Table 6.4). Subjective feelings of mood valence were also analysed due to the significant
correlation with memory. Plots of good versus poor free recall and recognition performers for each of
the emotion measures are presented in Appendix J. Between-subjects t-tests were conducted to test
13
The median was calculated to account for missing physiological data.
145
___________________________________________________________________________________
whether differences in the emotion responses according to memory performance differed between
conditions. The results of the tests are presented in Appendix K.
When comparing free recall memory performers, differences failed to reach statistical
significance across all measures for the Traffic Mix and Music Mix conditions. For the PA Music
condition, there were trends towards significance for RSP (t (12) = 1.95, p = .07, d = 1.14 (Figure 6.3a)
and Frontal Theta (t (12) = 2.14, p = .05, d = 1.15 (Figure 6.3b). When comparing good versus poor
recognition performers, there was a significant difference in emotional valence in the Traffic Mix
condition (t (11) = -2.73, p < .05, d = 1.53, Figure 6.4a), and IBI in the Music Mix condition (t (13) = 3.59, p
< .01, d = 1.88, Figure 6.4b).
Poor Recall
Good Recall
RSP
p = .07
p = .05
0.2
2
ln)
8
Frontal Theta
(b)
Theta Power (µV pk-pk
vs. baseline
RSP (per minute) vs. baseline
(a)
6
4
2
0
Traffic Mix
Music Mix
0.0
-0.2
-0.4
Traffic Mix
PA Music
Music Mix
PA Music
Figure 6.3 Mean differences between good and poor free recall performers for (a) RSP and (b) frontal
theta activity. Values represent change from baseline and errors bars represent standard error.
146
___________________________________________________________________________________
Poor Recognition
Good Recognition
5
4
3
2
1
0
-1
-2
-3
-4
-5
IBI
Valence
(b)
**
60
IBI (ms) vs. baseline
Subjective valence
(a)
*
40
20
0
-20
-40
-60
Traffic Mix
Music Mix
Traffic Mix
PA Music
Music Mix
PA Music
Figure 6.4 Mean differences between good and poor word recognition performers for (a) subjective
valence and (b) IBI. Values represent change from baseline and errors bars represent standard error.
6.3.4 Individual differences
Music enjoyment and BIS sensitivity moderated the relationship between BVA and word free
recall. These results indicated that individual differences in music engagement and arousal sensitivity
influenced the emotion-memory relationship. See Appendix L for a full report of the individual
differences results.
6.4 Discussion
The aim of this experiment was to determine whether emotion elicited by positive and arousing
music presented after learning a wordlist could modulate consolidation of memory for the wordlist. A
first pass analysis of the memory results revealed that music had no unique effect on memory
consolidation. Analysis of emotion component activation revealed that all three experimental
conditions had similar activating effects on subjective, autonomic, and cortical arousal, but differed in
emotional valence. The traffic-mix elicited negative valence ratings, and the music-mix and positive
music elicited positive valence ratings. These results indicated that either memory was facilitated by
147
___________________________________________________________________________________
emotional arousal elicited by all three conditions (i.e., EEM for all conditions), both negative and positive
in valence, or that the post-encoding emotional arousal treatment was ineffective.
6.4.1 Emotion effect on memory
In support of an EEM interpretation for all three conditions, memory scores in the current study
were higher than those reported in previous experiments using the same word list learning procedure.
It can be observed in Figure 6.5 that the percentage of words recalled after a one-week delay in the
current study was greater for both the control (Traffic Mix) and arousal (PA Music) conditions than that
obtained by Nielson and Bryant’s (2005) control (no reward post word list learning) and arousal ($1
reward post word list learning) conditions. Similarly, the percentage of words recognised (error
corrected) for the control and arousal conditions was greater than that obtained by Nielson and Bryant
(2005) and Nielson and Powless (2007, control, no intervention; arousal, comedy or dental surgery video
after a 30 min delay). Although Judde and Rickard (2010) used the same word list learning procedure,
comparisons could not be made with the current results as they did not test word free recall and the
recognition data was not corrected for error.
70
Control
Arousal
% Correct
60
50
40
30
20
Free Recall
Exp 3
N&P
N&B
Exp 3
N&B
10
Recognition
(corrected for error)
Figure 6.5 One-week delay percentage of a 30 item word list freely recalled and recognised (error
corrected) for control (light bars) and post-learning arousal (dark bars) across previously reported
studies and the current study (Exp 3). N & B = Nielson and Bryant (2005), N & P = Nielson and Powless
(2007). Note that the Nielson et al values were approximations obtained from published figures.
148
___________________________________________________________________________________
Further support for emotion to be the cause of the higher memory scores in the current study
was obtained from the emotion component data. The activation of multiple emotion components was
evidence of changes in both the arousal and valence dimensions of emotion in all three conditions. The
stimuli were rated as subjectively arousing, and there were significant increases from baseline in the
peripheral efference (IBI, %HF HRV, and RSP) and cognitive components (alpha and theta
desynchronisation). The conditions differed in emotional valence and in one of the peripheral efference
measures of arousal (BVA). Responses to the I-PANAS-SF revealed decreased positive affect from
commencement to completion of the experiment for those exposed to the traffic-mix, and increased
positive affect for those exposed to the positive and arousing music. The music-mix and positive music
also elicited more post-listening positive valence ratings and chills (an index of music enjoyment) than
the traffic-mix.
Decreases in BVA from baseline levels were observed for the music-mix and positive music but
not for the traffic-mix. As decreased BVA can be interpreted as noradrenergic modulation of
vasoconstriction (Grote, Zou, Kraiczi, & Hedner, 2003), this result indicated that the music-mix and
positive music may have activated neurobiological mechanisms that underlie EEM. However, the
significant difference between conditions in this index of noradrenergic activation was not associated
with enhanced memory performance. The failure to detect memory differences between conditions
may have been due to insufficient levels of activation (i.e., an insufficient dose response, reviewed in
Section 2.1). Alternatively, over-arousal elicited by the music-mix and positive music may have reduced
memory performance to that of the traffic-mix condition. The latter interpretation is consistent with
the inverted U arousal dose-response revealed in EEM research (see reviews by Baldi & Bucherelli, 2005;
McGaugh, 2004). The higher memory scores in the current study relative to previous studies using the
same word list learning paradigm, combined with the evidence that emotion was elicited in all three
conditions, provides qualified support for an EEM interpretation of these results.
6.4.2 Emotion components of good and poor memory performance
To further explore whether memory performance in each condition could be differentiated by
the various emotion components, the memory data were split into good and poor performers. The
analyses revealed that poor recognition performers in the traffic-mix condition were characterised by
149
___________________________________________________________________________________
negative mood valence compared to positive mood valence for good performers; that poor free recall
performers in the music-mix condition were characterised by decreases from baseline heart rate levels
compared to increases for good performers; and that poor free recall performers in the positive music
condition were characterised by increased frontal theta oscillations compared to decreases for good
performers.
Supplementary correlational analyses elucidated the possible mechanisms underlying these
differences between conditions. A positive correlation between emotional valence and recognition
memory indicated that as post-learning mood valence ratings became more positive, memory
performance improved. This relationship was explicit in the traffic-mix condition. Good performers had
relatively neutral valence responses to the traffic-mix, whereas poor performers had negative
responses. The aversive nature of the traffic-mix in this sub-group of memory performers may therefore
have interfered with the ongoing processing of the wordlist, thus attenuating the facilitatory effects of
arousal on memory.
The effect of the music-mix on memory may be explained by reductions in pre-existing arousal
levels. There were positive correlations between heart beat intervals and parietal alpha and theta
oscillations, indicating that decreases in heart rhythm were associated with decreased cortical activity.
Concomitant changes in both autonomic and cortical indices of arousal thus improved the confidence in
which heart period reductions could be attributed to reductions in emotional arousal. As heart period
is regulated by the noradrenergic system, heart period decreases for poor memory performers in the
music-mix condition were interpreted as the music-mix decreasing emotional arousal, thus attenuating
EEM. The music-mix memory attenuating effect was consistent with studies that reveal attenuation of
EEM with drugs that block β-adrenergic receptors in the central and peripheral nervous systems (Cahill
et al., 1994; O'Carroll et al., 1999b; van Stegeren et al., 1998), and more recently by anxiolytic music
(Rickard, Wong, & Velik, 2012). In the current study, autonomic and cortical arousal levels were
generally higher than baseline levels across the three conditions. When the music-mix reduced arousal
during the consolidation period, memory performance was impaired. The memory attenuating effect of
the anxiolytic music-mix was thus consistent with neurobiological modulation of EEM.
The increased frontal theta response to positive music for poor performers was consistent with
cognitive processing of emotional information interfering with consolidation. Frontal theta was
positively correlated with subjective and phasic arousal (SCR Min) levels. The concomitant increase in
frontal theta and phasic arousal is consistent with focused attention, effective information processing,
150
___________________________________________________________________________________
and emotion responding (Dawson et al., 2007; Pizzagalli, 2007; Sammler et al., 2007). The frontal theta
increase for poor memory performers in the positive music condition could therefore be interpreted as
increased concentration and emotion processing elicited by the positive music interfering with the
ongoing processing of the wordlist. Thus, as with the traffic-mix, post-encoding interference may have
attenuated the facilitatory effects of emotional arousal on memory; much like a distractor presented
after a working memory task prevents ongoing rehearsal. These conclusions should, however, be
considered with caution given the weak effect sizes and low participant numbers in each cell. It is
therefore imperative that further testing of the effects of post-encoding positive, negative, and neutral
stimuli on memory consolidation be conducted.
6.4.3 Limitations
The finding that the traffic-mix and music-mix had similar effects on arousal and memory as
positive and arousing music was unexpected. The traffic-mix was designed to be a non-music control
that required participants to actively listen for the random placement of certain elements within the
stimulus. Memory differences between traffic-mix and positive and arousing music caused by decreased
levels of attention could therefore be more readily excluded. The music-mix was designed to have the
same sound qualities as the positive and arousing music without being considered particularly ‘musical’.
Indeed, in some parts of the mix, the sounds were quite chaotic. The traffic-mix was a recording of
traffic sounds and a radio interview, sounds that would not normally be considered emotional.
The unique nature of these comparison stimuli relative to the somewhat familiar positive and
arousing music may partly explain the failure to detect emotion and memory difference between
conditions. Perhaps the unpredictable and attention demanding nature of the two control conditions
evoked the brain stem reflex mechanism of music induced emotion (described in Section 2.5), and in the
case of the traffic-mix, non-music induced emotion, thus increasing arousal levels. Subjective reports
indicated that the traffic-mix was appraised as negative, and the music-mix was appraised as positive,
thus eliciting responses on the arousal and valence dimensions of emotion. Furthermore, the chaotic
musical features of the music-mix may have evoked the musical expectancy mechanism, in which
interfering with the normal progression of familiar music (particularly Also Sprach Zarathustra and O
Fortuna) violated expectancy and elicited surprise. Based on the subjective valence reports, this
151
___________________________________________________________________________________
surprise was appraised as positive. If this interpretation is correct, then presentation of any unique or
unpredictable stimulus after learning could modulate memory consolidation.
Alternatively, facilitated memory across all three conditions relative to previous studies using
the same word list learning task may be explained by the context in which the experiment was
conducted, and/or to participant reactivity14 to the wordlist learning task and short-term memory test.
A well known example of participant reactivity is the Hawthorne effect, where it was observed that
participant behaviour was changed by being involved in research and not by the experimental
manipulation (changing the lighting conditions of the work environment). Considering first the
experimental context; testing participants alone and with complicated physiological recording
equipment may have had the effect of elevating baseline arousal levels, which in turn would have
increased the efficiency in which the wordlist was encoded (see also Section 4.5.1 for the interpretation
of differences between Cahill and colleagues memory results and the Exp. 1 memory results). Longterm memory would therefore have been strengthened for participants who were reactive to the
experimental context. Participants were also aware that their memory for the word list would be tested
immediately after the listening condition. This may have prompted them to exert more effort on
memorising the word list, and thus improved long-term retention. The combination of elevated baseline
arousal levels with knowledge of the short-term wordlist memory test may have compounded
participant reactivity. Observation of heart rhythm increases during the wordlist encoding task supports
this explanation15. Nevertheless, these explanations do not account for the decrement to memory with
the post-encoding presentation of the music-mix. Future research that attempts to manipulate emotion
should therefore carefully consider contextual factors that can influence the intended emotional
outcome.
6.4.4 Conclusion
Positive and arousing music presented after learning a word list had similar effects on LTM as
music and non-music controls. Activation of multiple components of emotion with all three
14
Participant reactivity is defined as the phenomenon were participants “react to features of an experiment so
that the process of making observations can change observations” (Haslam & McGarty, 2000, p. 75).
15
Due to movement artefact caused by the use of a finger plethysmograph to record heart rhythm, the data were
not submitted to parametric testing. Researchers interested in measuring physiological reactivity to experimental
tasks may wish to consider using electrocardiogram (ECG) as a more stable alternative to plethysmography.
152
___________________________________________________________________________________
experimental stimuli indicated that emotions had been elicited. Furthermore, memory scores were
generally higher than previous studies using the same word list learning procedure. The relationship
between these two factors tends to support an emotion-enhanced memory interpretation of the
results. However, care must be taken with this interpretation as the difference from previous research
may also be explained by differences in participant motivation. The subtle difference in the emotion
profiles of good and poor memory performers indicated the potential for post-encoding emotion
treatments to interfere with memory consolidation, either by placing more demand on cognitive
resources to appraise the emotion stimulus (positive and arousing music and the traffic-mix condition),
or by attenuating noradrenergic modulation of memory consolidation (the music-mix condition). The
unpredicted results of this study have illustrated the complexity of emotion and memory phenomena. It
is recommended that future research considers controlling not only the neurobiological and cognitive
variables that may influence EEM, but also contextual variables that may influence the intensity of the
emotion response.
153
7 GENERAL DISCUSSION
Research lead by James McGaugh and colleagues has demonstrated that arousal hormones and
neuromodulators released during an emotional event can strengthen consolidation of a memory
trace. The current research project set out to investigate whether emotional arousal elicited by music
yielded the same memory strengthening properties. Specifically, the aim was to determine whether
emotionally arousing music could strengthen the consolidation of unrelated information. Successful
facilitation of memory with music in the research laboratory would pave the way for generalisation of
EEM to the population at large. The anticipated complexity of the results was reduced by applying
stringent control to the experimental design, while the implementation of a broad range of measures
enabled testing of alternative explanations of music effects on memory. Theories from the domains of
emotion, neurobiology, and cognitive processing were drawn upon to determine the extent of the
emotion response elicited by positive and negative experimental stimuli, to test McGaugh’s adrenalarousal hypothesis of EEM, and to determine the extent to which cognitive processes independent of
the adrenal response accounted for facilitated memory.
The effect of music on memory varied widely across the three experiments. In Experiment 1,
experimenter-selected negative background music had no effect on emotion or memory. In Experiment
2, participant-selected enjoyed music presented before learning elicited multiple components of
emotion and facilitated the early stages of LTM. In Experiment 3, experimenter-selected positive music
presented after learning elicited multiple components of emotion, but had similar effects on memory as
music and non-music controls. Further investigation of the music-induced emotion effect on memory in
this study revealed that emotion elicited by the music impaired LTM. The music-induced memory effect
observed in Experiment 2 was consistent with increased noradrenergic control of heart rhythm. The
memory mechanism was therefore attributed to increased arousal as proposed by McGaugh and
colleagues. In contrast, the memory attenuating influence of positive and arousing music in Experiment
3 was attributed to cognitive processes that interfered with consolidation of the memory trace.
The non-music effects on memory revealed in this project were also noteworthy. In Experiment
1, modification of a widely used methodology to test EEM in humans yielded memory enhancement that
was most parsimoniously explained by cognitive processes that were independent of an adrenal arousal
response. Indeed, there was no evidence of increases in autonomic activity, argued to be an indirect
index of the amygdala-mediated, adrenergic modulation of synaptic plasticity in the hippocampus, to
account for the observed memory enhancement. This raises the possibility that previous research
154
CHAPTER 7. GENERAL DISCUSSION
___________________________________________________________________________________
investigating EEM failed to adequately distinguish between purely arousal-based mechanisms and the
influence of cognitive processes on memory, such as associated neural network activation and moodcongruence, more efficient encoding of an engaging story, or post-encoding elaboration. In Experiment
3, positive and arousing music had similar enhancing effects on memory as a random mixed-up version
of the same music, and a non-music control. There was, however, more direct support for McGaugh’s
model of EEM in this study. When the mixed version of music reduced heart rhythm, which is
modulated by noradrenergic receptors, there were concomitant decreases in memory scores. This type
of memory-arousal relationship has been revealed in studies that block noradrenergic receptors with
beta-blockers (e.g., propranolol), and has been interpreted as attenuation of neurobiological modulation
of EEM. The music-mix thus manipulated EEM, although in the direction opposite to expected
(attenuated rather than facilitated). A summary of the variables tested in each experiment and the
outcomes is presented in Table 7.1.
The influence of positive and arousing music on memory in Experiment 2 supports the
hypothesis that positive emotions can modulate memory. Due to the failure to differentiate broad from
narrow memory bias in Experiment 2, it was not possible to determine the extent to which positive
emotions influenced memory scope. The influence of positive emotions on the range of information
retained therefore remains unresolved. A comprehensive analysis of moderator variables (presented in
Appendix L) revealed that individual differences in arousal responsiveness moderated the relationship
between various indices of arousal and memory. However, the memory moderating effects of individual
differences were not consistent across experiments, nor were they consistent across multiple indices of
physiological arousal. Given the sporadic nature of the moderators, theses data were not presented in
the body of the thesis. It is recommended that future research considers their investigation with larger
sample sizes.
155
___________________________________________________________________________________
Table 7.1
Summary of Tested Variables and Outcomes for the Three Experiments
Experiment 1
Experiment 2
Experiment 3
Timing of music
exposure
Music selected by
During
Before
After
Experimenter
Participant
Experimenter
Emotion stimulus
Negative music
Enjoyed music a b c
Positive and arousing music a b c
Radio interview
Radio/traffic mixa b c
Others' musica b
Music mixa b c
MTBR
Neutral storyno music
Neutral storyneutral music
Images
Images
Words
Participant naivety
Complete
None
Partial
Test delay
1 week
10 - 60 minutes
1 week
Memory modulated
Recognition
Free recall
Recognition and free recall
Strength of memory
modulation
Memory predictors
↑ 11%d
↑ 27%e
↑ 35%f
Negative mood
Increased SCL
Negative valence (traffic-mix)
Memory narrowing
Decreased SCR Min
Decreased HR (music-mix)
None
Approach motivation
Increased RSP and
frontal theta (PA music)
None
Cognitive
processing
Arousal modulation
of memory
consolidation
Arousal modulation of memory
consolidation and cognitive
processing interference
Negative story
Controls
Emotion/memory
moderators
Proposed mechanism
ab
Note. Emotion components activated by the emotional or control stimuli are denoted by a = subjective, b
= peripheral efference, and c = cognitive. Comparisons for the strength of memory modulation are
denoted by d = negative narrative compared to a neutral narrative, e = participant selected music
compared to a radio interview, and f = Experiment 3 grand mean compared to Nielson and Bryant (2005)
and Nielson and Powless (2007) grand mean.
156
___________________________________________________________________________________
7.1 Emotion elicitation
Multiple components of emotion were successfully activated by a variety of musical and nonmusical stimuli in this project. A negative story without music presented in Experiment 1 elicited
emotions that were negatively valent and low in arousal. Music that participants selected themselves
elicited the highest post-listening positive mood valence ratings across experiments, the highest
frequency of chills16, and was the only music stimulus that simultaneously activated more emotion
components than the controls. There were, however, inconsistencies between predicted and actual
emotional responses to some of the stimuli (summarised in Table 7.2). There was no evidence of an
emotional response to background experimenter-selected music in Experiment 1, and the response to
experimenter-selected positive music in Experiment 3 was similar to a randomly mixed version of the
same music. Nevertheless, experimenter-selected positive music presented in its original form, or in
mixed form, elicited changes in multiple components of emotion that were consistent with a positive
and aroused emotional state. Relative to the non-music control, both versions of experimenter-selected
music elicited feelings of positive valence, induced chills, and increased vasoconstriction. It should be
noted, however, that the original form of positive and arousing music elicited longer lasting effects on
positive mood valence, indexed by positive adjective ratings at the end of the experimental procedure,
relative to the same music randomly mixed. It should also be noted that the traffic mix control elicited
lasting effects on negative mood valence, indexed by negative adjective ratings at the end of the
experimental procedure. It is therefore concluded that music is an effective method of eliciting positive
emotion, more so if the music is personally selected, and that experimenter-selected positive and
arousing music elicited a positive and aroused affective state, while a sound recording of local street
traffic mixed with a radio interview elicited a negative and aroused affective state.
16
Note that chills were not recorded in Experiment 1.
157
___________________________________________________________________________________
Table 7.2 Predicted Arousal and Valence and Actual Arousal and Valence for Control and Emotional
Music and Non-Music Stimuli Used in Each Experiment
Stimulus
Control
Exp. 1
Nonmusic
Emotional
Predicted
arousal
Actual
arousal
Predicted
valence
Actual
valence
0
0
0
-
↑
0
-
-
0
0
0
-
↑
0
+
-
Neutral
narrative
Negative
narrative
Music
Gymnopedies
#1
Mars, The
Bringer of War
Exp. 2
Nonmusic
Radio
interview
0
0
0
0
Music
Others’ music
rated as
neutral
0
↑
0
0
↑
↑
+
+
Own music
rated as
enjoyed
Exp. 3
Nonmusic
Traffic-radio
mix
0
↑
0
-
Music
Music-mix
0
↑
0
+
↑
↑
+
+
1. Cold Play
2. Also Sprach
Zarathustra
3. O Fortuna
158
___________________________________________________________________________________
The location of each of the experimental stimuli in two-dimensional emotion space is presented
Figure 7.1. The Experiment 1 negative story (Exp 1t) was negative and low in arousal, thus occupying Q3.
Experiment 2 control music (Exp 2c) increased arousal but was neither negative nor positive, thus
occupying a location in the top centre of the emotional space. The Experiment 2 enjoyed music (Exp 2t)
was positive and arousing, thus occupying Q1. The Experiment 3 non-music control (Exp 3c1) was
negative and arousing, thus occupying Q2, while both versions of the positive and arousing music, either
randomly mixed (Exp 3c2) or in its original form (Exp 3t) were positive and arousing, thus occupying Q1.
Figure 7.1 Two-dimension emotional space; valence (negative to positive), and arousal (deactivated to
activated); and the four corresponding emotion quadrants: Q1 (e.g., joy and excitement), Q2 (e.g.,
distress and anger), Q3 (e.g., depression and sadness), and Q4 (e.g., contentment and tenderness). Exp
1t = negative story emotion treatment, Exp 2t = participant selected enjoyed music treatment, Exp 2c =
music control, Exp 3t = positive and arousing music treatment; Exp 3c1 = non-music control, and Exp 3c2 =
music control.
Elicitation of synchronised changes in emotion components with participant-selected music in
Experiment 2 lends support to the argument that music emotions are authentic. Further support can
be inferred from the memory effects observed in this research project. Emotional events are more
159
___________________________________________________________________________________
memorable because it is adaptive for them to be so. Survival is enhanced if details surrounding a
positive or negative event are remembered and sought out or avoided in the future. In Experiment 2,
participant-selected music activated multiple components of emotion, while increases in the peripheral
efference component predicted the early stage of LTM. The observed memory effect indicates that
music-emotion may influence memory, and thus may be argued to have adaptive value.
7.1.1 Effective control of the emotional response
Predicted versus actual effects of some of the stimuli used in this research project varied
somewhat. In Experiment 1, background music was expected to enhance the emotionality of the
slideshow. However, valence and arousal ratings for all four conditions were similar irrespective of the
presence of background music (see Table 7.2). It was proposed that the failure to elicit emotion with
music was due to the dispersion of attention across visual and auditory modalities, thus decreasing
focus on the music and diminishing the influence of music-emotion evoking mechanisms. The small
memory advantage for those who viewed a neutral slideshow combined with neutral music was
attributed to the congruence of the two stimuli increasing engagement with the story, thus improving its
memorability. In Experiment 2, actual valence and arousal responses were in the predicted direction for
the non-music control and participant selected music. However, there was an unexpected increase in
arousal for others’ music that was previously rated as neutral. This unexpected increase in arousal after
listening to music previously rated as neither arousing nor calming was proposed to be caused primarily
by the experimental context. Again, attention focus may have influenced the mechanisms that underlie
music-evoked emotions. In this experiment, participants’ attention may have been divided when first
exposed to others’ music (in their own environment), and more focussed when in the experimental
setting, thus increasing the influence of music-emotion evoking mechanisms, particularly brain stem
reflexes.
In Experiment 3, the emotional response to the two control conditions was more problematic.
The absence of a non-emotion control condition constrained interpretation of emotion effects on
memory. This was overcome to a certain degree by comparing good and poor memory performers in
each condition. However, given the small sample size in these analyses, the results need to be
interpreted with caution. The control conditions were deemed emotional according to Scherer’s
160
___________________________________________________________________________________
component process model of emotion (2001). Emotion was confirmed when the stimuli elicited
synchronised changes in subjective feeling (encompassing the conscious experience of changes in
arousal and mood valence), peripheral efference (encompassing changes in the central and autonomic
nervous systems to support an appropriate adaptive response), and cognitive processing (encompassing
the evaluation of the stimulus and its potential to threaten or benefit well-being and goal attainment,
and increased cortical activation leading to more efficient information processing). The component
process model accounts for the arousal (subjective and peripheral efference) and valence (subjective
and cognitive processing) dimensions of emotion. Experiment 3 participants presented with the
controls experienced increased central and autonomic activity, with those presented with the non-music
control experiencing negative subjective mood valence and those presented with the music control
experiencing positive subjective mood valence.
The emotion response to the non-music control, which was a sound recording of local street
traffic with a radio interview randomly fading in and out, illustrated the difficulty in manipulating
emotion for the purpose of experimentation. Pre-testing of the sound recording during the design of
the experiment indicated that the stimulus was relatively innocuous, thus suitable as a non-music
emotionally neutral control. The innocuous nature of the stimulus clearly changed in the experimental
context. Given the nature of the experimental task and the measurement of brain and body responses
(as described in the Explanatory Statement), participants would have been cognisant that their memory
ability was being evaluated. This may have influenced the way they reacted to the sound recording. As
such, sounds that were of little consequence to listeners in the pre-experimental design stage, when no
monitoring was taking place, may have become emotionally activating during the experimental stage,
when the listener was under full observation and their ability to perform on certain tasks was being
tested. Confusion arising from the sound recording, which was purposely designed to be random to
ensure participants continued to pay attention, may therefore have had no emotional impact on the
pre-experiment listener, while eliciting aversive feelings in the evaluated listener. The participant
reactivity to the control condition in this experiment thus highlights the importance of accounting for
the context of the experimental manipulation when researching emotion.
The design of the music control, consisting of three music tracks played in unison two at a time,
was inspired by the music control designed by Menon and Levitin (2005). The use of the experimental
music to create the control thus controlled differences in pitch and loudness. The music control differed
from that of Menon and Levitin’s in the time fragments used and the source of the music ambiguity.
161
___________________________________________________________________________________
Menon and Levitin introduced ambiguity by presenting 350 ms randomly placed excerpts of the
experimental music continuously, whereas ambiguity in the Experiment 3 music control was introduced
by dividing each experimental music track into three excerpts, then playing two excerpts concurrently
until all tracks had been presented in full. Although similar in principle, the major difference between
these music controls was the duration of the excerpts. It is speculated that the music control in
Experiment 3 elicited an emotional response because the temporal structure of the tracks remained
relatively intact, thus the unpredictable nature of two excerpts played in unison and the change from
excerpt to excerpt violated musical expectations (Juslin & Vastfjall, 2008). Violation of musical
expectations has been cited as a significant contributor to pleasure elicited by music (Huron, 2006). In
future, it is recommended that excerpts used in this type of music control are short enough to eliminate
recognisable structural progressions.
Valence has been defined as a fundamental dimension of emotion experience (Russell, 1980).
The relatively valence neutral effect of the music control in Experiment 2 (others’ music selections)
indicated that although activating, with the absence of positive or negative feelings emotion was not
elicited. Others’ music that was rated as ‘neither enjoyed nor not enjoyed’ thus acted as an effective
music control. Using others’ music as the music control thus enabled control of structural elements of
the experimental music, such as loudness, tempo, pitch, or melodic contour, which may have influenced
arousal and memory.
7.1.2 Chills as a measure of music induced emotion
A review of the music induced chills, thrills, and frisson literature conducted by Huron and
Margulis (2010) indicates that the phenomenon is related to pleasure and autonomic nervous system
activation. The chills response could thus potentially be a simple measure of music induced ANS
activation in ecological settings. Given the theoretical importance of the autonomic nervous system in
neurobiological modulation of LTM consolidation, the physiological correlates of chills were examined in
the second and third experiments of the current project. Analysis of the chills data revealed that
personally selected enjoyed music elicited the most chills, which were associated with enjoyment and
subjective feelings of arousal. Chills were also experienced for positive and arousing experimenterselected music, played in its original form and when overlayed to create a random music mix, and
162
___________________________________________________________________________________
others’ enjoyed music. However, despite the frequency of chills reported, there was no evidence of
chills-related increases in autonomic activity. Of note, chills experienced while listening to others’ music
was not associated with music enjoyment. These results indicated that a) chills could be experienced
while listening to music that was not necessarily enjoyed; and b) the chills experienced were not
detected by the autonomic measures.
The experience of chills in response to music without concomitant feelings of enjoyment may be
explained by structural elements of the music. Huron and Margulis (2010) point to a number of musical
features that can elicit chills17. For instance, chills can be elicited by unexpected changes in the
progression of the music (abrupt change in tempo or rhythm or a new harmony), or by changes in
energy (rapid increase in loudness or broadening of the frequency range). Chills experienced while
listening to others’ music could therefore have been caused by the unpredictable/unexpected
perceptual elements of the music. This non-emotional aspect of the chills response needs to be
carefully considered when using music-induced chills as an index of emotion.
The failure to detect a relationship between chills and autonomic nervous system activity may
be due to the method of measurement used in the current study, and to the intensity of the chillsinducing music. Chills research using time-series analysis has consistently identified significant
correlations between the chills inducing epochs within music excerpts and physiological responses (e.g.
Grewe et al., 2007; Guhn, Hamm, & Zentner, 2007; Salimpoor et al., 2009). The analysis of means for
the entire music excerpt in the current research may therefore have failed to detect fluctuations in
physiological responses that accompanied specific chills sections of the music. Notwithstanding,
detection of physiological change using means for full music excerpts has been demonstrated to be
effective in previous research in our laboratory (Rickard, 2004).
Chills intensity may have also been a factor that prevented detection of autonomic activation in
the current research. Rickard (2004), for example, instructed participants to select one musical piece
that was ‘emotionally powerful and moving, and personally meaningful’, with no other criteria for
selection. Participants who provided their own music in the current research project were instructed to
select two music tracks that were intensely and consistently enjoyed, preferably unusual or eclectic,
17
Note that the phenomenon is described by Huron and Margulis as frisson, experienced as thrills from
music that are not associated with piloerection (gooseflesh or goosebumps). The phenomenon will be
described as chills in this research project to be consistent with the majority of studies.
163
___________________________________________________________________________________
chills inducing, without lyrics, not associated with episodic memory, and similar in energy level. The
selection criteria were therefore quite stringent compared to that set by Rickard, and may have
prevented participants from selecting music that induced intense emotional responses. Informal
feedback from study participants on the difficulty of the music selection task is consistent with this
interpretation. The Experiment 2 music selections may have therefore been less intense than those
elicited in Rickard’s study.
Further analysis of the Experiment 2 and 3 chills data partially supports the chills intensity
hypothesis. When the chills data were split into high and low chills frequency responders, there was a
trend for high chills responders to experience faster heart rhythm and greater vasoconstriction relative
to low responders (see Appendix M for the statistical test results). Based on previous reports of the
positive relationship between chills and skin conductance responses (e.g. Grewe et al., 2007), it was
expected that high chills responders would also experience more skin conductance responses than low
chills responders. The data revealed effects opposite to those expected. Low chills responders to both
personally selected music and others’ music experienced more skin conductance responses than high
chills responders. This result implies that music that elicits intense chills reduces phasic arousal (mean
decreases in SCR) while simultaneously increasing tonic arousal (mean increases in heart rhythm). It
therefore appears that music that elicits a high chills response is physiologically activating without being
startling. This could be a physiological profile that is unique to emotion elicited by music.
The relationship between chills and noradrenergic modulation of heart rhythm and
vasoconstriction indicates that chills music activates the mechanism that underlies EEM. Perhaps then,
an effective emotion stimulus for memory enhancement is intense chills music. Some caveats should,
however, be addressed. First, chills were also experienced by participants in this study while listening to
music that was rated as neutral. Thus, the experience of chills does not always index emotional arousal.
Second, to elicit sufficient ANS activation for EEM, the chills experience must be intense. The most
effective method of eliciting intense chills responses is to ask participants to provide their own music
(without restriction). This, however, raises the problem of experimental control as it is difficult to
determine whether emotion is elicited by the music of itself, or other variables such as associated
emotional memories or familiarity. In this case, music may not be unique as any stimulus that was
associated with emotional memories or was familiar could be used (see for examples the research
conducted by Thompson, Schellenberg and colleagues who demonstrate that non-music enjoyed stimuli
can have similar effects as music on cognitive tasks). A final caveat is that musical chills are not
164
___________________________________________________________________________________
experienced by everyone, and when they are, they are more likely to be experienced by females (Huron
& Margulis, 2010), thus limiting the generalizability of chills music as an emotion stimulus.
Nevertheless, the influence of intense chills music on EEM for a restricted population may still be a
fruitful avenue of future investigation.
This section has demonstrated that the emotion stimuli used in this research project have
effectively elicited subjective, autonomic, and cognitive processing responses that are consistent with
emotion. The unexpected emotion response to the control stimuli in Experiment 3 was attributed to the
qualitatively different experience of listening to a stimulus while behavioural responses were being
closely monitored, and to pleasure elicited by violation of musical expectations. Chills were reliably
elicited by participant-selected enjoyed music, and when sufficiently intense, elicited autonomic activity.
Participant-selected enjoyed music was also the most effective emotion stimulus, indicating that chills
music may be a valid emotion stimulus for future EEM research.
7.2 Neurobiological modulation of memory
One of the aims of this research project was to determine whether emotional music enhanced
memory consolidation via a physiological arousal mechanism. There was some evidence of
noradrenergic modulation of memory in Experiments 2 and 3, indexed by changes in cardiac activity. In
Experiment 2, reductions in heart rate variability predicted increasing early stage LTM scores, and in
Experiment 3, reduced heart rate during the consolidation period was associated with attenuation of
long-term EEM. The use of participant-selected enjoyed music in Experiment 2 was the most effective
method of facilitating memory, and as noted in the previous section, also elicited the highest frequency
of chills (an index of autonomic activation). Although the interval between encoding and memory
testing was within 1 hour, the use of distractor tasks after each condition ensured that the pattern of
memory performance could be attributed to the early stages of LTM consolidation, and not to
maintenance of the MTBR in short-term memory. Participant-selected music was the only condition of
the three used in this experiment to reveal a significant relationship between physiological arousal,
indexed by reduced parasympathetic nervous system control of heart rate, and memory (refer to
Appendix F). The observed relationship indicates that facilitated memory could be attributed to
165
___________________________________________________________________________________
McGaugh’s model of EEM. The results of Experiment 2 thus raise the possibility that participant-selected
music can be used as an effective emotion treatment to facilitate memory consolidation.
Nevertheless, cortical and autonomic activation elicited by other participants’ music (and
previously rated with emotional indifference) also predicted memory. Cortical and autonomic
activation devoid of emotional valence could therefore also account for facilitated early stage LTM. This
finding may be explained by an attention mechanism, whereby interesting or novel information elicits
an orienting reflex, perhaps regulated by the amygdala (see Section 2.1.2), leading to activation of brain
stem structures that increase cortical arousal and facilitate stimulus processing. However, emotional
arousal elicited by participant-selected music, indexed by increased autonomic arousal and positive
mood valence, was more effective at enhancing memory than non-emotional increases in arousal. The
long term effects of music induced general arousal (in contrast to emotional arousal) on memory are in
need of further investigation.
The Experiment 3 conditions yielded some interesting findings. When a stimulus presented
soon after learning was emotional, indexed by increases in negative mood valence or increases in frontal
theta activity, memory consolidation was impaired. This implies that post-encoding emotion processing
can interfere with the labile memory trace. It may thus be pertinent to have a buffer period after
learning to enable the memory trace to establish before attempting to strengthen its consolidation with
an emotion treatment. For instance, Judde and Rickard (2010) found that post-encoding presentation of
emotionally powerful positive and negative music immediately after wordlist learning had no influence
on LTM, whereas memory benefits were observed when the music was presented after a 20 minute
delay.
Based on the results of Experiment 3 and results presented by Judde and Rickard (2010), there
may be a critical period after learning that the emotion manipulation could occur. If the emotion
manipulation occurs immediately after learning it may interfere with the labile memory trace, and if too
delayed (30-45 minutes), the neuromodulation-sensitive phase of the memory trace may have decayed.
The ideal post-encoding emotion treatment period may therefore be between 5 and 30 minutes.
Arousal levels during the encoding period may also influence the effectiveness of the post-encoding
emotion treatment. According to the neurobiological model of EEM presented in Section 2.1, emotional
responses are regulated by the amygdala, which has projections to brain stem structures that regulate
noradrenergic projections throughout the cortex, including the BLA, and the sympathetic nervous
system (refer to Figure 2.1). Noradrenergic projections from the brain stem to the BLA strengthen
166
___________________________________________________________________________________
amygdala-hippocampus connections – which could be considered the ‘first wave’ of the emotional
arousal response. A second wave of activation occurs when brain stem activation leads to the release of
arousal hormones and neuromodulators from the adrenal gland, ultimately activating the BLA and
further strengthening amygdala-hippocampus connections. If the initial memory trace was weak, as
would be expected for everyday information which by necessity decays quickly, the second wave of
arousal hormones elicited by the post-encoding emotion stimulus would have little or no memory trace
to consolidate (see the memory tagging literature, e.g., Bergado, Lucas, & Richter-Levin, 2011; Cahill &
Alkire, 2003). It is therefore recommended that there be some degree of arousal associated with the
information to be remembered.
The combined results of this research project and those of Judde and Rickard (2010) provide
support for music induced neuromodulation of EEM. In the current research, memory changes were
associated with music-induced physiological activity, thus supporting McGaugh’s neurobiological model.
In Judde and Rickard’s study, LTM was successfully facilitated with a post-learning emotional music
treatment. There is therefore promise for emotionally powerful music to be an effective LTM
modulator.
7.3 Free recall and recognition memory differences
Little consistency in free recall and recognition memory was observed across the three
experiments. In Experiment 1, a recognition memory test was sensitive to changes in individual memory
performance, whereas free recall was not. The reverse occurred in Experiment 2, and in Experiment 3,
differences were observed in both types of memory test. The failure to detect differences in recognition
memory in Experiment 2 was attributed to an image memory ceiling effect. One hundred and twenty
images were presented in five separate image collages, and recognition of the images was tested within
60 minutes. Participants were presented with one matched foil per target image. It is possible that this
method of memory testing was not challenging enough to differentiate good and poor memory
performers. It is therefore recommended that image recognition memory tests either increase the ratio
of foils to target images, or extend the delay period before testing.
Recognition memory testing was more fruitful in Experiment 3, in which recognition for target
words embedded within matched word foils was tested. Clear trends in the good versus poor memory
data in this experiment included the relationship between heart rhythm and recognition memory, and
167
___________________________________________________________________________________
frontal cortical activation and free recall memory. These simple differences in memory type and
physiological response provide some support for the argument that noradrenergic modulation of
memory is best detected with recognition memory tests, while the influence of cognitive processes on
memory are best detected with free recall tests (Eich & Forgas, 2003; Kenealy, 1997; Rimmele, Davachi,
& Phelps, 2012).
7.4 Methodological considerations
There were a number of statistical and methodological limitations across the three experiments
that challenged interpretation of the outcomes (see Haslam & McGarty, 2000, for a description of
statistical and methodological uncertainty). The single most influential contributor to statistical
uncertainty was sample size. There were nine participants per condition in Experiment 1, and 15 per
condition in Experiment 3, which reduced to seven after splitting the data according to good and poor
memory performers. The power of the inferential tests to detect effects was thus low. Statistical power
was improved in Experiment 2 by using a within-subjects design and larger sample (N = 36).
Nevertheless, due to the nature of neurobiological modulation of EEM, the influence of circulating
arousal hormones elicited by one condition on memory for subsequent conditions could not be entirely
eliminated. Therefore, the increase in statistical certainty when using a within-subjects design was
offset by the increase in methodological uncertainty caused by the potential confound of latent arousal
hormone effects on memory. The obvious solution to this methodological limitation was to increase
sample size. However, the sample size obtained was the maximum achievable given time and resource
constraints. The current data should thus be considered pilot in nature.
Methodological limitations identified across the three studies were sample population, study
design intensity, participant reactivity, and artificiality. Study design intensity refers to the number of
variables measured and controlled in each experiment. For instance, previous studies of EEM have at
the minimum tested the effect of subjective feelings of emotion on memory, a sample of these have
also tested physiological arousal effects, and others have tested the influence of cognitive processes on
memory. It is rare to find studies that investigate all three constructs within the one experiment.
Participant reactivity refers to the phenomenon of changes in behaviour elicited by involvement in an
experiment that influence the construct under investigation (described in Section 6.4.3). Artificiality is
168
___________________________________________________________________________________
defined as the oversimplification of complex phenomena (e.g., emotional responses to music), thereby
depriving them of their meaning and richness (Haslam & McGarty, 2000).
The methodological limitations of population sample and design intensity were understood
during the study design stage, whereas the effects of participant reactivity and artificiality on study
outcomes were unexpected. With reference to the sample population, the participant pool was
comprised mostly of female undergraduate psychology students. The nature of the study would have
also attracted a higher proportion of individuals with an invested interest in music. Generalisation of
the results of this research are therefore limited to younger, educated, mostly female individuals who
are likely to have an interest in music. Further research is required using more representative samples
from the general population, particularly at the level of music interest.
The experiments conducted in this research project were more intensive than those previously
conducted. There is an abundance of research investigating the relationship between music and
emotion. Likewise, there is an abundance of research investigating the relationship between emotion
elicited by stories and images, and memory. It is less common to find studies that investigate the
combined effects of music and emotion on memory (Judde & Rickard, 2010; Rickard et al., 2012).
Intensive study design was therefore necessary given the range of variables that can influence both
emotional responses to music and memory performance. A wide range of emotion responses were thus
measured across the three studies, the timing of the emotion treatment was varied, as was the type of
information to be remembered and the delay interval before memory testing. Included in the design of
each experiment were several measures of individual differences. The influence of individual
differences on emotion and memory were scrutinised in detail (presented in the supplementary analysis
in Appendix L), but failed to yield any consistent or meaningful results. Changing key variables across
experiments allowed a logical development of the research, but compromised interpretation of the
effects. As a result, the research may be considered more exploratory in nature. Nevertheless, trends in
the research outcomes have been illuminating and could pave the way for further development of
theories of emotional response to music and EEM.
The methodological limitation of participant reactivity was unexpected. In Experiment 2 and 3,
participants were informed that the research was designed to examine the relationship between music
induced emotion and memory. Knowledge of memory testing may therefore have increased motivation
and effort to perform well, thus decreasing our ability to detect effects elicited by the experimental
stimuli. Increased motivation to perform well may also explain memory differences between
169
___________________________________________________________________________________
Experiment 3 and previous research using the same word-list learning procedure. It is therefore
recommended that participants remain naïve to the memory aims of the research and are debriefed
after memory testing has taken place.
Another aspect of participant reactivity that was not expected was the emotional response to
stimuli that were considered neutral during the study design stage. In Experiment 2, there was an
arousal response to others’ music selections previously rated by the same participants as neutral, and in
Experiment 3, there was an emotional response to the non-music and music controls. The emotional
response may have been caused by the context in which the stimulus was presented. For instance, the
experimental laboratory may have contributed to participants feeling nervous, uncertain, judged,
watched, determined, curious, uncomfortable, or confused. Perhaps these feelings would be absent if
participants were unaware or unconcerned with the response monitoring. A possible resolution to this
difficult to resolve aspect of experimental design is to use the Experience-Sampling Method (ESM), in
which individuals can be monitored without substantial interference in their normal lifestyles. ESM has
already been used to great effect in the music-emotion research field (e.g., Juslin et al., 2008; Sloboda &
O'Neill, 2001). The ESM could be complemented with ambulatory heart rate monitoring to determine
whether memory for specific episodes was related to concurrent changes in autonomic activity.
The experimental method also raises the problem of artificiality, in which reduction of the
phenomenon under investigation to measurable components can render the phenomenon meaningless.
The most obvious example of artificiality in the current research project was the emotional story used in
Experiment 1. In an attempt to control visual and linguistic confounds of EEM, the emotional narrative
became a shadow of its intended emotional intensity. In Experiment 3, control conditions designed to
account for differences in attention, engagement, and music structure became emotionally arousing and
may have facilitated memory. It has therefore proven to be difficult to find the right balance of
experimental control and ecological realism to successfully manipulate emotion and memory in the
research laboratory.
In sum, the degree of statistical and methodological uncertainty in the findings of this research
project was higher than expected. The power of the inferential tests was compromised by low
participant numbers, and generalising the results to the general population is limited by the narrow
demographic of the participant cohort. The aim of developing a methodology for facilitating memory
with music as an emotional stimulus in the general population has therefore not been fully achieved.
Despite these limitations, the intensive nature of the research resulted in identification of a broad range
170
___________________________________________________________________________________
of promising trends in the data which can be used for further theory development and hypothesis
testing. These will be discussed in the next section.
7.5 Conclusions and future directions
The primary aim of this research project was to explore whether positive emotion elicited by an
emotionally powerful stimulus (music) could be utilized to facilitate long-term memory for unrelated
material. The underlying motivation for this research was to extend previous research on EEM into a
context that would have greater ecological application, such that practitioners could use emotional
stimuli in an evidence-guided manner to promote enhanced consolidation of everyday events (for
instance, in age-related impairment). A range of methodologies and theories were drawn upon to
elucidate the causes of emotion effects on memory. A number of variables that could moderate the
relationship between emotion and memory were also identified and controlled. With stringent
methodology, the outcomes of this research revealed that even though music appeared to elicit
authentic emotions, the effects of emotional music on LTM were inconsistent and could not be precisely
defined. The clearest results were gained from Experiment 2, in which the emotion inducing properties
of music were established, as defined by Scherer’s multiple component model of emotion (2001), with
participant-selected emotional music enhancing memory for subsequently presented images. There
were trends in the Experiment 3 data that suggested music presented during the consolidation period
modulated memory. This result was consistent with studies in which neuromodulators administered
after learning enhance or attenuate emotional memory.
Across experiments, there were recurring trends in the physiological data to suggest that the
noradrenergic system, indexed by increased cardiac activity, was associated with long-term memory. As
the noradrenergic system has been strongly implicated in the neurobiological modulation of EEM, this
outcome supports neurobiological modulation of memory consolidation in the current project.
Nevertheless, the reliability of these music-induced noradrenergic-memory associations needs to be
established with replication.
During the course of this research project, some prevailing assumptions were challenged. The
assumption of previous EEM in humans has been that a three-story-phase emotional slideshow
depicting a boy who is critically injured elicited an emotion, which is regulated by the amygdala, and
171
___________________________________________________________________________________
resulted in ANS activation and adrenal release of arousal hormones necessary for neurobiological
modulation of memory. The methodology was adapted in the current project to accommodate music,
to control for an orienting reflex explanation of memory, and to apply stricter control on the linguistic
properties of the stories. In doing so, the distinct narrowing of memory for the emotional phase of the
slideshow was eliminated. It was therefore proposed that previous memory effects could have been
caused by an amygdala mediated orienting reflex, rather than a full emotional response. As an orienting
reflex can be argued to be a component of emotion, this proposal should be of little concern. However,
when considering replication and generalisation of the findings, eliciting an orienting reflex may be a
simpler and more reliable task than eliciting a full emotional response. It may therefore be worthwhile
extending this avenue of research by controlling the intensity of the stimulus to determine the influence
on LTM (refer to Table 7.2 for a summary of recommendations arising from this project). The same
three-story-phase methodology developed in the current research also revealed that cognitive
processes that were independent of increased physiological arousal could account for the observed
memory improvement. This raises the possibility that memory improvements that have previously been
attributed to the neurobiological model of EEM may not have fully accounted for the influence of
cognitive processes on memory.
One of the most promising results of this research project has been the consistent evidence of
music effects on multiple components of emotion, especially noradrenergic modulation of cardiac
activity. Of particular interest was the chills response. The reliable relationship between chills
responses and autonomic activity reported in previous research, combined with evidence that intense
chills were associated with increased cardiac activity in the current research, indicates that chills music
may be a reliable source of emotional arousal to manipulate EEM. Future EEM research could thus limit
the emotion stimulus to music that reliably induces a chills response. This will limit generalisation to a
population that experience chills from music. Nevertheless, initial testing in this sub population may
reduce the variability in music-induced emotional responding and improve the probability of detecting
music-induced memory effects.
Participant naivety was observed as being an important factor to consider when testing
memory. It appeared that in this project, knowledge of memory testing had an adverse effect on
responses to the experimental stimuli. Participants were made aware of the memory testing tasks in
the second and third experiments to increase engagement with the task and as a method of controlling
for differences in memory ability (the ‘memory test then re-test after a delay’ procedure). In this
172
___________________________________________________________________________________
context, naivety was not considered essential as rehearsal, and thus maintenance of material in shortterm memory, was prevented by other means. However, knowledge of the memory testing may have
increased endogenous arousal levels and therefore influenced the manipulation of emotional arousal
with the experimental stimuli. It is therefore recommended that participants remain naïve to the true
memory aims of the study, even if rehearsal has been controlled.
A final noteworthy observation was the memory-impairing influence of emotional stimuli
presented soon after learning. In terms of neurobiological modulation of memory, it is possible that the
newly acquired memory trace requires uninterrupted time to establish before it becomes available for
strengthening by the post-encoding arousal treatments. The exact duration of this interval has yet to be
established, although research conducted by Nielson and Lorber (2009) and Judde and Rickard (2010)
implies that the window could be between 5 and 30 minutes post-encoding. The arousal-inducing
properties of the remembered material should also be considered, given evidence that arousal elicited
during encoding can interact with post-encoding arousal treatments (Cahill & Alkire, 2003). Future
research could test the interaction between arousal elicited by the remembered material and the timing
of the post-encoding arousal treatment to yield a more fine-grained temporal profile of the
neurobiological memory effect.
173
___________________________________________________________________________________
Table 7.3
Summary of recommendations
Recommendations for music use in experimental studies
Music


Emotion
measurement




Memory
measurement


Controls





Design



Context


Enjoyable and chills inducing
Valence expressed by the music can be positive or negative
Use music induced chills as an index of ANS activation
Measures state changes in affective valence with psychometrically validated
scales (e.g., PANAS)
Reduce noise in cardiac activity data by using ECG
Confirm the effectiveness of the emotion manipulation with measures of
multiple components of emotion, particularly those related to emotional
valence
Use free recall memory tests to measure top-down effects of emotion on
memory
To avoid memory ceiling effects, increase the ratio of foils to targets in
recognition memory tests or extend the delay period before testing
Use effective active listening non-music control conditions to establish that
observed effects can be attributed to the unique properties of music
Comprehensively pilot test control conditions to ensure they do not
inadvertently elicit emotion responses during the experimental procedure
Use a repeated measures design so that participants act as their own control
Music that is restructured so that it can act as a music control should use
excerpts that eliminate recognisable structural progressions
Control individual differences in music enjoyment
Ensure participants are naive to the intended emotion manipulation and to
memory measurement
Use large samples to enable examination of sub-samples that differ according to
extraneous variables (e.g., memory ability, arousal sensitivity, and music
engagement).
To ensure participants can fully engage with and respond to the music, present
music before or after the dependent variable
Reduce extraneous arousal by allowing sufficient time for participants to
familiarise themselves with the experimental environment
Use ambulatory methods to measure emotional responses and memory in
ecological settings (e.g., wireless hand held reporting devices, heart rate
monitor)
174
___________________________________________________________________________________
Recommendations for music use to facilitate memory in practice

To ensure the emotion treatment is sufficiently activating (arousal dose-response), consider using
participant-selected enjoyed music

Where possible, use music that reliably induces a chills response

Allow individuals to fully attend to the music by presenting it before or after the information to be
remembered.
Recommendations for future research

Manipulate the timing of the post-encoding emotion treatment to yield a more fine-grained
temporal profile of the neurobiological memory effect

Examine whether appraisal of emotional information presented immediately after learning
interferes with memory consolidation

Examine whether arousal during encoding is a necessary element of emotion-enhanced memory by
varying the arousal of the remembered material and of the emotion manipulation

Examine whether an orienting reflex is a sufficient element of emotion-enhanced memory by
varying the perceptual intensity of the emotion stimulus

Examine whether intense chills elicited by music predicts amygdala activation and long-term
memory

Develop an effective method of operationalizing memory scope to enable the testing of positive
emotion on the breadth of information remembered
Anecdotal reports indicate that music is widely used in everyday settings to improve cognitive
processing performance. However, the evidence base to support the efficacy of music use in this way is
still in its infancy and in need of validation. This research project has added to the evidence base by
demonstrating that music does indeed have utility in enhancing performance of cognitive processes, but
there is still much to be learnt. Despite the complex nature of emotion, further complicated by the
experimental context in which it is manipulated, this research project has yielded some clear effects of
music on emotion and memory. The commonly agreed upon notion that music elicits emotions was
supported, thereby adding further empirical evidence of the emotion-inducing properties of music.
175
___________________________________________________________________________________
Emotional music may therefore be a powerful stimulus for further understanding the characteristics of
emotion. Music also had a modulating influence on memory. However, one of the effects observed was
the potential for emotional music to interfere with mechanisms that underlie memory consolidation.
The search must therefore continue to articulate the conditions under which music is of benefit or
detriment to cognitive processes. The methodological rigor applied in conducting this research has
provided an essential foundation from which this search can proceed.
176
8 REFERENCES
Abercrombie, H. C., Chambers, A. S., Greischar, L., & Monticelli, R. M. (2008). Orienting,
emotion, and memory: phasic and tonic variation in heart rate predicts memory for
emotional pictures in men. [Research Support, Non-U.S. Gov't]. Neurobiology of
Learning & Memory, 90(4), 644-650.
Adolphs, R., & Damasio, A. R. (2001). The interaction of affect and cognition: A neurobiological
perspective. In J. P. Forgas (Ed.), Handbook of affect and social cognition (pp. 27-49). NJ,
US: Lawrence Erlbaum Associates Publishers.
Adolphs, R., Denburg, N. L., & Tranel, D. (2001). The amygdala's role in long-term declarative
memory for gist and detail. Behavioral Neuroscience, 115(5), 983-992.
Adolphs, R., Tranel, D., & Denburg, N. (2000). Impaired emotional declarative memory following
unilateral amygdala damage. Learning & Memory, 7(3), 180-186.
Altenmuller, E., Schurmann, K., Lim, V. K., & Parlitz, D. (2002). Hits to the left, flops to the right:
different emotions during listening to music are reflected in cortical lateralisation
patterns. Neuropsychologia, 40(13), 2242-2256.
Anderson, A. K., Wais, P. E., & Gabrieli, J. D. (2006). Emotion enhances remembrance of neutral
events past. Proc Natl Acad Sci U S A, 103(5), 1599-1604.
Ashby, F., Isen, A. M., & Turken. (1999). A neuropsychological theory of positive affect and its
influence on cognition. Psychological Review, 106(3), 529-550.
Bachorik, J. P., Bangert, M., Psyche, L., Larke, K., Berger, J., Rowe, R., & Schlaug, G. (2009).
Emotion in motion: Investigating the time-course of emotional judgments of musical
stimuli. Music Perception, 26(4), 355-364.
Baddeley, A. (2005). Human Memory: Theory and Practice. Hove: Psychology Press Ltd.
Balch, W. R., Bowman, K., & Mohler, L. A. (1992). Music-dependent memory in immediate and
delayed word recall. Memory & Cognition, 20(1), 21-28.
Balch, W. R., & Lewis, B. S. (1996). Music-dependent memory: The roles of tempo change and
mood mediation. Journal of Experimental Psychology: Learning, Memory, and Cognition,
22(6), 1354-1363.
Balch, W. R., Myers, D. M., & Papotto, C. (1999). Dimensions of mood in mood-dependent
memory. Journal of Experimental Psychology: Learning, Memory & Cognition, 25(1), 7083.
Baldi, E., & Bucherelli, C. (2005). The inverted "u-shaped" dose-effect relationships in learning
and memory: modulation of arousal and consolidation. Nonlinearity Biol Toxicol Med,
3(1), 9-21. doi: 10.2201/nonlin.003.01.002
Ball, P. (2010). The Music Instinct: How music works and why we can't do without it. Oxford;
New York: Oxford University Press.
Barbas, H., Zikopoulos, B., & Timbie, C. (2011). Sensory Pathways and Emotional Context for
Action in Primate Prefrontal Cortex. Biological Psychiatry, 69(12), 1133-1139. doi:
10.1016/j.biopsych.2010.08.008
177
___________________________________________________________________________________
Baumgartner, T., Esslen, M., & Jancke, L. (2006a). From emotion perception to emotion
experience: Emotions evoked by pictures and classical music. [peer reviewed].
International Journal of Psychophysiology, 60, 34-43.
Baumgartner, T., Lutz, K., Schmidt, C. F., & Jancke, L. (2006b). The emotional power of music:
How music enhances the feeling of affective pictures. Brain Research, 1075, 151-164.
Benzon, W. (2001). Beethoven's Anvil: Music in Mind and Culture. New York: Basic Books.
Bergado, J. A., Lucas, M., & Richter-Levin, G. (2011). Emotional tagging-A simple hypothesis in a
complex reality. Progress in Neurobiology, 94(1), 64-76.
Bernardi, L., Porta, C., & Sleight, P. (2006). Cardiovascular, cerebrovascular, and respiratory
changes induced by different types of music in musicians and non-musicians: The
importance of silence. Heart, 92(4), 445-452.
Berntson, G. G., Quigley, K. S., & Lozano, D. (2007). Cardiovascular Physiology. In J. T. Cacioppo,
L. G. Tassinary & G. G. Bernston (Eds.), Handbook of Psychophysiology (3rd ed., pp. 182210). New York: Cambridge University Press.
Bigand, E., & Poulin-Charronnat, B. (2006). Are we "experienced listeners"? A review of the
musical capacities that do not depend on formal musical training. Cognition Vol 100(1)
May 2006, 100-130.
Bigand, E., Vieillard, S., Madurell, F., Marozeau, J., & Dacquet, A. (2005). Multidimensional
scaling of emotional responses to music: The effect of musical expertise and of the
duration of the excerpts. Cognition & Emotion, 19(8), 1113-1139.
Bloise, S. M., & Johnson, M. K. (2007). Memory for emotional and neutral information: gender
and individual differences in emotional sensitivity. Memory, 15(2), 192-204.
Blood, A. J., & Zatorre, R. J. (2001). Intensely pleasurable responses to music correlate with
activity in brain regions implicated in reward and emotion. Proc Natl Acad Sci U S A,
98(20), 11818-11823.
Blumstein, D. T., Bryant, G. A., & Kaye, P. (2012). The sound of arousal in music is
contextdependent. Biology Letters, 8(5), 744-747.
Boltz, M., Schulkind, M., & Kantra, S. (1991). Effects of background music on the remembering
of filmed events. Memory & Cognition, 19(6), 593-606.
Boltz, M. G. (2001). Musical soundtracks as a schematic influence on the cognitive processing of
filmed events. Music Perception, 18(4), 427-454.
Bornstein, R. F. (1989). Exposure and affect: Overview and meta-analysis of research, 19681987. Psychological Bulletin, 106(2), 265-289.
Bouhuys, A. L., Bloem, G. M., & Groothuis, T. G. G. (1995). Induction of depressed and elated
mood by music influences the perception of facial emotional expressions in healthy
subjects. Journal of Affective Disorders, 33(4), 215-226.
Bouhuys, A. L., Geerts, E., & Gordijn, M. C. (1999). Depressed patients' perceptions of facial
emotions in depressed and remitted states are associated with relapse: A longitudinal
study. Journal of Nervous and Mental Disease, 187(10), 595-602. doi:
http://dx.doi.org/10.1097/00005053-199910000-00002
Bower, G. H. (1981). Mood and memory. American Psychologist, 36(2), 129-148.
178
___________________________________________________________________________________
Bower, G. H. (1992). How might emotions affect learning? In S.-A. Christianson (Ed.), The
handbook of emotion and memory: Research and theory (pp. 3-31). Hillsdale, NJ,
England: Lawrence Erlbaum Associates, Inc.
Boyle, R., & Coltheart, V. (1996). Effects of irrelevant sounds on phonological coding in reading
comprehension and short-term memory. The Quarterly Journal of Experimental
Psychology A: Human Experimental Psychology, 49A(2), 398-416.
Bradley, M. M. (2009). Natural selective attention: Orienting and emotion. Psychophysiology,
46(1), 1-11. doi: http://dx.doi.org/10.1111/j.1469-8986.2008.00702.x
Bradley, M. M., & Lang, P. J. (2000). Measuring Emotion: Behavior, Feeling, and Physiology. In R.
D. Lane & L. Nadel (Eds.), Cognitive Neuroscience of Emotion (pp. 242-276). NY: Oxford
University Press.
Brown, S. (2000). The "Musilanguage" model of music evolution. In W. Nils L, Bjorn Merker, and
Steven Brown (Ed.), The origins of music (pp. 271-300). Cambridge: MIT Press.
Brown, S., Martinez, M. J., & Parsons, L. M. (2004). Passive music listening spontaneously
engages limbic and paralimbic systems. Neuroreport: For Rapid Communication of
Neuroscience Research, 15(13), 2033-2037.
Buchanan, T. W., & Adolphs, R. (2004). The neuroanatomy of emotional memory in humans. In
D. Reisberg & P. Hertel (Eds.), Memory and Emotion (pp. 42-75). NY,US: Oxford
University Press.
Buchanan, T. W., Denburg, N. L., Tranel, D., & Adolphs, R. (2001). Verbal and nonverbal
emotional memory following unilateral amygdala damage. Learn Mem, 8(6), 326-335.
Buchanan, T. W., Etzel, J. A., Adolphs, R., & Tranel, D. (2006). The influence of autonomic
arousal and semantic relatedness on memory for emotional words. International Journal
of Psychophysiology, 61(1), 26-33.
Buchanan, T. W., Karafin, M. S., & Adolphs, R. (2003). Selective effects of triazolam on memory
for emotional, relative to neutral, stimuli: Differential effects on gist versus detail.
Behavioral Neuroscience, 117(3), 517-525.
Buchanan, T. W., & Lovallo, W. R. (2001). Enhanced memory for emotional material following
stress-level cortisol treatment in humans. Psychoneuroendocrinology Vol 26(3) Apr
2001, 307-317.
Burke, A., Heuer, F., & Reisberg, D. (1992). Remembering emotional events. Memory &
Cognition, 20(3), 277-290.
Cacioppo, J. T., Berntson, G. G., Larsen, J. T., Poehlmann, K. M., & Ito, T. A. (2000). The
Psychophysiology of Emotion. In M. Lewis & J. M. Haviland-Jones (Eds.), Handbook of
Emotions (2nd ed., pp. 173-191). NY, London: Guilford Press.
Cacioppo, J. T., & Gardner, W. L. (1999). Emotion. Annual Review of Psychology, 50, 191-214.
Cahill, L., & Alkire, M. T. (2003). Epinephrine enhancement of human memory consolidation:
interaction with arousal at encoding. Neurobiology of Learning and Memory, 79(2), 194198.
Cahill, L., Gorski, L., Belcher, A., & Huynh, Q. (2004a). The influence of sex versus sex-related
traits on long-term memory for gist and detail from an emotional story. Consciousness &
Cognition, 13(2), 391-400.
179
___________________________________________________________________________________
Cahill, L., Gorski, L., & Le, K. (2003). Enhanced human memory consolidation with post-learning
stress: Interaction with the degree of arousal at encoding. Learning and Memory, 10(4),
270-274.
Cahill, L., Haier, R. J., Fallon, J., Alkire, M. T., Tang, C., Keator, D., . . . McGaugh, J. L. (1996).
Amygdala activity at encoding correlated with long-term, free recall of emotional
information. Proc Natl Acad Sci U S A, 93(15), 8016-8021.
Cahill, L., & McGaugh, J. L. (1995). A novel demonstration of enhanced memory associated with
emotional arousal. Consciousness and Cognition: An International Journal, 4(4), 410-421.
Cahill, L., & McGaugh, J. L. (1998). Mechanisms of emotional arousal and lasting declarative
memory. Trends in Neurosciences, 21(7), 294-299.
Cahill, L., Prins, B., Weber, M., & McGaugh, J. L. (1994). -Adrenergic activation and memory for
emotional events. Nature, 371(6499), 701-704.
Cahill, L., Uncapher, M., Kilpatrick, L., Alkire, M. T., & Turner, J. (2004b). Sex-related
hemispheric lateralization of amygdala function in emotionally influenced memory: an
FMRI investigation. Learn Mem, 11(3), 261-266.
Cahill, L., & van Stegeren, A. (2003). Sex-related impairment of memory for emotional events
with beta-adrenergic blockade. Neurobiology of Learning & Memory, 79(1), 81-88.
Campolongo, P., Roozendaal, B., Trezza, V., Cuomo, V., Astarita, G., Fu, J., . . . Piomelli, D.
(2009). Fat-induced satiety factor oleoylethanolamide enhances memory consolidation.
PNAS Proceedings of the National Academy of Sciences of the United States of America,
106(19), 8027-8031.
Canli, T., Zhao, Z., Brewer, J., Gabrieli, J. D., & Cahill, L. (2000). Event-related activation in the
human amygdala associates with later memory for individual emotional experience.
Journal of Neuroscience, 20(19), RC99.
Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and affective
responses to impending reward and punishment: The BIS/BAS Scales. J Pers Soc Psychol,
67(2), 319-333.
Cassidy, G., & MacDonald, R. A. (2007). The effect of background music and background noise
on the task performance of introverts and extraverts. Psychology of Music, 35(3), 517537.
Chamberlain, S. R., Müller, U., Blackwell, A. D., Robbins, T. W., & Sahakian, B. J. (2006).
Noradrenergic modulation of working memory and emotional memory in humans.
Psychopharmacology, 188(4), 397-407.
Chapados, C., & Levitin, D. J. (2008). Cross-modal interactions in the experience of musical
performances: physiological correlates. Cognition, 108(3), 639-651.
Charles, S. T., Mather, M., & Carstensen, L. L. (2003). Aging and emotional memory: the
forgettable nature of negative images for older adults. Journal of Experimental
Psychology: General, 132(2), 310-324.
Chin, T. C., & Rickard, N. S. (2010). Nonperformance, as well as performance, based music
engagement predicts verbal recall. Music Perception, 27(3), 197-208.
Christianson, S.-A. (1984). The relationship between induced emotional arousal and amnesia.
Scandinavian Journal of Psychology, 25(2), 147-160.
180
___________________________________________________________________________________
Christianson, S.-A. (1992a). Emotional stress and eyewitness memory: A critical review.
Psychological Bulletin, 112(2), 284-309.
Christianson, S.-A. (1992b). Remembering emotional events: Potential mechanisms. In S.-A.
Christianson (Ed.), The handbook of emotion and memory: Research and theory (pp.
307-340). Hillsdale, NJ, England: Lawrence Erlbaum Associates, Inc.
Christianson, S.-A., Loftus, E. F., Hoffman, H., & Loftus, G. R. (1991). Eye fixations and memory
for emotional events. Journal of Experimental Psychology: Learning, Memory, and
Cognition Vol 17(4) Jul 1991, 693-701.
Coan, J. A., & Allen, J. J. (2004). Frontal EEG asymmetry as a moderator and mediator of
emotion. Biol Psychol, 67(1-2), 7-49.
Cohen, A. J. (2001). Music as a source of emotion in film. Juslin, Patrik N (Ed); Sloboda, John A
(Ed). (2001). Music and emotion: Theory and research.
Craik, F. I. M., & Tulving, E. (1975). Depth of processing and the retention of words in episodic
memory. Journal of Experimental Psychology: General, 104(3), 268-294.
Crawford, H. J., & Strapp, C. M. (1994). Effects of vocal and instrumental music on visuospatial
and verbal performance as moderated by studying preference and personality.
Personality and Individual Differences, 16(2), 237-245.
Davidson, R. J. (1992). Anterior cerebral asymmetry and the nature of emotion. Brain Cogn,
20(1), 125-151.
Davis, W. B., & Thaut, M. H. (1989). The influence of preferred relaxing music on measures of
state anxiety, relaxation, and physiological responses. Journal of Music Therapy, 26(4),
168-187.
Dawson, M. E., Schell, A. M., & Filion, D. L. (2007). The Electrodermal System. In J. T. Cacioppo,
L. G. Tassinary & G. G. Bernston (Eds.), Handbook of Psychophysiology (3rd ed., pp. 159181). New York: Cambridge University Press.
de Groot, A. M. B. (2006). Effects of Stimulus Characteristics and Background Music on Foreign
Language Vocabulary Learning and Forgetting. Language Learning. Vol, 56(3), 463-506.
de l'Etoile, S. K. (2002). The effect of musical mood induction procedure on mood statedependent word retrieval. Journal of Music Therapy, 39(2), 145-160.
de Quervain, D. J., Kolassa, I. T., Ertl, V., Onyut, P. L., Neuner, F., Elbert, T., & Papassotiropoulos,
A. (2007). A deletion variant of the alpha2b-adrenoceptor is related to emotional
memory in Europeans and Africans. Nat Neurosci, 10(9), 1137-1139.
De Quervain, D. J. F., Roozendaal, B., Nitsch, R. M., McGaugh, J. L., & Hock, C. (2000). Acute
cortisone administration impairs retrieval of long-term declarative memory in humans.
Nature Neuroscience, 3(4), 313-314.
Deffenbacher, K. A., Bornstein, B. H., Penrod, S. D., & McGorty, E. (2004). A Meta-Analytic
Review of the Effects of High Stress on Eyewitness Memory. Law and Human Behavior,
28(6), 687-706.
Duncko, R., Cornwell, B., Cui, L., Merikangas, K. R., & Grillon, C. (2007). Acute exposure to stress
improves performance in trace eyeblink conditioning and spatial learning tasks in
healthy men. Learning & Memory, 14(5), 329-335.
181
___________________________________________________________________________________
Eich, E., & Forgas, J. P. (2003). Mood, cognition, and memory. In A. F. Healy & R. W. Proctor
(Eds.), Handbook of psychology: Experimental psychology (Vol. 4, pp. 61-83). Hoboken,
New York: John Wiley & Sons, Inc.
Eich, E., Ng, J. T. W., Macaulay, D., Percy, A. D., & Grebneva, I. (2007). Combining music with
thought to change mood. In J. A. Coan & J. J. Allen (Eds.), Handbook of emotion
elicitation and assessment (pp. 124-136). Oxford; New York: Oxford University Press.
Eich, E., & Schooler, J. W. (2000). Cognition/emotion interactions. Eich, Eric; Kihlstrom, John F;
Bower, Gordon H; Forgas, Joseph P; Niedenthal, Paula M, (2000). Cognition and emotion.
(pp. 3-29). xi, NY, US: Oxford University Press.
Ekman, P. (1992). Facial expressions of emotion: an old controversy and new findings. Philos
Trans R Soc Lond B Biol Sci, 335(1273), 63-69.
Ekman, P., Levenson, R. W., & Friesen, W. V. (1983). Autonomic nervous system activity
distinguishes among emotions. Science, 221(4616), 1208-1210.
Ellis, R. J., & Simons, R. F. (2005). The Impact of Music on Subjective and Physiological Indices of
Emotion While Viewing Films. Psychomusicology, 19(1), 15-40.
Etzel, J. A., Johnsen, E. L., Dickerson, J., Tranel, D., & Adolphs, R. (2006). Cardiovascular and
respiratory responses during musical mood induction. International Journal of
Psychophysiology. Vol, 61(1), 57-69.
Eysenck, H. J. (1987). Arousal and personality: The origins of a theory. In J. Strelau & H. J.
Eysenck (Eds.), Personality Dimensions and Arousal (pp. 1-13). New York: Plenum Press.
Fernald, A. (1993). Approval and disapproval: infant responsiveness to vocal affect in familiar
and unfamiliar languages. Child development, 64(3), 657-674.
Flores-Gutierrez, E. O., Diaz, J. L., Barrios, F. A., Favila-Humara, R., Guevara, M. A., Del RioPortilla, Y., & Corsi-Cabrera, M. (2007). Metabolic and electric brain patterns during
pleasant and unpleasant emotions induced by music masterpieces. Int J Psychophysiol,
65(1), 69-84.
Forgas, J. P. (1995). Mood and judgment: the affect infusion model (AIM). Psychological
Bulletin, 117(1), 39-66.
Frazier, P. A., Tix, A. P., & Barron, K. E. (2004). Testing Moderator and Mediator Effects in
Counseling Psychology Research. Journal of Counseling Psychology, 51(1), 115-134.
Frijda, N. H. (2009). Action tendencies. In D. Sander & K. R. Scherer (Eds.), The Oxford
Companion to Emotion and the Affective Sciences (pp. 1-2). Oxford: Oxford University
Press.
Frijda, N. H., & Scherer, K. R. (2010). Emotion definitions (psychological perspectives). In D.
Sander & K. R. Scherer (Eds.), The Oxford Companion to Emotion and the Affective
Sciences (pp. 142-144). New York: Oxford University Press.
Furnham, A., & Allass, K. (1999). The influence of musical distraction of varying complexity on
the cognitive performance of extroverts and introverts. European Journal of Personality,
13(1), 27-38.
Gable, P. A., & Harmon-Jones, E. (2008a). Approach-motivated positive affect reduces breadth
of attention. Psychol Sci, 19(5), 476-482.
182
___________________________________________________________________________________
Gable, P. A., & Harmon-Jones, E. (2008b). Approach-motivated positive affect reduces breadth
of attention. Psychol Sci, 19(5), 476-482.
Gabrielsson, A. (2001). Emotions in strong experiences with music. Juslin, Patrik N (Ed); Sloboda,
John A (Ed). (2001). Music and emotion: Theory and research.
Gabrielsson, A., & Lindstrom-Wik, S. (2003). Strong experiences related to music: A descriptive
system. Musicae Scientiae Vol 7(2) Fal 2003, 157-217.
Garson, J. (2010). Connectionism. The Stanford Encyclopedia of Philosophy Winter 2010 from
http://plato.stanford.edu/archives/win2010/entries/connectionism
Gasbarri, A., Arnone, B., Pompili, A., Marchetti, A., Pacitti, F., Calil, S. S., . . . Tomaz, C. (2006).
Sex-related lateralized effect of emotional content on declarative memory: an event
related potential study. Behav Brain Res, 168(2), 177-184.
Gerra, G., Zaimovic, A., Franchini, D., Palladino, M., Giucastro, G., Reali, N., . . . Brambilla, F.
(1998). Neuroendocrine responses of healthy volunteers to 'techno-music': relationships
with personality traits and emotional state. [Clinical Trial]. International Journal of
Psychophysiology, 28(1), 99-111.
Gibbs, M. E., & Summers, R. J. (2002). Role of adrenoceptor subtypes in memory consolidation.
Progress in Neurobiology, 67(5), 345-391.
Gold, P. E., Hankins, L., Edwards, R. M., & McGaugh, J. L. (1975). Memory interference and
facilitation with posttrial amygdala stimulation: effect of memory varies with footshock
level. Brain Research, 86(3), 509-513.
Gold, P. E., & McGaugh, J. L. (1975). A single-trace, two-process view of memory storage
processes. In D. Deutsch & J. A. Deutsch (Eds.), Short-term Memory (pp. 355-378). New
York: Academic Press.
Goldberg, L. R., Johnson, J. A., Eber, H. W., Hogan, R., Ashton, M. C., Cloninger, C. R., & Gough,
H. C. (2006). The International Personality Item Pool and the future of public-domain
personality measures. Journal of Research in Personality, 40, 84-96.
Gomez, P., & Danuser, B. (2004). Affective and physiological responses to environmental noises
and music. International Journal of Psychophysiology, 53(2), 91-103.
Grady, C. L., Hongwanishkul, D., Keightley, M., & Lee, W. (2007). The Effect of Age on Memory
for Emotional Faces. Neuropsychology, 21(3), 371-380.
Grewe, O., Kopiez, R., & Altenmuller, E. (2009). The chill parameter: Goose bumps and shivers
as promising measures in emotion research. Music Perception, 27(1), 61-74. doi:
http://dx.doi.org/10.1525/mp.2009.27.1.61
Grewe, O., Nagel, F., Kopiez, R., & Altenmuller, E. (2005). How does music arouse "chills"?
Investigating strong emotions, combining psychological, physiological, and
psychoacoustical methods. Annals of the New York Academy of Sciences, 1060, 446-449.
Grewe, O., Nagel, F., Kopiez, R., & Altenmuller, E. (2007). Emotions over time: synchronicity and
development of subjective, physiological, and facial affective reactions to music.
Emotion, 7(4), 774-788.
Grote, L., Zou, D., Kraiczi, H., & Hedner, J. (2003). Finger plethysmography - A method for
monitoring finger blood flow during sleep disordered breathing. Respiratory Physiology
and Neurobiology, 136(2-3), 141-152.
183
___________________________________________________________________________________
Guhn, M., Hamm, A., & Zentner, M. (2007). Physiological and musico-acoustic correlates of the
chill response. Music Perception, 24(5), 473-483.
Haberlandt, K. (1997). Cognitive Psychology (2nd ed.). Sydney: Allyn and Bacon.
Hallam, S., Price, J., & Katsarou, G. (2002). The effects of background music on primary school
pupils' task performance. Educational Studies. Vol, 28(2), 111-122.
Hamann, S., Monarch, E. S., & Goldstein, F. C. (2002). Impaired fear conditioning in Alzheimer's
disease. Neuropsychologia, 40(8), 1187-1195.
Hamann, S. B., Ely, T. D., Grafton, S. T., & Kilts, C. D. (1999). Amygdala activity related to
enhanced memory for pleasant and aversive stimuli. Nature Neuroscience 2(3), 289-293.
Hamill, R. W., & Shapiro, R. E. (2004). Peripheral Autonomic Nervous System. In D. Robertson
(Ed.), Primer on the Autonomic Nervous System (2nd ed., pp. 20-28). Boston: Academic
Press.
Hargreaves, D. J., & North, A. C. (2010). Experimental aesthetics and liking for music. In P. N.
Juslin & J. A. Sloboda (Eds.), Music and Emotion: Theory, Research, Applications (pp. 515546). Oxford: Oxford University Press.
Haslam, S. A., & McGarty, C. (2000). Doing Psychology: An introduction to research
methodology and statistics. London: Sage Publications Ltd.
Heuer, F., & Reisberg, D. (1990). Vivid memories of emotional events: the accuracy of
remembered minutiae. Memory & Cognition, 18(5), 496-506.
Hirokawa, E., & Ohira, H. (2003). The Effects of Music Listening after a Stressful Task on Immune
Functions, Neuroendocrine Responses, and Emotional States in College Students.
Journal of Music Therapy, 40(3), 189-211.
Hodges, D. A. (2010). Psycho-physiological Measures. In P. N. Juslin & J. A. Sloboda (Eds.), Music
and Emotion: Theory, Research, Applications (pp. 279-311). Oxford: Oxford University
Press.
Howell, D. C. (2002). Statistical Methods for Psychology (5th ed.). Pacific Grove, CA:
Duxbury/Thomson Learning.
Huron, D. (2003). Is music an evolutionary adaptation? In I. Peretz & R. Zatorre (Eds.), The
Cognitive Neuroscience of Music (pp. 57-75). New York, NY: Oxford University Press.
Huron, D. (2006). Sweet Anticipation: Music and the psychology of expectation. Cambridge: MIT
Press.
Huron, D., & Margulis, E. H. (2010). Musical expectancy and thrills. Juslin, Patrik N [Ed], 575604.
Husain, G., Forde Thompson, W., & Schellenberg, E. G. (2002). Effects of Musical Tempo and
Mode on Arousal, Mood, and Spatial Abilities. Music Perception, 20(2), 151.
Iwaki, T., Hayashi, M., & Hori, T. (1997). Changes in alpha band EEG activity in the frontal area
after stimulation with music of different affective content. Perceptual and Motor Skills,
84(2), 515-526.
Iwanaga, M., & Ito, T. (2002). Disturbance effect of music on processing of verbal and spatial
memories. Perceptual and Motor Skills, 94(3), 1251-1258. doi: 10.2466/pms.94.3.12511258
184
___________________________________________________________________________________
Iwanaga, M., & Moroki, Y. (1999). Subjective and physiological responses to music stimuli
controlled over activity and preference. Journal of Music Therapy, 36(1), 26-38.
Janata, P., Tomic, S. T., & Rakowski, S. K. (2007). Characterisation of music-evoked
autobiographical memories. Memory, 15(8), 845-860. doi:
10.1080/09658210701734593
Jancke, L., & Sandmann, P. (2010). Music listening while you learn: No influence of background
music on verbal learning. Behavioral and Brain Functions, 6(1), 3.
Jasper, H. H. (1958). The ten-twenty electrode system of the International Federation.
Electroencephalography and Clinical Neurophysiology 10(2), 371-375.
Jones, M. H., West, S. D., & Estell, D. B. (2006). The Mozart effect: Arousal, preference, and
spatial performance. Psychology of Aesthetics, Creativity, and the Arts, S(1), 26-32. doi:
http://dx.doi.org/10.1037/1931-3896.S.1.26
Judde, S., & Rickard, N. S. (2010). The effect of post-learning presentation of music on longterm word-list retention. Neurobiology of Learning and Memory, 94(1), 13-20.
Juslin, P. N., & Laukka, P. (2004). Expression, Perception, and Induction of Musical Emotions: A
Review and a Questionnaire Study of Everyday Listening. Journal of New Music
Research, 33(3), 217-238.
Juslin, P. N., Liljestrom, S., Vastfjall, D., Barradas, G., & Silva, A. (2008). An experience sampling
study of emotional reactions to music: Listener, music, and situation. Emotion, 8(5), 668683.
Juslin, P. N., Liljestrom, S., Vastfjall, D., & Lundqvist, L.-O. (2010). How does music evoke
emotions? Exploring the underlying mechanisms. In P. N. Juslin & J. A. Sloboda (Eds.),
Handbook of Music and Emotion: Theory, research, applications (pp. 605-642).
Melbourne: Oxford University Press.
Juslin, P. N., & Vastfjall, D. (2008). Emotional responses to music: the need to consider
underlying mechanisms. Behav Brain Sci, 31(5), 559-575.
Kenealy, P. M. (1997). Mood-state-dependent retrieval: The effects of induced mood on
memory reconsidered. The Quarterly Journal of Experimental Psychology A: Human
Experimental Psychology, 50A(2), 290-317.
Kensinger, E. A. (2009). Phases of Influence: How Emotion Modulates the Formation and
Retrieval of Declarative Memories. In M. S. Gazzaniga (Ed.), The Cognitive Neurosciences
(4th ed., pp. 725-737). Cambridge, Mass: MIT Press.
Kensinger, E. A., Garoff-Eaton, R. J., & Schacter, D. L. (2007a). Effects of emotion on memory
specificity in young and older adults. The Journals of Gerontology Series B: Psychological
Sciences and Social Sciences, 62(4), 208-215.
Kensinger, E. A., Garoff-Eaton, R. J., & Schacter, D. L. (2007b). Effects of emotion on memory
specificity: Memory trade-offs elicited by negative visually arousing stimuli. Journal of
Memory and Language, 56(4), 575-591.
Khalfa, S., Dalla Bella, S., Roy, M., Peretz, I., & Lupien, S. J. (2003). Effects of relaxing music on
salivary cortisol level after psychological stress. Annals of the New York Academy of
Sciences, 999, 374-376.
185
___________________________________________________________________________________
Kiger, D. M. (1989). Effects of music information load on a reading comprehension task.
Perceptual & Motor Skills, 69, 531-534.
Kihara, M., Sugenoya, J., & Low, P. A. (2004). Temperature Regulation. In D. Robertson (Ed.),
Primer on the Autonomic Nervous System (2nd ed., pp. 127-129). Boston: Academic
Press.
Kim, W.-S., Yoon, Y.-R., Kim, K.-H., Jho, M.-J., & Lee, S.-T. (2003). Asymmetric activation in the
prefrontal cortex by sound-induced affect. Perceptual and Motor Skills, 97(3, Pt 1), 847854.
Kivy, P. (1990). Music alone: Philosophical reflection on the purely musical experience. Ithaca,
NY: Cornell University Press.
Knight, W. E., & Rickard, N. S. (2001). Relaxing music prevents stress-induced increases in
subjective anxiety, systolic blood pressure, and heart rate in healthy males and females.
Journal of Music Therapy Vol 38(4) Win 2001, 254-272.
Koelsch, S., Fritz, T., Cramon, D. Y. v., Muller, K., & Friederici, A. D. (2006). Investigating emotion
with music: an fMRI study. Hum Brain Mapp, 27(3), 239-250.
Koelsch, S., Siebel, W. A., & Fritz, T. (2010). Funtional Neuroimaging. In P. N. Juslin & J. Sloboda
(Eds.), Handbook of Music and Emotion: Theory, research, applications (pp. 313-344).
Konecni, V. J. (2008). Does music induce emotion? A theoretical and methodological analysis.
Psychology of Aesthetics, Creativity, and the Arts, in press.
Kreutz, G., Bongard, S., Rohrmann, S., Hodapp, V., & Grebe, D. (2004). Effects of Choir Singing or
Listening on Secretory Immunoglobulin A, Cortisol, and Emotional State. Journal of
Behavioral Medicine, 27(6), 623-635.
Krumhansl, C. L. (1997). An exploratory study of musical emotions and psychophysiology.
Canadian Journal of Experimental Psychology Vol 51(4) Dec 1997, 336-352.
Kuhlmann, S., & Wolf, O. T. (2006). Arousal and cortisol interact in modulating memory
consolidation in healthy young men. Behavioral Neuroscience, 120(1), 217-223.
Laney, C., Campbell, H. V., Heuer, F., & Reisberg, D. (2004). Memory for thematically arousing
events. Memory & Cognition, 32(7), 1149-1159.
Laney, C., Heuer, F., & Reisberg, D. (2003). Thematically-induced arousal in naturally-occurring
emotional memories. Applied Cognitive Psychology, 17(8), 995-1004.
Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (2005). International affective picture system
(IAPS): Digitized photographs, instruction manual and affective ratings. Technical Report
A-6. Gainesville, FL: University of Florida.
LeDoux, J. (2007). The amygdala. Current Biology, 17(20), R868-R874.
LeDoux, J. E., & Phelps, E. A. (2000). Emotional networks in the brain. In M. Lewis & J. M.
Haviland-Jones (Eds.), Handbook of Emotions (2nd ed., pp. 157-172). NY: Guilford Press.
Levine, L. J., & Edelstein, R. S. (2009). Emotion and memory narrowing: A review and goalrelevance approach. Cognition & Emotion, 23(5), 833 - 875.
Levine, L. J., & Pizarro, D. A. (2004). Emotion and memory research: A grumpy overview. Social
Cognition, 22(5), 530-554.
186
___________________________________________________________________________________
Levitin, D. J. (2008). The World in Six Songs: How the musical brain created human nature. New
York: Penguin Group.
Lin, Y. P., Duann, J. R., Chen, J. H., & Jung, T. P. (2010). Electroencephalographic dynamics of
musical emotion perception revealed by independent spectral components.
Neuroreport, 21(6), 410-415.
Lingham, J., & Theorell, T. (2009). Self-selected "favourite" stimulative and sedative music
listening--how does familiar and preferred music listening affect the body? Nordic
Liu, D. L. J., Graham, S., & Zorawski, M. (2008). Enhanced selective memory consolidation
following post-learning pleasant and aversive arousal. Neurobiology of Learning and
Memory Vol 89(1) Jan 2008, 36-46.
Low, P. A. (2004). The Sweat Gland. In D. Robertson (Ed.), Primer on the Autonomic Nervous
System (2nd ed., pp. 124-126). Boston: Academic Press.
Lundqvist, L.-O., Carlsson, F., Hilmersson, P., & Juslin, P. N. (2009). Emotional responses to
music: Experience, expression, and physiology. Psychology of Music, 37(1), 61-90.
Maheu, F. S., Joober, R., Beaulieu, S., & Lupien, S. J. (2004). Differential Effects of Adrenergic
and Corticosteroid Hormonal Systems on Human Short- and Long-Term Declarative
Memory for Emotionally Arousing Material. Behavioral Neuroscience Vol 118(2) Apr
2004, 420-428.
Mather, M. (2004). Aging and Emotional Memory. In D. Reisberg & P. Hertel (Eds.), Memory and
Emotion (pp. 272 - 307). NY, US: Oxford University Press.
McClelland, J. L., & Rumelhart, D. E. (1985). Distributed memory and the representation of
general and specific information. Journal of Experimental Psychology: General Vol 114(2)
Jun 1985, 159-188.
McGaugh, J. L. (1989). Involvement of hormonal and neuromodulatory systems in the
regulation of memory storage. Annual Review of Neuroscience, 12, 255-287.
McGaugh, J. L. (2000). Memory: A century of consolidation. Science Vol 287(5451) Jan 2000,
248-251.
McGaugh, J. L. (2004). The amygdala modulates the consolidation of memories of emotionally
arousing experiences. Annual Review of Neuroscience, 27, 1-28.
McGaugh, J. L., Introini-Collison, I. B., Cahill, L. F., Castellano, C., Dalmaz, C., Parent, M. B., &
Williams, C. L. (1993). Neuromodulatory systems and memory storage: role of the
amygdala. Behav Brain Res, 58(1-2), 81-90.
McGaugh, J. L., & Roozendaal, B. (2002). Role of adrenal stress hormones in forming lasting
memories in the brain. Current Opinion in Neurobiology, 12(2), 205-210.
Menon, V., & Levitin, D. J. (2005). The rewards of music listening: response and physiological
connectivity of the mesolimbic system. NeuroImage, 28(1), 175-184.
MindMedia. (2004). User Manual for the BioTrace+ Software. Netherlands: Author.
Mockel, M., Rocker, L., Stork, T., Vollert, J., Danne, O., Eichstadt, H., . . . Hochrein, H. (1994).
Immediate physiological responses of healthy volunteers to different types of music:
cardiovascular, hormonal and mental changes. Eur J Appl Physiol Occup Physiol, 68(6),
451-459.
187
___________________________________________________________________________________
Nagel, F., Kopiez, R., Grewe, O., & Altenmuller, E. (2005). 'EMuJoy' - Software for continuous
measurement of perceived emotions in music: Basic aspects of data recording and
interface features. Submitted Manuscript. Hanover University of Music and Drama,
Germany.
Nagel, F., Kopiez, R., Grewe, O., & Altenmuller, E. (2007). EMuJoy: Software for continuous
measurement of perceived emotions in music. Behavior Research Methods, 39(2), 283290.
Nagel, F., Kopiez, R., Grewe, O., & Altenmuller, E. (2008). Psychoacoustical correlates of
musically induced chills. Musicae Scientiae, 12(1), 101-113.
Nantais, K. M., & Schellenberg, E. (1999a). The Mozart effect: An artifact of preference. Psychol
Sci, 10(4), 370-373.
Nantais, K. M., & Schellenberg, E. (1999b). The Mozart effect: An artifact of preference. Psychol
Sci, 10(4), 370-373.
Nater, U. M., Abbruzzese, E., Krebs, M., & Ehlert, U. (2006). Sex differences in emotional and
psychophysiological responses to musical stimuli. Int J Psychophysiol, 62(2), 300-308.
Neisser, U., & Harsch, N. (1992). Phantom flashbulbs: False recollections of hearing the news
about Challenger. New York, NY: Cambridge University Press.
Nielson, K. A., & Bryant, T. (2005). The effects of non-contingent extrinsic and intrinsic rewards
on memory consolidation. Neurobiology of Learning and Memory, 84(1), 42-48.
Nielson, K. A., & Lorber, W. (2009). Enhanced post-learning memory consolidation is influenced
by arousal predisposition and emotion regulation but not by stimulus valence or
arousal. Neurobiol Learn Mem, 92(1), 70-79.
Nielson, K. A., & Powless, M. (2007). Positive and negative sources of emotional arousal
enhance long-term word-list retention when induced as long as 30 min after learning.
Neurobiology of Learning and Memory Vol 88(1) Jul 2007, 40-47.
Nielson, K. A., Yee, D., & Erickson, K. I. (2005). Memory enhancement by a semantically
unrelated emotional arousal source induced after learning. Neurobiol Learn Mem, 84(1),
49-56.
Nittono, H. (1997). Background instrumental music and serial recall. Perceptual & Motor Skills,
84, 1307-1313.
North, A. C., & Hargreaves, D. J. (1997). Liking, arousal potential, and the emotions expressed
by music. Scandinavian Journal of Psychology, 38(1), 45-53.
North, A. C., Hargreaves, D. J., & Hargreaves, J. J. (2004). Uses of music in everyday life. Music
Perception Vol 22(1) Fal 2004, 41-77.
Nyklicek, I., Thayer, J. F., & Van Doornen, L. J. (1997). Cardiorespiratory differentiation of
musically-induced emotions. Journal of Psychophysiology, 11(4), 304-321.
O'Carroll, R., Drysdale, E., Cahill, L., Shajahan, P., & Ebmeier, K. (1999a). Memory for emotional
material: A comparison of central versus peripheral beta blockade. Journal of
Psychopharmacology, 13(1), 32-39.
O'Carroll, R. E., Drysdale, E., Cahill, L., Shajahan, P., & Ebmeier, K. P. (1999b). Stimulation of the
noradrenergic system enhances and blockade reduces memory for emotional material
in man. Psychological Medicine, 29(5), 1083-1088.
188
___________________________________________________________________________________
Oldfield, R. (1971). The assessment and analysis of handedness: The Edinburgh inventory.
Neuropsychologia, 9(1), 97-113.
Olsen, K. N., & Stevens, C. J. (2013). Psychophysiological response to acoustic intensity change
in a musical chord. Journal of Psychophysiology, 27(1), 16-26.
Palomba, D., Angrilli, A., & Mini, A. (1997). Visual evoked potentials, heart rate responses and
memory to emotional pictorial stimuli. Int J Psychophysiol, 27(1), 55-67.
Panksepp, J. (1998). Affective neuroscience : the foundations of human and animal emotions.
Peretz, I., Gagnon, L., & Bouchard, B. (1998a). Music and emotion: Perceptual determinants,
immediacy, and isolation after brain damage. Cognition, 68(2), 111-141.
Peretz, I., Gaudreau, D., & Bonnel, A. M. (1998b). Exposure effects on music preference and
recognition. Mem Cognit, 26(5), 884-902.
Perlovsky, L. (2010). Musical emotions: Functions, origins, evolution. Physics of Life Reviews,
7(1), 2-27.
PhotodexCorporation. (2004). ProShow Gold (Version 3.2.2040). Retrieved from
www.photodex.com
Pinker, S. (1997). How the Mind Works. New York: W. W. Norton.
Pizzagalli, D. A. (2007). Electroencephalography and high-density electrophysiological source
localization. In J. T. Cacioppo, L. G. Tassinary & G. G. Bernston (Eds.), Handbook of
Psychophysiology (3 ed., pp. 56-84). New York: Cambridge University Press.
Posner, J., Russell, J. A., & Peterson, B. S. (2005). The circumplex model of affect: an integrative
approach to affective neuroscience, cognitive development, and psychopathology.
Development & Psychopathology, 17(3), 715-734.
Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects in
simple mediation models Behavior Research Methods, Instruments, and Computers, 36,
717-731.
Preuss, D., & Wolf, O. T. (2009). Post-learning psychosocial stress enhances consolidation of
neutral stimuli. Neurobiology of Learning and Memory, 92(3), 318-326.
Rauscher, F. H., Shaw, G. L., & Ky, K. N. (1993). Music and spatial task performance. Nature,
365(6447), 611. doi: 10.1038/365611a0
Rauscher, F. H., Shaw, G. L., & Ky, K. N. (1995). Listening to Mozart enhances spatial-temporal
reasoning: towards a neurophysiological basis. Neurosci Lett, 185(1), 44-47. doi:
0304394094112214 [pii]
Reisberg, D. (2006). Memory for Emotional Episodes: The strengths and limits of arousal-based
accounts. In B. Uttl, N. Ohta & A. L. Siegenthaler (Eds.), Memory and Emotion:
Interdisciplinary Perspectives (pp. 15-36). Carlton, Victoria: Blackwell Publishing.
Reisberg, D., & Heuer, F. (2004). Memory for emotional events. In D. Reisberg & P. Hertel (Eds.),
Memory and Emotion (pp. 3-41). NY, US: Oxford University Press.
Reymann, K. G., & Frey, J. U. (2007). The late maintenance of hippocampal LTP: Requirements,
phases, 'synaptic tagging', 'late-associativity' and implications. Neuropharmacology,
52(1), 24-40.
189
___________________________________________________________________________________
Richter-Levin, G., & Akirav, I. (2003). Emotional tagging of memory formation - In the search for
neural mechanisms. Brain Research Reviews, 43(3), 247-256.
Rickard, N. S. (2004). Intense emotional responses to music: A test of the physiological arousal
hypothesis. Psychology of Music Vol 32(4) Oct 2004, 371-388.
Rickard, N. S., Wong, W. W., & Velik, L. (2012). Relaxing music counters heightened
consolidation of emotional memory. Neurobiology of Learning and Memory, 97(2), 220228. doi: 10.1016/j.nlm.2011.12.005
Rimmele, U., Davachi, L., & Phelps, E. A. (2012). Memory for time and place contributes to
enhanced confidence in memories for emotional events. Emotion, 12(4), 834-846.
Roozendaal, B. (2002). Stress and memory: Opposing effects of glucocorticoids on memory
consolidation and memory retrieval. Neurobiology of Learning and Memory Vol 78(3)
Nov 2002, 578-595.
Russell, J. A. (1980). A circumplex model of affect. Journal of Personality and Social Psychology
Vol 39(6) Dec 1980, 1161-1178.
Saarikallio, S. (2011). Music as emotional self-regulation throughout adulthood. Psychology of
Music, 39(3), 307-327.
Sacks, O. (2006). The Power of Music. [Commentary]. Brain, 129, 2528-2532.
Sah, P., Faber, E. S. L., Lopez De Armentia, M., & Power, J. (2003). The amygdaloid complex:
anatomy and physiology. Physiological Reviews, 83(3), 803-834.
Salame, P., & Baddeley, A. D. (1989). Effects of background music on phonological short-term
memory. The Quarterly Journal of Experimental Psychology A: Human Experimental
Psychology. Vol, 41(1-A), 107-122.
Salimpoor, V. N., Benovoy, M., Larcher, K., Dagher, A., & Zatorre, R. J. (2011). Anatomically
distinct dopamine release during anticipation and experience of peak emotion to music.
[10.1038/nn.2726]. Nat Neurosci, 14(2), 257-262.
Salimpoor, V. N., Benovoy, M., Longo, G., Cooperstock, J. R., & Zatorre, R. J. (2009). The
rewarding aspects of music listening are related to degree of emotional arousal. PLoS
One, 4(10), e7487. doi: 10.1371/journal.pone.0007487
Sammler, D., Grigutsch, M., Fritz, T., & Koelsch, S. (2007). Music and emotion:
electrophysiological correlates of the processing of pleasant and unpleasant music.
Psychophysiology, 44(2), 293-304.
Schellenberg, E. G. (2012). Cognitive performance after music listening: A review of the Mozart
effect. In R. A. R. MacDonald, G. Kreutz & L. Mitchell (Eds.), Music, health and wellbeing
(pp. 324-338). Oxford, UK: Oxford University Press.
Schellenberg, E. G., & Hallam, S. (2005). Music listening and cognitive abilities in 10- and 11year-olds: the blur effect. Ann N Y Acad Sci, 1060, 202-209.
Scherer, K. R. (2001). Appraisal considered as a process of multilevel sequential checking.
Scherer, Klaus R [Ed], 92-120.
Scherer, K. R. (2004). Which emotions can induced by music? What are the underlying
mechanisms? And how can we measure them? Journal of New Music Research, 33(3),
239-251.
190
___________________________________________________________________________________
Scherer, K. R. (2009). Emotion theories and concepts (psychological perspectives). In D. Sander
& K. R. Scherer (Eds.), The Oxford Companion to Emotion and the Affective Sciences (pp.
145-151). Oxford: Oxford University Press.
Schmidt, B., & Hanslmayr, S. (2009). Resting frontal EEG alpha-asymmetry predicts the
evaluation of affective musical stimuli. Neurosci Lett, 460(3), 237-240.
Schmidt, L. A., & Trainor, L. J. (2001). Frontal brain electrical activity (EEG) distinguishes valence
and intensity of musical emotions. Cognition & Emotion. , 15(4), 487-500.
Schneider, W., Eschman, A., & Zuccolotto, A. (2002). E-Prime User's Guide. Pittsburgh:
Psychology Software Tools Inc.
Schubert, E. (1999). Measuring emotion continuously: Validity and reliability of the twodimensional emotion-space. Australian Journal of Psychology Vol 51(3) Dec 1999, 154165.
Schubert, E. (2004). Modeling Perceived Emotion With Continuous Musical Features. Music
Perception Vol 21(4) Sum 2004, 561-585.
Schubert, E. (2007). Locus of emotion: the effect of task order and age on emotion perceived
and emotion felt in response to music. Journal of Music Therapy, 44(4), 344-368.
Segal, S. K., & Cahill, L. (2009). Endogenous noradrenergic activation and memory for emotional
material in men and women. Psychoneuroendocrinology, 34(9), 1263-1271. doi:
http://dx.doi.org/10.1016/j.psyneuen.2009.04.020
Severiens, S. E., & Dam, G. T. M. t. (1994). Gender Differences in Learning Styles: A Narrative
Review and Quantitative Meta-Analysis. Higher Education, 27(4), 487-501. doi:
10.2307/3448301
Sharot, T., Martorella, E. A., Delgado, M. R., & Phelps, E. A. (2007). How personal experience
modulates the neural circuitry of memories of September 11. Proc Natl Acad Sci U S A,
104(1), 389-394.
Sloboda, J. A. (1991). Music structure and emotional response: Some empirical findings.
Psychology of Music, 19(2), 110-120.
Sloboda, J. A., & O'Neill, S. A. (2001). Emotions in everyday listening to music. Juslin, Patrik N
(Ed); Sloboda, John A (Ed), (2001). Music and emotion: Theory and research. (pp. 415429). viii, NY, US: Oxford University Press.
Smeets, T., Otgaar, H., Candel, I., & Wolf, O. T. (2008a). True or false? Memory is differentially
affected by stress-induced cortisol elevations and sympathetic activity at consolidation
and retrieval. Psychoneuroendocrinology, 33(10), 1378-1386.
Smeets, T., Sijstermans, K., Gijsen, C., Peters, M., Jelicic, M., & Merckelbach, H. (2008b). Acute
consolidation stress enhances reality monitoring in healthy young adults. Stress, 11(3),
235-245.
Soetens, E., Casaer, S., D'Hooge, R., & Hueting, J. E. (1995). Effect of amphetamine on long-term
retention of verbal material. Psychopharmacology, 119(2), 155-162.
Sousou, S. D. (1997). Effects of melody and lyrics on mood and memory. Perceptual & Motor
Skills, 85, 31-40.
Southwick, S. M., Davis, M., Horner, B., Cahill, L., Morgan, C. A., 3rd, Gold, P. E., . . . Charney, D.
C. (2002). Relationship of enhanced norepinephrine activity during memory
191
___________________________________________________________________________________
consolidation to enhanced long-term memory in humans. Am J Psychiatry, 159(8), 14201422.
Spanagel, R., & Weiss, F. (1999). The dopamine hypothesis of reward: past and current status.
Trends in Neurosciences, 22(11), 521-527.
Squire, L. R., & Kandel, E. R. (1999). Memory: From mind to molecules. New York: Scientific
American Library.
Standing, L. (1973). Learning 10,000 pictures. Q J Exp Psychol, 25(2), 207-222.
Stefano, G. B., Zhu, W., Cadet, P., Salamon, E., & Mantione, K. J. (2004). Music alters
constitutively expressed opiate and cytokine processes in listeners.[see comment].
Medical Science Monitor, 10(6), 18-27.
Steinbeis, N., Koelsch, S., & Sloboda, J. A. (2006). The role of harmonic expectancy violations in
musical emotions: Evidence from subjective, physiological, and neural responses. J Cogn
Neurosci, 18(8), 1380-1393.
Strelau, J., & Eysenck, H. J. (1987). Personality dimensions and arousal. New York, NY: Plenum
Press.
Surwillo, W. W. (1963). The relation of simple response time to brain-wave frequency and the
effects of age. Electroencephalography and Clinical Neurophysiology, 15(1), 105-114.
Talarico, J. M., & Rubin, D. C. (2003a). Confidence, not consistency, characterizes flashbulb
memories. Psychol Sci, 14(5), 455-461.
Talarico, J. M., & Rubin, D. C. (2003b). Confidence, not consistency, characterizes flashbulb
memories. Psychol Sci, 14(5), 455-461.
Tan, S.-L., Spackman, M. P., & Bezdek, M. A. (2007). Viewers' interpretations of film characters'
emotions: Effects of presenting film music before or after a character is shown. Music
Perception Vol 25(2) Win 2007, 135-152.
Taylor, S. E. (1991). Asymmetrical Effects of Positive and Negative Events: The MobilizationMinimization Hypothesis. Psychological Bulletin, 110(1), 67-85.
Tesoriero, M., & Rickard, N. (2011). ? Musicae Scientiae(submitted).
Thaut, M. H., & de l'Etoile, S. K. (1993). The effects of music on mood state-dependent recall.
Thompson, E. R. (2007). Development and Validation of an Internationally Reliable Short-Form
of the Positive and Negative Affect Schedule (PANAS). Journal of Cross-Cultural
Psychology, 38(2), 227-242.
Thompson, W. F. (2009). Music, Thought, and Feeling: Understanding the Psychology of Music.
Thompson, W. F., Schellenberg, E. G., & Husain, G. (2001). Arousal, mood, and the Mozart
effect. Psychol Sci, 12(3), 248-251.
Touitou, Y., & Haus, E. (2000). Alterations with aging of the endocrine and neuroendocrine
circadian system in humans. Chronobiology International, 17(3), 369-390. doi:
doi:10.1081/CBI-100101052
Trainor, L. J., Austin, C. M., & Desjardins, R. N. (2000). Is infant-directed speech prosody a result
of the vocal expression of emotion? [Research Support, Non-U.S. Gov't]. Psychol Sci,
11(3), 188-195.
192
___________________________________________________________________________________
van Stegeren, A. H., Everaerd, W., Cahill, L., McGaugh, J. L., & Gooren, L. J. (1998). Memory for
emotional events: Differential effects of centrally versus peripherally acting beta blocking agents Psychopharmacology (Vol. 138, pp. 305-310).
Van Stegeren, A. H., Goekoop, R., Everaerd, W., Scheltens, P., Barkhof, F., Kuijer, J. P. A., &
Rombouts, S. A. R. B. (2005). Noradrenaline mediates amygdala activation in men and
women during encoding of emotional material. NeuroImage, 24(3), 898-909.
van Stegeren, A. H., Wolf, O. T., Everaerd, W., Scheltens, P., Barkhof, F., & Rombouts, S. A.
(2007). Endogenous cortisol level interacts with noradrenergic activation in the human
amygdala. Neurobiol Learn Mem, 87(1), 57-66.
Vanderark, S. D., & Ely, D. (1992). Biochemical and galvanic skin responses to music stimuli by
college students in biology and music. Perceptual & Motor Skills, 74, 1079-1090.
Vanderark, S. D., & Ely, D. (1993). Cortisol, biochemical, and galvanic skin responses to music
stimuli of different preference values by college students in biology and music.
Perceptual & Motor Skills, 77(1), 227-234.
vanGoethem, A., & Sloboda, J. A. (2011). The functions of music for affect regulation. Musicae
Scientiae, 15, 208-228.
Vecchiato, G., Astolfi, L., De Vico Fallani, F., Cincotti, F., Mattia, D., Salinari, S., . . . Babiloni, F.
(2010). Changes in brain activity during the observation of TV commercials by using EEG,
GSR and HR measurements. Brain Topography, 23(2), 165-179.
Vieillard, S., Roy, M., & Peretz, I. (2011). Expressiveness in musical emotions. Psychological
Research, 1-13.
Vuoskoski, J. K., & Eerola, T. (2011). The role of mood and personality in the perception of
emotions represented by music. Cortex, 47(9), 1099-1106.
Watkins, P. C., Vache, K., Verney, S. P., Muller, S., & Mathews, A. (1996). Unconscious moodcongruent memory bias in depression. Journal of Abnormal Psychology, 105(1), 34-41.
Werner, P. D., Swope, A. J., & Heide, F. J. (2006). The Music Experience Questionnaire:
development and correlates. Journal of Psychology, 140(4), 329-345.
Witvliet, C. V. O., & Vrana, S. R. (2007). Play it again Sam: Repeated exposure to emotionally
evocative music polarises liking and smiling responses, and influences other affective
reports, facial EMG, and heart rate. Cognition & Emotion, 21(1), 3 - 25.
Wolters, G., & Goudsmit, J. J. (2005a). Flashbulb and event memory of September 11, 2001:
consistency, confidence and age effects. Psychol Rep, 96(3 Pt 1), 605-619.
Wolters, G., & Goudsmit, J. J. (2005b). Flashbulb and event memory of September 11, 2001:
consistency, confidence and age effects. Psychol Rep, 96(3 Pt 1), 605-619.
Woo Ee, W., & Kanachi, M. (2005). The Effects of Music Type and Volume on Short-Term
Memory. Tohoku Psychologica Folia. Vol, 64, 68-76.
Yamamoto, M., Naga, S., & Shimizu, J. (2007). Positive musical effects on two types of negative
stressful conditions. Psychology of Music, 35(2), 249-275.
Zencor. (1998). Bioview (Version 2).
Zentner, M., Grandjean, D., & Scherer, K. R. (2008). Emotions evoked by the sound of music:
characterization, classification, and measurement. Emotion, 8(4), 494-521.
193
___________________________________________________________________________________
194
9 APPENDICES
195
APPENDIX A
A. Definitions of information categories across studies
Author/s
Term
Definition
Stimuli
Memory test
Adolphs Denberg &
Tranel (2001)*
Gist
Salient, general information of the scene
that was sufficient to distinguish that
particular stimulus from all the other
stimuli and did not depend on
remembering details of the scene.
Images and
single
sentence
narrative
24hr 4AFC
(written
questionnaire)
for narrative
and images
Visual Detail
Information that could only accessed from
a detailed memory of the visual image.
Gist
* as per Adolphs et al., 2001
Background
Visual Detail
* as per Adolphs et al., 2001
ImplicitPerceptual
Bottom-up, data driven.
Adolphs, Tranel &
Buchanan (2005)
Arntz, de Groot &
Kindt (2005)
Images and
single
sentence
narrative
Cahill et al.
(1994)
narrative
modified
196
Effect of
Emotion
Enhances
4AFC (images)
for images
Impairs
24 hr
written 4Amultiple choice
Enhances
Written 4Amultiple choice
Impairs
1 week, free
recall,
perceptual id,
recognition,
word stem
completion
Enhances
APPENDIX A
Author/s
Buchanan, Karafin
& Adolphs (2003)
Burke, Heuer &
Riesberg, (1992)
Term
Definition
Explicitconceptual
Meaning of the information processed.
Gist
Salient, general information about the
stimulus that could not be changed or
excluded without changing the basic story
line.
Visual Detail
No definition provided.
Plot relevant
Items relevant to how the story unfolds.
Gist
Items that essentially define the story.
197
Stimuli
Memory test
Effect of
Emotion
No change
* as per
Adolphs et
al., 2001
48 hr
multiple choice
Placebo group no change,
negative vs.
neutral
Multiple choice
Placebo group –
impaired,
negative vs.
neutral
Images and
single
sentence
narrative
1 week 4AFC
P1 weakly
enhances; P2
enhances; P3
weakly
enhances
APPENDIX A
Author/s
Cahill et al, (2004)
Term
Definition
Stimuli
Basic level
visual
Broad description of what the slide
showed.
Plot irrelevant
Items not relevant to how the story
unfolds.
Central details
Plot irrelevant but spatially associated with
plot relevant details.
P1 no effect; P2
enhances; P3
weakly impairs.
Peripheral
Items truly in the background.
Impairs
Central
Any information that cannot be removed
or altered without changing the
fundamental story line.
Peripheral
All other information.
Images and
single
sentence
narrative
Memory test
1 week
multiple choice
Effect of
Emotion
P1 weakly
enhances; P2
enhances; P3
weakly
enhances
Enhances
(BEM males
only).
Enhances
(BEM males and
females)
198
APPENDIX A
Author/s
Term
Definition
Stimuli
Memory test
Cahill & van
Stegeren, (2003)
Central
* as per Cahill et al, 2004
* as per
Cahill et al
(2004)
* as per Cahill et
al (2004)
Peripheral
* as per Cahill et al, 2004
Christianson &
Loftus (1991)
Enhances for
women
Central
Images
Immediate
cued recall,
4AFC
Peripheral
Heuer & Reisberg
(1990)
Effect of
Emotion
Enhances for
men
Enhances
Impairs
Central
Any fact or element pertaining to the
“basic story” that could not be “changed or
excluded” without changing the basic story
line.
Peripheral
Anything below a ‘basic level of
description’ (Rosch, 1978).
199
Images and
single
sentence
narrative
2 weeks
free recall,
4AFC,
recognition
memory for
narrative
Enhances
Enhances
APPENDIX A
Author/s
Term
Definition
Stimuli
Memory test
Kensinger, GaroffEaton, Schacter
(2007b)
Central
Emotionally salient information.
Images
Immediate
recognition
(same, similar,
new)
Peripheral
Information outside the focus of emotional
attention.
Impairs
Gist
General theme
Conditional
Detail
Specific visual details
Conditional
Gist
Elements in the story line, rather than the
specific contents of the slides.
Central details
Visually close to likely focus of attention.
Plot-irrelevant details that happened to be
associated with plot-central objects or
players within the slides.
No change
Peripheral
details
Far from likely focus. Details of objects
truly in the background of the slide
sequence.
Enhances
Laney et al (2004)
200
Nonarousing
images and
arousing
narrative
(thematic
emotion
induction)
Immediate
4AFC
Effect of
Emotion
Enhances
Enhances
APPENDIX A
Author/s
Term
Definition
Stimuli
Memory test
Libkuman, NicholsWhitehead, Griffith
& Thomas (1999)
Gist
Gist (items that essentially define the
story) and basic level visual information
(broad description of what the slide
showed).
Images and
single
sentence
narrative
(Heuer &
Reisberg,
1990,
modified)
Immediate
cued recall,
4AFC
Central detail
Elements central to the story.
Enhanced
Background
detail
Details peripheral to the story.
Enhanced
Central
Detail that was associated with the central
characters.
Detail that was not associated with the
central characters.
Information that is connected with the
source of the emotional arousal.
Libkuman, Stabler
& Otani (2004)
Background
Wessel et al (2000)
Gist and central
details
Peripheral
Information that is irrelevant or spatially
peripheral to the source of the emotional
arousal.
201
Images (high
arousal)
Images
Immediate
cued recall
Immediate
free recall,
cued recall,
detailed cued
recall
Effect of
Emotion
No change
(except exp 2,
impaired, no
statistical
results
provided)
Enhanced
Enhanced for
positive images
Exp 1 no effect
Exp 2 enhanced
Exp 1 no effect
APPENDIX B
B. Summary of aims and hypotheses
Experiment 1 non-music aims and hypotheses
Aim 1: Determine whether the newly developed materials replicated the intrinsic
emotion-memory effects reported by previous researchers
Hypothesis
1a
1b
Emotion elicited by a negative narrative in Phase 2 of a three-phase
slideshow would facilitate one week delayed free recall and recognition of
Phase 2 slideshow images relative to Phase 1 and Phase 3.
The increase in Phase 2 memory would account for greater total memory
scores for the negative slideshow relative to the neutral slideshow.
Supported
NO
NO
Aim 2: Determine whether any observed emotion-enhanced memory elicited by the
narrative could be accounted for by emotion or by mood-congruence effects on
encoding.
Hypothesis
2a
Supported
For emotion to be the underlying mechanism, any increase in memory
caused by the Phase 2 negative narrative would be associated with
activation of the emotion components of:
Increased subjective arousal ratings
Decreased mood valence ratings
NO
TREND
Increased skin conductivity (SCL)
Increased skin conductance responses (SCR)
Decreased skin temperature (TEMP)
Increased heart rate (decreased IBI)
NO
NO
NO
NO
Increased frown muscle activity (EMG)
NO
Increased memory narrowing
YES
202
APPENDIX B
Experiment 1 music aims and hypotheses
Aim 3: Determine whether extrinsic arousal elicited by emotionally powerful background
music facilitated memory for a neutral slideshow.
Hypothesis
3
Emotion elicited by arousing music played continuously in the background
of a neutral story would facilitate one-week delayed free recall and
recognition of all slideshow images relative to the neutral music and no
music conditions.
Supported
NO
Aim 4: Determine whether any observed emotion-enhanced memory elicited by music
was best explained by emotion or mood-congruence effects on encoding.
Hypothesis
4a
4b
Supported
For emotion to be the underlying mechanism, any increase in memory
caused by the emotional music would be associated with activation of the
emotion components of:
Decreased mood valence ratings
NO
NO
Increased skin conductivity (SCL)
NO
NO
NO
NO
Increased frown muscle activity (EMG)
NO
Increased memory narrowing
NO
For mood-congruence to be responsible for facilitated memory, greater
memory for the slideshow with emotional background music would be
attributed to more negative image details retained relative to the no-music
or neutral music conditions.
203
NO
APPENDIX B
Experiment 2
Aim: Determine whether participant-selected enjoyed music (PsM) facilitated memory.
Hypothesis
1
Supported
Relative to non-music and neutral music controls, PsM would activate the
emotion components of:
SUBJECTIVE FEELING
More positive mood valence ratings
YES
YES
PERIPHERAL EFFERENCE
Increased chills frequency
Increased skin conductance levels (SCL)
Decreased HRV (IBIsd)
YES
PARTIAL
NO
NO
NO
PARTIAL
COGNITIVE PROCESSES
Change in attention scope
Decreased judgment response times
NO
PARTIAL
2
PsM would elicit the highest memory scores.
PARTIAL
3
If emotional arousal was the causal mechanism, then any observed
facilitated memory effect would be predicted by increased autonomic
activity.
YES
4
If mood-congruence was responsible for facilitated memory, then any
observed facilitated memory effect would be explained by greater memory
for images congruent with self-reported mood.
NO
Note. Partial indicates that the comparison between PsM and control was significant while the
comparison between PsM and OsM was not significant.
204
APPENDIX B
Experiment 3
Aim: Determine whether arousal elicited by positive and arousing (PA) music presented
after learning facilitated the consolidation of LTM.
Hypothesis
1
Supported
Relative to music and non-music active listening controls, PA music would
activate the emotion components of:
SUBJECTIVE FEELING
More positive mood valence ratings
NO
PARTIAL
PERIPHERAL EFFERENCE
Increased chills frequency
Decreased heart rate variability (%HF HRV)
Decreased blood volume amplitude (BVA)
Increased respiratory rate (RSP)
PARTIAL
NO
NO
NO
NO
NO
COGNITIVE PROCESSES
Increased frontal midline theta activity
General decreases in alpha activity
2
Post-learning emotional arousal elicited by PA music would facilitate longterm (one week) word list memory relative to music and non-music active
listening controls.
NO
NO
NO
Note. Partial indicates that the comparison between PA Music and Traffic Mix was significant while the
comparison between PA Music and Music Mix was not significant.
205
APPENDIX C
C. Digital copy of Experimental Stimuli
Folder
Exp1
Content
4 slideshows:
- neutral narrative-no music
- emotional narrative-no music
- neutral narrative-neutral music
- neutral narrative-emotional music
Mouse click on the image to start the slideshow. Note that the first 15 seconds is silent.
Exp2
PsM samples
Radio interview control
Exp3
PAMusic
MusicMix
TrafficMix
CD located inside the back cover of the thesis.
206
APPENDIX D
D. Exp. 1 Linguistic Properties of the Narratives
Phase
Neutral Narrative
1
A man used to work at a busy
port as a shipping container
administrator. (14)
He often went to a restaurant
for lunch before it burnt down.
(12)
His wife is the primary carer of
their two school aged children.
(12)
He takes his children to visit his
brother who is in jail for fine
evasion. (15)
On the way they go walking in a
park and the father drops the
daughter’s jumper. (16)
Linguistic
Properties
F<M=5, >M=9
C<M=12, >M=2
I<M=11, >M=3
F<M=5, >M=7
C<M=10, >M=2
I<M=8, >M=4
F<M=4, >M=8
C<M=9, >M=3
I<M=9, >M=3
F<M=5, >M=10
C<M=12, >M=3
I<M=11, >M=4
F<M=5, >M=11
C<M=13, >M=3
I<M=13, >M=3
When they return home the
mother wants to wash the
jumper but can’t find it. (15)
F<M=5, >M=10
C<M=13, >M=2
I<M=12, >M=3
Some time later a woman finds
the daughter’s jumper near her
country house. (13)
The jumper is labeled with the
family details so the woman
contacts the mother. (14)
After a long delay they meet
while the mother is visiting the
cemetery. (13)
F<M6=, >M=7
C<M=10, >M=3
I<M=9, >M=4
F<M=6, >M=8
C<M=11, >M=3
I<M=11, >M=3
F<M=4, >M=9
C<M=12, >M=1
I<M=11, >M=2
2
3
Emotional Narrative
Linguistic
Properties
(as per neutral)
He is in jail for beating
his wife and plots his
revenge on her. (14)
When he gets out the
father collects his
children and takes them
into dense bush. (15)
When they don’t return
home the mother
becomes frantic with
concern for her
children. (14)
(as per neutral)
F<M=5, >M=9
C<M=12, >M=2
I<M=10, >M=4
F<M=4, >M=11
C<M=12, >M=3
I<M=10, >M=5
F<M=5, >M=9
C<M=10, >M=4
I<M=10, >M=4
Note. Numbers in brackets equal word count. F = familiarity, C = concreteness, I = imaginability. <M and >M =
number of words less than or greater than the mean for each word category (F, C, and I) (MRC Psycholinguistic
Database (http://www.psy.uwa.edu.au/mrcdatabase/uwa_mrc.htm).
207
APPENDIX E
E. Exp. 1 Narrative Pilot Testing
A pilot test of the story narratives was conducted with two groups of undergraduate university
students (neutral, n = 22, emotional, n = 23) as a class exercise. Participants viewed the slideshow
containing the nine IAPS images combined with the neutral or negative narrative. One week later, a 72
item four-alternative forced-choice (4AFC) surprise memory test was administered. Differences in
emotion ratings between the two groups revealed that, overall, students who heard the negative
narrative were more likely to disagree with the statement that the story made them feel happy, and
agree with the statement that the story made them feel sad (illustrated in Figure E1). A series of
independent samples t-tests confirmed that these differences in the mean sadness ratings were
significant. There were also significant differences in ratings of fear, anger, and disgust. Probably levels
for each test are presented in the figure below.
The two groups were similar in their ratings of story understanding (neutral M = 2.04, SD = 1.13,
emotional M = 2.65, SD = 1.11), complexity (neutral M = 3.86, SD = 0.71, emotional M = 3.51, SD = 0.89),
and emotional intensity (neutral M = 3.36, SD = 2.51, emotional M = 3.82, SD = 2.25). These ratings
indicate that the stories were relatively similar in linguistic properties and that the emotional story was
no more intense than the neutral story.
A
Mean Agreement Rating
*
**
*
**
Disagree (4)
Neither (3)
Agree (2)
C
om
pl
ex
nd
ab
le
10
Neutral
Negative
8
6
4
2
0
Intensity
U
nd
er
s
ta
gu
s
te
d
ed
D
is
ry
Su
rp
ris
A
ng
l
fu
Fe
ar
Sa
d
H
ap
py
Strongly Agree (1)
0 = not intense, 10 = very intense
B
Strongly Disagree (5)
Figure E1 Mean agreement ratings (A) for the statement ‘the story made me feel (insert emotion
adjective)’,’ the story was easy to understand’, and ‘the story was complex’; and story intensity (B) for
the neutral and negative slideshows. Error bars represent standard error of the mean.
* = p < .05, ** = p < .01
208
APPENDIX E
Memory testing revealed that slideshow images presented with the negative narrative in Phase
2 (M = 9.6, SD = 3.36) were more memorable than those in Phase 1 (M = 9.1, SD = 4.42) and Phase 3 (M
= 7.3, SD = 2.59). This pattern of facilitated memory for Phase 2 images was not present for the neutral
slideshow (Phase 1, M = 10.68, SD = 4.01; Phase 2, M = 8.81, SD = 3.70; and Phase 3, M = 8.68, SD = 2.43,
see Figure E2). A narrative (neutral vs. negative) by phase (3) mixed ANOVA revealed that the trend for
an interaction between narrative type and slideshow phase was not significant, F(2,80) = 2.58, p = .08,
ηp2 = .06.
Neutral Narrative
Emotional Narrative
4AFC Items Correct
16
14
12
10
8
6
1
2
3
Phase
Figure E2 Mean number of correct 4AFC recognition memory test items for each phase of the
slideshow. Dashed blue lines represent the neutral narrative and solid red lines represent the emotional
narrative. Error bars represent standard error.
A possible explanation for the failure of the trends in the data to reach statistical significance
was high variability caused by the testing environment. It was reported that a number of students had
difficulty hearing or viewing the slideshows. Additional variation could have been caused by low
incentive for the cohort of students to attend to the task. As the trends in the memory data were in the
expected direction, the decision was made to continue using the slideshows and memory test.
209
APPENDIX F
F. Exp. 2 Emotion variables Pearson’s Correlation Results
CONTROL
Enjoy
Chills
Valence
Arousal
FaceRT
SCL
SCRmin
TEMP
IBIm
IBIsd
Free
recall
r
N
r
N
r
N
r
N
r
N
r
N
r
N
r
N
r
N
r
N
OsM
Enjoy
Chills
Valence
Arousal
FaceRT
SCL
SCRmin
TEMP
IBIm
IBIsd
r
N
r
N
r
N
r
N
r
N
r
N
r
N
r
N
r
N
r
N
Enjoy
Chills
Valence
Arousal
FaceRT
.03
37
-.03
37
-.06
35
.14
35
-.20
35
-.13
30
-.14
30
-.17
37
-.01
35
-.05
35
-.02
35
.05
30
-.12
30
-.11
35
.26
35
-.16
35
-.23
30
-.12
30
Valence
Arousal
FaceRT
.03
37
.16
37
-.44**
35
-.25
35
.12
35
-.20
30
-.13
30
.06
37
-.03
35
.14
35
-.16
35
-.28
30
-.40*
30
-.32
35
-.07
35
-.10
35
-.180
30
-.38*
30
SC
SCRmin
TEMP
IBIm
-.06
37
-.01
37
-.17
37
-.08
37
.37*
35
.15
35
-.08
35
-.17
30
-.09
30
.47**
37
.27
37
.04
37
-.17
35
.06
35
-.28
35
.09
30
-.29
30
Free
recall
Enjoy
-.27
37
.21
37
-.15
37
-.03
37
-.26
37
.42*
35
.23
35
-.04
35
-.07
30
-.15
30
.19
37
.56***
37
.45**
37
.35*
37
-.29
35
-.10
35
.04
35
-.39*
30
-.19
30
Chills
.10
37
.19
37
-.13
37
-.13
35
-.27
35
.26
35
.30
30
.12
30
210
.17
35
-.07
35
.01
30
.15
30
SCL
.39*
35
.09
35
.10
30
.10
30
.08
35
.14
30
.20
30
SCRmin
.10
35
.06
30
-.02
30
.01
30
.02
30
TEMP
.02
30
.15
30
.56**
30
IBIm
.55**
30
APPENDIX F
Free
recall
PsM
Enjoy
Chills
Valence
Arousal
FaceRT
SCL
SCRmin
TEMP
IBIm
IBIsd
r
N
r
N
r
N
r
N
r
N
r
N
r
N
r
N
r
N
r
N
.16
37
.05
37
-.02
37
.17
37
-.29
37
.43*
35
-.33†
35
.03
35
-.01
30
-.40*
30
Enjoy
.42**
37
.30
37
.02
37
-.01
37
-.05
35
-.45**
35
-.23
35
-.06
30
.10
30
Chills
.15
37
.40*
37
-.07
37
.13
35
-.13
35
-.15
35
-.21
30
.16
30
Valence
Arousal
FaceRT
.01
37
-.13
37
-.10
35
.01
35
.22
35
.09
30
.23
30
-.26
37
.04
35
-.07
35
.22
35
-.09
30
-.15
30
-.37*
35
.04
35
-.15
35
-.07
30
-.11
30
SCL
.31
35
.08
35
-.01
30
.05
30
SCRmin
.04
35
.27
30
.36†
30
TEMP
.19
30
-.04
30
IBIm
.28
30
Note. r = Pearson’s product moment coefficient. N = sample size. * p < .05. ** p < .01. *** p < .001. † p =
.05.
211
APPENDIX G
G. Exp. 3 Music selection pilot
Purpose: To use an objective means of selecting three music tracks from a pool of six for use in
the main study.
Table G1
Music selections pilot tested
Music track and composer
Justification
Life in Technicolor, Coldplay
Also Sprach Zarathustra, Richard Strauss
Previous studies
The Mission, Ennio Morricone
Exp 2, participant selected music with highest peer
arousal ratings
O Fortuna, Carl Orff
Exp 2, participant selected music with highest peer
arousal ratings
Tuba Mirum, Wolfgang Armadeus Mozart
Previous studies
Ride of the Valkyries, Richard Wagner
Parameters measured
Arousal was the main parameter of interest and was measured by self-report of arousal, chills,
and liking. The experience of chills has been reported to be associated with physiological arousal
(detected by phasic skin conductance increases, Grewe et al., 2009). Liking has been reported to be
association with arousal in an inverted U relationship, with liking ratings highest for moderately arousing
music (Hargreaves & North, 2010). Familiarity may also be associated with liking (more familiar stimuli
are more liked, Bornstein, 1989), and therefore indirectly related to arousal. Affective valence, whether
participants responded positively or negatively to the music, was also measured. Differences in
affective valence could potentially induce differential effects on cognitive processing, such as fast and
broad processing strategies when in a positive mood, or slow and analytical strategies when in a
negative mood (Levine & Pizarro, 2004). For consistency, valence was measured in order to select music
that induced positive affect.
212
APPENDIX G
Procedure
Seventeen colleagues, friends, and family members listened to a compact disc recording of the
six music tracks at their own convenience. Participants were instructed to rate how they felt after each
track on scales of arousal, valence, liking, chills quantity, and familiarity. All rating scales were the same
as those used in the main study. Arousal and valence scales were presented as two dimensional grids
with axes ranging from -5 to 5. Liking and familiarity on Likert scales ranging from 1 to 7. Chills quantity
was an open box.
Results
One-way repeated measures ANOVAs were used to analyse differences across music tracks (6)
for each of the five rating scales. Mean differences between tracks were significant for all scales except
arousal. Consequently, a process of selection was used based on (in order of importance), valence,
chills, liking and familiarity. Tuba Mirum was excluded due to negative valence ratings compared to
positive valence ratings for the remaining five tracks. O Fortuna was selected due to significantly higher
chills ratings than the remaining four tracks. Ride of the Valkyries was excluded due to significantly
lower liking ratings than the remaining four tracks. The Mission was excluded based on the lowest
familiarity ratings of the remaining tracks. The remaining tracks were O Fortuna, Also Sprach
Zarathustra, and Life in Technicolor.
213
APPENDIX H
H. Exp. 3 Equipment set-up
Nexus data acquisition system set-up
Frontal EEG at sites Fp1 and Fp2, and Pz, and
EOG sensors; skin conductivity, heart
rhythm and blood volume amplitude
recorded from the non-dominant hand; and
respiratory rate recorded from below the
ribcage (not visible in this image). All leads
were connected to the receiver held by the
participant and transmitted wirelessly to a
computer hard drive.
The experimenter was
seated in front of two
computer monitors, one
controlling stimulus
presentation, the other
controlling and
monitoring the
physiological
recordings.
The participant was
seated in front of a
computer monitor and
was separated from the
rest of the room by the
blue room divider.
214
APPENDIX I
I. Exp. 3 Correlation matrix for the memory and emotion measures
.41**
41
.26
41
-.10
41
.13
41
.18
42
.42**
42
.54**
42
.42**
42
.01
42
-.13
42
-.03
42
.21
42
.09
44
.18
43
-.05
42
.05
43
-.04
42
-.19
42
-.14
42
-.14
42
.35*
42
.23
44
.00
43
-.09
42
.32*
43
.01
42
.06
44
.42**
42
.05
43
.32*
43
.16
43
-.29
29
-.04
42
.01
42
-.06
42
-.26
29
.06
44
.72**
44
-.05
30
.26
44
-.20
30
Chills
.09
42
-.10
42
.29
42
.08
42
.35*
43
.50**
43
.24
42
Liking
.24
42
.54**
42
-.05
42
-.09
42
-.06
42
-.29
29
Arousal
.50**
42
.50**
43
.42**
42
.18
43
.00
43
.13
43
-.23
29
Valence
.23
43
.02
42
.35*
43
.18
42
.09
44
.23
44
.17
44
-.11
30
.55**
41
.04
41
.01
41
-.12
41
.02
42
.50**
42
ThetaPz
.16
42
.08
42
-.12
41
.08
42
.13
41
.21
42
.35*
42
.22
42
.01
29
.30
42
-.14
42
.29
42
.08
42
.23
43
ThetaF
-.13
42
-.02
42
.29
42
.01
41
.29
42
-.10
41
-.03
42
-.14
42
-.12
42
-.29
29
-.03
42
-.44**
42
-.02
42
.16
42
AlphaPz
-.03
42
-.09
42
-.44**
42
-.14
42
.04
41
-.10
42
.26
41
-.13
42
-.14
42
-.25
42
.03
29
-.03
42
-.09
42
-.13
42
AlphaF
-.11
42
-.03
42
SCR min
.24
42
RSP
.24
42
-.11
42
-.03
42
-.03
42
.30
42
.55**
41
.09
42
.41**
41
.01
42
-.19
42
-.07
42
-.13
29
BVA
-.27
41
.07
41
.22
41
.02
41
-.03
43
-.03
42
-.09
41
.06
42
-.12
41
.31*
43
-.07
43
.04
43
-.20
30
HRV
-.26
42
-.10
42
.09
42
-.16
42
.09
44
-.25
43
-.10
42
-.09
43
-.18
42
.11
44
.07
44
.02
44
-.10
30
IBI
Recognition
IBI
N
HRV
N
BVA
N
RSP
N
SCRmin
N
AlphaF
N
AlphaPz
N
ThetaF
N
ThetaPz
N
Valence
N
Arousal
N
Liking
N
Chills
N
Free
Emotion
element
-.07
42
-.25
42
-.12
42
.22
42
.17
44
.13
43
-.06
42
.16
43
-.06
42
.72**
44
.26
44
.13
29
.03
29
-.29
29
.01
29
-.11
30
-.23
29
-.29
29
-.20
29
-.26
29
-.05
30
-.20
30
.08
30
.08
30
Note. Pearson’s correlations and N. Physiological values were change from baseline. EEG values were natural log transformed. Chills
correlations exclude the Traffic Mix condition for which no chills were reported. * p < .05. ** p < .01. *** p < .001.
215
APPENDIX J
J. Exp. 3 Good vs. poor memory performers
FREE RECALL % WORDS RETAINED
Self Report
Music Mix
Poor Free Recall
Good Free Recall
5
4
3
2
1
0
-1
-2
-3
-4
-5
Subjective valence
Subjective arousal
5
4
3
2
1
0
-1
-2
-3
-4
-5
Traffic Mix
Valence
(b)
Arousal
(a)
Traffic Mix
PA Music
Music Mix
PA Music
PNS
(e)
IBI
vs. baseline
40
20
0
-20
-40
-60
BVA
(f)
5
(d)
60
0
-5
-10
Traffic Mix
Music Mix
PA Music
Traffic Mix
Music Mix
RSP
p = .07
Music Mix
PA Music
8
6
4
2
0
PA Music
Traffic Mix
SCR/min
(g)
SCR/min vs. baseline
2
1
0
-1
-2
Traffic Mix
Music Mix
PA Music
CNS
Frontal Theta
(h)
(j)
p = .05
0.2
-0.2
-0.4
-0.6
ln)
2
vs. baseline
2
Alpha Power (µV pk-pk
vs. baseline
0.0
0.0
-0.2
-0.4
Music Mix
PA Music
0.0
-0.2
-0.4
-0.6
-0.6
Traffic Mix
Posterior Theta
0.2
ln)
ln)
2
vs. baseline
(i)
Posterior Alpha
0.2
Traffic Mix
Music Mix
216
PA Music
Traffic Mix
Music Mix
PA Music
APPENDIX J
RECOGNITION CORRECTLY IDENTIFIED AS PRESENT
Self Report
Subjective valence
(d)
40
20
0
-20
-40
-60
Traffic Mix
(g)
Traffic Mix
Music Mix
Music Mix
(e)
**
60
*
PA Music
IBI
Poor Recognition
Good Recognition
Valence
5
4
3
2
1
0
-1
-2
-3
-4
-5
PA Music
BVA
(f)
4
PNS
Music Mix
vs. baseline
Subjective arousal
5
4
3
2
1
0
-1
-2
-3
-4
-5
Traffic Mix
(b)
Arousal
(a)
2
0
-2
-4
-6
-8
PA Music
Traffic Mix
Music Mix
PA Music
RSP
8
6
4
2
0
Traffic Mix
Music Mix
PA Music
SCR/min
SCR/min vs. baseline
2
1
0
-1
-2
Traffic Mix
Music Mix
PA Music
CNS
(h)
(i)
Frontal Theta
(j)
Posterior Theta
0.2
0.0
-0.2
-0.4
-0.6
0.0
-0.2
-0.4
-0.6
Traffic Mix
Music Mix
PA Music
Posterior Alpha
0.2
Alpha Power (µV pk-pk 2 ln)
vs. baseline
Theta Power (µV pk-pk 2 ln)
vs. baseline
Theta Power (µV pk-pk 2 ln)
vs. baseline
0.2
0.0
-0.2
-0.4
-0.6
Traffic Mix
Music Mix
217
PA Music
Traffic Mix
Music Mix
PA Music
APPENDIX K
K. Exp. 3 Good vs. Poor memory t-test results for each emotion measure and each group
FREE RECALL
Arousal
Valence
IBI
BVA
RSP
SCR
AlphaPz
ThetaF
ThetaPz
P
G
P
G
P
G
P
G
P
G
P
G
P
G
P
G
P
G
N
7
7
7
7
6
7
6
7
6
7
7
7
6
7
7
7
6
7
M
0.00
1.71
-0.86
-0.71
-32.52
-22.20
-1.43
1.32
1.08
3.28
-0.54
-0.15
-0.14
-0.16
-0.10
-0.06
-0.12
-0.30
Traffic Mix
SD
t
2.00
-1.77
1.60
2.54
-0.11
2.29
41.51 -0.51
31.19
5.06
-0.86
6.30
5.24
-1.00
2.42
1.50
-0.57
0.86
0.40
-0.08
0.32
0.38
-0.21
0.26
0.42
0.79
0.38
p
.10
d
0.95
.91
0.06
.62
0.28
.41
0.48
.34
0.57
.56
0.33
.94
0.05
.84
0.12
.44
0.45
N
7
8
7
8
7
8
7
8
7
8
7
8
7
8
7
8
7
8
M
2.29
2.88
3.29
1.25
10.46
-29.30
-4.15
-4.65
5.39
3.36
0.24
0.89
-0.04
-0.23
0.00
-0.06
-0.07
-0.16
Music Mix
SD
t
1.80
-0.51
2.53
1.11
1.85
2.71
55.39
1.81
26.61
2.56
0.31
3.57
3.58
1.08
3.70
2.02
-0.70
1.55
0.25
1.73
0.17
0.12
0.92
0.11
0.16
1.28
0.14
p
.62
d
0.27
.09
1.07
.09
0.97
.76
0.16
.30
0.56
.50
0.36
.11
0.90
.37
0.52
.22
0.60
N
7
8
7
8
7
7
7
7
7
7
7
8
7
7
7
7
7
7
M
3.00
2.13
0.57
2.06
-7.82
-14.99
-1.49
-4.34
5.71
2.65
0.90
0.02
-0.19
-0.17
0.06
-0.15
-0.10
-0.30
PA Music
SD
t
1.22
1.05
1.89
2.11
-1.81
0.94
29.19
0.45
30.25
5.76
1.07
4.08
3.86
1.95
1.51
0.91
1.42
1.41
0.23
-0.13
0.30
0.16
2.14
0.20
0.17
1.52
0.30
p
.31
d
0.56
.09
0.98
.66
0.24
.31
0.58
.07
1.14
.18
0.76
.90
-0.07
.05
1.15
.15
0.84
Note. P = poor memory performer, G = good memory performer, N = number of participants in each cell, M = mean, SD = standard deviation, t =
between-subjects t value, p = probability, and d = Cohen’s d.
218
APPENDIX K
RECOGNITION
Arousal
Valence
IBI
BVA
RSP
SCR
AlphaP
ThetaF
ThetaP
P
G
P
G
P
G
P
G
P
G
P
G
P
G
P
G
P
G
N
7
6
7
6
6
6
6
6
6
6
7
6
6
6
7
6
6
6
M
1.00
0.67
-2.28
0.33
-32.62
-10.62
-0.48
0.30
1.41
2.73
-0.18
-0.70
-0.10
-0.11
0.03
-0.19
-0.16
-0.18
Traffic Mix
SD
t
1.82
0.28
2.42
1.98
-2.73
1.37
36.50
-1.28
21.15
4.88
-0.21
7.30
5.15
-0.54
2.98
0.50
0.77
1.74
0.42
0.01
0.20
0.20
1.23
0.41
0.53
0.11
0.11
p
.78
d
0.15
.02
1.53
.23
0.76
.83
0.13
.60
0.32
.46
0.46
.99
0.03
.24
0.72
.91
0.06
N
7
8
7
8
7
8
7
8
7
8
7
8
7
8
7
8
7
8
M
2.57
2.63
2.28
2.12
22.58
-39.91
-3.74
-5.01
4.65
4.01
-0.06
1.16
-0.03
-0.23
-0.06
-0.01
-0.06
-0.17
Music Mix
SD
t
1.72
-0.05
2.62
2.14
0.13
2.59
41.61 3.59
24.80
2.98
0.80
3.17
3.80
0.32
3.78
1.83
-1.39
1.57
0.21
1.85
0.21
0.11
-0.78
0.12
0.16
1.30
0.14
p
.96
d
-0.02
.90
0.07
.00
1.88
.44
0.41
.75
0.17
.19
-0.72
.09
0.96
.45
-0.41
.22
0.67
N
7
8
7
8
7
7
7
7
7
7
7
8
7
7
7
7
7
7
M
2.64
2.44
0.78
1.87
-3.41
-19.41
-3.90
-1.93
4.05
4.32
0.84
0.07
-0.20
-0.15
-0.12
0.02
-0.11
-0.30
PA Music
SD
t
1.75
0.24
1.61
2.29
-1.25
0.88
26.04
1.04
31.14
6.14
-0.72
3.82
4.07
-0.15
2.45
0.92
1.22
1.44
0.25
-0.34
0.28
0.16
-1.35
0.23
0.17
1.37
0.31
p
.82
d
0.12
.23
0.69
.32
0.56
.48
-0.40
.88
-0.08
.25
0.66
.74
-.018
.20
-0.74
.20
0.76
Note. P = poor memory performer, G = good memory performer, N = number of participants in each cell, M = mean, SD = standard deviation, t =
between-subjects t value, p = probability, and d = Cohen’s d.
219
APPENDIX L
L. Supplementary investigation of individual differences
A variety of extraneous variables have been identified that could moderate the relationship
between emotional arousal and memory. These variables, such as individual differences in arousal
sensitivity, gender, and engagement with music, may contribute to the high variability endemic to
emotional responses to music, and to memory performance. This supplementary analysis of individual
differences was conducted to determine their impact on memory. The primary purpose was to control
variation in emotional arousal caused by individual differences to determine whether they had a
significant influence on memory performance.
Arousal Sensitivity
Individual differences in trait arousal and behavioural responses to arousing situations have
been demonstrated to moderate emotional response, learning, and memory (reviewed in Chapter 2).
Trait arousal was therefore measured in Experiment 1 using the Extraversion scale of the International
Personality Item Pool (IPIP) Big 5 Personality Questionnaire (Goldberg et al., 2006). Situation specific
arousal response tendencies were measured in Experiments 2 and 3 using the Behavioural Inhibition
System and Activation System scales (BIS/BAS) developed by Carver and White (1994). The Extraversion
scale of the Big 5 Personality Questionnaire was selected to measure tonic arousal, which has been
posited by Eysenck (1987) to be reflective of trait extraversion. According to Eysenck’s theory,
extraverts have lower tonic cortical arousal levels than introverts, which manifests behaviourally as the
seeking out of stimulating experiences. In contrast, introverts have higher cortical arousal levels, are
overstimulated by sensory stimuli, and thus withdraw from stimulating experiences. The IPIP Big 5
Personality Questionnaire contains 50 items measuring the factors Surgency/Extraversion,
Agreeableness, Conscientiousness, Emotional Stability, and Intellect/Imagination. To avoid interfering
with the psychometric properties of the questionnaire, all items were administered to participants,
however only the Surgency/Extraversion factor was analysed. Items included statements like “I am the
life of the party” and “I keep in the background” with the response options of 1, very inaccurate to 5,
very accurate, centred at 3, neither inaccurate nor accurate. Half of the items were reversed scored.
High values therefore reflected high trait extraversion. Internal consistency for this factor was very good
(Cronbach’s alpha =.87).
The behavioural preference to approach or withdraw from arousing situations has been
demonstrated to influence attention (Gable & Harmon-Jones, 2008a) and memory (Judde & Rickard,
220
APPENDIX L
2010). The BIS/BAS scales were utilised in Experiments 2 and 3 to detect behavioural arousal responses
to specific situations (Carver & White, 1994), like those elicited by the experimental manipulations. The
BIS/BAS scales thereby provide a more ecologically valid measure of behavioural motivation. The
BIS/BAS scales measure an individual’s propensity to approach or withdraw from arousing situations.
The behavioural inhibition system controls the experience of anxiety in response to anxiety-relevant
cues (punishment, non-reward, and novelty) and inhibits movement towards goals. The behavioural
activation system is sensitive to cues that signal reward, non-punishment, and escape from punishment,
and increases movement towards goals (see review by Carver & White, 1994). In an experimental
setting, it would be predicted that high BIS scoring individuals would exhibit higher arousal responses to
negative emotional stimuli, and high BAS scoring individuals would exhibit higher arousal responses to
positive emotional stimuli. BIS scale items included the statements “I worry about making mistakes”
and “I have very few fears compared to my friends” (reverse scored). The BAS scale was comprised of
the subscales; reward responsiveness, drive, and fun seeking. Examples of an item from each subscale
are, in consecutive order; “It would excite me to win a contest”, “I go out of my way to get what I want’,
and “I crave excitement and new sensations”. Response options were 1, very true for me to 4, very false
for me. All but two items were reversed scored. High values therefore reflected high scores on each of
the four scales. Internal reliability for the scales was acceptable (Cronbach’s alpha = .66 to .76). As the
motivation to approach or withdraw was the main behaviour of interest, scores from the BAS subscales
were summed.
Gender
Empirical studies have identified a multitude of gender differences that are relevant to emotion
and memory. Males and females differ in the way they respond to emotional stimuli in terms of
amygdala activity, reaction time, hemispheric asymmetry, physiological reactivity, hormonal reactivity,
attention, and more. For instance, in a review of physiological studies of emotion, Bradley and Lang
(2000) revealed that females had greater smile muscle activity than males, females were more facially
expressive than males; and that males had greater SCR response than females. Emotion-memory
studies have consistently found differences in memory performance between males and females (Cahill,
Gorski, Belcher, & Huynh, 2004a; Cahill et al., 2004b; Cahill & van Stegeren, 2003; Gasbarri et al., 2006),
which may have been caused by these differences in emotional responsiveness.
221
APPENDIX L
Music Experience
Music experience has been demonstrated to increase arousal responsiveness to music (Steinbeis
et al., 2006; Vanderark & Ely, 1992, 1993). However, there are a number of studies that also report that
there are no differences between musicians and non-musicians in their ability to detect or experience
emotions in music (Bachorik et al., 2009; Bigand & Poulin-Charronnat, 2006), or to differ physiologically
in response to emotional music (Bernardi, Porta, & Sleight, 2006; Vieillard, Roy, & Peretz, 2011). Given
the theoretical significance of emotional arousal on memory, combined with the variable effects music
experience can have on emotional responses, it was deemed prudent to measure and analyse the
potential moderating effect of music experience on the relationship between arousal and memory.
Music experience was measured in Experiment 1 using Werner, Swope, and Heide’s (2006) Brief
Music Experience Questionnaire (BMEQ). The BMEQ consists of 53 items, derived from an original set of
141 that measure responses to music on the scales of commitment to music, innovative musical
aptitude, social uplift, affective reactions, positive psychotropic effects, and reactive musical behaviour.
Replication testing conducted by Werner et al. (2006) revealed that all scales except ‘social uplift’ had
reasonable internal consistency (alpha coefficients ranged from .69 to .86), and retest reliability was
acceptable for all scales (Pearson correlations ranged from .60 to .74). Two scales from the BMEQ were
used in Experiment 1 to measure musicianship and musical engagement. Musicianship was derived
from the ‘innovative musical aptitude’ scale, item number 1, “I can easily improvise on an instrument
without having music in front of me”; and item number 28, “People have applauded my performance of
music”, both measured on a 5-point scale from 1 (very untrue) to 5 (very true). The remaining four
items of this scale were not used as they could equally have applied to non-musicians (e.g. item number
23, “Totally new tunes that I’ve never heard before, sometimes pop into my head”). The ‘commitment to
music’ scale was used as an index of non-performance based musical engagement. Items included
statements like “Music is the most important thing in my life”. All seven items from this scale were
used.
Music experience was measured more directly in Experiments 2 and 3 using the criteria set by
Chin and Rickard (2010). One performance based and one non-performance based music variable was
analysed. The performance based variable was time since last played a musical instrument (5 point
scale ranging from ‘less than a week’ to ‘more than 10 years’). Data were split into less than or more
than one year. The non-performance based variable was purposeful music listening (5 point scale
ranging from ‘less than 1 hour’ to ‘more than 6 hours’).
222
APPENDIX L
Music Characteristics
Music distraction, enjoyment (or liking/preference) and familiarity are all interrelated, and in
turn related to arousal and memory. For example, the more familiar the music is, the more likely it is to
be liked (Peretz, Gaudreau, & Bonnel, 1998b). The more liked music is, the more it will be enjoyed and
therefore improve mood. Music that is familiar, liked or enjoyed is more likely to be more preferred;
and familiar, liked, enjoyed or preferred music can increase arousal (Thompson et al., 2001). Arousal
during learning can be detrimental to memory, while arousal before or after learning can facilitate
memory (reviewed in Chapter 2). Music familiarity, liking/enjoyment, and distraction were therefore
measured in Likert type scale format, depending on the aims of each experiment.
L.1 Method of analysis
Moderator analyses were conducted according to the guidelines proposed by Frazier, Tix and
Barron (2004) to examine the influence of individual differences on the relationship between emotional
arousal and memory. Continuous variables were preserved by using hierarchical multiple regression.
The analyses aimed to detect interaction effects of predictor (indices of emotion) and moderator
(individual differences) variables on the outcome variable, in all cases memory. Continuous predictor
and moderator variables were first standardised to simplify interpretation of interaction plots. Gender
and music instrument use in the last 12 months were dichotomous variables coded as ‘-1’ for males and
no music instrument use, and ‘1’ for females and yes, music instrument use. Interactions between the
predictor and moderator variables were calculated by multiplying the predictor z-score by the
moderator z-score or coded variable. The hierarchical multiple regression analysis was conducted in
two steps. In step one, the memory variable of interest was entered as the outcome variable, and
standardised predictor and moderator variables were entered as the predictor variables (enter method).
Step one of the analysis determined the main effect of the emotion variable and individual difference
variable on memory (t value of the coefficient). In step two, the predictor by moderator product
variable was entered. A significant improvement in the ANOVA F value (F change) with the entry of the
interaction variable indicated that an interaction between the predictor and moderator variables
improved the predictive utility of the model. A measure of individual differences thus influenced the
relationship between emotion and memory. Whether addition of the interaction term improved the
223
APPENDIX L
predictive utility of the model beyond chance levels was determined by the omnibus ANOVA F value (p <
.05)
The data for Experiment 1 and Experiment 3 were pooled to enable sufficient sample size for
conducting regression analyses. The categorical variable of sex was excluded from the Experiment 1 and
3 analyses due to unequal sample sizes (Exp. 1. 25% males, Exp. 3. 27% males), as was music
instruments use in Experiment 2 (30% yes responses). The proportion of males in Experiment 2 was
41%, and the proportion of individuals who had played a musical instrument in the previous 12 months
in Experiment 3 was 50%. Note that the sample size was low for regression analyses and results should
be considered with caution.
L.2 Results
A summary of the multiple regression analyses across the three experiments is presented in
Table L1. Highlighted values indicate that the addition of the interaction term in step two of the analysis
improved the predictive utility of the model (omnibus ANOVA F value). In some models, the F change
value of the interaction term was significant, yet chance level for the omnibus ANOVA was greater than
.05. As such, interaction values are highlighted only when probability for the omnibus ANOVA is less
than .05.
224
Table L1. Moderator matrices across the three experiments.
Experiment 1
Music enjoy/like
Music distraction
Music familiarity
Musicianship
Music commitment
Extraversion
Arousal
3,14
3,14
3,14
3,32
3,32
3,32
df
df
df
df
df
df
Valence
3,14
3,14
3,14
3,32
3,32
3,32
SCRmax
3,14 a
3,14 a
3,14
3,27 a
3,27 a
3,27
Experiment 2
SCL
3,14 a
3,14 a
3,14 a
3,28 a
3,28 a
3,28 a
Temp
3,14
3,14
3,14
3,30
3,30
3,30
IBI
3,13
3,13
3,13
3,28
3,28
3,28
EMG
3,14
3,14
3,14
3,27
3,27
3,27
Control
Osm
3,32
3,32
3,30
3,25
3,25
3,32
3,33
3,26
3,26
df
3,33
3,33
3,26
3,26
3,31
3,33
axb
3,33
3,33
BIS
3,31
b
3,31
3,32
b
3,33
3,32
3,33
3,30
axb
3,31
3,32
df
3,33
3,33
3,33
BAS
df
3,33
3,33
3,30
a, b
3,31
a, b
3,31
a
3,31
a
3,31
axb
3,26
axb
3,26
3,31
3,33
axb
3,33
3,33
3,33
Experiment 3
Arousal
Valence
ChillsQ
SCRmin
3,33
IBI
TEMP
df
RT
Purposeful
music listening
Gender
IBIsd
3,33
IBI
3,33
axb
SCRmin
ChillsQ
3,33
SCL
Valence
3,33
ChillsQ
Arousal
3,31
Valence
RT
3,26
Arousal
TEMP
3,26
RT
IBIsd
3,31
TEMP
IBI
3,31
a
IBIsd
SCRmin
3,33
IBI
SCL
3,33
SCRmin
Valence
df
SCL
Arousal
Music enjoy/
like
PsM
3,31
a,
axb
3,30
a, b
3,31
a
3,31
a
3,31
a
3,31
3,26
3,26
3,31
3,33
3,33
3,33
3,33
3,31
a
3,31
3,26
3,26
3,33
3,31
3,30
3,25
3,25
3,30
3,32
3,32
3,32
3,25
3,30
3,26
3,31
3,33
3,33
3,33
3,31
3,26
3,26
3,31
3,33
3,33
3,33
3,33
3,26
axb
3,26
3,33
3,31
3,31
a
3,31
3,25
a
3,26
3,32
3,26
3,31
3,26
3,26
3,31
3,33
3,33
3,33
axb
3,33
3,31
a
3,26
3,25
a
3,26
a
3,33
3,31
3,30
a
3,31
a
3,31
a
3,31
a
3,30
3,31
3,32
b
3,33
3,33
3,31
HFHRV
BVA
RSP
AlphaF
AlphaP
ThetaF
ThetaP
Free
Rec
Free
Rec
Free
Rec
Free
Rec
Free
Rec
Free
Rec
Free
Rec
Free
Rec
Free
Rec
Free
Rec
Free
Rec
Free
Rec
3,40
a
3,40
3,39
3,40
3,39
3,40
3,39
3,38
3,37
3,38
3,37
3,37
3,39
3,38
3,38
3,37
3,39
3,38
3,38
3,37
3,40
3,39
3,40
3,39
3,38
3,37
3,38
3,37
3,37
axb
3,37
3,38
3,39
3,38
axb
3,38
3,38
3,37
3,39
3,38
3,38
3,37
3,39
3,38
3,38
3,37
3,39
3,40
3,39
3,40
3,39
3,38
3,37
3,38
3,37
3,38
3,37
3,38
3,37
3,39
3,38
3,38
3,37
3,39
3,38
3,38
3,37
3,39
3,40
3,39
3,40
3,39
3,38
3,37
3,38
3,37
3,38
3,37
3,38
3,37
3,39
3,38
3,38
3,37
3,39
3,38
3,38
3,37
3,39
a
3,39
3,40
3,39
3,40
3,39
3,38
3,37
3,38
3,37
3,37
3,38
3,37
3,39
3,38
3,38
3,37
3,39
3,38
3,38
3,37
3,40
3,39
3,40
3,39
3,38
3,37
3,38
3,37
3,38
axb
3,38
3,37
3,38
3,37
3,39
3,38
3,38
3,37
3,39
3,38
3,38
3,37
Music enjoy/like
df
3,40
3,39
Music familiarity
df
3,40
3,39
Inst. use in 12 month
df
3,40
3,39
Purposeful listening
df
3,40
3,39
3,40
a
3,40
BIS
df
3,40
3,39
3,40
BAS
df
3,40
3,39
3,40
Note. df degrees of freedom from step two of the regression analysis. Highlighted cells indicate that the addition of the interaction term
improved the predictive utility of the ANOVA model (omnibus ANOVA p < .05). a = significant prediction of memory by a standardised emotion
measure (i.e. main effect of emotion). b = significant prediction of memory by a standardised moderator variable (i.e. main effect of individual
differences). a x b = significant prediction of memory by the interaction between emotion and individual difference measures.
225
APPENDIX L
Of all 266 possible emotion by individual difference interaction effects on memory across the
three experiments, only five reached statistical significance. There were, however, several significant
main effects that were consistent with an arousal-performance relationship. There were significant
associations between increasing arousal levels and higher memory scores in Experiment 1 (SCRmax and
SCL) and Experiment 2 (SCR, SCL, and IBIsd), but not in Experiment 3. The latter result may be due to
lower variability in arousal levels in this cohort of participants relative to those in Experiment 1
(expectation of a disturbing film) and Experiment 2 (listening to personally selected music).
Nevertheless, there was a positive association between valence ratings and memory in Experiment 3,
indicating that as mood improved, so did memory performance. Associations between individual
differences and memory were only observed in Experiment 2, with purposeful music listening having a
positive association with memory for images presented after the control and others’ music conditions,
and females recalling more control condition images than males. It is possible that this result was
attributed to general motivation levels, both of musically engaged individuals given this was a musicbased experiment, and of females. Females tend to be more intrinsically motivated to learn than males
(Severiens & Dam, 1994), which may have led to greater commitment to remember the control
condition images.
The moderating influence of individual differences on the relationship between emotion and
memory was powerful enough to reach statistical significance in Experiment 2 and 3. In Experiment 2,
hours of purposeful music listening and BAS sensitivity moderated the relationship between skin
temperature and response times (respectively) on memory for control condition images. Furthermore,
enjoyment of others’ music in Experiment 2 moderated the relationship between SCL and memory. In
Experiment 3, both music enjoyment and BIS sensitivity moderated the relationship between BVA and
free recall.
Arousal Sensitivity
For high BAS scorers in Experiment 2, increasing response times (an index of cortical
deactivation) were associated with higher memory scores (illustrated in Figure L1A). The reverse
pattern was observed for low BAS scorers. This result indicated that the motivation to approach
arousing situations moderated the relationship between cortical arousal and memory. However,
individuals that scored high in approach motivation and were cortically activated had lower memory
scores than those who scored low in approach motivated and were cortically activated. It therefore
226
APPENDIX L
appears that high BAS interfered with memory when arousal was high, and enhanced memory when
arousal was low. This result was in contrast to that expected for behavioural arousal sensitivity effects
on the arousal-memory relationship. Perhaps high BAS individuals were already at optimal baseline
arousal levels and the increased arousal elicited by the non-music control interfered with memory.
However, this seems unlikely given the neutral nature of the control stimulus.
The moderating effect of BIS on the relationship between arousal and memory in Experiment 3
was also not as expected (illustrated in Figure L1B). For individuals who tended to avoid arousing
situations (high BIS), increasing arousal levels (indexed by lower BVA) were associated with increasing
memory scores. It therefore appears that increasing arousal levels benefited memory performance for
these individuals. In contrast decreasing arousal levels benefited memory performance for those who
scored low on the BIS scale. Both of these arousal sensitivity effects on memory were opposite to
expected. Further testing is therefore warranted to confirm whether these results are reliable.
B
Exp. 2 BAS
35
30
25
20
15
10
5
0
Free Recall % Retained
Detail Free Recall
A
BAS Low
(-1SD)
BAS High
(+1SD)
Fast (-1SD) Slow (+1SD)
Control RT
Exp. 3 BIS
75
70
65
60
55
50
45
40
35
BIS Low
(-1SD)
BIS High
(+1SD)
Low (-1SD) High (+1SD)
BVA
Figure L1 (A) Influence of BAS sensitivity on the relationship between response time (RT) and non-music
control image free recall. (B) Influence of BIS sensitivity on the relationship between blood volume
amplitude (BVA) and word free recall.
Music Experience
Experiment 2 individuals who were more musically engaged, indexed by time spent purposefully
listening to music each day, showed a positive relationship between skin temperature and memory for
the non-music control images (illustrated in Figure L2). This result indicated that decreases in arousal
levels (increased skin temperature) were associated with increases in memory performance. In
contrast, there was a negative relationship between skin temperature and memory for those who were
less inclined to music listening in this way. The latter result is consistent with an arousal-performance
relationship, in which decreasing arousal levels are associated poorer memory performance. It is
227
APPENDIX L
difficult to make any meaningful inferences from this result as it is unclear why the music engagement
variable moderated memory for the non-music control condition and not for the music conditions.
Perhaps musically engaged individuals were over-aroused and listening to the non-music control
reduced arousal to optimum performance levels. The failure to detect a consistent memory moderating
influence of music engagement across the three conditions in Experiment 2, or in Experiments 1 and 3
thus casts doubt on the reliability of this result.
Detail Free Recall
Exp. 2 Hours of Purposeful Music
Listening per Day
35
30
25
20
15
10
5
0
Hours Low
(-1SD)
Hours High
(+1SD)
Low (-1SD)
High (+1SD)
Control Temp
Figure L2 Influence of hours of purposeful music listening each day on the relationship between skin
temperature and free recall of the non-music control images.
Music Characteristics
The regression analyses revealed that music enjoyment in Experiment 2 and music liking in
Experiment 3 moderated the relationship between physiological arousal and memory (illustrated in
Figure L3). For individuals who had higher enjoyment or liking ratings, there was a positive relationship
between arousal levels (indexed by increasing SCL in Exp. 2, Figure L3A; and decreasing BVA in Exp. 3,
Figure L3B). In contrast, for those with lower music enjoyment ratings, there was either no evidence of
a relationship (Exp. 2) or a negative relationship (Exp. 3) between arousal and memory. These results
are consistent with those reported by Schellenberg, Thompson, Jones, and others (Jones et al., 2006;
Nantais & Schellenberg, 1999b; Schellenberg & Hallam, 2005; Thompson et al., 2001), who
demonstrated that positive affect elicited by music was a significant contributing factor to the
relationship between arousal and cognitive performance.
228
APPENDIX L
Exp. 2 Music Enjoyment
35
30
25
20
15
10
5
0
B
Free Recall % Retained
Detail Free Recall
A
Enjoy Low
(-1SD)
Enjoy High
(+1SD)
Low (-1SD)
Exp. 3 Music Liking
75
70
65
60
55
50
45
40
35
Like Low
(-1SD)
Like High
(+1SD)
Low (-1SD) High (+1SD)
High (+1SD)
BVA
OsM SCL
Figure L3 (A) Influence of music enjoyment on the relationship between SCL and OsM condition image
free recall. (B) Influence of music liking on the relationship between blood volume amplitude (BVA) and
word free recall.
L.3 Testing information processing speed mediation of arousal and memory
Analysis of the Experiment 2 PsM effects on positive emotion reported in Section 6.3.2 revealed
that as SCL increased, time to respond (RT) to the task of judging whether a line drawing of a face was
positive or negative reduced. Correlational analysis revealed a significant correlation between SCL and
RT, confirming that as SCL increased, RT decreased. There was also a weak association between
decreasing RT and increasing memory scores. Further analysis of the relationship between the arousal
measures and memory after controlling for individual differences (reported in the previous section)
revealed that RT was significantly associated with memory after controlling for BAS sensitivity. The
combination of these results raised the possibility that RT mediated the relationship between SCL and
memory. The mechanism underlying increased SCL may have also improved information processing
speed (indexed by RT). This distinction may be of some importance as SCL is a measure of PNS
activation, and RT is a measure of CNS activation. Increased PNS activation may therefore have resulted
in increased CNS activation, and CNS activation may account for increased memory performance.
The possibility that RT mediated the relationship between SCL and memory was tested using a
series of simultaneous regression models, with free recall of images in the collage presented after PsM
the outcome variable. As the relationship between RT and memory was moderated by BAS in the
control condition, two sets of mediation analyses were conducted, each on data split by high and low
229
APPENDIX L
BAS scorers. The significant reduction in the relationship between SCL and memory after controlling for
the relationship between RT and memory would confirm a mediation effect (Frazier et al., 2004). Beta
values from the mediation analysis are presented in Figures L4 (low BAS) and L5 (high BAS).
The ‘low BAS’ model revealed a weak trend for increases in SCL to predict increased memory
scores (Figure L4A, Path c), SCL increases to significantly predict decreased RT (Figure L4B, Path a), and
decreased RT to significantly predict increased memory scores after controlling for SCL (Figure L4B, Path
b). There was a reduction in SCL prediction of memory (Figure L4A, Path c) after controlling for the
relationship between RT and memory (Figure L4B, Path c’), thus demonstrating partial RT mediation of
memory. However, Sobel’s test (Preacher & Hayes, 2004) revealed that the difference between the
direct effect (Path c) and indirect effect (Path c’) failed to reach statistical significance, Z = 1.57, p = .11
(LL 95 CI = -.12, UL 95 CI = 1.07). The ‘high BAS’ model revealed that increases in SCL significantly
predicted increased memory scores (refer to Figure L5A, Path c). However, SCL increases did not predict
decreased RT (Figure L5B, Path a), nor did decreased RT predict increased memory scores after
controlling for SCL (Figure L5B, Path b). Sobel’s test revealed that the small reduction in SCL prediction
of memory (Figure L5A, Path c) after controlling for the relationship between RT and memory (Figure
L5B, Path c’) failed to reach statistical significance, Z = -.08, p = .93 (LL 95 CI = -.20, UL 95 CI = .35). The
results thus demonstrated partial RT mediation of the relationship between SCL and memory for low
BAS scorers, and no RT mediation effect for high BAS scorers.
230
APPENDIX L
Figure L4 Testing RT mediation of the relationship between SCL and memory for low BAS scorers.
Figure L5 Testing RT mediation of the relationship between SCL and memory for high BAS scorers.
The moderating effect of BAS sensitivity on arousal (indexed by shorter response times to judge
faces as positive or negative) and collage recall further supported arousal mediation of memory.
Increasing arousal levels for individuals low in BAS sensitivity were related to increasing memory scores,
231
APPENDIX L
whereas increases in arousal levels for high BAS sensitive individuals had mildly negative associations
with memory. The association between increasing arousal levels and memory scores for low BAS
sensitive individuals may therefore be consistent with an inverted U-shaped relationship between
arousal and performance. The mild and slightly negative association between arousal and memory for
BAS sensitive individuals (higher BAS scorers) may have reflected an impairing effect of over-arousal on
memory. Perhaps individuals high in BAS sensitivity were already operating at peak arousal levels. The
increase in arousal levels elicited by the experimental conditions may therefore have had little additive
effect on memory performance. In contrast, those who were low in BAS sensitivity, and therefore less
susceptible to increased arousal elicited by extraneous factors, benefited from arousal elicited by the
music conditions, leading to increased memory performance.
L.4 Conclusion
Individual differences in arousal sensitivity, music experience, and music characteristics
moderated the relationship between arousal and memory. However, the effects were sparse and
inconsistent (as illustrated Table L1). Furthermore, it was unclear why individual differences would
moderate the relationship between arousal and memory in only one of the three Experiment 2
conditions, and why only one of the several indices of arousal where influenced at any one time. There
was general support for the hypothesis that arousal sensitivity moderates the relationship between
arousal and memory. However, the relationships were not always in the expected direction. The
moderating effect of music enjoyment/liking on memory was consistent with previous music-cognitive
performance research. It is thus recommended that future researchers measure enjoyment of the
experimental stimuli. The potential influence of enjoyment on arousal and cognitive performance can
thus be controlled and variability in the data reduced.
232
APPENDIX M
___________________________________________________________________________________
M. High versus low chills responder analysis
OsM
Chills
Enjoy
Valence
Arousal
FaceRT
SCL
SCRmin
TEMP
IBI
IBIsd
PsM
Chills
Enjoy
Valence
Arousal
FaceRT
SCL
SCRmin
TEMP
IBI
IBIsd
Chills
Low
High
Low
High
Low
High
Low
High
Low
High
Low
High
Low
High
Low
High
Low
High
Low
High
N
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
9
10
9
Mean
0.00
1.45
5.05
5.00
0.85
1.90
1.95
0.40
6.42
6.35
19.42
16.73
1.18
0.64
30.88
32.83
791.45
872.23
49.56
56.12
SD
0.00
0.44
0.76
0.97
1.67
1.33
1.42
2.13
0.21
0.22
10.72
4.75
0.48
0.62
4.40
1.84
82.88
104.27
15.15
18.51
T
-10.47
df
18
Sig
.00
0.13
18
.90
-1.56
18
.14
1.91
18
.07
0.73
18
.47
0.73
18
.48
2.18
18
.04
-1.29
18
.21
-1.88
17
.08
-0.85
17
.41
Low
High
Low
High
Low
High
Low
High
Low
High
Low
High
Low
High
Low
High
Low
High
Low
High
14
14
14
14
14
14
14
14
14
14
14
13
14
13
14
13
14
13
14
13
0.98
4.57
5.93
6.82
2.07
2.57
0.68
0.89
1.67
0.73
0.32
1.31
2.54
2.35
1.58
0.23
0.25
9.27
7.10
0.58
0.57
3.80
6.40
110.84
75.25
15.28
19.11
-7.08
26
.00
-4.20
26
.00
-0.65
26
.52
-3.45
26
.00
-0.07
26
.95
-0.37
25
.71
2.45
25
.02
1.11
25
.28
2.00
25
.06
0.28
25
.79
6.35
6.35
20.90
22.10
1.40
0.86
31.66
29.43
861.25
787.72
50.02
48.20
233
APPENDIX M
___________________________________________________________________________________
Music-Mix and
PA Music
Chills
Chills
Low
High
Liking
Low
High
Valence
Low
High
Arousal
Low
High
SCRminChange
Low
High
IBIchange
Low
High
pcHFHRVchange
Low
High
BVAchange
Low
High
RSPchange
Low
High
AlphaFrontLnChange Low
High
AlphaPzLnChange
Low
High
ThetaFrontLnChange Low
High
ThetaPzLnChange
Low
High
N
15
15
15
15
15
15
15
15
15
15
14
15
14
15
14
15
14
15
14
15
14
15
14
15
14
15
Mean
SD
0.33
2.77
4.73
5.30
1.43
2.13
2.87
2.27
0.45
0.57
-4.11
-17.56
3.22
6.27
-1.72
-5.53
3.95
4.52
-0.06
-0.16
-0.08
-0.23
0.00
-0.08
-0.13
-0.19
0.52
1.03
1.51
1.13
2.08
2.01
1.25
2.36
1.39
1.64
37.32
38.40
18.89
21.39
3.70
3.69
2.78
3.98
0.20
0.30
0.19
0.26
0.18
0.14
0.18
0.24
t
df
Sig
-8.14
28
.00
-1.16
28
.26
-0.94
28
.36
0.87
28
.39
-0.23
28
.82
0.96
27
.35
-0.41
27
.69
2.78
27
.01
-0.45
27
.66
1.08
27
.29
1.85
27
.08
1.29
27
.21
0.80
27
.43
Experiment 2 high and low chills responders were selected according to the medians for others’ music
(OsM) and participant selected music (PsM). Attempts were made to ensure that the full set of
physiological data was included in the two groups, therefore not all participant data were included.
Experiment 3 high and low chills responders were selected according to the median for both the musicmix and positive and arousing music.
The non-music control in both experiments was not analysed.
234
ERRATA
___________________________________________________________________________________
235
ERRATA
___________________________________________________________________________________
236
ERRATA
___________________________________________________________________________________
237

from monash.edu.au - Monash University Research Repository

Transcription

Similar documents

How to live to 100 and stay out of Emergency Department

sections in the Publication Manual.)

Two Dimensional Shapes for Emotional Interfaces

body horror

The Laws of Emotion

PlayStation 2 Architecture

Competitiveness Up, Anxiety Down Using a

You, Me, and My Brain: Self and other representations

Chapter 3 Personalized Music Emotion Prediction

The Sociology of Emotions [Thoits]