Frequency of gamma oscillations in humans is modulated by

Transcription

J Neurophysiol 114: 244 –255, 2015.
First published April 29, 2015; doi:10.1152/jn.00232.2015.
Frequency of gamma oscillations in humans is modulated by velocity of
visual motion
Elena V. Orekhova, Anna V. Butorina, Olga V. Sysoeva, Andrey O. Prokofyev,
Anastasia Yu. Nikolaeva, and Tatiana A. Stroganova
The MEG Centre, Moscow State University of Psychology and Education, Moscow, Russia
Submitted 9 March 2015; accepted in final form 23 April 2015
magnetoencephalography; visual cortex; frequency of gamma oscillations; electroencephalography; spatial suppression
are generated in excitatory-inhibitory neural circuits and mutually connected inhibitory interneuron networks. These oscillations depend crucially on
inhibitory GABA pathways, with fast-spiking (FS) parvalbumin-expressing (PV) interneurons playing a central role (Cardin et al. 2009; Carlen et al. 2012; Gulyas et al. 2010; Sohal et
al. 2009; Volman et al. 2011). High-frequency gamma oscillations seem to be of particular interest for understanding
various brain dysfunctions in neuropsychiatric disorders (Uhlhaas et al. 2011).
Gamma oscillations are reliably induced in the visual cortex
of human subjects by drifting gratings (Hoogenboom et al.
2006; Muthukumaraswamy and Singh 2013; Rice et al. 2013).
NEURAL GAMMA OSCILLATIONS
Address for reprint requests and other correspondence: E. V. Orekhova,
MEG Centre, Moscow State University of Psychology and Education, Shelepihinskaja Embankment 2a, 123290 Moscow, Russia (e-mail:
[email protected]).
244
The peak frequency of magnetoencephalography (MEG)-recorded visually-induced gamma oscillations is an individually
stable (Muthukumaraswamy et al. 2010) and highly genetically
determined (van Pelt et al. 2012) trait. However, the neurophysiological mechanisms that control the frequency of gamma
oscillations in human subjects have been intensely debated
(Brunet et al. 2014; Cousijn et al. 2014; Schwarzkopf et al.
2012; Swettenham et al. 2009).
Some insight into the neural basis of gamma frequency
control is coming from recent animal studies. It has been
shown that gamma frequency is affected by genetic and pharmacological manipulations of the excitability of FS PV interneuron (Anver et al. 2011; Ferando and Mody 2015; Mann
and Mody 2010). In particular, Mann and Mody (2010) found
that the frequency of hippocampal gamma oscillations can be
modulated through altering the balance between tonic excitation of inhibitory interneurons through the specific NR2D
subunit-containing N-methyl-D-aspartate receptors (NMDARs)
and tonic inhibition of these interneurons through ␦-GABAa
receptors.
Changes to the properties of sensory stimulation also may
affect gamma frequency. Gray and Di Prisco (1997) observed
a substantial increase in visual gamma oscillation frequency
with increasing velocity of motion of high-contrast gratings in
cats. These velocity-related modulations of gamma frequency
might reflect the intensity of the excitatory drive to FS PV
interneurons and provide valuable information on FS PV dysfunction in brain disorders. However, the effect of motion
velocity on gamma frequency has not been tested in human
subjects to date.
In the present study, we aimed to investigate the frequency
modulation of gamma oscillations in typically developing
children using visual gratings that move with a range of
velocities close to those previously used in animal studies
(1.2– 6.0°/s). We expected to find an increase in gamma frequency with increasing stimulus velocity similar to that found
for the local field potential (LFP) in animals (Gray and
DiPrisco 1997). If present in humans, the velocity-related
changes in gamma frequency might be less affected by morphological differences between individuals and more sensitive
to functional aspects of inhibitory circuitry than the gamma
frequency measured in response to a single type of stimulation.
Apart from controlled experimental manipulations, the frequency of visual gamma oscillations in pediatric populations
may be affected by development of GABAergic inhibitory
circuitry that continues up to early adulthood (Le Magueresse
and Monyer 2013). Indeed, synaptic elimination proceeds beyond adolescence into the third decade of life and occurs
0022-3077/15 Copyright © 2015 the American Physiological Society
www.jn.org
Downloaded from http://jn.physiology.org/ by 10.220.32.246 on October 20, 2016
Orekhova EV, Butorina AV, Sysoeva OV, Prokofyev AO, Nikolaeva
AY, Stroganova TA. Frequency of gamma oscillations in humans is modulated by velocity of visual motion. J Neurophysiol 114: 244–255, 2015.
First published April 29, 2015; doi:10.1152/jn.00232.2015.—Gamma
oscillations are generated in networks of inhibitory fast-spiking (FS)
parvalbumin-positive (PV) interneurons and pyramidal cells. In animals, gamma frequency is modulated by the velocity of visual motion;
the effect of velocity has not been evaluated in humans. In this work,
we have studied velocity-related modulations of gamma frequency in
children using MEG/EEG. We also investigated whether such modulations predict the prominence of the “spatial suppression” effect
(Tadin D, Lappin JS, Gilroy LA, Blake R. Nature 424: 312-315, 2003)
that is thought to depend on cortical center-surround inhibitory mechanisms. MEG/EEG was recorded in 27 normal boys aged 8 –15 yr
while they watched high-contrast black-and-white annular gratings
drifting with velocities of 1.2, 3.6, and 6.0°/s and performed a simple
detection task. The spatial suppression effect was assessed in a
separate psychophysical experiment. MEG gamma oscillation frequency increased while power decreased with increasing velocity of
visual motion. In EEG, the effects were less reliable. The frequencies of
the velocity-specific gamma peaks were 64.9, 74.8, and 87.1 Hz for the
slow, medium, and fast motions, respectively. The frequency of the
gamma response elicited during slow and medium velocity of visual
motion decreased with subject age, whereas the range of gamma frequency modulation by velocity increased with age. The frequency
modulation range predicted spatial suppression even after controlling
for the effect of age. We suggest that the modulation of the MEG
gamma frequency by velocity of visual motion reflects excitability of
cortical inhibitory circuits and can be used to investigate their normal
and pathological development in the human brain.
GAMMA FREQUENCY IS MODULATED BY VELOCITY
A
245
B
1.2 °/s
3.6 °/s
6.0 °/s
Fig. 1. A: stimulus used in the study. B: grand average time-frequency plots of the baseline-normalized gamma power (in dB) in response to gratings moving
with 3 velocities. A selection of posterior channels is shown at far right. Vertical lines mark stimulus onset. The most prominent gamma increase is observed
at the middle occipital gradiometer sensor pair (MEG2112-MEG2113).
METHODS
Subjects. Twenty-seven boys (8 –15 yr; mean age ⫽ 136.1 mo,
SD ⫽ 23 mo) were recruited from local schools via advertisements.
The criteria for inclusion were the absence of neurological abnormalities in the subject’s medical history and normal or corrected to
normal vision.
This study was approved by the local ethics committee of the
Moscow University of Psychology and Education and was conducted
following the ethical principles regarding human experimentation
(Helsinki Declaration). Written informed consent was obtained from
a parent/guardian of each child.
Experimental task. Subjects sat in a magnetically shielded room
(AK3b; Vacuumschmelze, Hanau, Germany), with the back of the
head resting against the helmet-shaped surface of the helium dewar.
The stimulus comprised a black and white annular sine wave grating
with a spatial frequency of 1.66 circles per degree and a diameter of
18° of visual angle (Fig. 1A). The stimulus was generated using
Presentation software (Neurobehavioral Systems) and projected to a
screen located 1.1 m in front of the participants with a Panasonic
PT-D7700E-K DLP projector. The moving image was presented with
1,280 ⫻ 1,024 screen resolution and a 60-Hz refresh rate. The grating
appeared in the center of the screen over a black background and
“contracted” to the central point at one of three velocities, 1.2, 3.6, or
6°/s, referred to below as “low,” “medium,” and “fast,” respectively.
The frequency of black to white transitions at a given position in the
visual field produced by these velocities was 4, 12, and 20 Hz,
respectively. In the intervals between the stimuli, a small white
fixation cross was presented on the black background in the center of
the screen. The participants were instructed to constantly maintain
their gaze at the fixation cross between stimuli.
Each trial began with the presentation of the fixation cross for 1,200
ms, followed by the moving grating. The moving grating appeared for
an interval that randomly varied between 1,200 and 3,000 ms. The
stimulus then stopped moving and remained on the screen for an
additional individually adjusted time (“response period:” ⬃900 ms)
during which the participant was instructed to detect the termination
of the motion and respond by pressing a button as quickly as possible.
If no response occurred within the response period, the grating was
replaced with a discouraging message “too late!” that remained on the
screen for 2,000 ms. The next trail began immediately after the button
press or the display of the discouraging message. To minimize visual
and mental fatigue, short (3– 6 s) animated cartoon characters were
presented during the rest period between every 2 and 5 grating stimuli.
The length of the response period was adjusted individually for
each participant during a short training session that directly preceded
the main experiment. During this training session, 18 stimuli (6 of
each type) were presented and the mean reaction time and its standard
deviation calculated. The individual response time for the main
experiment was then set at the mean reaction time (RT) plus three
standard deviations.
Individual RT and commission/omission errors were measured for
all but one participant. The experiment included three blocks of trials;
stimuli of different velocities were presented in all three blocks in a
pseudorandom order. The participants responded with either the right
or left hand in a sequence that was counterbalanced between blocks
and participants. Each stimulus velocity was presented 30 times
within each experimental block, resulting in 90 trials per stimulus
velocity.
MEG and electroencephalography recording. Although MEG is
more sensitive to visual gamma oscillations than electroencephalography (EEG) (Muthukumaraswamy and Singh 2013), its application is
limited in some groups of subjects (e.g., young children and individuals with low IQ). In the present study, we compared the effectiveness
of MEG and EEG for detection of the effects of stimulus velocity on
sustained visual gamma oscillations induced by moving gratings.
Neuromagnetic activity was recorded with a helmet-shaped 306channel detector array (Vectorview; Neuromag, Helsinki, Finland)
that comprised 102 identical triple sensor elements. Each sensor
element consisted of three superconducting quantum interference
devices, two with orthogonal planar gradiometer pickup coils and one
with a magnetometer pickup coil configuration. In this study, the
recordings from the 204 planar gradiometers were used for analyses.
J Neurophysiol • doi:10.1152/jn.00232.2015 • www.jn.org
predominantly in supragranular cortical layers that are thought
to be the main source of gamma oscillations (reviewed in Cho
et al. 2015; also see Oke et al. 2010). Investigations of gamma
oscillations in young participants may cast light on the typical
development of inhibitory cortical circuits and their pathology
in neurodevelopmental disorder. To assess maturational
change in gamma frequency, we included children and adolescents aged 8 to 15 yr in this study.
The other aim of this study was to investigate whether
interindividual variations in frequency characteristics of visual
gamma oscillations have consequences for visual perception,
particularly for the spatial suppression effect. It has been
shown that it is normally more difficult to detect direction of
motion of large (i.e., filling ⱖ5° of the visual field) rather than
small (⬃1° of the visual field) high-contrast visual stimuli
(Foss-Feig et al. 2013; Tadin et al. 2003; Tadin et al. 2011).
This difficulty with detecting of motion direction (the so-called
spatial suppression effect) is thought to result from stronger
surround inhibition with large high-contrast stimuli (Tadin et
al. 2003; Tadin et al. 2011). It is likely that the interindividual
differences in neuronal inhibition (and in gamma oscillation
frequency) may affect the magnitude of this spatial suppression
effect. Therefore, we have also tested for a putative link
between gamma frequency parameters and the spatial suppression effect assessed in a separate psychophysical experiment as
a part of this study.
246
same location. Based on previous neuroimaging studies (Hoogenboom et al. 2006; Muthukumaraswamy and Singh 2013), we expected
the visual sustained gamma response to be restricted to posterior MEG
sensors overlaying the occipital lobe. Therefore, we used a posterior
selection of planar gradiometers (Fig. 1B) for subsequent analysis.
Assessment of gamma response. We expected that the faster motion
of the annual grating would result in higher frequency of gamma
response. To reveal velocity-specific power changes at each of the
three stimulus velocities, we performed group level statistical analysis
of the stimulus-related changes in spectral power. The analysis proceeded in the following steps. First, we calculated average logtransformed baseline-normalized power (in dB) in 400- to 1,000-ms
poststimulus intervals for each frequency bin within 50 –120 Hz. This
time window contains a strong sustained gamma response to visual
motion and is not contaminated by stimulus onset response (Hoogenboom et al. 2006; Muthukumaraswamy and Singh 2013). Second, we
selected a gradiometer pair that displayed the highest gamma power
increase in response to motion in this time-frequency range. For all
three velocities, this was the gradiometer pair channel nos. 2,112 and
2,113 that correspond to the sensor element located approximately at
the middle occipital position. Third, we searched the data from this
gradiometer pair for velocity-specific frequencies that demonstrate
baseline-normalized spectral power for a given stimulus velocity
exceeding that of the two other velocities (while properly correcting
for family-wise error using the permutation approach). To correct for
multiple comparisons, the baseline-normalized spectral power values
at each frequency bin within the 50- to 120-Hz range were randomly
permuted 1,000 times between the two data sets (e.g., low- and
medium-velocity conditions). The maximum t-value from each permuted data set was taken across all frequency values when forming
the empirical null distribution and comparing it with original statistics,
with P ⬍ 0.05 as the significance threshold. The velocity-specific
frequencies were computed as an overlap between the results of the
two unidirectional pairwise comparisons (e.g., “low ⬎ medium” and
“low ⬎ fast” for the slow velocity; see the blue bar under the top right
graph in Fig. 4).
In each subject and for each stimulus velocity, we also calculated
individual velocity-specific peak frequencies (VSPFs). Considering
possible individual variations in head position and cortical morphology, the individual VSPF was always defined at the gradiometer pair,
where the maximal gamma response was observed in this particular
subject for a particular velocity. To ensure that the individual values
comply with the group gamma response, the search for the gamma
VSPF was limited to a certain frequency window. This window was
centered at the frequency of the group average gamma peak that was
closest to the velocity-specific frequencies. This peak overlapped with
“velocity specific” frequencies (see Fig. 4) for the slow and medium
velocities (62.5 Hz for the slow and 80 Hz for the medium velocities)
and was close to the velocity-specific frequencies for the fast velocity
(where it constituted 92.5 Hz). The width of each velocity-specific
window was set at the full width at half-maximum. This resulted in
velocity-specific frequency windows of 52.5–77.5 Hz for the slow
velocity, 55.0 – 87.5 Hz for the medium velocity, and 55.0 –97.5 Hz
for the high velocity. The individual spectra were smoothed with a
three-point moving average filter. After the smoothing, the algorithm
searched for the peaks of the baseline-normalized power within the
velocity-specific windows. Given that a specified frequency window
may cover several local maxima, the peak of the highest frequency
was always taken from all of the peaks that exceeded the baseline
value by at least two standard deviations. This approach makes
explicit the assumption that there may be several spectral peaks in
gamma response (see, e.g., Gray and DiPrisco 1997) and is less
susceptible to arbitrary detection of a local maximum. As we
discuss later, the applied approach to the gamma peak frequency
definition reduces the possible contribution from the photic driving
effect at 60 Hz.
We also recorded EEG (in all but 1 subject) at the Fz, Pz, Oz, O1, and
O2 positions referenced to the left earlobe. EEG impedance was kept
below 15 k⍀. The Oz electrode (that demonstrated the highest power
of gamma response on the group level) was the only one used for the
EEG analysis.
Prior to the MEG session, the positions of a set of four HPI coils
were digitized together with fiducial points using a Polhemus Fastrak
3D digitizer (Fastrak; Polhemus, Colchester, VT); the HPI coil positions were further used to track each subject’s head position inside the
helmet during recordings. The spatiotemporal signal space separation
method method (Taulu et al. 2004) implemented by MaxFilter (Elekta
Neuromag software) was used to suppress interference signals generated outside of the helmet/sensor array volume. Head movement
compensation was used as well. The data were converted to standard
head position (x ⫽ 0, y ⫽ 0, and z ⫽ 45 mm) across all experimental
blocks.
Four electrooculogram (EOG) electrodes were used to record
horizontal and vertical eye movements. EOG electrodes were placed
at the outer canti of the eyes and above and below the left eye. EEG
and EOG signals were recorded with a high-pass filter of 0.1 Hz.
MEG, EEG, and EOG signals were recorded with a band-pass filter of
0.03–330 Hz, digitized at 1,000 Hz, and stored for off-line analysis.
MEG and EEG data preprocessing. The correction of vertical eye
movement artifacts was performed for the continuous MEG data with
Brainstorm (http://neuroimage.usc.edu/brainstorm) (Tadel et al.
2011), using the SSP algorithm (Tesche et al. 1995; Uusitalo and
Ilmoniemi 1997). The subsequent analyses were done using the
SPM12 toolbox (http://www.fil.ion.ucl.ac.uk/spm) (Litvak et al.
2011). The pipeline was optimized for analysis of visual gamma
oscillations through defining a “spatio-spectral” region of interest over
the occipital cortex.
Data were epoched from ⫺0.5 to 1.2 s relative to the stimulus
onsets. The MEG and EEG epochs containing strong muscle artifacts
were excluded by thresholding the mean absolute values of highfrequency signals. The threshold was set at five standard deviations
from the amplitude of the 70-Hz high-passed signal averaged across
channels. The remaining epochs were further visually inspected in
SPM12, and the epochs that displayed longer than 100-ms episodes of
MEG average absolute amplitude exceeding the baseline value approximately three times or more were also excluded from the analysis.
The average number of artifact-free epochs per condition was 83
(range: 56 –90).
For efficient spectral estimation from a relatively small number of
trials, we used multitaper spectral analysis (Thomson 1982). This
method is based on premultiplying the data with a series of tapers
optimized for producing uncorrelated estimates of the spectrum in a
given frequency band. This approach sacrifices some of the frequency
resolution for an increase in signal-to-noise ratio. It does so by
effectively multiplying the number of trials by the number of tapers
used. We estimated the spectra in overlapping windows of 400 ms
(shifted by 50 ms). The frequency resolution was set to the inverse of
the time window (2.5 Hz) for ⱕ25 Hz, 0.1 times the frequency for
25–50 Hz. A constant 5-Hz resolution was used for 50 –100 Hz;
7.5-Hz resolution was used for 102.5–120 Hz. These settings resulted
in a single taper being used for 2.5–30 Hz, two tapers for 32.5– 42.5
Hz, three tapers for 45–100 Hz, and four tapers for 102.5–120 Hz. The
resulting time-frequency images had no discontinuities thanks to the
continuous frequency resolution function.
The time-frequency data were averaged across epochs using a
robust averaging procedure (Holland and Welsch 1977; Litvak et al.
2011). To reduce intersubject variability and to normalize power
changes across different frequency bands, the averaged power was log
transformed and baseline corrected using a ⫺500- to ⫺100-ms interval before stimulus onset as the baseline (LogR option in SPM); this
resulted in power changes in dB relative to baseline. Planar MEG
gradiometer channels were then combined by adding time-frequency
data for pairs of channels corresponding to orthogonal sensors at the
For the purpose of the present study, we calculated the spatial
suppression (SS) index as the difference between thresholds for the
large and small high-contrast stimuli normalized by their sum. The
greater SS index values correspond to a stronger SS effect.
RESULTS
MEG experiment: behavioral performance. The behavioral
data were available from 26 of 27 participants. The number of
omission and commission errors was low (omission: mean ⫽
2, range 0 –19 per condition; commission: mean ⫽ 0.6, range
0 –5 per condition) and did not differ between conditions. The
reaction time significantly decreased with increasing stimulus
velocity [ANOVA: F(2, 48)⫽ 18.2, P ⬍ 1e-05], being on
average 435.0 ms for the slow, 413.1 ms for the medium, and
406.3 for the fast velocities.
Gamma response to visual motion. In line with the previous
studies in adults, we observed a strong increase in gamma
power in response to visual motion in children. All three types
of stimuli (”slow,“ ”medium,“ and ”fast“) caused maximal
response over the occipital midline (Fig. 1B). The gamma
response to moving gratings was observed also in EEG, but its
signal-to-noise ratio was much lower than in MEG channels
(Fig. 2).
Effect of stimulus velocity on the magnitude of gamma
response. To test for the effect of velocity on gamma power,
we calculated mean baseline-normalized power (in dB) in the
50- to 120-Hz/400- to 1,000-ms range. Repeated-measures
ANOVA revealed strong reduction of the power of the gamma
response with increasing stimulus velocity [F(2, 52) ⫽ 44.8, P ⬍
10e-5; Fig. 3]. A smaller but still significant effect of velocity
was found also for the EEG gamma power [in Oz: F(2, 50) ⫽
5.0, P ⬍ 0.05].
Velocity-specific increases in gamma power. To check
whether the visual gamma response differentiated reliably
between the three types of stimuli at some ”velocity-specific“
frequencies, we used permutation analysis (see METHODS). The
significant velocity-specific increase of MEG gamma power
was observed at 50 – 67.5 Hz for the slow motion velocity, at
77.5– 82.5 Hz for the medium velocity, and at 95–97.5 Hz for
the fast velocity (Fig. 4). ANOVA with factors ”velocity“ and
velocity-specific frequency range (”range“) revealed the presence of the strong velocity ⫻ range interaction [F(4, 104) ⫽
27.9, ␧ ⫽ 0.64, P ⬍ 1e-5].
MEG
Fig. 2. Grand average time-frequency plots of gamma
power changes (in dB) elicited by gratings drifting at 1.2,
3.6, and 6.0°/s. Top: magnetoencephalographic (MEG) sensors with maximal gamma response (MEG2112 and
MEG2113). Bottom: electroencephalographic (EEG) channel Oz. Note different scales for MEG and EEG.
EEG
The average baseline-normalized MEG gamma power in the 50- to
120-Hz band was always calculated for the same gradiometer pair that
was used to define VSPF. A similar strategy was used to define EEG
VSPF and the average gamma power in the Oz derivation.
Only those participants for which the sustained gamma response
met the criterion for a peak for all three stimuli velocities (19 of 27
participants) were included in the VSPF analysis.
Psychophysical task. Detection of motion direction is hampered for
large, as opposed to small, high-contrast gratings (Betts et al. 2009;
Betts et al. 2005; Tadin et al. 2007; Tadin et al. 2003; Tadin et al.
2011). This decrease in motion sensitivity or spatial suppression was
suggested to reflect surround inhibition of motion-direction-selective
neurons. We tested this spatial suppression effect on a separate day,
within 2 mo, after the main MEG experiment. Visual stimuli were
presented with PsychoToolbox (Brainard 1997), a free Matlab application. To assess spatial suppression, we used the experimental
approach described by Foss-Feig et al. (2013). The psychophysical
data were available from 22 participants, 16 of which were also
included in VSPF analysis.
The stimuli in this task consisted of drifting vertical sine wave
gratings (1 cycle/degree, 4°/s), covered by a two-dimensional Gaussian envelope whose width defined the stimulus size. There were four
types of stimuli, low (1%) and high contrast (100%) stimuli of either
small (2° of visual field/Gaussian width) or large (12°) size. The
direction of motion (left or right) was determined randomly for each
trial.
Participants sat 60 cm from a monitor (Benq XL2420T, 24”W
LED, 1,920 ⫻ 1,080 resolution, 120 Hz). An assistant seated next to
the participant controlled for the correct distance from the monitor,
vertical head position, and adequate task performance. Participants
were asked to make an unspeeded, two-alternative, forced-choice
response indicating the perceived direction of motion by pressing the
left or right arrows on the keyboard. Each block consisted of either
high- or low-contrast trials, with the block order counterbalanced
across participants. The intertrial interval was 400 ms. In the beginning of each trial, a flicking central dot appeared for 100 ms, followed
by the stimulus. The initial stimulus duration was set to 150 ms. The
duration was further adjusted depending on the participant’s response
using two (1 for small and 1 for large stimuli) interleaved one-up
two-down staircases (8.3-ms step) that converged on 71% correct
performance. The block continued until both staircases completed
nine reversals typically lasting ⬃4 min. The first two reversals of each
staircase were excluded from the analysis; the threshold was computed by taking the mean over the remaining reversals. Most participants (and 14 of those 16 that had the VSPFs defined for all stimuli
types) performed two runs of low- and high-contrast blocks. The final
thresholds were averaged over the two runs, when available. A
training block was presented at the beginning of the session to
acquaint subjects with the task and stimulus types.
247
248
Log normalized gamma power (dB)
2.0
1.4
0.8
0.2
3.6 °/sec
6 °/sec
Velocity
Fig. 3. Effect of stimulus velocity on the magnitude of the visual gamma
response. Vertical bars denote 0.95 confidence intervals.
Velocity-specific gamma responses in EEG (channel Oz)
occurred at the same frequencies as in MEG, but they were less
prominent, and no velocity-specific frequencies were detected
after correction for multiple comparisons. Considering that no
significant differences were found with EEG, we further limited analysis to the MEG data.
Possible confounding effects of photic driving. The visual
stimuli were projected at the 60-Hz refresh rate. It is possible
that oscillations synchronized to the refresh frequency could
account for some percentage of gamma synchronization during
stimuli presentation (photic driving effect). To assess the
contribution of the 60-Hz refresh rate in the motion-related
gamma response, we computed evoked activity phase-locked
to the start of the visual stimulation separately for the three
Fig. 4. Gamma responses to moving gratings measured with MEG (sensors MEG2112-MEG2113; top) and EEG (Oz derivation; bottom). Grand average spectra
of normalized gamma power are shown for the three stimulus velocities: ”slow“ (1.2°/s; blue line), ”medium“ (3.6°/s; green line), and ”fast“ (6°/s; red line). The
shadows mark 95% confidence intervals. The ”velocity-specific“ gamma frequencies (P ⬍ 0.05, corrected for multiple comparisons) are marked at top far right
by the corresponding color on the horizontal bars. Insets show spatial distribution of the mean MEG gamma power (in dB) in the 50- to 120-Hz range.
1.2 °/sec
stimulus velocities. We then estimated the reliability of the
60-Hz evoked response using the same ”maximal threshold“
permutation approach that was used to estimate the reliability
of VSPFs on the group level (see METHODS). The effect of the
60-Hz refresh rate was significant for all three velocities (slow,
P ⬍ 0.003; medium, P ⬍ 0.0004; fast, P ⬍ 10e-5). To assess
possible differences in contribution of the evoked activity at
different velocities, we further performed repeated-measures
ANOVA with factor velocity. The effect of velocity was
significant [F(2, 52) ⫽ 20.8, P ⬍ 10e-5]. Planned comparisons
revealed an increase in evoked 60-Hz power from the medium
to the high velocity [F(1, 26) ⫽ 25.3, P ⬍ 0.0001], whereas no
difference between the slow and the medium velocities was
found (P ⫽ 0.2). The significantly higher 60-Hz evoked response to the fast velocity condition could be explained by
contribution of the third harmonic of its black-to-white transition frequency (⬃20 Hz for the fast velocity).
Worthy of note is the fact that our approach of defining the
gamma VSPF minimized the contribution of the 60-Hz evoked
activity because it was based on the choice of the ”right-most“
peak (see METHODS). Indeed, inspection of the individual VSPFs
in Fig. 6A shows that only three subjects displayed VSPFs at
60 Hz (2 for the slow and 1 for the medium velocities).
Gamma VSPFs. Inspection of individual spectra revealed a
velocity-related shift of gamma frequency in virtually all participants who displayed reliable gamma peaks. Figure 5 shows
normalized power spectra for six representative individuals.
The VSPFs constituted on average 64.9 Hz (SD ⫽ 5.2) for the
slow, 74.8 Hz (SD ⫽ 6.0) for the medium, and 87.1 Hz (SD ⫽
6.9) for the fast velocity, thus suggesting a 22-Hz increase in
frequency with increasing velocity of visual motion from 1.2 to
6.0°/s (later addressed as VSPF modulation range).
249
Swettenham et al. (2009) have observed previously that the
individuals with a higher peak frequency in their gamma
response to stationary gratings also displayed a smaller frequency increase in response to motion. These authors suggested the presence of a saturation effect that limited the
frequency of gamma oscillations. To test this hypothesis, we
checked whether the VSPF modulation range could be predicted from the VSPF to the slow velocity. We also included
age as an independent factor. The regression with factors ”age“
and ”slow-velocity VSPF“ was significant [F(2, 17) ⫽ 8.3,
adjusted to r2 ⫽ 0.43, P ⬍ 0.01]. However, the inspection of
partial correlations has shown that the slow velocity VSPF did
not predict the VSPF modulation range independently from age
(P ⫽ 0.13).
B
90
Gamma VSPF (Hz)
Rho = 0.24
80
Rho = -0.58*
70
60
50
100
6.0 °/s
3.6 °/s
1.2°/s
120
Rho = -0.80***
140
160
180
Gamma VSPF modulation range (Hz)
A
Regarding the response to the fast velocity stimuli, 50% of
participants had VSPF of 92.5 Hz, and none of the participants
displayed VSPF that exceeded this value (Fig. 6A). This
finding cannot be explained by the ”border effect“ because the
upper border of the velocity-specific gamma frequency range
for the fast velocity was 97.5 Hz. On the other hand, this
92.5-Hz frequency corresponds to the frequency of the ”highest-frequency gamma peak“ on the group spectrum (Fig. 4C).
This observation confirms the presence of a saturation effect
for the highest motion velocity.
Effect of age. No correlations with age were found for the
power of the gamma response within the 50- to 120-Hz range
at any stimulus velocity (Pearson r, all P ⬎ 0.28). The
correlations between gamma power measured at the VSPFs
40
30
20
Rho = 0.56*
10
0
100
120
140
160
180
Age (months)
Age (months)
Fig. 6. Age-related changes in VSPF of gamma response elicited by visual stimuli moving with different velocities (A) and in gamma VSPF modulation range
(B). *P ⬍ 0.01, ***P ⬍ 0.0001.
Fig. 5. Baseline-normalized power spectra of gamma oscillations induced by the 3 types of stimuli (slow, medium, and fast) in 6 representative participants.
Individual velocity-specific peak frequenies (VSPFs) are marked as follows: Œ, slow; 䊐, medium; {, fast velocities. Note the shift of frequency maximum with
increasing stimulus velocity.
250
Table 1. Prediction of the spatial suppression from the gamma frequency characteristics and age: results of the multiple regression
analysis
SS index (n⫽16)
VSPF Slow, Age
VSPF Medium, Age
VSPF Fast, Age
VSPF Range, Age
r2adj ⫽ 0.04
␤age ⫽ 0.07
␤VSPF slow ⫽ ⫺0.36
r2adj ⫽ 0.10
␤age ⫽ 0.60#
␤VSPF medium ⫽ 0.40
r2adj ⫽ 0.21#
␤age ⫽ 0.14
␤VSPF fast ⫽ 0.48#
r2adj ⫽ 0.29*
␤age ⫽ ⫺0.17
␤VSPF range ⫽ 0.73*
SS, spatial suppression; VSPF, velocity-specific peak frequency; r2adj, modified version of r2 that has been adjusted for the no. of predictors in the model.
#P ⬍ 0.1; *P ⬍ 0.05, VSPF range ⫽ (VSPF fast ⫺ VSPF slow).
DISCUSSION
In the present study, we recorded MEG and EEG in children
during visual presentation of annular gratings contracting at
three different velocities and analyzed gamma oscillations
elicited by this stimulation. Our results can be summarized as
follows. First, the frequency of gamma oscillations increased
with increasing velocity of the visual motion. Second, the older
children demonstrated a wider range of velocity-related modulation of their gamma peak frequency. The frequency of
gamma induced by the slow (1.2°/s) and medium (3.6°/s)
velocities decreased in children from 8 to 15 yr of age, whereas
no age-related changes were observed for the fast (6.0°/s)
motion. Third, the broader range of gamma frequency modulation predicted stronger spatial suppression measured in children in a separate psychophysical experiment.
Modulation of gamma oscillations by velocity. The increase
of peak frequency of gamma oscillations with increasing velocity of visual motion has been reported previously in LFP
recordings in animals (Gray and DiPrisco 1997). The present
study is the first to demonstrate this effect in human subjects.
The effect of velocity on gamma frequency was visible in
both MEG and EEG (Figs. 2 and 4). However, the estimation
of individual gamma peaks in EEG was problematic due to the
low signal-to-noise ratio. Therefore, we limit our further discussions to the analysis of gamma oscillations as recorded by
MEG. The substantially higher sensitivity of MEG than EEG
to gamma oscillations has been reported previously in adults
(Muthukumaraswamy and Singh 2013). The advantage of
MEG over EEG might be even higher in children because they
are more likely to move during the experiment, and EEG was
suggested to be more sensitive to muscular artifacts than MEG
(Muthukumaraswamy and Singh 2013). However, because we
used a single Oz EEG electrode in our study, it is possible that
future multichannel EEG recordings could provide a better
signal-to-noise ratio than what we report herein.
We note that although the 60-Hz photic driving effect was
present in our data and its magnitude depended on stimulus
velocity, this effect could only reduce the observed velocityrelated modulation of gamma frequency. Indeed, the 60-Hz
photic driving effect was highest for the fast velocity. As such,
photic driving cannot be used to explain the observed shift to
higher velocity-specific peak frequencies (74.8 Hz for medium
and 87.1 Hz for fast velocities) or velocity-specific frequency
ranges (77.5– 80 Hz for medium and 95–97.5 Hz for fast
velocities) with increasing velocity of visual motion (Fig. 4).
Unlike numerous findings regarding gamma power (e.g., see
Hermes et al. 2014; Koelewijn et al. 2013; Uhlhaas et al. 2009),
the reports about changes in gamma frequency produced by
manipulating features of visual stimuli in humans are rare. For
example, the effect of stimulus size previously reported in
intracranial studies in animals (Ray and Maunsell 2010) was
shown to be either small (van Pelt and Fries 2013) or absent
(Perry et al. 2013) in MEG recordings of human subjects. A
slightly (on average 5 Hz) lower gamma peak frequency was
observed in MEG recordings on adult subjects in response to
stimuli presented in the periphery (6° of visual angle) compared with stimuli presented in the center of the visual field
(van Pelt and Fries 2013). The studies linked most closely to
ours have investigated how the transition from stationary to
moving gratings changed MEG-recorded gamma activity in the
visual cortex of healthy adults (Muthukumaraswamy and Singh
2013; Naue et al. 2011; Swettenham et al. 2009). There are
notable distinctions between results of the present study and
these previous studies. First, we observed a large increase in
VSPF (22 Hz on average) when the stimulus velocity increased
and subjects’ age were also not significant (slow: Rho ⫽
⫺0.12, P ⫽ 0.6; medium: Rho ⫽ ⫺0.12, P ⫽ 0.6; fast: Rho ⫽
⫺0.36, P ⫽ 0.13). The gamma VSPF reliably decreased with
age for the slow (Rho ⫽ ⫺0.8, P ⬍ 0.0002) and medium
(Rho ⫽ ⫺0.6, P ⬍ 0.005) velocities but did not change with
age for the fast velocity (Rho ⫽ 0.26, P ⫽ 0.26) (Fig. 6A). The
VSPF modulation range reliably increased with age (Rho ⫽
0.56, P ⬍ 0.01, Fig. 6B).
Gamma frequency and spatial suppression. Spatial suppression results were available from 22 participants. In 16 of these
participants, VSPF has been detected for all three stimulus
velocities. The mean time required for correct detection (TCD)
of motion direction was 92.2 ms (SD ⫽ 45.0) for the large
stimuli and 40.8 ms (SD ⫽ 15.6) for the small stimuli. The
increase of the detection time with increasing stimulus size was
highly significant [F(1, 21) ⫽ 54.9, P ⬍ 10e-5], indicating the
presence of a strong spatial suppression effect. We further
calculated an SS index as a difference between TCD for the
high-contrast stimuli of large and small size normalized to the
subject’s mean TCD. The correlation between this SS index
and age was nearly significant (n ⫽ 22, Pearson r ⫽ 0.41, P ⫽
0.061), suggesting a tendency toward a developmental increase
in spatial suppression.
To test whether the visual gamma peak frequency predicts
this SS index, we performed an exploratory regression analysis
with one of the gamma parameters (VSPF for the slow,
medium, or fast velocity or the VSPF range) and age as
independent variables. Results of these analyses are summarized in Table 1. The analysis revealed that the broader VSPF
range predicts greater SS index independently of age. There
was also a tendency for a greater SS index in children with
higher VSPF values to the fast velocity (P ⫽ 0.08). No link
between SS index and frequency of the VSPF to the slow and
medium velocities was found.
velocity of visual motion can be explained by augmented
excitatory drive to the FS PV interneurons.
It is also likely that modulations of gamma frequency by
other properties of visual stimuli (e.g., contrast, eccentricity,
size) are mediated through the stimulus capacity to activate the
inhibitory circuitry in visual cortex. Indeed, the higher frequency of LFP gamma oscillations in response to gratings of
relatively higher contrasts in monkey V1 (Jia et al. 2013; Ray
and Maunsell 2010) and in human MEG (Hadjipapas et al.
2015) can be explained by a greater capacity of higher-contrast
stimuli to activate neurons’ inhibitory surround (Paffen et al.
2005). In addition, the decrease in gamma peak frequency with
decreasing visual eccentricity observed in MEG on humans
(van Pelt and Fries 2013) and in LFP of monkeys (Lima et al.
2010) is consistent with the eccentricity-dependent scaling of
intensity of surround inhibition in the areas V1 and V2 (Zuiderbaan et al. 2012). The modulation of LFP gamma frequency
by size of a visual stimulus also depends on stimulation of
inhibitory interneurons (Gieselmann and Thiele 2008). Gieselmann and Thiele (2008) demonstrated that the gratings centered at the receptive field of the recorded neurons induced
gamma oscillations in LFP only if they also stimulated a
neuron’s inhibitory surround. They also observed that the
frequency of oscillations depended on the area of the stimulated inhibitory surround.
It has been suggested that apart from the strength of the
inhibition, the frequency of gamma oscillations might depend
on some constitutional factors, such as the size of the visual
cortical areas (Schwarzkopf et al. 2012) and conduction delays
(Gieselmann and Thiele 2008). Relative changes in gamma
frequency caused by changes in motion velocity are likely to be
less sensitive to these factors than the absolute values, and
therefore, they may more closely reflect functioning of FS PV
neurons, one of the major classes of inhibitory neurons in the
mammalian cortex. In particular, experimental modulations of
gamma frequency by motion velocity may appear useful to
probe the function of inhibitory networks in disorders characterized by abnormalities of FS PV neurons, e.g, schizophrenia
(Gonzalez-Burgos and Lewis 2012; Lewis 2014) and autism
(Lawrence et al. 2010).
Although a shift in FS PV neurons’ excitability is the most
likely cause of the velocity-related increase in gamma peak
frequency observed in our study, other factors might also
contribute. In particular, stimulation with different velocities
may result in different involvement of cortical regions. For
example, neurons in the area MT are more sensitive to higher
velocities (Yang et al. 2009), whereas the motion-sensitive
neurons in the posteriomedial visual area (PM) are ”tuned“ to
slower drift rates compared with those in V1 (Roth et al. 2012).
Therefore, MT and PM may differently modulate the frequency of gamma oscillations in V1, depending on the speed of
visual motion. It has been shown, however, that feedback
projections to V1 from extrastriate areas target mostly pyramidal cells, whereas terminals on inhibitory interneurons are
rare (see discussion in Gieselmann and Thiele 2008). Therefore, considering the role of inhibitory cells in regulating
gamma frequency, the contribution of feedback projections
into gamma frequency modulation is unclear. Future source
localization and connectivity studies may help to investigate
the contribution of different cortical sources in the modulation
of gamma frequency by the velocity of visual motion.
from 1.2 to 6.0°/s. The previous studies, on the other hand,
reported smaller increases in frequency. Swettenham et al.
(2009) found an average 10-Hz difference in gamma peak
frequency when they compared static gratings with those
moving at 1.33°/s (that is close to the slow velocity presented
herein). The same group later reported an even more moderate
increase of 5 Hz for a larger group of participants using the
same paradigm (Muthukumaraswamy and Singh 2013). Second, we observed that increasing the velocity from 1.2 to 6°/s
was associated with a drop in gamma power (Fig. 3) in addition
to the increase in gamma frequency. Although the transition
from the stationary to the slowly moving grating led to an
increase in gamma frequency, it either did not affect (for
1.75°/s, as in Naue et al. 2011) or increased gamma power (for
1.33°/s, as in Muthukumaraswamy and Singh 2013). Such
uncorrelated changes in gamma frequency and power have
been found previously in animal studies (Jia et al. 2013) and
can be explained by different control mechanisms of these
properties of gamma oscillations.
What neural mechanism mediates changes in gamma frequency? The finding of a positive correlation between an
individual’s peak frequency of MEG-recorded visual gamma
oscillations and concentration of GABA in the visual cortex
gave rise to the hypothesis that higher gamma frequency
reflects stronger GABAergic inhibition (Edden et al. 2009).
However, the recent studies in humans have shown that facilitation of GABAergic transmission either did not affect
(Muthukumaraswamy et al. 2013; Saxena et al. 2013) or
decreased (Campbell et al. 2014; Lozano-Soldevilla et al.
2014) frequency of visual gamma oscillations as measured
with MEG. Importantly, nonspecific pharmacological GABA
modulations may affect GABA transmission on both inhibitory
FS PV interneurons and principle cells (Edden et al. 2009;
Muthukumaraswamy et al. 2013; Saxena et al. 2013; Campbell
et al. 2014; Lozano-Soldevilla et al. 2014). On the other hand,
experimental studies in animals and brain tissues in vitro
provide evidence that the frequency of gamma oscillations is
determined mainly by the net excitation of the inhibitory
circuitry. Indeed, manipulations that affect excitability of FS
PV interneurons effectively modulate gamma frequency,
whereas those selectively influencing principle cells do not
produce noticeable frequency changes (Mann and Mody 2010).
A large increase in gamma frequency may be elicited either by
excitation of FS PV cells through FS-specific NMDARs or by
a reduction in their inhibition through abolishment of FSspecific ␦-unit-containing GABA receptors (Ferando and
Mody 2015; Mann and Mody 2010). In contrast, the application of NMDA antagonists decreases the frequency of highfrequency gamma oscillations (Anver et al. 2011; Oke et al.
2010). Furthermore, alcohol, a drug that both increases
GABAergic inhibition at GABAA receptors and decreases
glutamatergic excitation at NMDARs, reliably reduces the
frequency of visual MEG gamma oscillations in humans (Lozano-Soldevilla et al. 2014). Considering the results of animal
studies, it is likely that alcohol might operate through either
potentiation of inhibition or reduction of excitation of the FS
PV cells. Modeling studies also support the role of inhibitory
neurons in the regulating frequency of gamma oscillations
(Kirli et al. 2014). Considering these previous animal and
computational studies, it is likely that our observation of a
robust increase in gamma frequency with an increase in the
251
252
related to a smaller frequency increase in response to motion in
adult subjects. These authors suggested the presence of an
”upper oscillation frequency“ for the motion-driven gamma
responses and proposed that some individuals are naturally
close to this limit. Our study in children indicates that the
frequency of gamma oscillations in response to the slowly
moving gratings does not predict a frequency increase in
response to the fast motion when age is accounted for. Moreover, the strong velocity-related increase of gamma frequency
suggests that if the ”limit of upper gamma frequency“ does
exist, it is clearly not reached in children in response to the
stimuli moving with velocities of 1.2 to 3.6°/s. On the other
hand, the observations that none of the participants displayed
gamma VSPF of ⬎92.5 Hz and that 50% of the children had
92.5 Hz VSPF for the fast velocity suggest that this frequency
may be close to a physiological limit for this type of
stimulation.
Development of gamma oscillations. Our results are in line
with the previous studies that reported a developmental decrease in gamma frequency in response to stationary visual
grating between 8 and 45 yr (Gaetz et al. 2012) and 20 and 45
yr (Muthukumaraswamy et al. 2010). We extended this finding
by showing that similar developmental dynamics characterize
the response to the gratings moving with slow and medium
velocities (1.2 and 3.6°/s) in boys aged 8 –15 yr. No such
age-related decline in frequency was observed in case of the
fast motion (6.0°/s). Similarly to Gaetz et al. (2012), we did not
find significant developmental changes in the power of sustained visual gamma oscillations.
Considering that increased myelination was suggested to
produce faster EEG frequencies (Nunez 2000), the observed
maturational decrease in gamma frequency seems counterintuitive. Indeed, developmental studies typically report stable
frequency growth of slow oscillations during the course of
maturation (Hudspeth and Pribram 1992; Orekhova et al. 2006;
Stroganova et al. 1999). Presumably, the different developmental courses of gamma and slow oscillations can be explained by differing mechanisms in their generation. Slow
oscillations arise in broadly synchronized cortico-subcortical
loops (Hughes and Crunelli 2007; Miller 1991). Myelination of
these loops undergoes a long developmental course, and their
conductivity increases with age (Paus et al. 2001). In contrast,
visual gamma oscillations depend to a greater extent on horizontal intrinsic connections within visual cortical areas (Ray
and Maunsell 2010) and may depend predominantly on the
properties of local inhibitory circuits.
Gaetz et al. (2012) tested whether the age-related changes in
morphology of visual cortical areas (i.e., cortical volume or
thickness) could account for reduction in gamma frequency,
but they found no significant links. Hence, it is unlikely that the
developmental decline in gamma frequency observed in case of
slow and medium velocities in our study can be explained by
changes in brain macroanatomy occurring during childhood
and adolescence. More likely, this decline reflects functional
maturation of visual cortical areas. Maturation of the FS PV
interneurons might be of particular relevance for the developmental changes in gamma frequency. Unlike other interneurons, the FS cells undergo protracted development that proceeds well into adolescence and involves both inhibitory and
excitatory transmission in FS cells (Pinto et al. 2010; Wang
and Gao 2009).
In our study, the velocity-related increase in gamma frequency was associated with a remarkable reduction of power in
the gamma oscillations (Fig. 3). Differing contributions of a
number of factors may explain the nonlinear dynamics of
gamma amplitude as well as the generally inconsistent relationship between gamma amplitude and frequency modulation
(Jia et al. 2013). Similar to the present study, some studies in
animals also reported a reduction in amplitude of visual gamma
oscillations with increasing gamma frequency (Gieselmann
and Thiele 2008; Lima et al. 2010). For example, Gieselmann
and Thiele (2008) have found that an increase in stimulus size
led to augmentation of the LFP gamma response with a
corresponding drop in frequency of gamma oscillations. These
authors suggested that the greater strength of gamma LFP
resulted from a bigger size of the stimulated area, whereas the
concomitant slowing down of gamma frequency was explained
by longer conduction times required for synchronization of
activity of relatively remote neural populations. The changes in
gamma frequency described by Lima et al. (2010), on the other
hand, are difficult to explain by changes in conduction times.
These authors observed that, when compared with a single
moving grating, superimposition of two moving gratings
(depth-order plaids) increased the gamma peak frequency
while reducing the power of gamma oscillations. They hypothesized that this finding reflects a competition and increased
mutual inhibition between subsets of neurons activated by the
two component gratings. In this case, the increase in frequency
of the gamma oscillations could be explained by a stronger
excitation of the inhibitory network in response to the plaids, as
opposed to the gratings.
Jia et al. (2013) reported that a transition from low to
medium (50%) contrast in visual gratings is accompanied by an
increase in both gamma frequency and power, whereas further
contrast increases, while inducing an increased gamma frequency, led to a reduction of the peak gamma power (Fig. 4)
(Jia et al. 2013). The effect of the velocity of visual stimuli on
gamma characteristics resembles that of the contrast. Indeed,
gamma frequency and power displayed the concordant changes
during transition from stationary to moving gratings in the
study by Muthukumaraswamy and Singh (2013) but changed
in opposite directions with an increase in stimulus motion
velocity in our experiment. Most probably, the concomitant
growth in gamma power and frequency resulted from a
strengthening of excitatory drive to both principal and inhibitory cells. Modulation of gamma frequency by feedback projections (Jia et al. 2013; Kang et al. 2010) could also contribute. Further strengthening of the input (and of the excitatory
drive to the inhibitory FS PV cells) could reduce the amplitude
of gamma oscillations via affecting the balance between excitation and inhibition (Atallah and Scanziani 2009). Another
possible explanation for the drop in gamma power is a reduced
neuronal synchronization at high gamma frequencies. It has
been suggested that the gamma cycle creates a ”window of
opportunity“ for synchronization between spatially distant neuronal assembles (Fries 2005). Conceivably, the shorter window
of opportunity at high gamma frequency reduces the number of
neuronal assemblies involved in synchronous membrane potential fluctuations, which may in turn lead to a reduction in
gamma power.
Swettenham et al. (2009) have observed previously that the
higher oscillation frequency for the static visual stimuli was
et al. 2009), contrast sensitivity (Benedek et al. 2010), contour
integration (Braunitzer et al. 2011; Kiorpes and Bassin 2003),
and perception of moving stimuli (Lewis 2014; Tadin et al.
2003). Indeed, studies in children and monkeys show that these
”low-order“ visual functions undergo long developmental trajectories that proceed into the teen years (Braunitzer et al.
2011; Kiorpes and Bassin 2003; Lewis 2014; Palomares et al.
2011). A wider gamma frequency modulation range may be
associated with greater maturity and/or efficiency of the visual
perceptual functions. In particular, we hypothesize that the
broader range of gamma frequency modulations might relate to
a more efficient neural inhibition as assessed through measuring the effect of spatial suppression in the visual cortex (Tadin
et al. 2003; Tadin et al. 2011).
Gamma frequency and spatial suppression. It has been noted
previously that detection of motion direction is hampered for
large, compared with small, high-contrast gratings (Tadin et al.
2003). This decrease in motion sensitivity or spatial suppression was suggested to reflect surround inhibition of directionselective neurons. The decline in neural inhibition related to
aging (Betts et al. 2009; Betts et al. 2005) or disease (Golomb
et al. 2009; Tadin et al. 2007) leads to a reduction of the spatial
suppression effect. Just the opposite, maturational increase in
GABAergic inhibition has been proposed to lead to greater
spatial suppression in adults compared with children (Lewis
2014). The tendency for a correlation between spatial suppression magnitude and a child’s age observed in our study is in
line with this previous finding.
Notably, we have found that the greater magnitude of the
spatial suppression effect correlated with the range of gamma
VSPF modulation and that this correlation was not explained
by age-related changes in the VSPF modulation range (Table
1). There was also a tendency for correlation between spatial
suppression and peak gamma frequency in response to the fast
motion of the gratings (Table 1). This pattern of results suggests that the more prominent spatial suppression might be
linked to faster kinetics of FS PV cells and greater diapason of
their modulation by excitation.
Although the intact surround inhibition in MT/V5 seems to
be critical for spatial suppression (Tadin et al. 2011), its exact
neural mechanisms are yet unknown. Our data suggest that
either the properties of FS PV neuron circuitry in V1 or some
external influences modulating their activity are among the
factors affecting the magnitude of spatial suppression in
children.
In conclusion, children in our study exhibited robust frequency modulation of MEG-recorded gamma oscillations by
velocity of visual motion. We propose that modulation of
gamma frequency by motion velocity may help to investigate
typical development of inhibitory networks in the visual cortex
and its disturbance in neurodevelopmental and psychiatric
disorders.
ACKNOWLEDGMENTS
We heartily thank the children and their families for their participation in
this study.
GRANTS
The study has been supported by Russian Science Foundation Grant no.
14-35-00060 and the Charity Foundation for Autism “Way Out.” The MEG
Concentration of the ␣1- and ␦-subunits of ␦-GABAA receptors (␦-GABAAR) on inhibitory neurons increases with age
(Le Magueresse and Monyer 2013; Pinto et al. 2010). Concentration of the GABA-synthesizing enzyme GAD65 also increases from early in life (⬍4 yr) to the teenage and adult
years, leading to increase in ”on-demand“ GABA release
(Pinto et al. 2010). Together, these factors may contribute to an
increase in ␦-GABAAR-mediated tonic inhibition of FS PV
interneurons.
Wang and Gao (2009) reported that expression of NMDARs
in PV interneurons is higher during development than in
adulthood. In the frontal cortex, the number of FS cells demonstrating NMDAR-mediated currents progressively drops
from ⬃75% in young to 25% in adult rats (Wang and Gao
2009). The similar developmental decline of NMDA-mediated
neurotransmission may also take place in the visual cortex
(Carmignoto and Vicini 1992). The developmental reduction in
the number of NMDARs may contribute to a reduction in FS
PV cells’ excitability with age (Anver et al. 2011; Mann and
Mody 2010).
As a result of the combination of the factors listed above
(and possibly also others), the excitability of PV neurons
decreases with age, and a greater level of input is needed to
recruit PV cells in the mature compared with the immature
brain (Le Magueresse and Monyer 2013). Considering that
higher excitability of FS PV neurons is associated with a higher
gamma frequency (Anver et al. 2011; Ferando and Mody 2015;
Mann and Mody 2010), the reduction in gamma oscillation
frequency in older children is well in line with the developmental decrease in FS PV interneurons’ excitability.
Despite the age-related slowing of gamma oscillations to
low/medium-velocity stimuli (1.2 and 3.6°/s), the frequency of
gamma response to the fast moving gratings was the same in
the younger and the older children. As a result, the range of
frequency modulation by velocity becomes higher with age.
Although the causes of such an increase are unknown, maturational changes in kinetics of the FS PV inhibitory neurons
may contribute. It has been found that duration of miniature
inhibitory postsynaptic potentials is shorter in postpubertal
compared with prepubertal monkeys (Hashimoto et al. 2009).
This shortening may appear to be a universal phenomenon in
all cortical neurons (Hashimoto et al. 2009; Le Magueresse and
Monyer 2013). The decay time of NMDAR-mediated synaptic
currents also decreases during development (Carmignoto and
Vicini 1992; Paoletti et al. 2013), probably as a result of a
developmental switch in NMDAR subunit composition
(Paoletti et al. 2013). The decrease in the ratio of slow NMDA
to fast AMPA glutamate receptors in FS neurons during the
adolescent period (Gonzalez-Burgos and Alejandre-Gomez
2005; Wang and Gao 2010) may further speed up the kinetics
of inhibitory interneurons. The faster kinetics of FS PV cells
may then refine their temporal integration properties and allow
greater gamma frequency modulation under an increasing
strength of visual input.
It has been suggested that developmental changes in inhibitory neurotransmission may have essential consequences for
information processing (Ferando and Mody 2015; Pinto et al.
2010). Considering the high concentration of GABA in the
visual cortex (Zilles et al. 2002), the developmental refinement
of inhibitory processes may be needed for optimal visual
perceptual functions such as orientation discrimination (Edden
253
254
Centre is supported by core funding from the Russian Ministry of Education
and Science (RFMEFI61914X0006).
DISCLOSURES
No conflicts of interest, financial or otherwise, are declared by the authors.
AUTHOR CONTRIBUTIONS
E.V.O., O.V.S., and T.A.S. conception and design of research; E.V.O.,
A.V.B., and T.A.S. analyzed data; E.V.O. and T.A.S. interpreted results of
experiments; E.V.O., A.V.B., and A.O.P. prepared figures; E.V.O. drafted
manuscript; E.V.O., A.V.B., O.V.S., A.O.P., A.Y.N., and T.A.S. approved
final version of manuscript; O.V.S., A.O.P., and A.Y.N. performed experiments; O.V.S. and T.A.S. edited and revised manuscript.
Anver H, Ward PD, Magony A, Vreugdenhil M. NMDA receptor hypofunction phase couples independent gamma-oscillations in the rat visual
cortex. Neuropsychopharmacology 36: 519 –528, 2011.
Atallah BV, Scanziani M. Instantaneous modulation of gamma oscillation
frequency by balancing excitation with inhibition. Neuron 62: 566 –577,
2009.
Benedek K, Janaky M, Braunitzer G, Rokszin A, Keri S, Benedek G.
Parallel development of contour integration and visual contrast sensitivity at
low spatial frequencies. Neurosci Lett 472: 175–178, 2010.
Betts LR, Sekuler AB, Bennett PJ. Spatial characteristics of center-surround
antagonism in younger and older adults. J Vis 9: 25.1–25.15, 2009.
Betts LR, Taylor CP, Sekuler AB, Bennett PJ. Aging reduces centersurround antagonism in visual motion processing. Neuron 45: 361–366,
2005.
Brainard DH. The Psychophysics Toolbox. Spat Vis 10: 433– 436, 1997.
Braunitzer G, Rokszin A, Kobor J, Nagy A, Sztriha L, Benedek G.
Development of visual contour integration in children with migraine without
aura. Cephalalgia 31: 1048 –1056, 2011.
Brunet N, Vinck M, Bosman CA, Singer W, Fries P. Gamma or no gamma,
that is the question. Trends Cogn Sci 18: 507–509, 2014.
Campbell AE, Sumner P, Singh KD, Muthukumaraswamy SD. Acute
effects of alcohol on stimulus-induced gamma oscillations in human primary
visual and motor cortices. Neuropsychopharmacol 39: 2104 –2113, 2014.
Cardin JA, Carlen M, Meletis K, Knoblich U, Zhang F, Deisseroth K, Tsai
LH, Moore CI. Driving fast-spiking cells induces gamma rhythm and
controls sensory responses. Nature 459: 663– 667, 2009.
Carlen M, Meletis K, Siegle JH, Cardin JA, Futai K, Vierling-Claassen D,
Ruhlmann C, Jones SR, Deisseroth K, Sheng M, Moore CI, Tsai LH. A
critical role for NMDA receptors in parvalbumin interneurons for gamma
rhythm induction and behavior. Mol Psychiatry 17: 537–548, 2012.
Carmignoto G, Vicini S. Activity-dependent decrease in NMDA receptor
responses during development of the visual cortex. Science 258: 1007–1011,
1992.
Cho RY, Walker CP, Polizzotto NR, Wozny TA, Fissell C, Chen CM,
Lewis DA. Development of sensory gamma oscillations and cross-frequency coupling from childhood to early adulthood. Cereb Cortex 25:
1509 –1518, 2015.
Cousijn H, Haegens S, Wallis G, Near J, Stokes MG, Harrison PJ, Nobre
AC. Resting GABA and glutamate concentrations do not predict visual
gamma frequency or amplitude. Proc Natl Acad Sci USA 111: 9301–9306,
2014.
Edden RA, Muthukumaraswamy SD, Freeman TC, Singh KD. Orientation
discrimination performance is predicted by GABA concentration and
gamma oscillation frequency in human primary visual cortex. J Neurosci 29:
15721–15726, 2009.
Ferando I, Mody I. In vitro gamma oscillations following partial and complete ablation of delta subunit-containing GABAA receptors from parvalbumin interneurons. Neuropharmacology 88: 91–98, 2015.
Foss-Feig JH, Tadin D, Schauder KB, Cascio CJ. A substantial and
unexpected enhancement of motion perception in autism. J Neurosci 33:
8243– 8249, 2013.
Fries P. A mechanism for cognitive dynamics: neuronal communication
through neuronal coherence. Trends Cogn Sci 9: 474 – 480, 2005.
Gaetz W, Roberts TP, Singh KD, Muthukumaraswamy SD. Functional and
structural correlates of the aging brain: relating visual cortex (V1) gamma
REFERENCES
band responses to age-related structural change. Hum Brain Mapp 33:
2035–2046, 2012.
Gieselmann MA, Thiele A. Comparison of spatial integration and surround
suppression characteristics in spiking activity and the local field potential in
macaque V1. Eur J Neurosci 28: 447– 459, 2008.
Golomb JD, McDavitt JR, Ruf BM, Chen JI, Saricicek A, Maloney KH,
Hu J, Chun MM, Bhagwagar Z. Enhanced visual motion perception in
major depressive disorder. J Neurosci 29: 9072–9077, 2009.
Gonzalez-Burgos G, Lewis DA. NMDA receptor hypofunction, parvalbuminpositive neurons, and cortical gamma oscillations in schizophrenia. Schizophrenia Bull 38: 950 –957, 2012.
González-Burgos I, Alejandre-Gómez M. Cerebellar granule cell and Bergmann glial cell maturation in the rat is disrupted by pre- and postnatal
exposure to moderate levels of ethanol. Int J Dev Neurosci 23: 383–388,
2005.
Gray C, DiPrisco G. Stimulus-dependent neuronal oscillations and local
synchronization in striate cortex of the alert cat. J Neurosci 17: 3239 –3253,
1997.
Gulyas AI, Szabo GG, Ulbert I, Holderith N, Monyer H, Erdelyi F, Szabo
G, Freund TF, Hajos N. Parvalbumin-containing fast-spiking basket cells
generate the field potential oscillations induced by cholinergic receptor
activation in the hippocampus. J Neurosci 30: 15134 –15145, 2010.
Hadjipapas A, Lowet E, Roberts M, Peter A, De Weerd P. Parametric
variation of gamma frequency and power with luminance contrast: A
comparative study of human MEG and monkey LFP and spike responses.
Neuroimage 112: 327–340, 2015.
Hashimoto T, Nguyen QL, Rotaru D, Keenan T, Arion D, Beneyto M,
Gonzalez-Burgos G, Lewis DA. Protracted developmental trajectories of
GABAA receptor alpha1 and alpha2 subunit expression in primate prefrontal cortex. Biol Psychiatry 65: 1015–1023, 2009.
Hermes D, Miller KJ, Wandell BA, Winawer J. Stimulus dependence of
gamma oscillations in human visual cortex. Cereb Cortex. First published
May 22, 2014; doi:10.1093/cercor/bhu091.
Holland PW, Welsch RE. Robust regression using iteratively re-weighted
leastsquares. Commun Stat Theory Methods 6: 813– 827, 1977.
Hoogenboom N, Schoffelen JM, Oostenveld R, Parkes LM, Fries P.
Localizing human visual gamma-band activity in frequency, time and space.
Neuroimage 29: 764 –773, 2006.
Hudspeth WJ, Pribram KH. Psychophysiological indices of cerebral maturation. Int J Psychophysiol 12: 19 –29, 1992.
Hughes SW, Crunelli V. Just a phase they’re going through: the complex
interaction of intrinsic high-threshold bursting and gap junctions in the
generation of thalamic alpha and theta rhythms. Int J Psychophysiol 64:
3–17, 2007.
Jia XX, Xing DJ, Kohn A. No consistent relationship between gamma power
and peak frequency in macaque primary visual cortex. J Neurosci 33: 17–25,
2013.
Kang K, Shelley M, Henrie JA, Shapley R. LFP spectral peaks in V1 cortex:
network resonance and cortico-cortical feedback. J Comput Neurosci 29:
495–507, 2010.
Kiorpes L, Bassin SA. Development of contour integration in macaque
monkeys. Vis Neurosci 20: 567–575, 2003.
Kirli KK, Ermentrout GB, Cho RY. Computational study of NMDA
conductance and cortical oscillations in schizophrenia. Front Comput Neurosci 8: 133, 2014.
Koelewijn L, Rich AN, Muthukumaraswamy SD, Singh KD. Spatial
attention increases high-frequency gamma synchronisation in human medial
visual cortex. Neuroimage 79: 295–303, 2013.
Lawrence YA, Kemper TL, Bauman ML, Blatt GJ. Parvalbumin-, calbindin-, and calretinin-immunoreactive hippocampal interneuron density in
autism. Acta Neurol Scand 121: 99 –108, 2010.
Le Magueresse C, Monyer H. GABAergic interneurons shape the functional
maturation of the cortex. Neuron 77: 388 – 405, 2013.
Lewis DA. Inhibitory neurons in human cortical circuits: substrate for cognitive dysfunction in schizophrenia. Curr Opin Neurobiol 26: 22–26, 2014.
Lima B, Singer W, Chen NH, Neuenschwander S. Synchronization dynamics in response to plaid stimuli in monkey V1. Cereb Cortex 20: 1556 –1573,
2010.
Litvak V, Mattout J, Kiebel S, Phillips C, Henson R, Kilner J, Barnes G,
Oostenveld R, Daunizeau J, Flandin G, Penny W, Friston K. EEG and
MEG data analysis in SPM8. Comput Intell Neurosci 2011: 852961, 2011.
Lozano-Soldevilla D, ter Huurne N, Cools R, Jensen O. GABAergic
modulation of visual gamma and alpha oscillations and its consequences for
working memory performance. Curr Biol 24: 2878 –2887, 2014.
Sohal VS, Zhang F, Yizhar O, Deisseroth K. Parvalbumin neurons and
gamma rhythms enhance cortical circuit performance. Nature 459: 698 –
702, 2009.
Stroganova TA, Orekhova EV, Posikera IN. EEG alpha rhythm in infants.
Clin Neurophysiol 110: 997–1012, 1999.
Swettenham JB, Muthukumaraswamy SD, Singh KD. Spectral properties
of induced and evoked gamma oscillations in human early visual cortex to
moving and stationary stimuli. J Neurophysiol 102: 1241–1253, 2009.
Tadel F, Baillet S, Mosher JC, Pantazis D, Leahy RM. Brainstorm: a
user-friendly application for MEG/EEG analysis. Comput Intell Neurosci
2011: 879716, 2011.
Tadin D, Kim J, Park S. Weakened center-surround interactions in visual
motion processing in schizophrenia. Schizophrenia Bull 33: 545, 2007.
Tadin D, Lappin JS, Gilroy LA, Blake R. Perceptual consequences of
centre-surround antagonism in visual motion processing. Nature 424: 312–
315, 2003.
Tadin D, Silvanto J, Pascual-Leone A, Battelli L. Improved motion perception and impaired spatial suppression following disruption of cortical area
MT/V5. J Neurosci 31: 1279 –1283, 2011.
Taulu S, Kajola M, Simola J. Suppression of interference and artifacts by the
Signal Space Separation Method. Brain Topogr 16: 269 –275, 2004.
Tesche CD, Uusitalo MA, Ilmoniemi RJ, Huotilainen M, Kajola M,
Salonen O. Signal-space projections of MEG data characterize both distributed and well-localized neuronal sources. Electroencephalogr Clin Neurophysiol 95: 189 –200, 1995.
Thomson DJ. Spectrum estimation and harmonic analysis. Proc IEEE 70:
1055–1096, 1982.
Uhlhaas PJ, Pipa G, Lima B, Melloni L, Neuenschwander S, Nikolic D,
Singer W. Neural synchrony in cortical networks: history, concept and
current status. Front Integr Neurosci 3: 17, 2009.
Uhlhaas PJ, Pipa G, Neuenschwander S, Wibral M, Singer W. A new look
at gamma? High- (⬎60 Hz) ␥-band activity in cortical networks: function,
mechanisms and impairment. Prog Biophys Mol Biol 105: 14 –28, 2011.
Uusitalo MA, Ilmoniemi RJ. Signal-space projection method for separating
MEG or EEG into components. Med Biol Eng Comput 35: 135–140, 1997.
van Pelt S, Boomsma DI, Fries P. Magnetoencephalography in twins reveals
a strong genetic determination of the peak frequency of visually induced
␥-band synchronization. J Neurosci 32: 3388 –3392, 2012.
van Pelt S, Fries P. Visual stimulus eccentricity affects human gamma peak
frequency. Neuroimage 78: 439 – 447, 2013.
Volman V, Behrens MM, Sejnowski TJ. Downregulation of parvalbumin at
cortical GABA synapses reduces network gamma oscillatory activity. J
Neurosci 31: 18137–18148, 2011.
Wang HX, Gao WJ. Cell type-specific development of NMDA receptors in
the interneurons of rat prefrontal cortex. Neuropsychopharmacol 34: 2028 –
2040, 2009.
Wang HX, Gao WJ. Development of calcium-permeable AMPA receptors
and their correlation with NMDA receptors in fast-spiking interneurons of
rat prefrontal cortex. J Physiol 588: 2823–2838, 2010.
Yang Y, Zhang J, Liang Z, Li GX, Wang YC, Ma YY, Zhou YF, Leventhal
AG. Aging affects the neural representation of speed in Macaque area MT.
Cereb Cortex 19: 1957–1967, 2009.
Zilles K, Palomero-Gallagher N, Grefkes C, Scheperjans F, Boy C,
Amunts K, Schleicher A. Architectonics of the human cerebral cortex and
transmitter receptor fingerprints: reconciling functional neuroanatomy and
neurochemistry. Eur Neuropsychopharmacol 12: 587–599, 2002.
Zuiderbaan W, Harvey BM, Dumoulin SO. Modeling center-surround
configurations in population receptive fields using fMRI. J Vis 12: 10, 2012.
Mann EO, Mody I. Control of hippocampal gamma oscillation frequency by
tonic inhibition and excitation of interneurons. Nat Neurosci 13: 205–212,
2010.
Miller R. Cortico-Hippocampal Interplay and the Representation of Contexts
in the Brain. Berlin and New York: Springer-Verlag, 1991.
Muthukumaraswamy SD, Myers JF, Wilson SJ, Nutt DJ, Hamandi K,
Lingford-Hughes A, Singh KD. Elevating endogenous GABA levels with
GAT-1 blockade modulates evoked but not induced responses in human
visual cortex. Neuropsychopharmacol 38: 1105–1112, 2013.
Muthukumaraswamy SD, Singh KD. Visual gamma oscillations: the effects
of stimulus type, visual field coverage and stimulus motion on MEG and
EEG recordings. Neuroimage 69: 223–230, 2013.
Muthukumaraswamy SD, Singh KD, Swettenham JB, Jones DK. Visual
gamma oscillations and evoked responses: variability, repeatability and
structural MRI correlates. Neuroimage 49: 3349 –3357, 2010.
Naue N, Strüber D, Fründ I, Schadow J, Lenz D, Rach S, Körner U,
Herrmann CS. Gamma in motion: pattern reversal elicits stronger gammaband responses than motion. Neuroimage 55: 808 – 817, 2011.
Nunez PL. Toward a quantitative description of large-scale neocortical dynamic function and EEG. Behav Brain Sci 23: 371–398; discussion 399 –
437, 2000.
Oke OO, Magony A, Anver H, Ward PD, Jiruska P, Jefferys JG,
Vreugdenhil M. High-frequency gamma oscillations coexist with lowfrequency gamma oscillations in the rat visual cortex in vitro. Eur J
Neurosci 31: 1435–1445, 2010.
Orekhova EV, Stroganova TA, Posikera IN, Elam M. EEG theta rhythm in
infants and preschool children. Clin Neurophysiol 117: 1047–1062, 2006.
Paffen CL, van der Smagt MJ, te Pas SF, Verstraten FA. Center-surround
inhibition and facilitation as a function of size and contrast at multiple levels
of visual motion processing. J Vis 5: 571–578, 2005.
Palomares M, Englund JA, Ahlers S. Patterns and trajectories in Williams
Syndrome: the case of visual orientation discrimination. Res Dev Disabil 32:
1021–1029, 2011.
Paoletti P, Bellone C, Zhou Q. NMDA receptor subunit diversity: impact on
receptor properties, synaptic plasticity and disease. Nat Rev Neurosci 14:
383– 400, 2013.
Paus T, Collins DL, Evans AC, Leonard G, Pike B, Zijdenbos A. Maturation of white matter in the human brain: a review of magnetic resonance
studies. Brain Res Bull 54: 255–266, 2001.
Perry G, Hamandi K, Brindley LM, Muthukumaraswamy SD, Singh KD.
The properties of induced gamma oscillations in human visual cortex show
individual variability in their dependence on stimulus size. Neuroimage 68:
83–92, 2013.
Pinto JG, Hornby KR, Jones DG, Murphy KM. Developmental changes in
GABAergic mechanisms in human visual cortex across the lifespan. Front
Cell Neurosci 4: 16, 2010.
Ray S, Maunsell JH. Differences in gamma frequencies across visual cortex
restrict their possible use in computation. Neuron 67: 885– 896, 2010.
Rice JK, Rorden C, Little JS, Parra LC. Subject position affects EEG
magnitudes. Neuroimage 64: 476 – 484, 2013.
Roth MM, Helmchen F, Kampa BM. Distinct functional properties of
primary and posteromedial visual area of mouse neocortex. J Neurosci 32:
9716 –9726, 2012.
Saxena N, Muthukumaraswamy SD, Diukova A, Singh K, Hall J, Wise R.
Enhanced stimulus-induced gamma activity in humans during propofolinduced sedation. PLoS One 8: e57685, 2013.
Schwarzkopf DS, Robertson DJ, Song C, Barnes GR, Rees G. The frequency of visually induced gamma-band oscillations depends on the size of
early human visual cortex. J Neurosci 32: 1507–1512, 2012.
255

Frequency of gamma oscillations in humans is modulated by

Transcription

Similar documents

Misty May-Treanor September 26, 2014 FREE

Meal Aktiv Fritid – ut på tur

Gamma Alpha Chapter News

Infinite Edge ParOOon Models for Overlapping Community

EXPANDING THE GAMMA FUNCTION NEAR ITS NEGATIVE

MLK YA UPCF program - Gamma Gamma Lambda Chapter

welcome to our - Synergy Health

Teambuilding Aktiv Fritid - Aktiv Fritid

meet the other side of the interface encourage people to