Effect of title on eye-movement exploration of cubist - IRIS

Perception, 2009, volume 38, pages 479 ^ 491
doi:10.1068/p6080
Effect of title on eye-movement exploration of cubist
paintings by Fernand Lëger
Zoi Kapoula, Gintautas Daunysô, Olivier Herbez, Qing Yang
Laboratoire IRIS, FRE3154, CNRS, Poªle de Stomatologie Ophtalmologie et ORL, Hoªpital Europëen
Georges Pompidou, 20 rue Leblanc, 75908 Paris Cedex 15, France;
e-mail: [email protected]; ô Department of Radio Engineering, University of Technology,
Siauliai, Lithuania
Received 30 April 2008, in revised form 12 August 2008; published online 26 March 2009
Abstract. Franklin et al (1993, Leonardo 26 103 ^ 108) reported that title information influenced
the interpretation of paintings but not the way observers explore and look at the paintings;
in their study subjects used a hand pointer to indicate where they looked. We used eye-movement
recording and examined the effect of title on eye-movement exploration of nonrealistic cubist
paintings giving rise to free interpretation. Three paintings by Fernand Lëger were used: The Wedding
contained high density of small fragments of real human faces, limbs, or arbitrary fragments mixed
with large plane surfaces; The Alarm Clock consisted of arbitrary fragments creating perception
of a person; Contrast of Forms contained forms and cylinders. Different groups of naive subjects
explored paintings without knowing the title (spontaneous condition), with the instruction to
invent a title (active condition), and after announcement of the authentic title (driven condition).
Exploration time was unrestricted and eye movements were recorded by Chronos video-oculography.
Fixation duration was found to increase in the driven condition relative to active condition;
such increase occurred for all paintings. In contrast, fixation-duration variability remained stable
over all title conditions. Saccade amplitude increased in the driven condition for Contrast of
Forms. Increase of fixation duration and of saccade size are attributed to additional cognitive
analysis, ie search fitting between the title and the painting. When comparing paintings within
each title condition, The Wedding produced different results than the other paintings: longer
exploration time (in spontaneous condition), higher fixation duration variability (in spontaneous
and driven conditions), but smaller saccade sizes (in active and driven conditions). The differences are attributed to visual aspects (high density of small fragments) but also to complex
semantic analysis of multiple segments of faces and limbs contained by this painting.
Spatial distribution of fixation time was highly selective, with a preponderance of the central
area that was the most fixated for all paintings and all title conditions. In the driven condition,
however, loci of most frequent fixations were different than in the other conditions from the first
5 s; particularly for The Alarm Clock the title drove the eyes rapidly on the inconspicuous
fragment of the clock. Our findings go against Franklin's conclusions. We conclude that title
information influences both physiological parameters of eye movements and the distribution of
fixation time over different selected areas of the painting.
1 Introduction
Saccades are the stereotyped fast eye movements which are used by us to explore the
visual world; each saccade is followed by a fixation period during which visual processing is performed, as clear vision is possible only when images in the eyes are stable
(Westheimer and McKee 1975). The way humans explore paintings is a fascinating field
which has attracted researchers in the last century, with Buswell (1935) and Yarbus
(1967) as pioneers in the USA and the Soviet Union, respectively. Buswell (1935) used
a variety of pictures: classic paintings as well as cubist artwork (eg Duchamp's Nude
Descending a Staircase). Yarbus used mainly photographs and paintings from the Soviet
realistic school. He reported that (i) the eyes fixate mainly the meaningful areas of
a realistic image, regardless of the density of details, colour, or contrast; (ii) the centre
of the image attracts the eyes more; (iii) with additional exploration time, the eyes repeat
fixations between the same meaningful points rather than explore new ones.
480
Z Kapoula, G Daunys, O Herbez, Q Yang
In the relatively recent field of psycho-aesthetics a major issue is the time-course
of aesthetic appreciation, and two largely opposite models have been postulated.
Locher et al (1996), on the basis of their studies of pictorial balance and perception,
emphasise the almost immediate holistic pre-attentive appreciation of paintings. Visual
processing would occur in parallel across the visual field, resulting in a representation
of the global structure organisation of the painting including balance and symmetry.
Eye movements would be driven by such global representation to process detailed
syntactic or semantic content within this holistic quasi-immediate representation. In
contrast, the model proposed by Leder et al (2004) suggests multiple rather successive
stages in the appreciation of artwork evolving over time: perceptual analysis, implicit
information integration, explicit classification, cognitive mastering, and, in parallel,
emotional evaluation; the final output of all the above stages being the aesthetic judgment, aesthetic experience, and pleasure. Leder et al (2006) studied the role of entitling
the artwork; they used a variety of artwork, and provided descriptive or elaborative
titles to the subject at different instants during painting exploration. They measured
subjective responses, such as liking and understanding the painting. Their results support
the idea that a certain time is needed to assign a meaning to the painting induced by the
title.
The question of the influence of the title on how paintings are seen and explored
visually has been addressed earlier by Franklin et al (1993). They elegantly reviewed
theoretical writings on the role of words and titles. Titles can function in different ways
depending upon the style and content of artwork: they may simply provide a straightforward description of the work adding little to its interpretation or meaning; or they
can provide a contextual factor which may highlight some aspect of the work with
respect to directing visual attention and conceptual interpretation. In the experimental
study by Franklin et al (1993) subjects viewed an image of the painting while hearing
the title spoken. Observers talked about the painting they were seeing, and used
a flashlight pointer to indicate where they were looking. For both paintings studied
(Monet's Terrace at Ste Adresse, Arshile Gork's Agony) the title affected the observers'
interpretative reading (ie the descriptions of the observers) but not where they looked.
Note, however, that there were no eye-movement recordings, and conclusions on where
subjects looked were based on their verbal responses and the flashlight pointer. Eye
movements are faster than hand movements and can be triggered quasi-reflexively
before even being conscious on where attention is directed. Therefore the question of
how the title information influences where observers look can be answered only with
objective eye-movement recordings. It is well known that eye-movement exploration
of an image is determined both by properties of the image (bottom ^ up) and by
top ^down factors. Title information acting in a top ^ down manner could modulate eyemovement exploration, namely the areas selected for frequent fixation, and the order
in which such areas are fixated. Moreover, we expected the title information to
influence temporal parameters such as fixation duration time due to the time required
by additional cognitive processes such as matching the title with the content of the
painting.
In the present study we used objective eye-movement recording. Our first goal
was to examine influence of information in the title on eye-movement characteristics,
namely on total exploration time, fixation duration, and saccade amplitude. Our
second goal was to examine the influence of the title on the distribution of fixation
time over different areas of the painting. Moreover, to link this goal with the models
of fast holistic versus successive stage processing of a painting, we examined fixation
time distribution in the first and last 5 s of exploration. The eventual influence of the
title even on the initial fixations could be compatible with the model of Locher (1996)
suggesting very rapid global visual and semantic processing of the painting.
Effect of title on eye-movement exploration of cubist paintings by Fernand Lëger
481
The study was carried out on three non-realistic paintings. Paintings were selected
to provide a continuum from face outlines to abstraction within the colourful,
dynamic, fragmented universe of Fernand Lëger's cubism: The Wedding, The Alarm
Clock, and Contrast of Forms (see figure 1a). Such paintings give rise to free interpretation. Three groups of subjects were studied: the first group explored each of the three
paintings spontaneously with no prior knowledge of the title (spontaneous condition);
the second group was asked to explore the painting and invent the title (active condition); the third group was given the authentic title upon presentation of each painting
(driven condition).
(a)
The Wedding
The Alarm Clock
The Contrast of Forms
(b)
Figure 1. [In colour online, see http://dx.doi.org/10.1068/p6080] (a) The paintings studied, The Wedding
(La Noce) 1910 ^ 1911, oil on canvas, 257 cm 6206 cm, Musëe National d'Art Moderne, Paris;
The Alarm Clock (Le Rëveille-Matin or La Rëveil-Matin; the two names are possible according
to the Musëe National d'Art Moderne where the painting is kept), 1914, oil on canvas,
100 cm681 cm; Contrast of Forms (Contraste de Formes) 1913, oil on canvas, 130 cm697 cm,
Philadelphia Museum of Art. (b) Patterns of eye-movement exploration for each painting from
the same subject in the spontaneous condition. The pattern is different for the different paintings.
Note the high concentration of eye positioning to certain areas, where faces or meaningful objects
could be perceived in The Wedding and in The Alarm Clock, and to a lesser extent in Contrast
of Forms.
482
Z Kapoula, G Daunys, O Herbez, Q Yang
2 Materials and methods
2.1 Characteristics of the paintings
The location and year of each painting are indicated in the legend of figure 1. All three
paintings are non-realistic, the most abstract being Contrast of Forms, composed of
fragments of forms and cylinders. The Alarm Clock is composed of arbitrary fragments
with colours similar to those in the Contrast of Forms, but the fragments are organised
in a way to promote perception of a person; an inconspicuous fragment, the alarm
clock is present in the bottom right of the painting. According to art historians the
painter's intention was to depict a woman seated close to nature morte holding her face
in her hand, perhaps a variation of the painting Femme Cousant (Laugier and Richet
1981). As we shall show, the majority of the subjects perceived instead a joyful male
musician. The Wedding is a mixture of cubism and figuration and contains many real
but incomplete human faces and limbs appearing through vaporous surfaces, and
surrounded by arbitrary fragments. Fernand Lëger said to Abert Elsen (historian of
art, University of Stanford, written communication 1979) that he tried to depict the
marriage in a non-emotional way and that he was inspired by pictures of the marriage;
perhaps he referred to the marriage of his friend Andrë Mare, married 25 July 1910,
at which Fernand Lëger was present (Laugier and Richet 1981). For the purpose of
our study, we take that The Wedding is the only painting in which Fernand Lëger
did include explicitly faces, albeit not complete and covered. Faces were seen by our
subjects viewing a projection of a high-quality slide of the painting with a size of
19 deg624 deg (see below). One should note that all three paintings are dominated by
edges and their distribution and density as well as the size of the fragments differs,
particularly between the third and the first two paintings. There is evidence that edges
and local contrast are involved in the guidance of eye movements, perhaps more than
luminance and chromaticity (Tatler et al 2005). The overall structural composition
is characterised by segregation of key points along the central vertical axis (white
dress, faces of the married couple in The Wedding, central person in The Alarm Clock,
interleaved cylinders providing depth perception in Contrast of Forms). The overall
structural complexity influences visual exploration (Locher 1996, 2006). In this case the
vertical composition of the paintings caused rather elongated patterns of eye-movement
exploration (see section 3).
2.2 Subjects
Twenty-three subjects (nine females, fourteen males) participated in the study; they
were from various research teams at the laboratory of Physiology of Perception and
Action at the Colle©ge de France; they were all naive with respect of the purpose of
the experiments and saw the paintings for the first time. Their ages ranged from 22
to 45 years. All subjects had normal ocular eye alignment and normal visual acuity for
both eyes.
2.3 Eye movement recording
Horizontal and vertical components of eye movements were recorded continuously
with a high-resolution video-based eye-tracker system (Chronos Skalar, www.skalar.nl);
this tracker represents a new generation in image-based eye-movement measurement
technology, based on programmable image sensors interfaced directly with digital
processing circuitry. This system allows high image sampling rate. The resolution of
the system is 5 0:1 deg, and it is linear over 208 for both horizontal and vertical
directions. Eye position images were sampled at 200 Hz and were analysed offline.
During the experiment the head was stabilised with a front- and a chin-support; this
stabilisation is effective and is currently used in our laboratory for several physiological
studies on adults or even on children (Yang et al 2002; Yang and Kapoula 2003).
Effect of title on eye-movement exploration of cubist paintings by Fernand Lëger
483
2.4 Painting exploration
The paintings were presented on a computer screen and had a viewing size of 19 deg
624 deg at a distance of 64 cm from the computer screen. Different groups of subjects
explored the paintings under three conditions: eleven were asked to explore each painting
with no knowledge of the title (spontaneous condition); six were given the instruction
to explore and invent a title for each painting (active condition); the remaining six
subjects were provided with the title given by the painter (driven condition). The order
of presentation of the paintings was counterbalanced among subjects. The duration of
each condition was always unrestricted. Before and after exploration of each painting
a calibration task was carried out, during which the subject fixated a target shifted
left, right, up, and down at eccentricities of 208; the target was a pair of nonius lines
(two vertical segments offset horizontally by 5 min of arc). Subjects were instructed to
fixate as accurately as possible between the two lines. Following exploration of all three
paintings the subjects were interviewed with regards to their perception and apprehension
of each painting, and the title they invented in the active condition.
2.5 Data analysis
The Chronos software was used to calibrate the data. Subsequent offline algorithms
were used to analyse eye movement parameters of the left eye. Eye movements associated with blinks were discarded from the analysis. The duration of exploration (free
from blinks) by each subject and in each condition was then evaluated. A standard
saccade-detector algorithm was used to detect saccades; the onset of the saccades was
determined at the point when eye velocity exceeded 158 sÿ1 ; saccade offset when eye
velocity dropped below 108 sÿ1. These criteria are similar to those used by others
(Kapoula et al 2001; Takagi et al 1995; Yang et al 2002) and allowed us to detect even
small saccades (0.158). The quality of these criteria was verified by an investigator
scrutinising on the computer screen eye position and velocity signals. Fixation duration was determined as the time period between two successive saccades; again, the
standard criteria used to exclude artifacts was a minimal duration of 50 ms and a
displacement of the eyes during the fixation lower than 18.
The distribution of the exploration time was found not to be normal for The Wedding
and for The Alarm Clock; moreover the test for homogeneity of variance prior to ANOVA
failed significance for some parameters (fixation duration, saccade amplitude, variability).
Consequently, we applied the non-parametric statistical tests. Comparisons between title
conditions run on different groups of subjects were made with the Krustal ^Wallis
test; the Mann ^ Whitney U test was used for comparisons between any two title conditions. The Friedman non-parametric test was used to examine the effect of painting
for the three conditions together. The Wilcoxon test was used for comparisons between
any two paintings.
3 Results
Subjective reports in the spontaneous condition revealed semantic interpretation of the
paintings ranging from social complex situations to abstraction. Briefly, in The Wedding,
almost all subjects described human faces, a social common activity around a person
or crowds (eg people around the food market, factory crowd, and crowd of people
from a village). In The Alarm Clock all subjects perceived a joyful man with beret
(unconventional man with a beret, man in fanfare with drums, man with black hat, etc).
The beret is an element recognised easily by French observers; note, however, that the
same painting shown during conferences to observers with other cultures (Lithuanian,
Greek) still evoked similar responses, eg a joyful musician. In Contrast of Forms subjects reported mostly a collection of objects (drums, ribbons, spotlights, food picnic).
484
Z Kapoula, G Daunys, O Herbez, Q Yang
Importantly, there were common elements in the interpretations by different subjects
despite the fact that the paintings were non-realistic and rather abstract.
Examples of eye-movement exploration are shown in figure 1b; the eye movement
pattern differs according to painting. Note the repetitive movements along the centre;
the pattern is particularly elongated for The Wedding which contains an elongated human
figure.
50
40
30
20
10
Mean fixation duration=ms
(a)
0
(d)
350
300
250
200
SD of fixation duration=ms
(b)
(e)
SD of saccade amplitude=deg
Exploration time=s
60
Mean saccade amplitude=deg
3.1 Effects of title
Figure 2a shows mean exploration times with their standard errors (SE) for the three
paintings in the three conditions (spontaneous, active, and driven). There was no significant effect of title for any of the paintings.
4
3
2
3
2
1
The Wedding The Alarm
Clock
Contrast of
Forms
150
Condition
spontaneous
active
driven
140
130
120
110
100
(c)
The Wedding The Alarm Contrast of
Clock
Forms
Figure 2. Group means of (a) exploration time, (b) fixation duration, (c) fixation duration variability, (d) saccade amplitude, and (e) saccade amplitude variability. Vertical bars are standard
errors. Data are shown for each painting and for each of the three conditions. In the spontaneous
condition group means are based on eleven subjects, in the active and the driven condition on
six subjects.
The data for mean fixation duration with SE are shown in figure 2b. The Kruskal ^
Wallis test applied to each painting showed the title effect for all three paintings
(all significant at p 5 0:05). The Mann ^ Whitney U test comparing any two title conditions showed significantly longer fixation in the driven condition than in the active
condition for all three paintings (all significant at p 5 0:05); for The Wedding, fixation
durations were also significantly longer relative to the spontaneous condition (U ˆ 13,
p 5 0:05).
Effect of title on eye-movement exploration of cubist paintings by Fernand Lëger
485
Group means of fixation-duration variability are shown in figure 2c. The Kruskal ^
Wallis test applied to each painting showed no significant effect for fixation-duration
variability. Such a negative result combined with the positive result for mean fixation
duration indicates that these two parameters can be controlled independently,
eg variability of fixation duration remains small even though mean fixation duration
increases.
Saccade amplitudes are shown in figure 2d. The Mann ^ Whitney U test showed
higher mean saccade amplitude for Contrast of Forms in the driven condition than in
the active condition (U ˆ 6, p 5 0:05). Saccade amplitude variability data are shown
in figure 2f; there was no title effect for any of the paintings.
To summarise, the results show that the authentic title in the driven condition
increases mean fixation duration for all paintings and does so without increasing
fixation-duration variability; it also renders saccade size significantly larger for Contrast
of Forms.
3.2 Effects of painting
The Friedman test applied to the exploration time for each condition separately
2
showed a significant effect of painting in the spontaneous condition only (w11
, 2 ˆ 8:91,
p 5 0:01); the Wilcoxon test used for comparisons between any two paintings revealed
that the exploration time was significantly longer for The Wedding than for The Alarm
Clock (T ˆ 2, p 5 0:01), or for Contrast of Forms (T ˆ 8, p 5 0:05).
The Friedman test applied to fixation-duration variability showed a significant
painting effect, particularly for the spontaneous condition (w62, 2 ˆ 6:72, p 5 0:05) and
for the driven condition (w62, 2 ˆ 7, p 5 0:05); in the driven condition the Wilcoxon
test indicated higher fixation-duration variability for The Wedding than for The Alarm
Clock (T ˆ 1, p 5 0:05), or for Contrast of Forms (T ˆ 0, p 5 0:05). This again
indicates that variability can be controlled to a certain degree independently from mean
fixation duration.
The Wilcoxon test indicated smaller saccade amplitude for The Wedding than for
The Alarm Clock (T ˆ 1, p 5 0:05) in the active condition; in the driven condition smaller
saccade amplitude for The Wedding than for Contrast of Forms (T ˆ 1, p 5 0:05).
In summary, comparison of the paintings with the same title condition shows
different results for The Wedding: longer exploration time (in spontaneous condition),
higher fixation-duration variability (in spontaneous and driven conditions), and smaller
saccade sizes (in active and driven conditions).
3.3 Spatial selectivityö distribution of fixation time
Each painting was arbitrarily divided into 12 areas and the percentage of fixation
time devoted to each area was calculated (see figure 3); results for each condition are
grouped over all subjects. If the fixation time had been distributed equally, each of
the areas would receive 8.33% of the fixation time. Here we will present only areas
receiving 12% or more of the fixation time, which is significantly different from the
theoretical value of 8.33% of equal distribution; such values ranged from 12% to 30%
indicating differences in viewing time among the preferred areas. The distributions are
different for the three paintings, and for the three conditions. In the spontaneous condition central area 8 is the most fixated in all three paintings as it contains the faces of
the married, the face of the person, and 3-D cylinders, respectively; the difference
between the paintings resides in the next most fixated areas (with values 4 12%). For
The Wedding, the next most fixated area is area 5 containing the dress; for The Alarm
Clock the next most fixated areas are 6, 9, and 11, and for Contrast of Forms areas 5,
9, and 6. The existence of several other frequently fixated areas indicates more than
simple central effects. Area 6 for this group of subjects attracts an important portion
of fixation time, like the neighbouring area 9.
486
Z Kapoula, G Daunys, O Herbez, Q Yang
The Wedding
Spontaneous condition
Active condition
Driven condition
50
50
50
40
40
40
30
30
30
20
20
20
10
10
10
0
0
0
50
50
50
40
40
40
30
30
30
20
20
20
10
10
10
0
0
0
50
50
50
40
40
40
30
30
30
20
20
20
10
10
10
0
0
0
The Alarm Clock
Contrast of Forms
Figure 3. [In colour online.] Distribution of fixation time over different areas.
In the active condition the data show particularly pronounced concentration on
the central area 8, and it is again the second most fixated area which differs among
paintings: namely area 5 for The Wedding and Contrast of Forms, but area 6 for The
Alarm Clock.
The distributions for the driven condition are also different for the three paintings
and can be summarised as follows. For The Wedding the most fixated areas are 8,
5, and 7, containing fragments of faces, the dress, and limbs. For The Alarm Clock the
areas are 8, 6, and 11önote that despite the global similarity of the distribution of
fixation time between the driven condition and the spontaneous condition there are
differences: area 6, ie the clock area, is now more frequently fixated while in the spontaneous condition this area is fixated at rates similar to area 9. This suggests more
clear perception of the clock in the driven condition owing to the knowledge of the title.
For Contrast of Forms, increased fixation time remained allocated to the same areas
for all conditions (8 and 5).
In summary, the results indicate both differences between paintings in each condition and differences across title conditions. The central area 8 attracts most of the
fixation time for all paintings and this is accentuated in the active condition, as if
the subjects had been spontaneously using this area as the vintage point to invent a title.
The title in the driven condition renders non-visually salient elements (such as the fragment
of The Alarm Clock) subject of distinct increases in fixation time.
3.4 Spatial selectivity over time
The aim of this analysis was to provide insight into how visual and semantic interpretation manipulated by the title influence spatial selectivity over time. Henderson and
Hollingworth (1998), as well as Itti and Koch's (2000) model of visual salience, suggest
Effect of title on eye-movement exploration of cubist paintings by Fernand Lëger
Spontaneous condition
First 5 s
50
The Wedding
Last 5 s
First 5 s
The Alarm Clock
Last 5 s
First 5 s
Contrast of Forms
Last 5 s
Active condition
First 5 s
50
Driven condition
First 5 s
487
50
40
40
40
30
30
30
20
20
20
10
10
10
0
0
0
50
Last 5 s
50
Last 5 s
50
40
40
40
30
30
30
20
20
20
10
10
10
0
0
0
50
First 5 s
50
First 5 s
50
40
40
40
30
30
30
20
20
20
10
10
10
0
0
0
50
Last 5 s
50
Last 5 s
50
40
40
40
30
30
30
20
20
20
10
10
10
0
0
0
50
First 5 s
50
First 5 s
50
40
40
40
30
30
30
20
20
20
10
10
10
0
0
0
Last 5 s
Last 5 s
50
50
40
40
40
30
30
30
20
20
20
10
10
10
0
0
0
50
Figure 4. [In colour online.] Distribution of fixation time over different areas during the first 5 s
and last 5 s of exploration.
488
Z Kapoula, G Daunys, O Herbez, Q Yang
that visual salience influences selection of fixated areas first, and semantic influences
follow. In contrast, the model of Locher (1996) suggests parallel rapid holistic processing of visual and semantic content. In figure 4 we plot the distribution of fixation
time during the first 5 s and during the last 5 s of exploration for each condition.
Inspection of the distributions shows that in most cases the distribution of fixation
time during the first 5 s is more concentrated on a few selected areas than during
the last 5 s; the three exceptions are the spontaneous condition for Contrast of Forms,
and the active and driven conditions for The Wedding for which fixation time remains
highly concentrated on the face and/or dress area (8 and 5). Extended spreading over
time is compatible with the idea of semantic influence and interpretation acting later
than visual salience. The spontaneous condition for The Alarm Clock is indicative of
such evolution: the clock area, area 6, initially moderately fixated, becomes later a
pole of intensive fixation. Yet, for the same painting in the driven condition, under the
top ^ down influence of the authentic title this area is intensively fixated in the first
5 s and rather ignored in the last 5 s. These observations suggest that visual and
semantic influences can act sequentially or simultaneously and very rapidly, depending
on the painting and on the title condition; in other words, the interplay between title
and visual processing depends on the painting.
4 Discussion
Our aim in this study was to determine the effects of the title for three non-realistic
paintings, more or less abstract. The results provide clear evidence that both the
physiological parameters of eye movements and the distribution of fixation time
over different areas of the paintings are influenced by the title information. Fixation
duration increases for all paintings in the driven condition; saccade amplitude also
increases in this condition, but only for Contrast of Forms. The Wedding produced
different results, particularly in the spontaneous and driven conditions. Each of these
results will be discussed next.
4.1 Effects of title
Fixation duration is an important parameter of visual exploration of natural scenes
reflecting cognitive processing (Henderson and Hollingworth 1998). Guo et al (2006)
reported longer fixation durations when viewing faces than when viewing natural scenes.
Longer fixations for all paintings in the driven condition could be due to higher cognitive
activity. The authentic title would activate an effort to fit the title to the painting;
this influences semantic analysis and leads to longer fixations. Interestingly, however,
the title condition had no significant effect on the variability of fixation durations. This
clearly indicates that the two parameters can be controlled independently. Moreover,
the title in the driven condition increased saccade sizes for Contrast of Forms; larger
saccade steps could be indicative of an effort to compare (link, oppose) fragments of
the paintings giving perception of contrasted forms. Gombrich (1985), like Levinson
(1985) cited by Franklin et al (1993), pointed out that title is a significant contextual
factor, and that different titles engage the viewer in different ways. In some cases titles
provide explicit directions for interpretations; the attempt to achieve fit between
the title and the visual image would become part of the viewer's experience. With the
paintings used by us here being non-realistic, the title information presumably influenced interpretation and stimulated the effort to search the fitting between title and
visual content of the painting. The increase of fixation durations and of saccade
amplitudes provide physiological support for the above-cited ideas of Gombrich (1985)
and Levinson (1985).
Effect of title on eye-movement exploration of cubist paintings by Fernand Lëger
489
4.2 Differences between paintings
Without title, the subjects spontaneously adopted different exploration time according
to painting; The Wedding got longer exploration time. The titles invented by the subjects
in the active condition are indicative of the complexity of social situation perceived for
this painting (see section 3). Recall that the painter intended to depict a marriage,
and human faces and limbs were really painted and seen by the subjects. We suggest
that multiple fragments of faces and limbs stimulate complex analysis, including recognition, memory, categorisation operations, and emotional evaluation of the depicted
faces and of the social situation; this leads to longer exploration time.
Fixation duration was also more variable for The Wedding. One could argue that
increased variability of fixation direction is due to a visual factor, eg colour, requiring
attention of spatial frequency for this painting. It is known, however, that such features
have little influence on eye movement. Indeed, Henderson and Hollingworth (1998)
reported no influence on the distribution of fixation durations of the type of graphic
support (drawings versus photographs of natural scenes). Another aspect to be considered is the higher density of small fragmented elements and edges in The Wedding.
Unema et al (2005) reported that increased density of elements in artificial visual
scenes (assembly of objects) influences fixation duration; however, fixation duration
and its variability were found to decrease when the density of elements increased, and
this is rather opposite to what we observed in The Wedding. Thus, visual factors alone
cannot explain our observations of more variable fixations in The Wedding. Most likely,
variability is due, at least partially, to complexity of semantic analysis. Saccade sizes
were found to be smaller for The Wedding, particularly in the active and driven conditions. First, we should consider again the visual factors, namely the higher density and
the smaller size of the fragments in this painting as being the cause of this decrease
in saccade sizes. Indeed, Unema et al (2005), using objects in artificial scenes, have
convincingly shown that the amplitude of saccades decreases as the number of elements
in the scene increases. A similar factor may be involved in our study.
In the study of Unema et al (2005) both saccade amplitude and fixation duration
decreased when the number of objects in the scene increased. Our observations (see
figure 2) go against the idea of a common mechanism controlling saccade amplitude
and fixation duration. Such a mechanism may exist for artificial scenes, but not for
paintings such as The Wedding (fixation durations are longer for the Wedding while
their saccade amplitude is smaller; see the driven condition).
4.3 Spatial selectivity, visual salience, power of centre, and title influences
Artists know now to combine in the same area visual aesthetic and semantic information. The power of the centre is the title of the book of Arnheim (1988); our findings
support the power of the centre as discussed by Arnheim, and are in line with several
other empirical studies, eg that of Locher (1996). Indeed, the central area 8 contained
most meaningful information: the faces of the married, the face of the person, or 3-D
cylinders; this area received more fixation time for all paintings. The differences
between paintings reside more in the next, less-central highly fixated areas. The title
condition did influence the evolution over time of spatial selection. In the driven condition, The Alarm Clock was highly fixated in the first 5 s, while in the spontaneous
conditions it is fixated at the end of the exploration time and not very distinctively
from the neighbouring areas (6 and 9, see figure 3). Our observations for the driven
condition suggest that top ^ down and bottom ^ up influences can act in parallel and
immediately. Potter (1976) [cited by Henderson and Ferreira (2004)] reported that viewers
can detect very rapidly a target scene identified by a verbal label. The authentic title in
the driven condition acted as a verbal label driving the eyes rapidly to The Alarm Clock.
490
Z Kapoula, G Daunys, O Herbez, Q Yang
On the other hand, the late fixation of the clock in the spontaneous condition is
compatible with the model proposed by Henderson and Hollingworth (1998), according
to which exploration, initially guided by visually salient elements, is followed by semantic
interpretation modifying their salience. Our findings showing almost immediate influence
of the title on the spatial distribution of fixation time over the painting clearly go
against the conclusion of Franklin et al (1993) based on light-pointer hand movements.
The present study shows, to our knowledge for the first time, that title information
does influence the way observers explore visually the painting, and does so in many
ways. As we make about 2 saccades per second, the first 10 saccades allow us to fixate
the majority of meaningful points of the painting that are subsequently refixated several
times. Thus, the first 5 s of exploration could correspond essentially to motor exploration
driven by the initial fast holistic visual processing proposed by Locher's model (1996);
we suggest that such motor activity in its turn consolidates the initial holistic visual
image and its storage in memory.
The results, particularly for The Alarm Clock, show that the influence of the
title on spatial selectivity depends also on the painting. This supports the theoretical
writings of Gombrich (1985) and Levinson (1985) emphasising that title is a significant
contextual factor engaging the viewer differently, depending also on the painting.
4.4 Face perception from non-realistic paintings
A final aspect revealed by this study is the power of attraction of the human face.
This is well known in human psychophysiology (Kanwisher 2000; Thorpe et al 1996).
Nevertheless, the faces studied here are peculiar and to our knowledge explored for
the first time. In fact, there is no face in The Alarm Clock, but a face is perceived
by almost all subjects. In The Wedding the faces are real albeit incomplete, covered by
transparent surfaces and surrounded by arbitrary elements. Yet, they are perceived,
and constitute the centre of eye movement exploration and the basis of underlying
semantic interpretation of the painting. Semantic interpretation of the observer does
not always meet the artist's intention. In fact, it is remarkable that Fernand Lëger
spent some effort, particularly in his latest figuration period, in depicting faces without
emotion, his artistic aim being to treat the human face like another object in order to
emphasise the plastic value and avoid sentimentality. As pointed out by Flam (1987),
fragments of human figures in Lëger's work, although depersonalised, cannot always
be regarded as `desentimentalised', and do evoke a new kind of human experience.
Acknowledgment. Qing Yang and Olivier Herbez were supported from the CEE Eurokinesis contract
(QLT-2000-00442).
References
Arnheim P, 1988 The Power of the Center (Berkeley, CA: University of California)
Buswell G T, 1935 How People Look at Pictures: A Study of the Psychology of Perception in Art
(Chicago, IL: University of Chicago)
Flam J, 1987 ``Important retrospective of Lëger's oil on canvas works from 1905 ^ 1950'', in
Catalogue of the Exhibition Fernand Lëger (New York: Acquavella Galleries, USA)
Franklin M B, Robert C, Becklen R, Doyle C, 1993 ``The influence of titles on how paintings
are seen'' Leonardo 26 103 ^ 108
Gombrich E H, 1985 ``Image and word in twentieth-century art'' Work and Image 1 213 ^ 241
Guo K, Mahmoodi S, Robertson R G, Young M P, 2006 ``Longer fixation duration while viewing
face images'' Experimental Brain Research 171 91 ^ 98
Henderson J M, Ferreira F, 2004 ``Scene perception for psycholinguists'', in The Interface of
Language, Vision, and Action: Eye Movements and the Visual World Eds J M Henderson,
F Ferreira (New York: Psychology Press)
Henderson J M, Hollingworth A, 1998 ``Eye movements during scene viewing: An overview'',
in Eye Guidance in Reading and Scene Perception Ed. G Underwood (Oxford: Elsevier)
Itti L, Koch C, 2000 ``A saliency-based search mechanism for overt and covert shifts of visual
attention'' Vision Research 40 1489 ^ 1506
Effect of title on eye-movement exploration of cubist paintings by Fernand Lëger
491
Kanwisher N, 2000 ``Domain specificity in face perception'' Nature Neuroscience 3 759 ^ 763
Kapoula Z, Isotalo E, Muri R M, Bucci M P, Rivaud-Pechoux S, 2001 ``Effects of transcranial
magnetic stimulation of the posterior parietal cortex on saccades and vergence'' NeuroReport
12 4041 ^ 4046
Laugier C, Richet M, 1981 ``Oeuvres de Fernand Lëger (1881 ^ 1955), catalogue de la collection
du Musëe National d'Art Moderne'', Centre Georges Pompidou (Paris: Editions Stock)
Leder H, Belke B, Oeberst A, Augustin D, 2004 ``A model of aesthetic appreciation and aesthetic
judgments'' British Journal of Psychology 95 489 ^ 508
Leder H, Carbon C C, Ripsas A L, 2006 ``Entitling art: Influence of title information on understanding and appreciation of paintings'' Acta Psychologica (Amsterdam) 121 176 ^ 198
Levinson J, 1985 ``Titles'' Journal of Aesthetics and Art Criticism 44 29 ^ 39
Locher P, 1996 ``The contribution of eye-movement research to an understanding of the nature
of pictorial balance perception: a review of the literature'' Empirical Studies of the Arts 14
143 ^ 163
Locher P J, 2006 ``The usefulness of eye movement recording to subject an aesthetic episode with
visual art to empirical scrutiny'' Psychology Science 48 106 ^ 114
Locher P, Gray S, Nodine C, 1996 ``The structural framework of pictorial balance'' Perception 25
1419 ^ 1436
Potter M C, 1976 ``Short-term conceptual memory for pictures'' Journal of Experimental Psychology:
Human Learning and Memory 2 509 ^ 522
Takagi M, Frohman E M, Zee D S, 1995 ``Gap-overlap effects on latencies of saccades, vergence
and combined vergence-saccades in humans'' Vision Research 35 3373 ^ 3388
Tatler B W, Baddeley R J, Gilchrist I D, 2005 ``Visual correlates of fixation selection: effects of
scale and time'' Vision Research 45 643 ^ 659
Thorpe S, Fize D, Marlot C, 1996 ``Speed of processing in the human visual system'' Nature 381
520 ^ 522
Unema P, Pannasch S, Joos M, Velichkovsky B M, 2005 ``Time-course of information processing
during scene perception: The relationship between saccade amplitude and fixation duration''
Visual Cognition 12 473 ^ 494
Westheimer G, McKee S P, 1975 ``Visual acuity in the presence of retinal-image motion'' Journal
of the Optical Society of America 65 847 ^ 850
Yang Q, Bucci M P, Kapoula Z, 2002 ``The latency of saccades, vergence, and combined eye movements in children and in adults'' Investigative Ophthalmology & Visual Science 43
2939 ^ 2949
Yang Q, Kapoula Z, 2003 ``Binocular coordination of saccades at far and at near in children
and in adults'' Journal of Vision 3 554 ^ 561
Yarbus A L, 1967 Eye Movements and Vision (New York: Plenum Press)
ß 2009 a Pion publication
ISSN 0301-0066 (print)
ISSN 1468-4233 (electronic)
www.perceptionweb.com
Conditions of use. This article may be downloaded from the Perception website for personal research
by members of subscribing organisations. Authors are entitled to distribute their own article (in printed
form or by e-mail) to up to 50 people. This PDF may not be placed on any website (or other online
distribution system) without permission of the publisher.