Another study looked into the effects of emotions on gaze

1
Being looked at and social anxiety
Judgments of being looked at is a marker of social anxiety in males; measuring the cone of
direct gaze.
Yae Young Juna
Isabelle Maraschelb, Colin W.G. Cliffordc
Mark R Daddsd, *
a
sdf
b
sdf
c
sdf
d
sdf
Abstract
Previous studies suggest that socially anxious individuals overestimate the number of people
looking at them. In order to test this, the current study measured the level of accuracy in
discerning who is looking at them and not (cone of direct gaze) by using the gaze
categorisation task. The CoDG measures the range of eye gaze directions that you perceive as
looking directly at you. Results revealed that socially anxious males had a wider CoDG than
low socially anxious males. High and low socially anxious females did not differ in their
CoDG. Furthermore, both low socially anxious males and females had an emotion effect on
CoDG, where fearful faces narrow cones than neutral or angry, but high socially anxious
males did not reflect these findings. Males then appear more susceptible for their gaze
perception to be affected by individual differences, whereas females are more influenced by
contextual information (emotions). We explain these results by the innate preferences for
social stimuli, like eye contact, in females allowing immunity to their individual differences.
Keywords: accuracy, gaze perception, eyes, gender, emotion
2
Being looked at and social anxiety
1.
Introduction
Socially anxious people differ from non-socially anxious people in that they have an
excessive fear of scrutiny – one of the most common underlying fears associated with social
anxiety (Barlow & Durand, 2009). A main feature that activates the feeling of being
scrutinised is whether that person is looking directly at you or not, or more specifically,
whether you think they are looking at you or not. This perception of feeling looked at can be
measured by using the Gaze Categorisation Task (GCT), devised by Calder and colleagues
(Mareschal, Calder, Dadds, & Clifford, in review). The index produced is commonly known
as the cone of direct gaze (CoDG). By using the CoDG, researchers are able to quantitatively
analyse what factors influence the accuracy in judging eye gaze direction. Throughout this
article, the gazer refers to the face stimuli or person that is doing the gazing. The term,
observer is defined as the individual who is judging where the gazer is looking.
CoDG is a range of gaze directions that an observer perceives as direct (i.e. looking
directly at them). Put in another way, it is the level of accuracy in discerning direct from
averted gazes. The edges of this cone and anywhere in between are where the observer judges
the eye gaze as looking at him/her. The GCT used to measure the gaze cone involves the
presentation of faces looking at varying directions. The subject is required to judge whether
each face is looking directly at them or not which generates the estimated width of when the
subject perceives a direct and averted gaze (i.e. the cone of direct gaze).
Previous studies that have used CoDG measurements investigated how emotions and
other contextual features affect how much we feel looked at by others. For instance,
Mareschal and colleagues (in review) found that gaze perception involves the combination of
head orientations with eye gaze. Specifically, participants were better at differentiating a
direct gaze from an averted one when shown just the eyes than when the eye gazes were
3
Being looked at and social anxiety
superimposed onto a head facing straight at the observer. This was because such head
orientation causes participants to perceive the face as looking at them, which has been
supported by other studies (Cline, 1967; Gibson & Pick, 1963; Langton, Honeyman, &
Tessler, 2004; Martin & Rovira, 1982; Symons, Lee, Cedrone, & Nishimura, 2004).
Another study looked into the effects of emotions on gaze perception by measuring
the CoDG of angry and fearful faces (Ewbank, Jennings, & Calder, 2009). They found that
observers had a wider gaze cone for angry faces than for fearful or neutral faces. In other
words, angry faces were thought to be looking at the participant over a wider range of gaze
directions. The shared-signal hypothesis lends support for this finding (Adams & Kleck,
2003), which states that eye gaze and emotional expressions are processed together and
consequently enhance the perception of each other. More specifically, approach-related
emotions such as happiness and anger enhance the processing of direct gazes because a direct
gaze implies an approaching motivation. Avoidant-related emotions, on the other hand,
enhance averted gaze processing as they both imply a withdrawing motivation.
Ewbank and colleagues (2009) also found great variations between subjects and
suggested individual differences to play a significant role. One possible individual factor, as
previously mentioned, includes social anxiety. The hypervigilance-avoidance theory has
implications on how social anxiety might play a role in gaze judgments. This theory suggests
that social anxiety is characterised by an initial hypervigilance to threatening stimuli, which
after a prolonged period (of about roughly 500ms) is followed by avoidance to the stimuli
(Onnis, Dadds, & Bryant, 2011; Wieser, Pauli, Weyers, Alpers, & Muhlberger, 2009).
Because socially anxious people have a fear of being observed, the initial attentional grab
towards the eyes increase the chances of evaluating eye gaze to be direct. Soon after, this
early vigilance is followed by avoidance of the eyes, preventing those who are socially
4
Being looked at and social anxiety
anxious from re disconfirming their perception that others are looking at them. Hence, people
who are socially anxious may have a wider CoDG or poorer accuracy in perceiving direct
gaze due to this gaze processing method.
One study tested this hypothesis (Gamer, Hecht, Seipp, & Hiller, 2011). Eight patients
with social phobia and eight healthy controls had their CoDG measured using the centeringdecentering task (Gamer et al., 2011). In the centering task, participants adjusted the eyes of a
face (that was originally looking extreme right or left) just so that the face was looking
directly at them. The decentering task was similar to the centering task, but the face began by
displaying a direct gaze, and participants were to adjust the eyes so that the face started to no
longer look at them. Each participant performed these two tasks when head orientation was
facing towards or away from the participant and sometimes in the presence of another head
(distractor head), which was displayed on a second monitor beside the first. Results revealed
the CoDG of social phobic patients widened only in the presence of distractor head.
Surprisingly, in the absence of a second gazer, Gamer et al.’s (2011) results found no
effect of social anxiety on perception of gaze direction. However, with only 16 participants in
total, there is great room for a type II error. Furthermore, the centring and decentring
adjustment method is not optimal for measuring gaze cones as these tasks are a lot less
reliable and prone to variability compared to the gaze categorisation procedure (Hock &
Schoner, 2010). It is also unrealistic that anyone would be asked to manipulate another
person’s eyes to centre or decentre their gaze, or gaze at a face for prolonged periods of time.
Thus, the task does not reflect an authentic situation and lacks statistical power and so further
investigation is needed to find whether social anxiety levels affect how much one perceives to
be looked at.
5
Being looked at and social anxiety
When considering sociality factors (such as eye gaze and social anxiety), it is
important to take gender into account because there is an innate preference for social stimuli
for females but not in males (Connellan, Baron-Cohen, Wheelwright, Batki, & Ahluwalia,
2000; Lutchmaya & Baron-Cohen, 2002; Lutchmaya, Baron-Cohen, & Raggatt, 2002). In
particular, females respond to eye contact and eye gaze differently to males. Studies have
shown that females – both as infants and adults– make more eye contact and prefer faces and
mutual direct gaze over mobiles, while male infants prefer mobiles (Bayliss, di Pellegrino, &
Tipper, 2005; Connellan et al., 2000; Lutchmaya & Baron-Cohen, 2002). Females are also
better at engaging in joint attention (looking in the direction another is looking at) and depend
on gaze direction to affect their attention more than males (Geary, 1999).This dependence on
and early experience of gaze direction would lead to better discriminating ability in whether
someone is looking at them or not. That is, females may have a narrower CoDG than males
due to their sensitivity to eye gaze.
Other socially-related variables, such as the familiarity of a face, affect gaze cueing
differently between genders as well. Female participants who were more familiar with the
faces in the task were using gaze cues to influence their decisions more than the unfamiliar
group (Deaner, Shepherd, & Platt, 2007). This pattern of results was not the case for males. If
Deaner et al.’s (2007) results generalise to other social factors, it is plausible that the effects
that emotions have on eye gaze (i.e. angry faces lead to an overestimation in the number of
direct gazes perceived), may be largely seen in females. This is a reasonable assumption as
females perform better on emotion recognition tasks than males (Baron-Cohen, Jolliffe,
Mortimore, & Robertson, 1997). Therefore, being able to recognise an emotion more
accurately suggests advanced emotions processing and enhances the ability to judge gaze
direction.
6
Being looked at and social anxiety
Gender differences can also be found in social anxiety symptoms (Pierce, 2009). For
instance, young socially anxious women are likely to have more physiological arousal, feel
more observed by others and be anxious on more social situations than males (Ingles et al.,
2011). Men, on the other hand, are more likely to escape and avoid social situations (Ingles et
al., 2011). Consequently, males might avoid eye gaze more intensely because direct eye gaze
signals the intention to engage in social interaction. As suggested by the hypervigilanceavoidance hypothesis, deliberately avoiding the eyes can increase the chances of mistaking
the gazer to be looking at them and hence, generate a wider CoDG. Thus, more avoidance in
high socially anxious males could be associated with overestimating the number of people
looking at them compared to high socially anxious females.
Therefore, there were two main aims of this study. Firstly, we wanted to determine
whether socially anxious individuals, in particular males, feel more looked at by measuring
their cones of direct gaze. Secondly, we wanted to see whether emotions had a greater effect
on gaze perception for females than males. Due to the novelty of this study, we also explored
possible mechanisms that underlie an effect between social anxiety and CoDG. This included
how comfortable or frequent participants feel looked at, the gender of the experimenter and
gazers and the level of threat perceived in each face. It was hypothesised that high social
anxiety (especially high socially anxious males) participants have a wider CoDG than low
social anxiety participants. Furthermore, we proposed that gaze cones would be wider for
angry faces and narrower for fearful faces (Ewbank et al., 2009), and that this effect will be
greater in females.
7
Being looked at and social anxiety
2.
Method
2.1. Participants
93 students (Male = 33; Mage = 19.18; SDage = 2.270), who were enrolled in courses
Psychology 1A or Psychology 1B at the University of New South Wales participated for
course credit. All participants had corrected-to-normal vision. Participants eligible for this
study were those who were high (reported “4” out of a total score of 4) or low (reported “1”)
on the 9th question of the Depression Anxiety and Stress Scale (DASS) measure (“I was
worried about situations in which I might panic and make a fool of myself”). Written and
informed consent were obtained from participants.
2.2. Materials and Stimuli
2.2.1. Face Stimuli. Four male faces from Daz software (Mareschal et al., in review)
were used for this study. These faces were all expressing a neutral emotion, presented at the
centre of the screen and had hair removed to avoid distraction from the eyes and facial
expressions. At a viewing distance of roughly 57cm, the faces subtended on average at
15.1x11.2 degrees. For each of the faces, angry, fearful and neutral faces were manipulated
using the software FaceGenerator, Adobe Photoshop and FaceFilter Studio 2. Figure 1a
shows each male face stimulus expressing each emotion. Ultimately, each face stimulus had
its eyes removed, was set to grey-scale and had an image size of 1024 x 767 pixels. To ensure
each face was conveying the right emotion, they were shown to 10 people (NMale = 5) who
were asked to state the emotion expressed out of 4 possible choices (unsure, neutral, fear and
anger). On average, participants correctly reported emotions 90% of the time.
2.2.2. Eye gaze. There were nine eye gaze directions which subtended at different
horizontal visual angles (-12º, -6º, -3º, -1º, 0º, 1º, 3º, 6º, 12º; see Figure 1b) These deviations
8
Being looked at and social anxiety
were generated following the method of Mareschal et al. (in review). An eyes-only stimulus
was superimposed on each face and independently manipulated using MATLAB™
(MathWorks Ltd) to rotate the eyes along the horizontal axis. Each of these nine eye
deviations was repeated three times for each of the four faces used. This was done for the
three emotion conditions: neutral, fear and anger. Therefore, in one run there was a total 432
faces.
2.2.3. Gaze Categorisation Task. A computer program, using MATLAB was used to
measure the cone of direct gaze. The laptop was always positioned facing the centre of the
chair. Each face was randomly presented for 500ms, followed by a 300ms inter-stimulus
interval (Figure 2). Participants were then required to indicate the gaze of the faces as averted
to the left, direct or right by using the keys “j”, “k”, and “l” respectively. The screen
remained grey and the next stimulus was presented after the participant pressed one of the
response keys. If a key other than the response keys was pressed, a short, unobtrusive sound
was set off and the next stimulus was shown. These data were neglected.
2.2.4. Measures of Social Anxiety. Both the Social Avoidance and Distress
(SAD;(Watson & Friend, 1969)) and Fear of Negative Evaluation (FNE; Watson, & Friend,
1969) scales are self-report questionnaires and were used to measure social anxiety levels.
2.3. Design
A 2 x 2 x (3) mixed design was used, where the two between-subjects factors were
Gender (Male and Female) and Social Anxiety (HSA – High Socially Anxious; LSA – Low
Socially Anxious), and the within-subjects design was the Emotions (N – Neutral; F – Fear;
A – Anger) expressed by the face stimuli. To categorise participants into either high or low
socially anxious groups, those who scored 16 or more (out of a total score of 30) on the FNE
were categorised into the HSA group, and 15 or below into the NSA group. This was because
9
Being looked at and social anxiety
cut-off for the top quartile of social anxiety levels in the non-clinical population is roughly
15-17 (Mansell, Clark, Ehlers, & Chen, 1999; Winton, Clark, & Edelmann, 1995). The
dependent variable measured was the cone of direct gaze, which was measured using the gaze
categorisation task (see below or Mareschal et al., in review).
2.3.1. Mechanisms of Gender Effects (MOGE). This questionnaire consisted of 19
items that addressed four possible mechanisms that account for gender differences in CoDG
and social anxiety levels: level of comfort when being looked at by others; level of threat
perceived in face stimuli; the gender of the gazer, and; awareness of the experimenter.
Responses to these questions were made on a 5-item Likert scale, where 1 stood for “not at
all”, 2 “only slightly”, 3 “moderately”, 4 “quite a lot” and 5 “extremely”. See Appendix B.1
for a sample of the questionnaire.
2.4. Procedure
Participants were given an information sheet to read. Once the consent form was
completed, participants were told a brief instruction of the tasks. When explaining the
instructions for the computer task, it was emphasised that participants were to sit still and to
minimise moving their heads in order to keep the visual angles consistent. A chin-rest was
not used as it has been reported to cause discomfort. The experimenter remained in the room,
but emphasised that she will not be watching them perform the task and sat with her back to
the participant. The computer task was run twice, allowing an un-timed break in between
until the participant wished to continue. After, the FNE and SADS questionnaires were asked
to be filled out. For 60 of these subjects (Male = 19, Mage = 19.03, SD = 2.456), the MOGE
questionnaire was also handed out. Upon completion, the researcher thanked the participant,
debriefed them.
10
Being looked at and social anxiety
2.5. Data analysis
2.5.1. Measuring CoDG. The gaze cones were measured using the method of
Mareschal et al. (in review). Firstly, responses from the two runs were compiled into one data
set for each participant. Figure 3 shows the data for one participant as a function of the eye
deviation for the neutral face condition. The data for the “left” (diamonds), “right” (triangles)
and “direct” (squares) responses are plotted as a proportion of the total number of responses
for each eye deviation. For example, when the eye deviation was -12, this participant always
responded “left”. Logistic functions were fit to the proportion of left and right responses
(dashed and dotted curves) and a function for “direct” (solid curbe) was fit by removing the
sum of the left and right curves from 1(Ewbank et al., 2009). Curve fits used the NelderMead simplex method (Nelder & Mead, 1965). This method minimises the residual variance
and can be operated through the fminsearch function on MatLab. The cone of direct gaze is
taken as the measure between the points of intersection between the two averted curves and
the “direct” curve: one where the left and direct responses crossed over and the other where
the direct and right responses crossed. These two cross-overs are termed categorical
boundaries.
2.5.2. Statistical Approach. Independent samples t-tests were used to compare genders
and social anxiety groups on their age, FNE scores, SAD scores and CoDG. To investigate
whether contextual and individual factors interacted to influence CoDG, MANOVAs were
used when Emotions variable were analysed, and ANOVA was used to analyse betweensubject factors Gender and Social Anxiety (High or Low). Paired sample t-tests were carried
out to observe simple main effects of emotions on CoDG. All correlations reported in this
experiment were a one-tailed bivariate Pearson correlation. Moderate correlations (greater
than .20) were reported. Alpha for all statistical analyses was controlled at .05.
11
Being looked at and social anxiety
3.
Results
3.1. Group demographics
Table 1 shows the group characteristics of High and Low Social Anxiety groups. Both
Social Anxiety groups did not differ in age, percentage of males in each group and FNE
scores. Table 2 shows the group characteristics of males and females. There were no
significant differences between gender groups on age and FNE scores.
3.2. Interaction between contextual and individual factors on CoDG
Emotions, gender and social anxiety interacted to affect CoDG. This was reflected by
a significant a three-way interaction MANOVA effect amongst the three variables (F(2,87) =
3.119, p < .05).
From observing Figure 4, HSA males appeared to have a wider CoDG than low
socially anxious males. However, this was not significant when analysing these groups via a
univariate ANOVA (F(1,31) = 1.906, p = .177). It also seemed like CoDGs in high socially
anxious males were wider than their female counterparts, but was also non-significant
(F(1,27) = 1.501, p = .231). Furthermore, effects of emotions on CoDG were not exacerbated
in HSA groups, determined by a non-significant result from MANOVA (F(2,87) = 1.608, p
= .206).
When analysing Figure 4, high socially anxious males clearly did not conform to the
general pattern of emotions effects on gaze perception. This was shown statistically by using
paired-samples t-tests to compare how two emotions affected CoDG. Fearful faces were
narrowing CoDG significantly more than neutral or angry faces for females (Fear-Neutral:
t(58) = -3.676, p < .005; Fear-Anger: t(58) = -2.284, p < .05), but not for males (Fear-Neutral:
t(32) = -2.009, p= .053; Fear-Anger: t(32) = -.581, p= .565). Fearful faces were also
12
Being looked at and social anxiety
narrowing CoDG significantly more than both neutral and angry faces in LSA group (FearNeutral: t(62) = -3.383, p < .005; -Fear t(62) = -2.031, p < .05), but in the HSA group, fear
only narrowed the CoDG more than neutral faces (Fear-Neutral: t(28) = -2.575, p<.05 FearAnger: t(28) = -1.083, p= .288). This trend is clearly visible in Figure 4, where high socially
anxious males are not following the same trend of different emotional expressions on CoDGs.
A MANOVA has also revealed a significant main effect for Emotions on CoDG
(F(2,87) = 7.875, p < .005). Paired-samples t-tests revealed that the CoDG width was
narrower for fearful faces compared to neutral (t(91) = -4.181, p < .001) or angry (t(91) = 2.207, p < .05) faces. Angry faces on the other hand, were not different to neutral faces in
affecting the CoDG (t(91) = .1692, p = .094).
3.3. Analysis of possible mechanisms underlying gender difference
Correlations between each subtype of the MOGE questionnaire on social anxiety
measures (FNE and SAD) are shown in Table 3. Levels of social anxiety were generally
correlating more on each subtype of the MOGE questionnaire for females than for males. For
females, high social anxiety was most correlated with their level of comfort of being looked
at by other females (where more discomfort was associated with higher social anxiety: r
= .302; p < .05). The level of awareness of the experimenter’s presence was most
significantly correlated with FNE scores in males, especially the awareness of the
experimenter’s gender (r = .302). It might be interesting to note that the relationship between
awareness of the experimenter and level of social anxiety in males was stronger in the SAD
measure (r = .519; p < .05)
Table 4 shows correlations between subtypes of the MOGE questionnaires and CoDG
across each emotion for males. For males, the gender of the face seemed relevant to their
CoDG, where more awareness led to a narrow gaze (r = -.447). Social anxiety levels in males
13
Being looked at and social anxiety
were moderately correlated in a positive fashion with CoDG. However, this was only
observed for neutral and fearful faces, but not angry faces, as suggested by the three-way
MANOVA.
Correlations between CoDG subtype measures from the MOGE and FNE scores were
generated for females and shown in Table 4. CoDG was strongly and negatively correlated
with how comfortable female participants feel at being looked at (r = -.536; p < .01). The
more they felt looked at by others was associated with a narrower CoDG (r = -.316),
especially for fearful faces (r = -.352; p < .05). Other variables had no noteworthy
associations with CoDG in females.
4.
Discussion
The aim of this study was to investigate whether the feeling of being looked at is a
marker of social anxiety. As predicted, high social anxiety levels were associated with an
overestimation in perceiving direct gazes and were only found in males. Furthermore, fearful
faces produced a narrow gaze cone than angry or neutral faces. This effect was more
prominent for females than males.
The trend indicating that a wider cone of gaze is a specific social anxiety marker in
males is supported by the hypervigilance-avoidance hypothesis (e.g. Onnis et al., 2011). That
is, socially anxious individuals find an initial grab towards social stimuli (such as the eyes)
and actively avoid them, preventing them from disconfirming their beliefs of being looked at.
Males in particular avoid anxiety-provoking situations more than females (Ingles et al., 2011),
explaining why gaze cones were wider for only socially anxious males. Females, on the other
hand, do not show such widening effect with increasing levels of social anxiety because of
their early experience to identifying eye gaze (Connellan et al., 2000; Lutchmaya & BaronCohen, 2002).
14
Being looked at and social anxiety
Possible mechanisms underlying social anxiety effects on gaze perception revealed
that social anxiety in females decreased their level of comfort of being looked at by others,
which consequently widened their CoDGs. However, social anxiety in females increased the
perceived frequency of being looked at, which narrowed their CoDGs. Hence, direct gaze
perception in high social anxiety in females is affected by these two opposing mechanisms.
As a result, these two mechanisms cancel out each other and generate no significant
correlation between social anxiety and CoDG. For males, social anxiety was not associated
with widening the perception of direct gaze through these mechanisms.
Once again, in accordance with the hypervigilance-avoidance hypothesis, a wider
cone of direct gaze is due to avoiding the gazing face before one can properly determine
whether the face was looking at them or not (Onnis et al., 2011). The recorded amount of
discomfort in women is relative to how likely they are to avoid the gazes of others, which
ultimately widens their gaze cones. However, when correlating social anxiety with CoDG,
there is barely any association. This is most likely because females have more experience in
engaging in eye gaze and discriminating gaze directions (Connellan et al., 2000; Lutchmaya
et al., 2002). An advantage in telling apart where someone is looking may have nullified any
effect of social anxiety on gaze perception.
In the study reported here, results have shown that fearful faces narrowed perception
of direct gaze more than angry and neutral faces, where effects of anger and neutral did not
differ. However, past research found that our ability to distinguish true direct gaze from
averted was worse when faced with angry faces compared to fearful and neutral, where
fearful was not different to neutral (e.g. Ewbank et al., 2009). A reasonable account for this
discrepancy is that the absence of an anger effect is due to the neutral faces looking almost as
15
Being looked at and social anxiety
threatening as angry faces. An alternative reason is that the neutral faces were ambiguous,
which also increases the chances of it being perceived as threatening (Georgiou et al., 2005).
Nevertheless, there was a significant difference between angry and fearful faces in
affecting how we perceive direct gaze as suggested by the shared signal hypothesis (Adams
& Kleck, 2003). Overall, emotions are one of the socially-relevant contextual cues that make
an impact on whether we judge a person to be looking at us or not. Hence, our current
findings support previous CoDG studies and the shared-signal hypothesis. Interestingly, gaze
perception for women as more affected by emotions than men. Though this could be due to a
larger sample size for women, this finding can also be explained by their earlier experience in
eye gaze. Such experience allows women to attenuate to finer details and contextual factors
without being interrupted by their individual differences.
There are a few limitations that should be considered. The most problematic was the
statistical power problem which reduced the study’s ability to detect any significant effects.
Furthermore, though we instructed participants the importance of keeping their head still, it is
difficult for them to remain in the same position for 15 minutes continuously. Despite
instructions, participants may have been moving their observing position within trials. This
increases the amount of variance generated by each person, and ultimately increases error
variance in each between-subjects group. Especially, socially anxious males may have been
subconsciously leaning backwards as they have a higher tendency to avoid social situations
than females (Ingles et al., 2011). This may have led them to systematically have a wider
CoDG than other groups. Hence, the small sample size and large variance may account for
the substantial standard errors and wider gaze cones in high socially anxious males.
The current study has opened up multiple paths that require further investigation.
Future studies should consider other factors that could affect CoDG. For instance, autism-
16
Being looked at and social anxiety
spectrum disorders are associated with a lack of eye contact and reduced joint attention
(Lutchmaya & Baron-Cohen, 2002; Lutchmaya et al., 2002). Hence, it may be of interest to
look into how autism-like traits have an effect on direct gaze perception. Other individual
differences affecting sociality, like conduct disorder or psychopathic-like traits may be
interesting to investigate.
Eye-tracking measures might be used to further support how wider gaze cones in high
socially anxious individuals were due to their conscious effort to avoid eye gaze, preventing
disconfirmation of whether they were being looked at. Physiological measures such as heart
rate, and skin conductance rates (sweating) might further investigate whether social anxiety
has a correlative relationship with gaze perception. Likewise, brain-imaging studies might be
helpful in finding gender or trait social anxiety differences in brain activation when
processing emotions and eye gaze.
The current findings of a large gender effect is an indication that other studies
measuring eye gaze or are using tasks involving eye gaze should be cautious with collapsing
data across gender. Gender differences should be analysed before combining data or else,
studies may be at risk of missing an important gender effect.
It is important to examine cognitive biases in anxious people and address them during
treatment. An act of avoidance to prevent disconfirmation of negative thoughts is commonly
found in anxious people. These cognitive biases and safety behaviours are the main causes
for maintaining high levels of anxiety (Foa, Gilboa-Schechtman, Amir, & Freshman, 2000).
The current study has indicated that such safety behaviour is the primary causal factor for
socially anxious males in thinking they were being looked at. Hence, to overcome this fear,
preventing their tendencies to avoid eye gaze can be a useful intervention. In addition, the
17
Being looked at and social anxiety
CoDG can be easily incorporated as a diagnostic tool for detecting thoughts of being
frequently looked at, even in children and intellectually disabled patients, due to its simplicity.
The present study found that male and females differ in how contextual and individual
factors affect perception of direct eye gaze. That is, high socially anxious males overestimate
the number of people looking at them than low socially anxious males. Direct gaze
perception in females, on the other hand, are not affected by their levels of social anxiety.
Furthermore, while females show that perception of direct gaze in fearful faces are more
accurate than neutral or angry faces, only low socially anxious males showed this effect. This
is because males are less experienced with discriminating eye gaze directions which makes
their perception of eye gazes more susceptible to individual factors. Hence, females are less
likely to be affected by individual differences such as traits of social anxiety because being a
woman alone gives them an advantage in eye gaze perception. Males, on the other hand, are
more influenced by their levels of social anxiety, due to their lack of early exposure to eye
gaze perception-enhancing experiences. Therefore, there are gender differences in what
contextual and individual factors are processed or given more weight when perceiving direct
gaze.
18
Being looked at and social anxiety
References
Adams, Reginald B., Jr., & Kleck, Robert E. (2003). Perceived gaze direction and the
processing of facial displays of emotion. Psychological Science, 14, 644-647.
Baron-Cohen, Simon, Jolliffe, Therese, Mortimore, Catherine, & Robertson, Mary. (1997).
Another advanced test of theory of mind: Evidence from very high functioning adults
with autism or Asperger Syndrome. Journal of Child Psychology and Psychiatry, 38,
813-822.
Bayliss, Andrew P., di Pellegrino, Giuseppe, & Tipper, Steven P. (2005). Sex differences in
eye gaze and symbolic cueing of attention. The Quarterly Journal of Experimental
Psychology A: Human Experimental Psychology, 58A, 631-650.
Cline, Marvin G. (1967). The Perception of Where a Person Is Looking. The American
Journal of Psychology, 80, 41-50.
Connellan, Jennifer, Baron-Cohen, Simon, Wheelwright, Sally, Batki, Anna, & Ahluwalia,
Jag. (2000). Sex differences in human neonatal social perception. Infant Behavior &
Development, 23, 113-118.
Deaner, Robert O, Shepherd, Stephen V, & Platt, Michael L. (2007). Familiarity accentuates
gaze cuing in women but not in men. Biology Letters, 3, 65-68.
Ewbank, Michael P., Jennings, Caroline, & Calder, Andrew J. (2009). Why are you angry
with me? Facial expressions of threat influence perception of gaze direction. Journal
of Vision, 9, 16.
Foa, Edna B., Gilboa-Schechtman, Eva, Amir, Nader, & Freshman, Melinda. (2000).
Memory bias in generalized social phobia: Remembering negative emotional
expressions. Journal of Anxiety Disorders, 14, 501-519.
Gamer, Matthias, Hecht, Heiko, Seipp, Nina, & Hiller, Wolfgang. (2011). Who is looking at
me? The cone of gaze widens in social phobia. Cognition and Emotion, 25, 756-764.
19
Being looked at and social anxiety
Georgiou, George A., Bleakley, Cheryl, Hayward, James, Russo, Riccardo, Dutton, Kevin,
Eltiti, Stacy, & Fox, Elaine. (2005). Focusing on fear: Attentional disengagement
from emotional faces. Visual Cognition, 12, 145-158.
Gibson, J. J., & Pick, A. D. (1963). PERCEPTION OF ANOTHER PERSONS LOOKING
BEHAVIOR. American Journal of Psychology, 76, 386-&.
Hock, Howard S., & Schoner, Gregor. (2010). Measuring perceptual hysteresis with the
modified method of limits: Dynamics at the threshold. Seeing and Perceiving, 23,
173-195.
Ingles, Candido J., Piqueras, Jose A., Garcia-Fernandez, Jose M., Garcia-Lopez, Luis J.,
Delgado, Beatriz, & Ruiz-Esteban, Cecilia. (2011). Gender and age differences in
cognitive, psychophysiological and behavioural responses of social anxiety in
adolescence. Psychology in Spain, 15, 80-86.
Langton, S. R. H., Honeyman, H., & Tessler, E. (2004). The influence of head contour and
nose angle on the perception of eye-gaze direction. Perception & Psychophysics, 66,
752-771.
Lutchmaya, Svetlana, & Baron-Cohen, Simon. (2002). Human sex differences in social and
non-social looking preferences, at 12 months of age. Infant Behavior & Development,
25, 319-325.
Lutchmaya, Svetlana, Baron-Cohen, Simon, & Raggatt, Peter. (2002). Foetal testosterone and
eye contact in 12-month-old human infants. Infant Behavior & Development, 25, 327335.
Mansell, Warren, Clark, David M., Ehlers, Anke, & Chen, Yi-Ping. (1999). Social anxiety
and attention away from emotional faces. Cognition and Emotion, 13, 673-690.
Martin, Wade, & Rovira, Marta. (1982). Response biases in eye-gaze perception. Journal of
Psychology: Interdisciplinary and Applied, 110, 203-209.
20
Being looked at and social anxiety
Onnis, Renzo, Dadds, Mark R., & Bryant, Richard A. (2011). Is there a mutual relationship
between opposite attentional biases underlying anxiety? Emotion, 11, 582-594.
Pierce, Tamyra. (2009). Social anxiety and technology: Face-to-face communication versus
technological communication among teens. Computers in Human Behavior, 25, 13671372.
Symons, L. A., Lee, K., Cedrone, C. C., & Nishimura, M. (2004). What are you looking at?
Acuity for triadic eye gaze. Journal of General Psychology, 131, 451-469.
Watson, David, & Friend, Ronald. (1969). Measurement of social-evaluative anxiety. Journal
of Consulting and Clinical Psychology, 33, 448-457.
Wieser, Matthias J., Pauli, Paul, Weyers, Peter, Alpers, Georg W., & Muhlberger, Andreas.
(2009). Fear of negative evaluation and the hypervigilance-avoidance hypothesis: An
eye-tracking study. Journal of Neural Transmission, 116, 717-723.
Winton, Emma C., Clark, David M., & Edelmann, Robert J. (1995). Social anxiety, fear of
negative evaluation and the detection of negative emotion in others. Behaviour
Research and Therapy, 33, 193-196.
21
Being looked at and social anxiety
Figure 4. Mean effects of gender, social anxiety and emotion and Cone of Direct Gaze width.
Bars represent standard errors of the means for each group.
22
Being looked at and social anxiety
Figure 3. Grand mean plot showing the measurement of the Cone of Direct Gaze (CoDG).
The proportion of responses plotted against the number of visual degrees of gaze deviation
away from direct. The triangles represent the mean proportion of responses the participant
said “right”. The squares represent the “direct” responses and the diamonds the “left”
responses. The CoDG is calculated by finding the difference between gaze deviations at the
intersection between the right and direct response line and the intersection between the left
and direct response line (as indicated by the vertical dotted lines). The three faces are visual
presentations of the gaze deviation at -12, 0 and 12 degrees (from left to right).
23
Being looked at and social anxiety
Figure 2. Illustration of one trial during the gaze categorization task. The face stimulus was
presented for 500ms, followed by a 300ms grey screen. During the 300ms period, the
computer did not register responses. After this, the program waited for the participant
response before presenting the next stimulus.
24
Being looked at and social anxiety
Figure 1a. Sample of each male face displaying the three emotions: i) neutral, ii) fear, and iii)
anger. 1b. Sample of one neutral male face stimuli displaying all nine eye gazes.
25
Being looked at and social anxiety
Age
High Social Anxiety
Low Social Anxiety
(n = 29)
(n = 63)
M
SD
M
SD
t(90)
p
18.900
1.291
19.330
2.609
-0.853
.396
-0.650
.517
15.098
< 0.001
% males
FNE
31
21.97
38
3.375
8.95
4.034
FNE = Fear of Negative Evaluation scores
Table 1: Means, standard deviations and significance of difference between social anxiety
groups on age, proportion of males (%), social anxiety scores and Cones of Direct Gaze.
26
Being looked at and social anxiety
Females
Low SA
Threat1
Males
High SA
Low SA
High SA
-0.299
0.000
-0.390
0.980
0.107
0.058
-0.580
0.300
Comfort
-0.681
-0.344
0.228
-0.048
Exp pres
0.176
-0.060
0.249
-0.075
Exp gen
0.030
0.200
-0.188
0.302
-0.400
-0.180
-0.208
0.780
Face gender aware
Freq looked
SA = social anxiety. Ratings of 1); Threat = how threatening the faces were rated; 2) Face
gender aware = how aware participants were of the gazers’ gender; 3) Comfort = how
comfortable participants feel when looked at by others; 4) Experimenter = how aware
participants were of the experimenter’s presence; 5) Awareness of experimenter’s gender
(female); 6) Frequency = how frequently participants feel they are being looked at.
Table 2: Correlations of each dimension of the MOGE (Mechanisms of Gender Effects)
questionnaire with the width of Cone of Direct Gaze for low and high anxious males and
females.