Content uploaded by Jasmine K.W. Lee
Author content
All content in this area was uploaded by Jasmine K.W. Lee on Aug 26, 2022
Content may be subject to copyright.
Consciousness and Cognition 105 (2022) 103400
1053-8100/© 2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
A more featural based processing for the self-face: An
eye-tracking study
Jasmine K.W. Lee
a
,
*
, Steve M.J. Janssen
a
, Alejandro J. Estudillo
a
,
b
,
*
a
University of Nottingham Malaysia, Malaysia
b
Bournemouth University, UK
ARTICLE INFO
Keywords:
Holistic processing
Featural processing
Self-face processing
Eye-tracking
ABSTRACT
Studies have suggested that the holistic advantage in face perception is not always reported for
the own face. With two eye-tracking experiments, we explored the role of holistic and featural
processing in the processing and the recognition of self, personally familiar, and unfamiliar faces.
Observers were asked to freely explore (Exp.1) and recognize (Exp.2) their own, a friend’s, and an
unfamiliar face. In Exp.1, self-face was xated more and longer and there was a preference for the
mouth region when seeing the own face and for the nose region when seeing a friend and un-
familiar faces. In Exp.2, the viewing strategies did not differ across all faces, with eye xations
mostly directed to the nose region. These results suggest that task demands might modulate the
way that the own face is perceived and highlights the importance of considering the role of the
distinct visual experience people have for the own face in the processing and recognition of the
self-face.
1. Introduction
The own face is strongly tied to one’s identity (e.g., Estudillo & Bindemann, 2017a), and the ability to recognize it helps in
maintaining a sense of self (Estudillo & Bindemann, 2016, 2017b; Estudillo et al., 2018; Tsakiris, 2008). Being a signicant stimulus
critical to one’s identity and the most relevant face to each individual (e.g., McNeill, 1998), there has been an increased interest in self-
face processing in recent years. However, there is little understanding of the cognitive processes involved in self-face processing and,
more specically, whether these processes differ from those for other familiar and unfamiliar faces. Using eye-tracking, the present
study addresses this question by exploring the quantitative and qualitative differences in the visual scanning between the own,
familiar, and unfamiliar faces.
1.1. Self-face processing
Although it is widely accepted that faces are processed at a global or holistic level (Estudillo, 2012; Maurer et al., 2002; Rossion,
2013; Wong et al., 2021), it has been suggested that, compared to other faces, the own face is processed in a more featural manner. For
instance, participants are faster at creating a mental image of a facial feature of their own face compared to a mental image of a facial
Abbreviations: AOI, Area of Interest; SF, Self-face; FF, Friend’s face; UF, Unfamiliar Face.
* Corresponding authors at: University of Nottingham Malaysia, School of Psychology, Jalan Broga, Semenyih 43500, Selangor, Malaysia (J.K.W.
Lee). Bournemouth University, Department of Psychology, Poole House Talbot Campus, BH12 5BB, UK (A.J. Estudillo).
E-mail addresses: khpy5jlk@nottingham.edu.my (J.K.W. Lee), aestudillo@bournemouth.ac.uk (A.J. Estudillo).
Contents lists available at ScienceDirect
Consciousness and Cognition
journal homepage: www.elsevier.com/locate/concog
https://doi.org/10.1016/j.concog.2022.103400
Received 27 December 2021; Received in revised form 15 August 2022; Accepted 18 August 2022
Consciousness and Cognition 105 (2022) 103400
2
feature of a familiar face but are slower in creating a mental image of the whole own face compared to a mental image of the whole
familiar face (Greenberg & Goshen-Gottstein, 2009). The authors hence concluded that the own face is processed in a more featural
based manner. In a different study, Keyes and Brady (2010) showed that participants were faster and more accurate at recognizing
their own face than friends’ and strangers’ faces, and interestingly, this processing advantage was observed for both upright and
inverted orientations. As inverting a face disrupts the holistic processing of faces (e.g., Rossion, 2009; Yin, 1969), this nding suggests
that the processing of own face relies on a more featural processing approach.
It is possible that the distinct visual experiences an individual has with their own and other faces might contribute to these pro-
cessing differences. For instance, most of the visual experience gathered with the own face is acquired through self-inspection in
mirrors (Br´
edart, 2003; Gregory, 2001) and thus the distribution of views for one’s own face is generally restricted to mirror-reversed
frontal views (Br´
edart, 2003). The effect of such exposure can be observed through an individual’s preference for mirror-reversed
images of the own face compared to non-reversed images. Importantly, this preference was not found for familiar faces (e.g., Brady
et al., 2005; Laeng & Rouw, 2001; Troje & Kersten, 1999). Additionally, when people perceive their own and other people’s faces, they
might have different processing goals. Specically, whereas individuals tend to perceive the face of other people for identication
purposes, they tend to perceive the own face for the detailed inspection of facial features (e.g., grooming purposes; see Estudillo &
Bindemann, 2017a, 2017b). Hence, the different demands associated with the perception of the own and other faces might partially
explain the processing differences between the own and other faces.
1.2. Eye-tracking measures in self-face recognition
Eye movements are thought to provide a sensitive measure of visual processing (Henderson, 2003) and an index of the cognitive
processes involved in the task at hand (Just & Carpenter, 1980). Fixations are generally referred to as “pauses over informative re-
gions” (Salvucci & Goldberg, 2000), such that these “pauses” are indicative of extracting or encoding information (Poole & Ball, 2006).
Amongst the face recognition literature, a specic facial region receiving a higher number of xations is generally conceived as an
indicator of its saliency or its informativeness compared to other xated regions (Holmqvist et al., 2011). Additionally, the duration of
xation is generally indicative of the amount of time used to process a xated region (Salvucci & Goldberg, 2000) and a longer xation
duration suggests a greater cognitive exertion when extracting information (i.e., information complexity; Rayner, 1998).
Eye movements are also postulated to indicate the processing style (e.g., Hills, 2018; Rossion, 2008). For example, holistic pro-
cessing is generally associated with longer central eye xations to the nose region and between the eyes (e.g., Blais et al., 2008), as this
strategy allows for the perception of the facial region as a whole (Van Belle, Ramon, et al., 2010). Featural processing, on the other
hand, is implied through a higher number of xations to individual facial features (e.g., Rossion, 2008). Furthermore, several notable
face-scanning strategies have been revealed through eye-tracking studies on face perception. For instance, when viewing faces, most
eye xations lands in between the eyes (Hsiao & Cottrell, 2008; Tyler & Chen, 2006) followed by fewer xations on the mouth and
other facial features (Bindemann et al., 2009; Stacey et al., 2005). Conversely, studies have shown that prosopagnosic patients who
rely on featural processing directed more eye xations to the mouth instead of to the eyes (e.g., Bukach et al., 2006; Orban de Xivry
et al., 2008; Ramon et al., 2010).
In recent years, a growing literature has explored the differences in gaze behaviour when looking at the own face compared to other
faces. For instance, using eye-movement measures, Chakraborty and Chakrabarti (2018) asked participants to identify their own face
from a series of self-other face morph images. Participants made longer xations to the lower part of the self-face compared to other
faces, whereas no differences were reported for the upper part of the face. Although these results may indicate a peculiar visual
scanning strategy for the own face, this study compared the gaze behaviour between the self-face and an unfamiliar face, so the re-
ported effects could be confounded with simple familiarity effects (see Estudillo, 2012). Furthermore, the face stimuli were presented
at the centre of the screen which could lead to the initial xation to coincide with the centre of the face (Bindemann et al., 2009).
In another study, Hills (2018) recorded the eye movements of children aged between 6 and 11 years when asked to perform a
familiarity judgement task with the own, a familiar, and an unfamiliar face. The ndings showed that the own face received signif-
icantly more xations which were directed to the diagnostic facial features (i.e., eyes, mouth, and nose), altogether suggesting an
overall enhanced use of featural processing for the self-face compared to other faces. Contrary to holistic processing, featural pro-
cessing is generally associated with a higher number of short xations to each feature (Bombari et al., 2009) and such a xation pattern
has also been observed when participants viewed inverted images (e.g., Hills et al., 2013). Interestingly, the self-face and the familiar
face also received longer central xations than the unfamiliar face, suggesting an enhanced use of holistic processing for familiar faces
(see Blais et al., 2008; Van Belle, Ramon, et al., 2010). These ndings indicate that the processing of the own face employs both holistic
and featural processing and this dual strategy ensures that the self-face is processed efciently (Hills, 2018). However, as this study was
only conducted with children, it is unknown whether the observed effects reect adults-like face processing strategies or, in contrast,
are a consequence of immature face processing strategies (Hills & Lewis, 2018). In fact, some research has found that xation duration
to natural scenes decreases with age and salient features have a stronger inuence on children compared to adults (Helo et al., 2014).
Other studies have not found different visual-scanning strategies for the self and other faces. For example, Kita et al. (2010) asked
participants to watch a morphing movie (e.g., self-face gradually changing into a familiar face) and to respond when they thought that
the initial face image had morphed into a target face image. Although self-face evoked increased oxyhaemoglobin changes in frontal
areas of the brain, they found no difference in the xation count and xation duration across face image conditions. The authors
suggested that irrespective of face identity, individuals employ similar strategies when sampling facial information. However, this
information is later processed differently wherein the oxyhaemoglobin activity around the right inferior frontal gyrus changes across
face identity, with increased activity in the self-face condition compared to the familiar face condition (Kita et al., 2010).
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
3
Two reasons might help explain the lack of consistency across the ndings of the aforementioned studies. First, several studies
above (e.g., Chakraborty & Chakrabarti, 2018; Hills, 2018) did not mention if the self-images were presented in a mirror-reversed or
normal orientation. Experience with the own face is mostly gathered through mirrors (i.e., in a mirror-reversed orientation). Thus,
when someone is presented with a photograph of their own face, the face image is always ipped (i.e., normally oriented) compared to
the usual mirror-reversed version of themselves that they see in the mirror. Consequently, photographs of one’s own face misplace
facial asymmetries (Frautschi et al., 2021) which affects self-perception (see Lu & Bartlett, 2014). In fact, individuals prefer mirror-
reversed images compared to normally oriented images of the own face and this effect is not found in familiar faces (Brady et al.,
2005; Br´
edart, 2003). Therefore, to ensure ecological validity, it is important to record eye movements when viewing a mirror-reversed
image of the own face as this view closely corresponds to the visual experience people have with the own face (i.e., when looking in the
mirror).
Second, the lack of consistency across studies could also be explained by the different task demands employed across these studies
(Kita et al., 2010). In line with this notion, Stacey et al. (2005) reported that the effects of familiarity on face processing were
inuenced by the type of task demands imposed. More specically, for tasks requiring the involvement of higher cognitive load (i.e.,
memory), such as recognition or familiarity judgement tasks, an individual’s attention window narrows, allowing only limited in-
formation to be processed, whereas, under low demand tasks, attention can be widely dispersed throughout a scene.
The sensitivity of eye-movement behaviour to task constraints has been well-established in previous face-recognition/perception
studies. For instance, Cook (1978) observed different visual sampling behaviours depending on whether participants were asked to
memorize a series of faces or to recognize them. Additionally, Walker-Smith et al. (1977) observed differences in eye-movement
behaviour of participants when asked to match either simultaneously or successively presented face images. Taken together,
different patterns of gaze behaviour would be expected under different task demands, as eye movements are thought to be goal-
directed and vary according to task constraints (Henderson, 2003). Consequently, we might expect that the processing of the own
face is modulated by the type of task employed. Indeed, Bortolon and Raffard (2018) took the view that self-face processing may be
inuenced by the type of task used for testing a self-face advantage (SFA). In a meta-analysis, the authors reported SFA effects for
memory and perceptual based identication tasks (i.e., determine face identity or head orientation), whereas no SFA effects were
reported for tasks which involve attentional processes (i.e., visual search or face detection tasks). Based on these discussed studies, it
seems reasonable to elucidate that task demands may modulate one’s self-face processing.
1.3. The current study
With two different experiments, the present study aims to explore the role of holistic and featural processing in the processing and
the recognition of own, familiar, and unfamiliar faces using the eye-tracking technique. Rather than restricting eye movements to a
specic task demand, Experiment 1 used instead a free-viewing task to observe spontaneous eye-movement behaviour while exploring
the own face and other faces. A free-viewing task was used as passive viewing of faces might imply a more direct index for face
processing compared to task-oriented viewing of faces (Scott et al., 2005). In particular, recognizing a face in our daily lives is more
often accompanied by a ‘passive’ recognition, wherein a facial representation is activated briey after unintended perception of a face
rather than by an ‘active’ recognition, wherein sustained attention is required to focus on a representation of a face (Sugiura et al.,
2000) which typically occurs in task-oriented viewing of faces.
Therefore, to explore this aim, participants were presented with one face (self, friend, or unfamiliar) at a time and were asked to
freely explore the images. Finally, to complement the free-viewing data, visual scanning behaviour for faces under the connement of
tasks was considered in Experiment 2, that is, participants were asked to make overt responses by judging the identity of faces with
differing levels of familiarity.
2. Experiment 1
With eye-tracking measures, Experiment 1 was conducted to explore the role of holistic and featural processing of the own face and
other faces in a free-viewing paradigm. This paradigm also allowed us to examine the relevance of each facial feature for each of the
three different identities by quantifying the facial features xated through the number of xations and average xation duration
without the restriction of task demands. Based on the literature on eye-tracking measurements and evidence suggesting that featural
processing supports self-face recognition (e.g., Greenberg & Goshen-Gottstein, 2009), we expected to nd that compared to other
faces, the self-face would receive a higher number of xations and a longer xation duration. Additionally, it is possible that most of
these xations are directed to mouth areas, as some studies with prosopagnosic patients have shown that featural processing is
associated with more xations to the mouth compared to eyes (e.g., Bukach et al., 2006; Ramon et al., 2010). For both friend and
unfamiliar faces, we expected to observe a xation pattern indicative of holistic processing and longer xations to the nose area (see
Van Belle, de Graef, et al., 2010).
We also manipulated the vertical orientation of face images. Previous studies had consistently shown that inverted faces receive
more xations than upright faces due to a more featural based processing for inverted faces (e.g., Hills et al. 2013; Van Belle, Ramon,
et al., 2010). Hence, we also expect a higher number of xations for the inverted friend face an unfamiliar face compared to when these
faces would be being presented in an upright manner. However, if the own face is processed in a more featural manner than other faces,
we expected that the self-face would receive a similar number of xations across its upright and inverted versions.
Finally, as the experience that we have with our face is mostly through mirrors, we also manipulated the horizontal orientation (i.e.,
normal oriented or mirror-reversed) of the face images. Studies have shown that compared to an unfamiliar face, participants preferred
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
4
mirror-reversed images of their face compared to normally oriented images (e.g., Brady et al., 2005; Laeng & Rouw, 2001).
Overall, for Experiment 1, we expected the own face to be sampled in a more featural manner with an overall higher number of
xations and longer xation durations, and to show no inversion effect. Conversely, we expected the friend and unfamiliar faces to be
sampled more holistically with an overall lesser number of xations and shorter xation durations compared to the own face, and to
show an inversion effect.
2.1. Method
2.1.1. Participants
Thirty Malaysian
1
participants (4 males; M
age
=20.53, SD =2.03) were recruited from the University of Nottingham Malaysia. All
participants were recruited in pairs matched by age, gender, and ethnicity, so each of them served as a friend to the other. The pairs had
to have known each other for at least 6 months and have met at least once a week. All participants had a normal or corrected-to-normal
vision. All participation was completely voluntary, and all gave informed consent after the experimental procedure had been fully
understood. Participants either received course credit or RM5 for their contribution. Ethics approval for the study was obtained from
the Science and Engineering Research Ethics Committee of the University of Nottingham Malaysia.
2.1.2. Apparatus
A desktop mounted EyeLink 1000+eye-tracking system with a sampling rate of 1000 Hz was used. The eye tracker was positioned
under the display screen at a distance of 75 cm from the participant. Participants were asked to position their heads on a chin rest to
minimize head movements.
2.1.3. Stimuli
Photograph stimuli (self and friend faces) were individually tailored for each participant. Each participant was photographed under
similar conditions (i.e., constant lighting conditions and a uniform grey background). Different images were used for each identity to
reduce image-specic learning. Participants were photographed in a frontal position while assuming neutral and happy expressions
and articulating three different speech sounds (i.e., A, O, and E; see Fig. 1a). All ve different images were used as “self-face” for the
participants themselves and as “friend’s face” for their friend, respectively. Six separate individuals (three males and three females)
matched in age and race were photographed under similar conditions to be used as “unfamiliar faces”. These unfamiliar faces were
counterbalanced across each participant. Each participant’s stimulus set consisted of three sets of identities: one self-face (5 different
images ×2 inversion ×2 orientation), one friend’s face (5 different images ×2 inversion ×2 orientation), and one unfamiliar face (5
different images ×2 inversion ×2 orientation).
Using Adobe Photoshop CS6, all photographs were resized to 401x 562 pixels, corresponding to an approximate visual angle of
8.09
◦
horizontally and 11.32
◦
vertically at a viewing distance of 75 cm. All photographs were rotated to ensure eyes were collinear. All
face stimuli were being cropped based on their individual contours to ensure that face shape information was available to participants.
Each face image was saved in an upright, normal; upright, mirror-reversed; inverted, normal; and inverted and mirror-reversed version
(see Fig. 1b). Images were ipped vertically downwards to create an inverted image and ipped horizontally across to create a mirror-
reversed version. All images were collected and processed at least one week before the experimental session.
Note. (a) The ve different images of each identity: from top left: “neutral” and “happy”; from bottom left: “A”, “O”, and “E”
expression; (b) an example of a face image presented in two different inversion conditions and two different orientation conditions.
From top left: upright and normal; mirror-reversed and normal; from bottom left: inverted and normal, and inverted and mirror-
reversed.
2.1.4. Procedure
After giving their informed consent, participants were individually tested in a dimly lit room. At the beginning of the experiment,
the standard nine-dots EyeLink calibration procedure was performed. This calibration was later validated with a second sequence of
nine xations. Calibration was repeated if the latter showed low measurement accuracy.
Each trial began by asking participants to xate on a single centred dot with an automatic drift correction. The experimenter
pressed a button to initiate a trial when participants were seen to be xating on the dot. Participants would rst see an average face
mask being presented to a similar location as the target face. The average face mask was only removed when participants xated on its
location and the target face would then be made visible to the participants. Target face stimulus was randomly presented to either the
top or bottom location of the screen. This method ensures that the critical face regions did not coincide with the centrally presented
xation cross at the beginning of each trial. Each target face stimulus was displayed for 3000 ms. Participants were asked to freely
explore the presented face images and feedback was provided when participants’ gazes would leave the screen.
Each participant completed a total of 120 trials (60 trials per block) with the four combinations of orientation and inversion of the
ve expressions of the three face identities being displayed twice. The presentation of trials was also counterbalanced across face
identity, inversion, and orientation, respectively. The experiment lasted for approximately 15 min, and participants were given a short
break between the two blocks, followed by a recalibration phase.
1
Based on the effect size from previous research (i.e.,
η
p
2
=0.11, Hills, 2018) and an alpha of 0.05, a power analysis performed in G*Power 3.1
(Faul et al., 2007) gives a required sample size of 28 participants to achieve 80% power.
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
5
2.1.5. Data analyses
Eye-movement data were processed from the target face onset to aggregate the total number of xations, their locations, and their
durations. Short continuous xations (i.e., shorter than 80 ms) were combined with the following xations if they would fall within
half a degree of the visual angle; otherwise, the xation was excluded. Such short xations were excluded due to the possibility of
incorrect saccade planning and less likely to reect meaningful processing of information (Pollastek et al., 1984). For cases where an
eye blink took place, its duration was integrated with the immediately preceding xation, as information processing is unlikely to
pause during a blink (Pollastek et al., 1984).
Predened AOI was generated individually for each face image, such that they outlined a region for the eyes, nose, mouth, and rest
of the face (see Fig. 2). The location of all AOIs were identical across all presented face stimuli. As each face consists of a different face
shape and different speech congurations, the AOI for the facial features differed across each individual (see Fig. 2). Due to differences
in size among facial features, the dimension of each AOI differs within and between faces. Hence, any xation data could simply reect
the relative size of the AOIs rather than the interest region held by an observer (see Bindemann et al., 2009). To address this issue, area-
normalization for the xation data was performed by dividing the proportion of xations to an AOI by the size of the AOI (i.e., the total
area of the screen occupied by a particular AOI). This procedure normalizes the size of each AOI so that a score larger than one in-
dicates that the AOI is specically targeted (see Fletcher-Watson et al., 2008). This normalization adjustment was only conducted for
analysis of the raw data involving “facial features.”.
Fig. 1. Example of Face Stimuli used in Experiment 1.
Fig. 2. Predened Area of Interests (AOIs) for Face Stimulus. Note. An example of face stimulus of Experiment 2 with its predened AOIs: a) eyes; b)
nose; c) mouth; and d) rest of the face. Excluding the AOI for the rest of the face (i.e., the outline of the face), the size and location of all AOIs were
identical across all presented face stimuli.
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
6
All raw eye-tracking data (number of xations and xation duration) over the predened AOIs within each face were collected for
each trial. For each face presentation, the number and duration of xations for each AOI were aggregated and summed and later
averaged across faces to provide indices of an average total number of xations and xation duration for each AOI.
A general analysis of the average total number of xations and average total xation duration was conducted using a 3 (identity:
self (SF), friend (FF), and unfamiliar (UF)) ×(inversion: inverted vs upright) ×2 (orientation: mirror-reversed vs normal) repeated-
measures ANOVA. Next, to assess the extent to which specic features were looked at when viewing different identities, the
normalized scores for the number of xations and xation duration were analysed with a 3 (identity: SF, FF, and UF) ×2 (inversion:
inverted vs upright) ×2 (orientation: mirror-reversed vs normal) ×4 (features: eyes, nose, mouth, and others). All post-hoc analyses
were Holm-Bonferroni corrected. Additionally, for all analyses with the variable ‘features’, the Mauchly’s test of sphericity showed
signicance, thus Huynh-Feldt corrections were applied to correct the degrees of freedom.
2.2. Results
2.2.1. Average total number of xations
Fig. 3 presents the average total number of xations across conditions. A 3 ×2 ×2 repeated-measures ANOVA, with the variables:
identity, inversion, and orientation, was conducted on the average total number of xations received by each face. The left side of
Table 1 reports the detailed ANOVA results. The analysis revealed a signicant main effect for identity, with Holm-Bonferroni cor-
rected pairwise comparisons indicating the SF (M =8.90, SD =1.40) being xated upon signicantly more than FF (M =8.45, SD =
1.55; p =.006, d =0.62) and UF (M =8.41, SD =1.62; p =.008, d =0.60) but no signicant difference between FF and UF (p =1.00, d
=0.05). The main effect of inversion was also signicant, with the upright faces (M =8.72, SD =1.57) receiving more xations than
inverted faces (M =8.45, SD =1.54). There was no signicant main effect of orientation and no signicant interactions.
For the following analysis, the main effects of identity, inversion, and orientation and the interactions between them will not be
described, as they have already been described in the previous analysis. Instead, only signicant main effects of the factor AOI or
interactions between the AOI and previously described factors will be reported.
Fig. 4 presents the normalized average total number of xations for each feature AOI across different factors. A 3 ×2 ×2 ×4
repeated-measures ANOVA, with the variables: identity, inversion, orientation, and features, was conducted on the normalized scores
for the number of xations. The detailed ANOVA results are summarized on the left side of Table 2. A main effect of features was found,
which was qualied by a signicant interaction with the identity factor. Simple main effects analyses followed by Holm-Bonferroni
corrected pairwise comparisons revealed that the nose was xated more often in FF (M =1.90, SD =0.88) and UF (M =1.93, SD
=0.90) compared to SF (M =1.73, SD =0.81). In contrast, the mouth was xated more often in SF (M =1.61, SD =0.51) compared to
FF (M =1.48, SD =0.52) and UF (M =1.45, SD =0.52). The rest of the face was xated signicantly more on SF (M =0.25, SD =0.11)
than UF (M =0.22, SD =0.12). Finally, there were no signicant differences in the number of xations for the eyes across all three
identities (see Table 3).
2.2.2. Average total xation duration
A 3 ×2 ×2 repeated-measures ANOVA, with the variables: identity, inversion, and orientation, was conducted on the average total
xation duration. This analysis revealed no signicant main effects and interactions. The summary of the ANOVA results is presented
on the right side of Table 1.
Fig. 5 presents the area-normalized average total xation duration to each feature AOI across all factors. A 3 ×2 ×2 ×4 repeated-
measures ANOVA, with the variables: identity, inversion, orientation, and features, was conducted on the normalized scores for the
average total xation duration. The summary of the ANOVA results is presented on the right side of Table 2. The analysis revealed a
Fig. 3. The Average Total Number of Fixations for Different Identities (Exp. 1). Note. The average total number of xations received by faces for
each identity (SF, FF, and UF) across different inversion and orientation conditions. Error bars represent the standard error of the mean.
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
7
signicant main effect of features, which was qualied by a signicant interaction with the factors: identity, inversion, and orientation
respectively, and was further qualied by a four-way interaction between these factors. A simple main effect analyses followed by
Holm-Bonferroni corrected comparisons revealed that the eyes were xated longer on both FF (M =1.98, SD =0.77) and UF (M =
2.06, SD =0.86) compared to SF (M =1.85, SD =0.68) and were also xated longer across upright faces (M =2.16, SD =0.93) than
inverted faces (M =1.76, SD =0.60). Next, the nose was xated longer for a mirror reversed UF in inverted condition (M =2.90, SD =
1.27) compared to in an upright condition (M =0.01, SD =0.001). The mouth was xated longer on SF (M =1.70, SD =0.70)
compared to FF (M =1.60, SD =0.62) and UF (M =1.47, SD =0.67) and was also xated longer on inverted faces (M =2.03, SD =
0.78) compared to upright faces (M =1.14, SD =0.55). Finally, the rest of the face was xated longer for both an upright SF (M =0.31,
SD =0.17) and an upright UF (M =0.46, SD =0.25) compared to in an inverted SF (M =0.23, SD =0.12) and an inverted UF (M =
0.22, SD =0.10), whereas for normal oriented faces, the rest of the face was xated for a longer duration on SF (M =0.27, SD =0.15)
Table 1
Statistical Analysis of Average Total Number of Fixations and Average Total Fixation Duration corresponding to Identity, Inversion, and Orientation
(Exp. 1).
Variables Average Total Number of Fixations Average Total Fixation Duration
df F
η
p
2
df F
η
p
2
Identity 2, 58 7.73
***
0.21 2, 58 2.99 0.09
Inversion 1, 29 16.81
***
0.37 1, 29 3.80 0.12
Orientation 1, 29 0.004 0.00 1, 29 0.27 0.01
Identity ×Inversion 2, 58 0.004 0.001 2, 58 0.73 0.03
Identity ×Orientation 2, 58 1.79 0.06 2, 58 0.38 0.01
Inversion ×Orientation 1, 29 0.04 0.001 1, 29 0.01 0.00
Identity ×Inversion ×Orientation 2, 58 0.30 0.01 2, 58 0.58 0.02
***
p <.001.
Fig. 4. Area-normalized Average Total Number of Fixations across Each Facial Feature (Exp. 1). Note. Area-normalized average total number of
xations received by (a) eyes, (b) nose, (c) mouth, and (d) other facial regions for each identity across different inversion conditions. The orientation
factor is collapsed across each AOI. Error bars represent the standard error of the mean.
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
8
compared to FF (M =0.22, SD =0.11; see Table 4).
2.3. Discussion
Findings from Experiment 1 can be summarized as follows: when asked to explore the faces, (1) the own face received a greater
number of xations compared to a friend’s face and an unfamiliar face; however, both the own face and friend’s face were xated on
for a longer duration than the unfamiliar face. Next, (2) we did not nd evidence supporting an inversion effect across all faces, such
that upright faces were xated more and longer compared to inverted faces. Lastly, (3) the mouth feature receives more xations and is
xated longer in one’s own face than in other faces.
The self-face was sampled more compared to a friend and unfamiliar face, and it was also xated for a longer duration compared to
an unfamiliar face. Whereas a higher number of xations may indicate that individual features of the own face are sampled more (Hills,
2018), longer xation durations may suggest a difculty in disengaging attention from the own face (Devue et al., 2009). Notably, we
observed no differences between the xation duration for the own face and the friend’s face. Next, contradicting previous studies that
had shown an increased number of xations for inverted faces due to the disruption of holistic processing for faces (i.e., an inversion
effect; Rossion, 2008, 2009; Yin, 1969), Experiment 1 did not observe such gaze pattern for inverted faces across all three identities,
such that upright faces were xated with a higher number of xations and a longer xation duration compared to inverted faces. To
compensate for the disruption in the extraction of holistic facial information in an inverted face, a more featural “scan path” is
employed to extract the necessary structural facial information slowly and partially for recognition purposes (see Barton et al., 2006).
Following this line of reasoning, we postulate that, as free-viewing tasks do not require any extraction of facial information, such a
viewing pattern was not observed for the inverted faces.
Additionally, compared to the friend and unfamiliar face, the nose was xated with a lesser number of xations whereas the mouth
was xated with a higher number of xations and a longer xation duration on the own face. Whereas long central xations to the nose
denote holistic processing of faces (e.g., Hills, 2018; Van Belle, de Graef, et al., 2010), a higher number of xations to the mouth might
indicate a more featural based processing of faces (e.g., Bukach et al., 2006; Ramon et al., 2010). These ndings may suggest that when
exploring faces, observers employ a more featural based processing with their own faces than with a friend’s or an unfamiliar face.
Table 2
Statistical Analysis of Average Total Number of Fixations and Average Fixation Duration corresponding to Features, Identity, Inversion, and
Orientation (Exp. 1).
Variables Average Total Number of Fixations Average Total Fixation Duration
df F
η
p
2
df F
η
p
2
Identity – – – 2, 58 9.80
***
0.25
Inversion – – – 1, 29 43.55
***
0.60
Orientation – – – 1, 29 43.121
***
0.60
Features 1.42, 41.20 90.75
***
0.76 1.57, 45.56 67.44
***
0.74
Features ×Identity 4.47, 129.69 3.81
**
0.12 3.57, 103.62 7.99
***
0.22
Features ×Inversion 1.35, 39.02 26.63
***
0.48 1.78, 51.53 11.87
***
0.29
Features ×Orientation 1.95, 56.44 0.28 0.01 1.43, 41.32 22.16
***
0.43
Features ×Identity ×Inversion 4.41, 127.73 1.49 0.05 4.30, 124.59 23.26
***
0.45
Features ×Identity ×Orientation 4.42, 122.33 0.37 0.01 3.66, 106.19 19.21
***
0.40
Features ×Inversion ×Orientation 1.63, 47.17 2.32 0.07 2.04, 59.06 24.00
***
0.45
Features ×Identity ×Inversion ×Orientation 4.66, 135.22 1.70 0.06 4.41, 127.90 23.16
***
0.44
Note. Huynh-Feldt corrections were applied to the df. for all analyses with the ‘features’ variable.
***
p <.001.
**
p <.01.
Table 3
Simple Main Effect Analysis of Identity Considered at Each Level of Features on the Average Total Number of Fixations (Exp. 1).
Simple Main Effects of Identity at each level of:
Features df F
η
p
2
Pairwise Comparisons
Eyes 2, 58 1.39 0.05 –
Nose 2, 58 4.07* 0.12 SF <FF (p =.04, d = − 0.49)
SF <UF (p =.05, d = − 0.46)
FF =UF (p =1.00, d = − 0.06)
Mouth 2, 58 4.68* 0.14 SF >FF (p =.03, d =0.43)
SF >UF (p =.04, d =0.51)
FF =UF (p =1.00, d =0.09)
Others 2, 58 4.12* 0.12 SF =FF (p =.52, d =0.26)
SF >UF (p =.04, d =0.47)
FF =UF (p =.30, d =0.31)
*
p <.05.
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
9
Conversely, Experiment 1 showed a similar viewing pattern for both friend and unfamiliar faces, such that there were no signicant
differences in the overall number of xations when exploring both faces. Furthermore, compared to the own face, the nose feature was
sampled with a higher number of xations whereas the mouth feature was sampled with a lesser number of xations across both the
Fig. 5. Area-normalized Average Total Fixation Duration for Each Facial Feature (Exp. 1). Note. Area-normalized average total xation duration
received by (a) eyes, (b) nose, (c) mouth, and (d) other facial regions for each identity across different inversion conditions. The orientation factor is
collapsed across each AOI. Error bars represent the standard error of the mean.
Table 4
Signicant Simple Main Effects of Identity, Inversion, and Orientation Considered at Each Level of Features on the Average Total Fixation Duration
(Exp. 1).
Simple main effects of Identity, Inversion, and Orientation at each level of
Features Sig. factors
/interactions
df F
η
p
2
Pairwise Comparisons
Eyes Identity 2, 58 6.20
**
0.18 SF <FF (p =.04, d = − 0.46)
SF <UF (p =.02, d = − 0.55)
FF =UF (p =.18, d = − 0.25)
Inversion 1, 29 5.27* 0.15 upright faces >inverted faces
Nose Identity ×Inversion ×Orientation 2, 58 39.07
***
0.57 inverted, mirror-reversed, UF >upright, mirror-reversed, UF
Mouth Identity 1.97, 57.07 7.02
**
0.20 SF >FF (p =.02, d =0.32)
SF >UF (p =.01, d =0.60)
FF =UF (p =.08, d =0.42)
Inversion 1, 29 39.69
**
0.58 inverted faces >upright faces (p <.001)
Others Identity ×Inversion ×Orientation 2, 58 18.82
**
0.39 upright SF >inverted SF (p <.05, d =0.43)
upright UF >inverted UF (p <.001, d =1.12)
normal oriented SF >normal oriented FF (p =.01, d =0.58)
***
p <.001.
**
p <.01.
*
p <.05.
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
10
friend and unfamiliar face. Such ndings may suggest that when asked to explore faces, the own face is processed distinctly compared
to other familiar or unfamiliar faces.
3. Experiment 2
Experiment 1 used a free-viewing task without the restrictions of task demands, allowing the capture of spontaneous eye move-
ments when exploring faces. However, as participants were only required to passively explore the face images without any specic
task, it is possible that different face types facilitate or even elicit different types of tasks. For instance, in an event-related potential
(ERP) study, Sui et al. (2006) presented evidence that even when participants were not asked to perform an explicit face-recognition
task, the own face was automatically recognized compared to a familiar face, suggesting that the self-face recognition was not
modulated by task demands. Consequently, one may ask whether the visual scanning behaviour for faces in a free-viewing task differs
from when being restricted by task demands. More specically, we asked in Experiment 2 if participants would show a different
viewing pattern for the faces when asked to make explicit responses regarding the identity of a certain face.
Experiment 2 was conducted to complement the free-viewing ndings, exploring the visual scanning behaviour for faces under the
restriction of tasks. Participants were asked to identify faces of differing levels of familiarity, and the eye movements before partic-
ipants reach their decision were recorded and analysed. Having an identity task demand, for instance, would then require participants
to extract facial information to facilitate their judgements in identity (Scott et al., 2005). Based on previous evidence (e.g., Kita et al.,
2010; Stacey et al., 2005), we hypothesized that when task demands were introduced, observers would adopt a similar scanning
strategy across the self-face, the friend’s face, and the unfamiliar face. More specically, to facilitate the extraction of facial infor-
mation, the gaze would be directed more and longer towards the centre of a face which allows for the simultaneous extraction of facial
information.
3.1. Method
3.1.1. Participants
Thirty Malaysian participants (4 males; M
age
=23.57, SD =1.90) were recruited from the University of Nottingham Malaysia. As
was the case in Experiment 1, all participants were recruited in pairs matched by age, gender, and race, so each of them served as a
friend for the other participant. They had to have known each other for at least 6 months and have met at least once a week. All
participants were right-handed and had a normal or corrected-to-normal vision. All participation was completely voluntary, and all
gave informed consent after the experimental procedure had been fully understood. Participants received either course credit or RM5
for their contribution. Ethics approval for this present study was obtained from the Science and Engineering Research Ethics Com-
mittee of the University of Nottingham Malaysia.
3.1.2. Apparatus and stimuli
The apparatus set-up and preparation of stimuli are similar to Experiment 1, such that the face stimuli were cropped based on their
individual contours. See Fig. 2 for examples of experimental stimuli.
3.1.3. Procedure
The procedure was similar to the preceding experiments, except for the following changes. Participants were required to indicate
whether the face presented was their own face, their friend’s face, or the stranger’s face by pressing a button on the keyboard (“J”, “K”,
Fig. 6. The Median Reaction Time to Identify Faces. Note. The median reaction time (ms) per participant for each identity (SF, FF, and UF) across
different conditions. Error bars represent the standard error of the mean.
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
11
and “L”, respectively). Face stimuli remained on the screen until the participants made a keyboard response. Twelve practice trials
were presented before the experiment to familiarize participants with the task.
Each participant completed two blocks (120 trials per block) of testing. In each block, fteen images (3 target identity ×5 different
expressions) were presented twice in each of the four different inversion and orientation conditions. The experiment lasted for
approximately 20 min, and participants were given a short break between the two blocks, followed by a recalibration phase.
3.2. Results
3.2.1. Behavioural performance
Recognition accuracy was high for all three faces in the face identity judgement task (self-face: 99.21 %; friend’s face: 98.75 %;
unfamiliar face: 98.88 %). Fig. 6 illustrates the median reaction time for each identity. The median reaction time (RT) was used instead
of the mean RT to remove the inuence of extreme values. Median RTs were subjected to a repeated-measures ANOVA with the factors:
identity, inversion, and orientation. The analysis revealed a signicant main effect of inversion, F(1, 29) =48.04, p <.001,
η
p
2
=0.62,
with a shorter median reaction time to upright faces compared to inverted faces. The analysis revealed no other signicant main or
interaction effects.
3.2.2. Average total number of xations
Fig. 7 shows the average total number of xations for each face identity across inversion and orientation conditions. A 3 ×2 ×2
repeated-measures ANOVA, with the variables: identity, inversion, and orientation, was conducted on the total number of xations.
The left side of Table 5 reports the summary of the ANOVA results. We found a signicant main effect of inversion, with the inverted
faces (M =2.94, SD =0.82) receiving signicantly more xations than upright faces (M =2.67, SD =0.58). The analysis revealed no
other signicant main or interaction effects.
Fig. 8 presents the normalized total number of xations for each feature across different factors. A 3 ×2 ×2 ×4 repeated-measures
ANOVA with the variables: identity, inversion, orientation, and features, was conducted on the normalized scores for the average total
number of xations. The left side of Table 6 shows the detailed ANOVA results. A main effect of features was found, which was
quantied by a signicant interaction with identity. Simple main effects analyses followed by Holm-Bonferroni corrected pairwise
comparisons revealed that the mouth was xated upon lesser for UF (M =0.80, SD =0.60) than SF (M =1.11, SD =0.69) or FF (M =
1.01, SD =0.68), but there were no signicant differences between SF and FF. Conversely, the rest of the face was xated more for UF
(M =0.41, SD =0.18) than SF (M =0.33, SD =0.16) and FF (M =0.35, SD =0.19), but there were no signicant differences between
SF and FF. Finally, there were no signicant differences in the number of xations between the eyes and nose across all identities (see
Table 7).
This analysis further revealed a signicant interaction between identity and orientation (see Table 6). Post-hoc analysis revealed
that FF was xated upon more in the normal oriented version (M =1.59, SD =0.79) than in a mirror-reversed orientation (M =1.53,
SD =0.73; p =.02, d =0.45), whereas the number of fixations on both SF (normal: M =1.54, SD =0.77; mirrored: M =1.53, SD =0.73;
p =.91, d =0.02) and UF (normal: M =1.50, SD =0.78; mirrored: M =1.55, SD =0.77; p =.07, d =0.34) did not differ signicantly
across the orientation conditions.
3.2.3. Average total xation duration
Fig. 9 shows the average total xation duration for each identity across inversion and orientation conditions. A 3 ×2 ×2 repeated-
measures ANOVA, with the variables: identity, inversion, and orientation, was conducted on the average xation duration. The right
Fig. 7. The Average Total Number of Fixations for Different Identities (Exp. 2). Note. The average total number of xations per participant received
by faces for each identity (SF, FF, and UF) across different inversions and orientation. Error bars represent the standard error of the mean.
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
12
side of Table 5 reports the summary of the ANOVA results. A main effect of identity was reported with Holm-Bonferroni corrected
pairwise comparisons revealing that participants xated shorter on SF (M =238.44, SD =36.50) than FF (M =247.10, SD =39.01; p
=.003, d = − 0.67) and UF (M =253.59, SD =44.44; p <.001, d = − 0.78). Contrariwise, the xation durations for FF and UF (p =.10, d
= − 0.41) did not differ signicantly. The analysis further revealed no other signicant main or interaction effects.
Fig. 10 presents the normalized average total xation duration for each feature across different factors. A 3 ×2 ×2 ×4 repeated-
measures ANOVA, with the variables: identity, inversion, orientation, and features, was conducted on the normalized scores for the
average total xation duration. The right side of Table 6 reports the summary of the ANOVA results. The analysis revealed a signicant
main effect for features, which was quantied by a signicant interaction with the identity factor. Simple main effects analyses fol-
lowed by Holm-Bonferroni corrected pairwise comparisons revealed that the mouth was xated shorter on UF (M =0.77, SD =0.67)
Table 5
Statistical Analysis of Average Total Number of Fixations and Average Total Fixation Duration corresponding to Identity, Inversion, and Orientation
(Exp. 2).
Variables Average Total Number of Fixations Average Total Fixation Duration
df F
η
p
2
df F
η
p
2
Identity 2, 58 0.73 0.02 2, 58 13.01
***
0.31
Inversion 1, 29 19.71
***
0.41 1, 29 0.06 0.002
Orientation 1, 29 2.00 0.07 1, 29 0.84 0.03
Identity ×Inversion 2, 58 1.31 0.04 2, 58 0.11 0.004
Identity ×Orientation 2, 58 1.00 0.03 2, 58 2.42 0.08
Inversion ×Orientation 1, 29 0.01 0.00 1, 29 0.86 0.03
Identity ×Inversion ×Orientation 2, 58 3.04 0.10 2, 58 0.62 0.02
***
p <.001.
Fig. 8. Area-Normalized Average Total Number of Fixations for Each Facial Feature (Exp. 2). Note. Area-normalized total number of xations
received by (a) eyes, (b) nose, (c) mouth, and (d) other facial regions for each identity across different inversion conditions. The orientation factor is
collapsed across each AOI. Error bars represent the standard error of the mean.
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
13
than SF (M =1.045, SD =0.78) or FF (M =1.053, SD =0.79), but there was no signicant difference between SF and FF. Conversely,
the rest of the face was xated upon longer for UF (M =0.30, SD =0.19) than SF (M =0.25, SD =0.16) or FF (M =0.26, SD =0.20),
and there were no signicant differences between SF and FF. Finally, there were no signicant differences in the xation duration for
the eyes and nose across all identities (see Table 8).
The analysis also revealed a signicant interaction between identity and orientation (see Table 6). Post-hoc analysis revealed that
FF was xated upon longer in the normal oriented version (M =1.65, SD =0.90) than in a mirror-reversed orientation (M =1.59, SD =
0.86; p =.04, d =0.40), whereas the number of xations on both SF (normal: M =1.48, SD =0.89; mirrored: M =1.71, SD =0.86; p
=.85, d =0.04) and UF (normal: M =1.54, SD =0.90; mirrored: M =1.69, SD =0.93; p =.15, d =0.27) did not differ signicantly
Table 6
Statistical Analysis of Average Total Number of Fixations and Average Total Fixation Duration corresponding to Identity, Inversion, Orientation and
Features (Exp. 2).
Variables Average Total Number of Fixations Average Total Fixation Duration
df F
η
p
2
df F
η
p
2
Features 1.62, 46.84 106.53
***
0.79 1.54, 44.68 95.97
***
0.77
Identity ×Orientation 2, 58 4.39* 0.13 2, 58 3.36* 0.10
Features ×Identity 3.66, 106.18 6.35
***
0.18 3.21, 90.45 4.74
***
0.14
Features ×Inversion 1.56, 45.14 3.71* 0.11 1.54, 44.51 4.24 0.13
Features ×Orientation 1.90, 55.15 0.73 0.02 1.64, 47.40 1.72 0.06
Features ×Identity ×Inversion 3.63, 105.14 1.27 0.04 3.33, 96.42 1.08 0.04
Features ×Identity ×Orientation 3.92, 113.74 1.58 0.05 3.40, 98.47 2.33 0.07
Features ×Inversion ×Orientation 1.59, 46.21 2.37 0.08 1.38, 39.96 2.37 0.08
Features ×Identity ×Inversion ×Orientation 3.07, 89.15 0.39 0.01 2.72, 78.79 0.81 0.03
Note. Huynh-Feldt corrections were applied to the df for all analyses with the ‘features’ variable.
***
p <.001.
*
p <.05.
Table 7
Simple Main Effects of Identity Considered at each level of Features on the Average Total Number of Fixations (Exp. 2)
Simple main effects of Identity at each level of:
Features df F
η
p
2
Pairwise Comparisons
Eyes 2, 58 1.95 0.15 –
Nose 2, 58 2.08 0.07 –
Mouth 2, 58 21.03
***
0.42 SF >UF (p <.001, d =1.00)
FF >UF (p <.001, d =1.11)
SF =FF (p =1.00, d =0.07)
Others 2, 58 11.21
***
0.28 SF <UF (p <.001, d = − 0.87)
FF <UF (p =.002, d = − 0.68)
SF =FF (p =1.00, d = − 0.11)
***
p <.001.
Fig. 9. Average Total Fixation Duration for Different Identities (Exp. 2). Note. The average total xation duration received by faces for each identity
(SF, FF, and UF) across different inversion and orientation. Error bars represent the standard error of the mean.
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
14
across the orientation conditions.
3.3. Discussion
Experiment 2 explored the effects of task demands, specically a face identication task, on the viewing pattern for self, friend, and
unfamiliar faces. We observed that when asked to recognize faces, (1) there were no signicant differences in the number of xations
across all faces, but the own face was xated for a shorter duration compared to the friend and the unfamiliar face. Next, (2) there were
no signicant differences in the number of xations and xation duration for the eyes and the nose across all face identities, whereas
the mouth and the rest of the face were xated with a lesser number of xations and a shorter amount of xation duration on
Fig. 10. Area-Normalized Average Total Fixation Duration for Each Facial Feature (Exp. 2). Note. Area-normalized average total xation duration
received by (a) eyes, (b) nose, (c) mouth, and (d) other facial regions for each identity across different inversion conditions. The orientation factor is
collapsed across each AOI. Error bars represent the standard error of the mean.
Table 8
Simple Main Effects of Identity Considered at each level of Features on the Average Total Fixation Duration (Exp. 2).
Simple main effects of Identity at each level of:
Features df F
η
p
2
Pairwise Comparisons
Eyes 2, 58 2.58 0.08 –
Nose 2, 58 2.31 0.10 –
Mouth 2, 58 14.71
***
0.34 SF >UF (p <.001, d =0.79)
FF >UF (p <.001, d =1.15)
SF =FF (p =1.00, d =0.02)
Others 2, 58 4.42* 0.13 SF <UF (p =.04, d =0.49)
FF <UF (p =.04, d =0.47)
SF =FF (p =1.00, d =0.11)
***
p <.001.
*
p <.05.
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
15
unfamiliar faces compared to both the friend face and self-face, and nally, (3) we reported an evident inversion effect on the viewing
patterns for all faces.
When asked to identify the faces, we observed no signicant differences in the number of xations across all face identity con-
ditions with the nose being sampled with a higher number of xations and the longest xation duration compared to all facial features.
To identify a face, individuals rst scan several facial features to extract facial information, followed by structural analysis and se-
mantic encoding (Kita et al., 2010). When completing tasks which require higher cognitive demands, such as memory, the attention
space of an individual narrows, allowing only limited information to be processed, therefore leading to a similar scanning strategy to
extract facial information from all faces (see Stacey et al., 2005). To further facilitate this strategy, by directing attention to the centre
of a face (i.e., the nose area), individuals are able to extract facial information simultaneously, as a “whole” face representation (Van
Belle, Ramon, et al., 2010), altogether further facilitating the face identication task.
Next, the own face received shorter individual xations compared to a friend and unfamiliar faces when asked to make identity
judgements suggest that compared to other faces, individuals spent less time acquiring sufcient facial information to identify their
own face (e.g., Hsiao & Cottrell, 2008). As people are more familiar with their own faces and also have a more robust mental rep-
resentation of their own face (Tong & Nakayama, 1999), less cognitive effort is needed when extracting facial information from their
own faces (see Rayner, 1998; Salvucci & Goldberg, 2000).
Overall, ndings from Experiment 2 showed that task demands modulated the viewing patterns for the own face. Specically, when
asked to freely view their own face in Experiment 1, individuals adopted a more featural processing strategy, whereas when prompted
to extract facial information, individuals adopted a more holistic approach instead. However, such a viewing pattern was not observed
for both familiar and unfamiliar faces, wherein individuals adopted a more holistic approach when asked to passively view and when
asked to identify faces.
4. General discussion
In two experiments, we explored the role of holistic and featural processing when viewing the own face and other faces in a free-
viewing task (Experiment 1) and a face-identication task (Experiment 2). Overall, in Experiment 1, the own face received a higher
number of xations than both the friend and unfamiliar face, with a higher proportion of xations and longer xation durations to the
lower regions of the own face as compared to the lower regions of the other faces. Interestingly, in Experiment 2, the number of
xations did not differ signicantly for all three faces, with the nose receiving a higher proportion of xations and being xated for a
longer duration than the other facial features.
4.1. A more featural based processing for the own face
First, when asked to freely explore the faces, a greater proportion of xations were allocated to the own face compared to both
familiar and unfamiliar faces. These results are in line with Hills (2018) who also showed a higher number of xations to the own face.
Featural processing is generally associated with a higher number of xations to individual facial features (e.g., Bombari et al., 2009)
than holistic processing (Hills, 2018). In line with this notion, studies have reported that such a viewing pattern was also observed
when individuals were asked to look at inverted faces (e.g., Hills et al., 2013; Van Belle, de Graef, et al., 2010). As facial features for the
own face might be overall sampled more often, each facial feature (i.e., eyes and mouth) may be focused and processed individually,
therefore resulting in an overall higher number of xations for the own face compared to other faces. Additionally, such ndings may
also imply a feature verication dependent process for the own face (Van Belle, Ramon et al., 2010). This process ensures a match
between a perceived stimulus and its stored representation in memory, through the comparison of feature by feature, resulting in an
overall more individual xations to the facial features.
Experiment 1 also revealed that in comparison to the friend and unfamiliar face, the nose on the self-face received a fewer number
of xations. Fixations to the nose have been associated with holistic processing as such xations would allow for a perception of the
whole face (Van Belle, Ramon, et al., 2010). Under this assumption, our results suggest that when asked to explore faces, the self-face is
processed less holistically compared to familiar and unfamiliar faces. Additionally, the mouth was xated more often and longer on the
own face compared to other faces. Chakraborty and Chakrabarti (2018) also observed a similar gaze allocation strategy to the lower
region of the own face. The authors suggested that the own face holds attention more compared to other faces as the own face triggers
more exploration of the facial features of the own face (see also Devue et al., 2009). More specically, despite the eyes providing ample
facial information, due to the “rewarding nature” of the own face to sustain attention, increased sampling of facial features could take
place (Chakraborty & Chakrabarti, 2018). Likewise, studies have also shown that individuals with prosopagnosia who rely on the
featural processing of faces, xated less on the eyes but directed a greater proportion of gaze towards the mouth (e.g., Bukach et al.,
2006; Ramon et al., 2010).
Experiment 1 also showed no differences in the proportion of xations across the friend’s face and the unfamiliar face, suggesting
no differences in the sampling manner between both faces. This nding is also in line with the study by Van Belle and colleagues
(2010), which showed no differences in the number of xations when viewing a friend and an unfamiliar face. More specically, we
observed that a higher number of xations are positioned on the nose for both a friend and an unfamiliar face compared to the self-face.
The observed differences in the gaze pattern between the own face and other faces suggest the own face is processed in a distinctive
manner compared to a personally familiar face and an unfamiliar face, at least in a free-viewing paradigm.
However, when asked to identify faces, we observed no differences in the viewing pattern for the own face, a friend’s face, and an
unfamiliar face, with a higher number of xations and a longer total xation time on the nose across all faces. Generally, familiarity
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
16
judgements are based on an appreciation of the face as a whole instead of focusing on detailed feature information (e.g., Van Belle,
Ramon, et al., 2010). Directing xations towards the centre of a face allows for extracting all facial information simultaneously to
facilitate the face identication task. Indeed, Hsiao and Cottrell (2008) demonstrated that face recognition can be achieved within two
xations, and these xations are generally allocated around the top of the nose. These ndings were consistent with the ndings by
Stacey et al. (2005) who reported that the effects of familiarity on face processing were inuenced by the task demands imposed.
More specically, the differences reported in the eye-movement patterns for the own face across a passive-viewing and recognition
task might also reect the different goals for the processing and the recognition of the own face. In particular, individuals generally
perceive others’ faces for identication purposes, whereas one does not aim to identify their own faces when looking in the mirror.
However, when the task demands were introduced and kept consistent for all faces, there were no differences in the eye-movement
patterns when recognizing the own, the personally familiar, and the unfamiliar face. Therefore, the ndings across the two experi-
ments seem to suggest an inuence of task demands on the viewing pattern for the own face, such that when asked to passively view or
explore their own face, individuals adopted a more featural processing strategy, whereas when asked to make identity judgement or to
recognize the own face, a more holistic approach is adopted. In particular, due to the personal signicance and relevance of the own
face, the processing and recognition of the own face may be supported by both featural and holistic processing (see Hills, 2018) and
these processes are employed depending on the task at hand.
Previous research found that two xations are enough for face recognition (e.g., Hsiao & Cottrell, 2008). Interestingly, when we
reanalysed the results of Experiment 1 including only the rst two xations (see Supplementary Material), we did not nd any dif-
ferences in the viewing patterns across all three faces. This nding replicates the results of Experiment 2 and conrm that the dif-
ferences in the viewing patterns between the self-face and other faces reported in Experiment 1 are not related to identication
processes. Instead, the differences found in Experiment 1 could reect a stronger attention holding property of the own face (e.g.,
Devue et al., 2009), which might cause individuals to spend more time exploring and looking at their own face.
4.2. The nose as a diagnostic facial feature
Contradicting the feature-saliency hypothesis which denotes the eyes as the most diagnostic feature when perceiving faces (e.g.,
Hsiao & Cottrell, 2008; Shepherd et al., 1981; Walker-Smith et al., 1977), our ndings suggest otherwise. We observed a preference for
the nose to be xated when perceiving faces. For instance, in a free-viewing task (Exp. 1), a similar number of xations and xation
duration was allocated to both the nose and the eyes; whereas the nose received a greater proportion of xation and longer xations
compared to the eyes when the task demands were introduced (Exp. 2).
Studies have consistently reported a hierarchy of features for upright faces, with the eyes identied as the most diagnostic feature
for face recognition, followed by the mouth and the nose (e.g., Ellis et al., 1979; Shepherd et al., 1981). Specically, the eyes are
focused upon more often when perceiving faces, at least for Western individuals (Blais et al., 2008). In fact, when asked to describe
faces, Western participants tend to describe the eyes more often compared to other facial features (Ellis et al., 1979). The upper face,
specically the eyes, automatically attracts attention due to its role in expressing social cues, such as emotions or direction of gaze
(Barton et al., 2006; Shepherd et al., 1981).
Despite the well-established role of eyes in the face processing literature, Kita et al. (2010) observed that when asked to view faces,
East Asian participants made more xations toward the nose instead of the eyes or mouth. Furthermore, Blais et al. (2008) showed that,
in comparison to Western Caucasian individuals, East Asian individuals showed a preference to “integrate information holistically”,
hence resulting in attention being directed to the centre of a face (i.e., the nose area), which allows the perception of the “whole” face.
Consistent with these studies, we obtained a similar proportion of xations and xation duration between the eyes and nose in
Experiment 1. Nevertheless, beyond the race of faces or participants, it is also worth taking note that the different diagnostic features in
Western and Asian populations may be attributed to the type of picture employed across each study. Specically, whereas Kita et al.
(2010) employed cropped faces in Asian populations (similar to our study), Blais et al. (2008) used headshots in Western populations.
These two studies found that people xate on the eyes and the nose, respectively. These features happen to be located at the centre of
the image in both types of pictures (i.e., eyes on the headshots and bridge of the nose on cropped images), and this location could thus
be the most optimal point of xation to gather the most visual information, regardless of whether the face appears at the centre of the
screen or not).
Nevertheless, the role of the nose in the perception of faces became more evident when participants were asked to make identity
judgements (Exp. 2). In this case, participants may adopt an efcient strategy by xating at the centre of a face (nose area) for facial
information to be extracted simultaneously (Hsiao & Cottrell, 2008; Van Belle, Ramon, et al., 2010), therefore resulting in the nose
being sampled more often and xated longer across all faces, regardless of their identity, in Experiment 2. Importantly, this xation
pattern cannot be explained by the observer’s tendency to xate in the centre of the screen, as the face stimuli were presented either to
the top or bottom of the screen in a pseudorandom order.
4.3. Modulation of task demands on an inversion effect for faces
Contradicting previous studies which showed a higher number of xations for inverted faces (e.g., Barton et al., 2006; Hills, 2018),
ndings from the free-viewing task reported no such observation for inverted faces whereas ndings from the identity judgement task
showed a higher proportion of gaze for inverted than upright faces. Inverting a face is known to disrupt the holistic processing of a face,
which affects the extraction of facial information (Rossion, 2008, 2009). To compensate for this disruption, featural processing is
employed, such that inverted faces receive a higher proportion of xations compared to upright faces (i.e., an inversion effect). It is
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
17
important to acknowledge that most studies which reported a higher proportion of xations for inverted faces included task demands,
requiring participants to make familiarity judgements and to extract facial information from these face stimuli (e.g., Hills, 2018).
Following this line of argument, we reported an inversion effect for Experiment 2 which participants were asked to make identity
judgements, whereas no such observation was reported for a free-viewing task wherein observers might make a ‘general sweeping
scan’ of faces instead of attending to critical facial features that facilitate face recognition tasks (Sammaknejad et al., 2017). Overall,
our ndings suggest that the inversion effect for faces may also be modulated by the presence of task demands.
4.4. Limitations
One limitation of this study that should be noted is the use of static faces rather than dynamic faces. Static faces are important to
shed light on cognitive mechanisms underlying face perception, as static stimuli are more controlled in their presentation. Never-
theless, these stimuli may not be an accurate depiction of real-life scenarios and it remains an articial method employed in laboratory
settings. The everyday-life interactions between individuals are a dynamic process, thus using dynamic stimuli in face research would
be more ecologically valid.
It is also important to note that the diagnostic facial feature information might not be necessarily reected by the location of
xations. For instance, predening the area of interest might mask important information about the potential differential scanning
behaviour for faces with different levels of familiarity through the reduction of xation points to a facial feature which is nearest to the
actual xation point (see Van Belle, Ramon, et al., 2010).
4.5. Conclusions
In conclusion, due to the personal signicance and relevance of the own face, we showed that the processing and recognition of self-
face may be supported by both a featural and holistic processing and these processes are employed depending on the nature of the
experiment: either the exploration or the recognition of faces. Specically, the own face may hold attention more than other faces,
allowing individuals to further explore their face (i.e., featural processing) compared to other faces, whereas when asked to identify
faces, holistic processing is employed across the self-face and other faces.
Declaration of Competing Interest
The authors declare that they have no known competing nancial interests or personal relationships that could have appeared to
inuence the work reported in this paper.
Acknowledgements
The authors are thankful to the participants who took part in this study.
Funding
This research did not receive any specic grant from funding agencies in the public, commercial, or not-for-prot sectors.
Appendix A. Supplementary material
Supplementary data to this article can be found online at https://doi.org/10.1016/j.concog.2022.103400.
References
Barton, J. J., Radcliffe, N., Cherkasova, M. V., Edelman, J., & Intriligator, J. M. (2006). Information processing during face recognition: The effects of familiarity,
inversion, and morphing on scanning xations. Perception, 35(8), 1089–1105. https://doi.org/10.1068/p5547
Bindemann, M., Scheepers, C., & Burton, A. M. (2009). Viewpoint and centre of gravity affect eye movements to human faces. Journal of Vision, 9(2). https://doi.org/
10.1167/9.2.7
Blais, C., Jack, R. E., Scheepers, C., Fiset, D., & Caldara, R. (2008). Culture shapes how we look at faces. PLoS ONE, 3(8). https://doi.org/10.1371/journal.
pone.0003022
Bombari, D., Mast, F. W., & Lobmaier, J. S. (2009). Featural, congural, and holistic face-processing strategies evoke different scan patterns. Perception, 38,
1508–1521. https://doi.org/10.1068/p6117
Bortolon, C., & Raffard, S. (2018). Self-face advantage over familiar and unfamiliar faces: A three-level meta-analytic approach. Psychonomic Bulletin & Review, 25(4),
1287–1300. https://doi.org/10.3758/s13423-018-1487-9
Brady, N., Campbell, M., & Flaherty, M. (2005). Perceptual asymmetries are preserved in memory for highly familiar faces of self and friend. Brain and Cognition, 58
(3), 334–342. https://doi.org/10.1016/j.bandc.2005.01.001
Br´
edart, S. (2003). Recognising the usual orientation of one’s own face: The role of asymmetrically located details. Perception, 32(7), 805–811. https://doi.org/
10.1068/p3354
Bukach, C. M., Gauthier, I., & Tarr, M. J. (2006). Beyond faces and modularity: The power of an expertise framework. Trends in Cognitive Sciences, 10(4), 159–166.
https://doi.org/10.1016/j.tics.2006.02.004
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
18
Chakraborty, A., & Chakrabarti, B. (2018). Looking at my own face: Visual processing strategies in self-other face recognition. Frontiers in Psychology, 9. https://doi.
org/10.3389/fpsyg.2018.00121
Cook, M. (1978). Eye movements during recognition of faces. In M. M. Gruneberg, P. E. Morris, & R. N. Sykes (Eds.), Practical aspects of memory (pp. 286–292).
Academic Press.
Devue, C., Van der Stigchel, S., Br´
edart, S., & Theeuwes, J. (2009). You do not nd your own face faster; you just look at it longer. Cognition, 111(1), 114–122. https://
doi.org/10.1016/j.cognition.2009.01.003
Ellis, H. D., Shepherd, J. W., & Davies, G. M. (1979). Identication of familiar and unfamiliar faces from internal and external features: Some implications for theories
of faces recognition. Perception, 8(4), 431–439. https://doi.org/10.1068/p080431
Estudillo, A. J. (2012). Facial memory: the role of the pre-existing knowledge in face processing and recognition. Europe’s Journal of Psychology, 8(2), 231–244.
https://doi.org/10.5964/ejop.v8i2.455
Estudillo, A. J., & Bindemann, M. (2016). Multisensory stimulation with other race faces and the reduction of racial prejudice. Consciousness and Cognition, 42,
325–339. https://doi.org/10.1016/j.concog.2016.04.006
Estudillo, A. J., & Bindemann, M. (2017a). A multi-sensory system for self-face learning. In M. Bindemann, & A. M. Megreya (Eds.), Face processing: Systems,
disorders and cultural differences (pp. 241–254). Nova Science Publisher.
Estudillo, A. J., & Bindemann, M. (2017b). Can gaze-contingent mirror-feedback from unfamiliar faces alter self-recognition? The Quarterly Journal of Experimental
Psychology, 70(5), 944–958. https://doi.org/10.1080/17470218.2016.1166253
Estudillo, A. J., Kaufmann, J. M., Bindemann, M., & Schweinberger, S. R. (2018). Multisensory stimulation modulates perceptual and post perceptual face
representations: Evidence from event-related potentials. European Journal of Neuroscience, 48(5), 2259–2271. https://doi.org/10.1111/ejn.14112
Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). G* Power 3: A exible statistical power analysis program for the social, behavioral, and biomedical sciences.
Behavior Research Methods, 39(2), 175–191. https://doi.org/10.3758/BF03193146
Fletcher-Watson, S., Findlay, J. M., Leekam, S. R., & Benson, V. (2008). Rapid detection of person information in a naturalistic scene. Perception, 37(4), 571–583.
https://doi.org/10.1068/p5705
Frautschi, R. S., Orfahli, L. M., & Zins, J. E. (2021). Reecting on your reection: Examining the effect of a non-reversing mirror on self-perception. Aesthetic Surgery
Journal, 41(12), NP1989–NP1993. https://doi.org/10.1093/asj/sjab179
Greenberg, S. N., & Goshen-Gottstein, Y. (2009). Not all faces are processed equally: Evidence for featural rather than holistic processing of one’s own face in a face-
imaging task. Journal of Experimental Psychology: Learning, Memory and Cognition, 35(2), 499–508. https://doi.org/10.1037/a0014640
Gregory, R. L. (2001). Seeing oneself. Perception, 30, 903–904. https://doi.org/10.1068/p3008ed
Helo, A., Pannasch, S., Sirri, L., & R¨
am¨
a, P. (2014). The maturation of eye movement behavior: Scene viewing characteristics in children and adults. Vision Research,
103, 83–91. https://doi.org/10.1016/j.visres.2014.08.006
Henderson, J. M. (2003). Human gaze control during real-world scene perception. Trends in Cognitive Sciences, 7(11), 498–504. https://doi.org/10.1016/j.
tics.2003.09.006
Hills, P. J. (2018). Children process the self-face using congural and featural encoding: Evidence from eye tracking. Cognitive Development, 48, 82–93. https://doi.
org/10.1016/j.cogdev.2018.07.002
Hills, P. J., & Lewis, M. B. (2018). The development of face expertise: Evidence for a qualitative change in processing. Cognitive Development, 48, 1–18. https://doi.org/
10.1016/j.cogdev.2018.05.003
Hills, P. J., Cooper, R. E., & Pake, J. M. (2013). First xations in face processing: The more diagnostic they are the smaller the face-inversion effect. Acta Psychologica,
142(2), 211–219. https://doi.org/10.1016/j.actpsy.2012.11.013
Holmqvist, K., Nystr¨
om, M., Andersson, R., Dewhurst, R., Jarodzka, H., & Van de Weijer, J. (2011). Eye tracking: A comprehensive guide to methods and measures. Oxford
University Press.
Hsiao, J. H., & Cottrell, G. W. (2008). Two xations sufce in face recognition. Psychological Science, 19(10), 998–1006. https://doi.org/10.1111/j.1467-
9280.2008.02191.x
Just, M. A., & Carpenter, P. A. (1980). A theory of reading: From eye xations to comprehension. Psychological Review, 87(4), 329–354. https://doi.org/10.1037/
0033-295X.87.4.329
Keyes, H., & Brady, N. (2010). Self-face recognition is characterized by “bilateral gain” and by faster, more accurate performance which persists when faces are
inverted. The Quarterly Journal of Experimental Psychology, 63(5), 840–847. https://doi.org/10.1080/17470211003611264
Kita, Y., Gunji, A., Sakihara, K., Inagaki, M., Kaga, M., Nakagawa, E., & Hosokawa, T. (2010). Scanning strategies do not modulate face identication: Eye-tracking
and near-infrared spectroscopy study. PLoS ONE, 5(6). https://doi.org/10.1371/journal.pone.0011050
Laeng, B., & Rouw, R. (2001). Canonical views of faces and the cerebral hemispheres. Laterality, 6(3), 193–224. https://doi.org/10.1080/713754410
Lu, S. M., & Bartlett, S. P. (2014). On facial asymmetry and self-perception. Plastic and Reconstructive Surgery, 133(6), 873e–881e. https://doi.org/10.1097/
PRS.0000000000000211
Maurer, D., Le Grand, R., & Mondloch, C. J. (2002). The many faces of congural processing. Trends in Cognitive Sciences, 6(6), 255–260. https://doi.org/10.1016/
S1364-6613(02)01903-4
McNeill, D. (1998). The face. Little, Brown and Company.
Orban de Xivry, J. J., Ramon, M., Lef`
evre, P., & Rossion, B. (2008). Reduced xation on the upper area of personally familiar faces following acquired prosopagnosia.
Journal of Neuropsychology, 2(1), 245–268. https://doi.org/10.1348/174866407X260199
Pollastek, A., Rayner, K., & Collins, W. E. (1984). Integrating pictorial information across eye movements. Journal of Experimental Psychology: General, 113(3),
426–442. https://doi.org/10.1037/0096-3445.113.3.426
Poole, A., & Ball, L. J. (2006). Eye tracking in HCI and usability research. In C. Ghaoui (Ed.), Encyclopedia of human computer interaction (pp. 211–219). IGI Global.
Ramon, M., Busigny, T., & Rossion, B. (2010). Impaired holistic processing of unfamiliar individual faces in acquired prosopagnosia. Neuropsychologia, 48(4),
933–944. https://doi.org/10.1016/j.neuropsychologia.2009.11.014
Rayner, K. (1998). Eye movements in reading and information processing: 20 years of research. Psychological Bulletin, 124(3), 372–422. https://doi.org/10.1037/
0033-2909.124.3.372
Rossion, B. (2008). Picture-plane inversion leads to qualitative changes of face perception. Acta Psychologica, 128(2), 274–289. https://doi.org/10.1016/j.
actpsy.2008.02.003
Rossion, B. (2009). Distinguishing the cause and consequence of face inversion: The perceptual eld hypothesis. Acta Psychologica, 132(3), 300–312. https://doi.org/
10.1016/j.actpsy.2009.08.002
Rossion, B. (2013). The composite face illusion: A whole window into our understanding of holistic face perception. Visual Cognition, 21(2), 139–253. https://doi.org/
10.1080/13506285.2013.772929
Salvucci, D. D., & Goldberg, J. H. (2000). Identifying xations and saccades in eye-tracking protocols (pp. 71–78). ACM Press.
Sammaknejad, N., Pouretemad, H., Eslahchi, C., Salahirad, A., & Alinejad, A. (2017). Gender classication based on eye movements: A processing effect during
passive face viewing. Advances in Cognitive Psychology, 13(3), 232–240. https://doi.org/10.5709/acp-0223-1
Scott, L. S., Luciana, M., Wewerka, S., & Nelson, C. A. (2005). Electrophysiological correlates of facial self-recognition in adults and children. Romanian Association of
Cognitive Sciences: Cognition, Brain, Behaviour, 9, 211–238.
Shepherd, J., Davies, G., & Ellis, H. D. (1981). Studies of cue saliency. In G. Davies, H. D. Ellis, & J. Shepherd (Eds.), Perceiving and remembering faces (pp. 105–131).
Academic Press.
Stacey, P. C., Walker, S., & Underwood, J. D. M. (2005). Face processing and familiarity: Evidence from eye-movement data. British Journal of Psychology, 96(4),
407–422. https://doi.org/10.1348/000712605X47422
Sugiura, M., Kawashima, R., Nakamura, K., Okada, K., Kato, T., Nakamura, A., , … Fukuda, H., et al. (2000). Passive and active recognition of one’s own face.
Neuroimage, 11(1), 36–48. https://doi.org/10.1006/nimg.1999.0519
J.K.W. Lee et al.
Consciousness and Cognition 105 (2022) 103400
19
Sui, J., Zhu, Y., & Han, S. (2006). Self-face recognition in attended and unattended conditions: An event-related brain potential study. NeuroReport, 17(4), 423–427.
https://doi.org/10.1097/01.wnr.0000203357.65190.61
Tsakiris, M. (2008). Looking for myself: Current multisensory input alters self-face recognition. PLoS ONE, 3(12). https://doi.org/10.1371/journal.pone.0004040
Tong, F., & Nakayama, K. (1999). Robust representations for faces: Evidence from visual search. Journal of Experimental Psychology: Human Perception and Performance,
25(4), 1016–1035. https://doi.org/10.1037/0096-1523.25.4.1016
Troje, N. F., & Kersten, D. (1999). Dependent recognition of familiar faces. Perception, 28(4), 483–487. https://doi.org/10.1068/p2901
Tyler, C. W., & Chen, C. C. (2006). Spatial summation of face information. Journal of Vision, 6(10). https://doi.org/10.1167/6.10.11
Van Belle, G., de Graef, P., Verfaillie, K., Rossion, B., & Lef`
evre, P. (2010). Face inversion impairs holistic perception: Evidence from gaze-contingent stimulation.
Journal of Vision, 10(5). https://doi.org/10.1167/10.5.10
Van Belle, G., Ramon, M., Lef`
evre, P., & Rossion, B. (2010). Fixation patterns during recognition of personally familiar and unfamiliar faces. Frontiers in Psychology, 1.
https://doi.org/10.3389/fpsyg.2010.00020
Walker-Smith, G. J., Gale, A. G., & Findlay, J. M. (1977). Eye movement strategies involved in face perception. Perception, 6(3), 313–326. https://doi.org/10.1068/
p060313
Wong, H. K., Estudillo, A. J., Stephen, I. D., & Keeble, D. R. (2021). The other-race effect and holistic processing across racial groups. Scientic Reports, 11(1). https://
doi.org/10.1038/s41598-021-87933-1
Yin, R. K. (1969). Looking at upside-down faces. Journal of Experimental Psychology, 81(1), 141–145. https://doi.org/10.1037/h0027474
J.K.W. Lee et al.
