J. Behav. Ther. & Exp. Psychiat.

50 (2016) 215e222

Contents lists available at ScienceDirect

Journal of Behavior Therapy and

Experimental Psychiatry
journal homepage: www.elsevier.com/locate/jbtep

Further insight into self-face recognition in schizophrenia patients:

Why ambiguity matters*
Catherine Bortolon a, b, *, Delphine Capdevielle b, c, Robin N. Salesse d, Stephane Raffard a, b
a
Epsylon Laboratory, EA 4556 Montpellier, France
b
University Department of Adult Psychiatry, CHU Montpellier, Montpellier, France
c
INSERM U-1061, Montpellier, France
d
Movement to Health Laboratory, EuroMov, Montpellier-1 University, Montpellier, France

a r t i c l e i n f o a b s t r a c t

Article history: Background: Although some studies reported specically self-face processing decits in patients with
Received 5 May 2015 schizophrenia disorder (SZ), it remains unclear whether these decits rather reect a more global face
Received in revised form processing decit. Contradictory results are probably due to the different methodologies employed and
3 August 2015
the lack of control of other confounding factors. Moreover, no study has so far evaluated possible daily
Accepted 12 September 2015
Available online 15 September 2015
life self-face recognition difculties in SZ. Therefore, our primary objective was to investigate self-face
recognition in patients suffering from SZ compared to healthy controls (HC) using an objective mea-
sure (reaction time and accuracy) and a subjective measure (self-report of daily self-face recognition
Keywords:
Schizophrenia
difculties).
Face recognition Method: Twenty-four patients with SZ and 23 HC performed a self-face recognition task and completed a
Self-face recognition questionnaire evaluating daily difculties in self-face recognition. Recognition task material consisted in
three different faces (the own, a famous and an unknown) being morphed in steps of 20%.
Results: Results showed that SZ were overall slower than HC regardless of the face identity, but less
accurate only for the faces containing 60%e40% morphing. Moreover, SZ and HC reported a similar
amount of daily problems with self/other face recognition. No signicant correlations were found be-
tween objective and subjective measures (p > 0.05).
Limitations: The small sample size and relatively mild severity of psychopathology does not allow us to
generalize our results.
Conclusions: These results suggest that: (1) patients with SZ are as capable of recognizing their own face
as HC, although they are susceptible to ambiguity; (2) there are far less self recognition decits in
schizophrenia patients than previously postulated.
2015 Elsevier Ltd. All rights reserved.

1. Introduction social interaction decits including face processing. Overall, studies

Schizophrenia is a heterogeneous psychiatry disorder that af-
fects approximately 0.5% of the general population (Goldner, Hsu,
Waraich, & Somers, 2002). Schizophrenia is characterized, among
other symptoms, by social interaction decits. In the last decades,
researchers started to focus on the different factors implicated in
*
This research was supported by AlterEgo, a project funded by the European
Union FP7 (grant #600610).
* Corresponding author. Laboratory Epsylon, EA 4556, Montpellier University 3, 1
University Department of Adult Psychiatry, 39 Avenue Charles Flahault, 34295
Montpellier Cedex 5, France.
E-mail address: catherine.bortolon@etu.univ-monpt3.fr (C. Bortolon).
suggest that schizophrenia patients are impaired in face recogni-
tion. Nevertheless, these decits might be better explained by other
confounding factors such as cognitive and sensorial decits (Darke,
Peterman, Park, Sundram, & Carter, 2013).
More recently, researchers have focused specically on self-face
processing decits. Results are, however, contradictory. Some
studies found that schizophrenia patients present a specic decit
in recognizing self faces compared to familiar and unknown faces
(Irani et al., 2006; Jiu et al., 2014; Kircher, Seiferth, Plewnia, Baar, &
Schwabe, 2007). Conversely, other studies did not nd evidence
supporting a specic self-face processing decit (Heinisch, Wiens,
Grundl, Juckel, & Brune, 2013; Lee, Kwon, Shin, Lee, & Park,
2007). The discrepancies between these results could be

http://dx.doi.org/10.1016/j.jbtep.2015.09.006
0005-7916/ 2015 Elsevier Ltd. All rights reserved.
216 C. Bortolon et al. / J. Behav. Ther. & Exp. Psychiat. 50 (2016) 215e222
explained by the different methodologies employed. While both
Irani et al. (2006) and Kircher et al. (2007) used a familiarity
judgment task, Lee et al. (2007) used a visual search task, and
Heinisch et al. (2013) and Jia, Yang, Zhu, Liu, and Barnaby (2013) a
morphing procedure. Moreover, the lack of control of other con-
founding factors, notably, cognitive decits, also deserves some
consideration when analyzing these discrepancies. First, Irani et al.
(2006) did not mask the external features of the face, which might
divert patients' attention from the relevant features of the face.
Second, both Kircher et al. (2007; two rst experiments) and Irani
et al. (2006) did not control for stimulus habituation (Kircher
et al., 2007). Third, the task used in the study by Lee et al. (2007)
was heavily dependent of attention capacity (visual search task).
Forth, the studies using a morphing procedure (Heinisch et al.,
2013; Jia et al., 2013; Kircher et al., 2007) failed to control for
several important factors: (1) they did not present results for the
100% self-face; and (2) they did not evaluate how patients deal with
the ambiguity of the stimulus and the mismatch between their
internal representation of the self face and the morphed self-face
displayed. Fifth, Heinisch et al. (2013) make their conclusions
based on reaction time data only. Nevertheless, it has been shown
that slowing of reaction time is a general feature of schizophrenia
(Schatz, 1998); therefore it may not be a good parameter to judge
the existence of decits in face processing. Finally, except for one
study (Lee et al., 2007), none of these studies included a control
task to verify whether the slow-down observed during the self-face
recognition task was specic to faces or represented a more global
decit of processing speed.
Self-face recognition tasks provide a direct measure of patients'
ability to recognize their own face. Nevertheless, no study has, so
far, measured how patients suffering from schizophrenia perceive
their daily self-face recognition difculties. Laroi, D'Argembeau,
Bredart, and van der Linden (2007) developed a questionnaire
assessing self-face recognition failures in everyday life, namely,
Self-face Recognition Questionnaire (SFRQ). This questionnaire
evaluates several types of self-face recognition difculties
including: (1) misidentication of one's own face as being that of
someone else; (2) failure of recognition of one's own face, and (3)
perception of one's own face as being different from the internal-
ized representation. They showed that individuals with higher
scores on schizotypal personality presented more daily life dif-
culties in recognizing their own face compared to those with lower
scores. Moreover, they showed that self-face recognition difculties
were correlated with disorganized schizotypal dimension.
In sum, the current literature provides contradictory results
regarding self-face recognition decits in patients with schizo-
phrenia in laboratory setting and daily life. Therefore, our primary
objective was to investigate self-face recognition in patients
suffering from schizophrenia disorder compared to healthy con-
trols using an objective measure (reaction time and accuracy) and
a subjective measure (self-report of daily self-face recognition
difculties). More specically, we aimed to investigate how pa-
tients suffering from schizophrenia deal with: (1) perceptual am-
biguity and (2) the mismatch between their self-face mental
representation and the image displayed when their own face was
morphed with someone else's face. A secondary aim of the present
study was to investigate the correlations between the subjective
measures and the objective measures of self-face recognition
and patients' symptomatology including positive and negative
symptoms, and insight.
Regarding the objective measure, we hypothesized that pa-
tients suffering from schizophrenia would be overall slower than
healthy controls, but not less accurate. More specically, we ex-
pected that patients suffering from schizophrenia would perform
worse than controls only under perceptual ambiguity, that is, when
they have to decide the identity of a face containing 60% of the
person and 40% of another one. In agreement, previous studies have
shown that perceptual ambiguity might impair patients' perfor-
mance on categorization of facial expression (de Gelder et al.,
2005). With respect to the subjective measure, we hypothe-
sized that patients with schizophrenia would score higher than
healthy controls suggesting more everyday life difculties in
recognizing their own face.
2. Materials and method
2.1. Participants
In total, 24 schizophrenia patients who fulll the DSM-IV
diagnostic criteria for schizophrenia, currently receiving inpatient
or outpatient care were included. Inclusion criteria were being
between 18 and 60 years of age, having a diagnosis of schizophrenia
and being capable of reading, understanding and speaking French.
Exclusion criteria were substance abuse other than cannabis or
alcohol, co-morbid neurological disorder, history of severe brain
trauma or current electro-convulsivotherapy. Twenty three healthy
subjects were recruited in Montpellier area. The control group was
screened for current psychiatric illness using the Mini-
International Neuropsychiatric Interview (Sheehan et al., 1998)
and participants were excluded if they met criteria for any current
axis I disorder of the DSM-IV-TR or if they were rst-degree rela-
tives of subjects with schizophrenia. The control participants were
matched on age, sex and education level with schizophrenia
patients.
All participants provided written consent. The study received
approval by the local ethics committee for medical research.
2.2. Measures
This research protocol is part of the European STREP project
AlterEgo (FP7-ICT-2011-9 00 Cognitive Systems and Robotics #
600610). This protocol implicated several evaluations, however,
only some of them will be presented here. More specically, it
implicated a social motor coordination task, evaluations of neuro-
cognitive functions such as cognitive exibility, inhibition, working
memory, evaluations of social cognition among other measures.
2.2.1. Cognitive and clinical measures
Premorbid intelligence was estimated using the fNART
(Mackinnon & Mulligan, 2005). Mean chlorpromazine equivalents
dosage was computed. Severity of schizophrenic symptoms was
evaluated using the Positive and Negative Syndrome Scale (PANSS;
(Kay, Fiszbein, & Opler, 1987)). Insight was measured with the
PANSS item (G12).
2.2.2. Self-face recognition questionnaire (SFRQ; (Laroi et al.,
2007))
The SFRQ consists of six items evaluating recognition failures or
anomalies concerning oneself. Two items concern mis-
identications, two recognition failures and two concern percep-
tion of unusual aspects. If participants answer Yes to an item, they
had to subsequently answer three additional questions regarding
the frequency, whether the image was clear and the degree of
stress/tiredness. Finally, a score for other-recognition was also
calculated from this questionnaire.
2.3. Stimuli
Frontal view pictures of each participant's face with a neutral
expression were taken the day before the experiment using an 8
 C. Bortolon et al. / J. Behav. Ther. & Exp. Psychiat. 50 (2016) 215e222
megapixel digital camera (Canon PowerShot SX40). We also added
a famous and an unknown face in order to test whether decits
in face processing are specic to self-face or rather represent a
global face recognition decit. For the famous face we used well-
known famous people that participants were able to correctly
identify. Photographs of unknown people with neutral expressions
were chosen from the NimStim Face Stimulus Set (www.macbrain.
org). Photographs were race and gender-matched.
All pictures were edited in the Adobe Photoshop in order to
match pictures for luminance, to crop each photograph into an oval
encircling the eyes, the nose, and mouth removing visual cues
about hair and clothing, and resize. Self-faces were mirror-
reversed.
Fantamorph software (Abrosoft V.4) was used to morph the
different faces two by two (self-famous, self-unknown, famous-
unknown) in steps of 20%, resulting in 4 pictures (Fig. 1). In total 15
photos were presented (12 morphing faces, 1 famous, 1 unknown
and 1 participants' face).
The experimental task was designed and presented using E-
prime software. All instructions and images were displayed on a
white background in the center of the monitor with a resolution of
1920x1080 pixels.
2.4. Procedure
The rst session was devoted to take the participant's photo-
graphs, evaluate patients' symptomatology (PANSS) and self-face
recognition decits using the SFRQ. Also during this section, we
conrmed with the participants whether s/he was able to recognize
the famous person.
The second session was destined to perform the face-
recognition task. Participants were tested individually. During the
tasks participants will be sited in a sound-attenuated room at a
distance of 60 cm to a 17-inch monitor. The task started with a
control task, in which participants had to press one of the two
buttons, as fast as possible, when they saw either a red or a blue car.
Subsequently, they performed the face recognition task. The task
was composed of 150 trials, divided in 5 blocks. Each face appeared
randomly10 times for a maximum of 4s in the center of the com-
puter screen one at a time. The faces were replaced by a xation
cross appearing for 1s.
Participants were instructed to respond as quickly and accu-
rately as possible. They were presented with instructions to press
one of three buttons to indicate which of the three face types (self,
famous or stranger) were presented. Stickers signaling the identity
were placed on each button to avoid confusion and forgetfulness.
They had to answer whether the face displayed looked more like
themselves, the famous or an unknown face. The identities
assigned to the three buttons were balanced across the
Fig. 1. Morphing continuum from Unknown face to Famous face in steps of 20%.
217
participants. The whole task took between 15 and 20 min. After the
experimental task, participants performed the cognitive evaluation
(fNART).
2.5. Statistical analyses
KolmogoroveSmirnov was used to check variables distributions.
Reaction time values below 0.1 s and above 2 s were excluded from
the analyses.
For normal distributions, comparisons and correlations were
performed using repeated Measures ANOVA, Independent t test
and Pearson correlations. Non parametric analyses were per-
formed when transformations failed to normalize the variable
distribution: ManneWhitney test for independent comparisons
and either Wilcoxon Signed Ranks Test or Friedman Test for paired
comparisons. Statistical analyses were performed using SPSS
version 17.0. The software Statistica version 8 was used for the
post-hoc analysis.
Instead of aggregating together all the faces along the contin-
uum under the labels of Self, Famous or Unknown, we
analyzed each face individually. For accuracy we analyzed the
number of self, famous and unknown responses for each face
along each one of the three morphing continuums. Therefore, we
performed three 2 (two groups) x 6 (different faces along the
morphing continuum) ANOVAs for each morphing continuum
(self-famous; self-unknown; famous-unknown). For reaction time,
we analyzed each face within the morphing continuum it belonged.
Thus, we performed three 2 (two groups) x 6 (different faces along
the morphing continuum; e.g. from 100% self face to 100% famous
face) ANOVAs, one for each morphing continuum (self-famous;
self-unknown; famous-unknown).
3. Results
3.1. Sociodemographic and clinical characteristics
Sociodemographic and clinical characteristics of the partici-
pants are presented in Table 1. No signicant differences were
found between patients and healthy controls in terms of age, ed-
ucation and gender.
3.2. Objective measure
3.2.1. Accuracy
When analyzing the amount of self responses (Fig. 2A) in the
continuum between self and famous faces, results revealed a
main effect of group, F(1, 45) 5.96, p 0.019, h2 0.177,
morphing continuum, F(5, 225) 298.97, p < 0.0001, h2 0.869,
and also an interaction between the two factors, F(5, 225) 2.98,
218 C. Bortolon et al. / J. Behav. Ther. & Exp. Psychiat. 50 (2016) 215e222

Table 1
Mean (M) and Standard Deviation (SD) of Sociodemographic and Clinical Characteristics of the sample.

Schizophrenia Patients (N 24) Healthy controls (N 23) Statistics

M SD M SD

Age 31.96 8.95 29.69 3.75 t(45) 1.121, p 0.268

Education (years) 11.96 2.46 11.17 1.15 t(45) 0.382, p 0.704
fNART 99.82 9.43 103.06 6.65 t(45) 1.357, p 0.179
PANSS
Positive symptoms 8.96 2.16
Negative symptoms 15.46 7.56
General psychopathology 27.00 6.01
Total score 51.42 1.91
Medicationa 758.75 47.45
Illness duration (years) 7.27 6.34
N % N %
Gender (Male) 21 87.5 21 91.3 c2 0.179, p 0.672
Note: PANSS: Positive and Negative Symptoms Scale (Kay et al., 1987).
a
Chlorpromazine equivalents.

Fig. 2. Mean Accuracy of Schizophrenia patients and Healthy controls on the different face morphing continuums.

p 0.013, h2 0.062. Post-hoc analysis for the interaction effect
revealed a signicant difference between groups only for the face
60% self - 40% famous. Healthy controls answered self more
often than schizophrenia patients (p 0.001).
Regarding the self-unknown continuum, analyses revealed a
main effect of morphing continuum, F(5, 225) 293.08, p < 0.0001,
h2 0.867, and also an interaction between morphing continuum
and group, F(5, 225) 2.26, p 0.028, h2 0.054. The main effect
of group was not signicant, F(1, 45) 1.04, p 0.314, h2 0.023.
Post-hoc analysis revealed, however, that differences did not reach
statistical signicance. The effect of morphing continuum was
similar for both continuums. As expected, the higher the proportion
of Self, the higher the number of responses self (see
supplementary data). For the continuum between famous and
unknown, no signicant effect was found on the amount of self
responses (p > 0.05).
Regarding the amount of famous response (Fig. 2B) across
each continuum, results revealed only a main signicant effect of
morphing continuum, F(5, 225) 335.59, p < 0.0001, h2 0.882,
for the famous-self morphing. The effect of group was no signi-
cant, F(1, 45) 0.06, p 0.802, h2 0.001. Likewise, only a main
effect of morphing continuum, F(5, 225) 356.41, p < 0.0001,
h2 0.888, was found for the continuum famous-unknown. The
effect of group was no signicant, F(1, 45) 1.17, p 0.285,
h2 0.025. As the proportion of Famous increased, the number of
responses famous increased (see supplementary data). For the
continuum between self and unknown, no signicant effect was
found (p > 0.05). A trend toward signicant was, however, observed
 C. Bortolon et al. / J. Behav. Ther. & Exp. Psychiat. 50 (2016) 215e222 219

(p 0.6) for faces containing 60% of the self and 40% of the
unknown face.
Finally, for the amount of unknown responses (Fig. 2C), a
signicant main effect of face, F(5, 225) 280.75, p < 0.0001,
h2 0.862, was observed in the continuum unknown-famous. No
main effect of group was observed, F(1, 45) 0.60, p 0.441,
h2 0.013. A similar pattern of results was found for the continuum
unknown-famous. A main effect of face, F(5, 225) 342.646,
p < 0.0001, h2 0.884, but not of group, F(1, 45) 0.39, p 0.532,
h2 0.009, was found. Post-hoc analyses also showed a similar
pattern of results for both continuums (see supplementary data).
Lastly, a signicant main effect of face, F(5, 225) 17.26, p < 0.0001,
h2 0.277, and group, F(1, 45) 4.08, p 0.049, h2 0.083 was
found, as well as a trend toward a signicant interaction effect, F(5,
225) 2.21, p < 0.054, h2 0.047. Even though we observed a
difference between patients with schizophrenia and healthy con-
trols for the faces containing 60% self and 40% famous (SZ:
M 2.83 0.63; HC: 1.52 0.64; MD 1.31; p 0.245) and 60%
famous and 40% self (SZ: M 2.46 0.5; HC: 1.00 0.51;
MD 1.46; p 0.121), post hoc analysis revealed no signicant
differences.

3.2.2. Reaction time

Fig. 3 displays mean reaction time of both healthy controls and
patients with schizophrenia for each face. Regarding self-famous
morphing, analyses revealed a main effect of morphing contin-
uum, F(5, 225) 61.49, p < 0.0001, h2 0.577. As the proportion of
self or famous face increased, the reaction time decreased
(Fig. 3A.). Those faces containing more than 80% of the self/
famous face elicited faster responses than those faces containing
60%. The effect of group was not signicant, F(1, 45) 3.382,
p 0.073, h2 0.070. An interaction effect of morphing continuum
and group, F(5, 225) 2.792, p 0.018, h2 0.058, was also found,
even though post-hoc analysis revealed no differences between
groups.
For the self-unknown morphing (Fig. 3B.), on the other hand,
analyses revealed a main effect of group, F(1, 45) 8.381, p 0.006,
h2 0.157, and morphing continuum, F(5, 225) 32.183,
p < 0.0001, h2 0.417. Schizophrenia patients were slower than
healthy controls. However, both groups were faster at responding
when the face contained 80% or more of the self face compared to
faces containing only 60%. Finally, for the famous-unknown
morphing (Fig. 3C.), a signicant main effect of group, F(1,
45) 5.563, p 0.023, h2 0.110, morphing continuum, F(5,
225) 38.05, p < 0.0001, h2 0.458, and an interaction between
morphing continuum and group, F(5, 225) 2.959, p 0.013,
h2 0.062, were found. Post-hoc analysis revealed no signicant
differences between groups.

3.3. Control task

Results revealed that even though schizophrenia patients

(M 679.62, SD 119.35) were slower than healthy controls
(M 588.20, SD 213.62), no signicant main effect of group was
found, t(37.15) 1.788, p 0.082, d 0.52.
3.4. Self-face recognition: subjective measure
Between groups comparisons of the subjective measures of self
recognition (Table 2) revealed no signicant differences between
groups (p > 0.05). Within groups comparisons, on the other hand,
showed a signicant effect of the type of self-face recognition dif-
culty, c2 22.271, p 0.0001. Additional analyses, after Bonfer-
roni correction, revealed that healthy controls reported more
unusual perceptions compared to both misidentications,
Fig. 3. Mean reaction time for schizophrenia patients and healthy controls on the
different face morphing continuums.
Z 3.658, p 0.0001, and recognition failures, Z 2.790,
p 0.005. Conversely, schizophrenia patients reported more un-
usual perceptions, Z 2.672, p 0.008, and recognition failures,
Z 2.740, p 0.006, compared to misidentications.
3.5. Correlation analyses
Results revealed no signicant correlations between accuracy,
reaction time and both medication and illness duration (p > 0.05).
220 C. Bortolon et al. / J. Behav. Ther. & Exp. Psychiat. 50 (2016) 215e222

Table 2 schizophrenia were overall slower, but they were able to recognize
Mean (M) and Standard Deviation (SD) of the Self Recognition Questionnaire for their own face when perceptual ambiguity was low.
Schizophrenia patients and Healthy Controls.
Previous studies have already shown that perceptual ambiguity
Schizophrenia Patients (N 24) Healthy controls (N 23) impairs patient's performance on categorization of facial expres-
M SD M SD sion (de Gelder et al., 2005). When the picture we are looking at is
well dened, we quickly recognize it, in particular when we are
Total Score 3.50 2.63 3.74 2.18
Misidentications 0.95 1.00 0.91 0.90 seeing a familiar stimulus. However, under ambiguous stimulus,
Frequency 1.32 1.46 1.26 1.18 this facilitation disappears and the decision making depends more
Clarity 2.41 2.87 1.87 1.91 on the observer than on the stimulus characteristics. Studies have
Associated stress 2.59 3.11 3.26 3.25
shown that schizophrenia present a more global decit in decision
Recognition Failures 1.27 0.88 1.22 0.90
Frequency 2.41 1.79 1.83 1.34 making under ambiguity (Fond et al., 2013) in addition to both
Clarity 3.55 2.84 2.57 2.27 cognitive and sensorial decits (Darke et al., 2013). These decits
Associated stress 3.45 2.87 3.91 3.33 could further impact on face processing. Moreover, this effect could
Unusual Perceptions 1.27 0.94 1.61 0.66 be heightened in a forced decision make context, under the pres-
Frequency 2.36 1.97 2.91 1.35
sure of the time constraint, which might aggravate more global
Clarity 4.05 3.20 4.52 2.39
Associated stress 3.41 2.87 4.22 2.33 sensory decits (Butler et al., 2008). Previous studies have shown
Others recognition 3.05 1.40 2.52 0.95 that patients can perform similarly to controls on face processing
Mirror (frequency) 2.05 0.84 2.64 0.79 when the time constraint is reduced (Goghari, Macdonald, &
Note: Self-face recognition Questionnaire (Laroi et al., 2007). Sponheim, 2011). The hesitation in the context of ambiguity
might lead individuals to make more false alarms by judging that a
face more similar to their own face is rather an unknown face. Even
No signicant correlations were found between symptoms and though this kind of force-choice task does not represent their daily
both accuracy and reaction time. Signicant, albeit moderate, cor- lives, difculties in making a decision and the hesitation in
relations were found between frequency of unusual perceptions matching a face with an identity might lead to real handicap during
and both hallucinations and disorganization. Participants who social interaction.
experienced more hallucinations and more disorganization prob- This difculty in dealing with perceptual ambiguity might also
lems also reported more frequently the experience of recognizing be understood in light of the predictor-error theory (Corlett, Taylor,
their own face but thinking it did not t well with their Wang, Fletcher, & Krystal, 2010) and an increased top-down
representations. inhibitory (e.g. prior knowledge and perceptual expectation) in-
No signicant correlations were found between objective and uence on perception (Aleman, Bocker, Hijman, de Haan, & Kahn,
subjective measures. 2003; Vercammen & Aleman, 2010). On the one hand, predictor-
error (the mismatch between expectancy and experience) is sug-
4. Discussion gested to be the basis of development and maintenance of de-
lusions (Corlett, 2015). More specically, authors argue that
This study primary aim was to further explore self-face recog- delusions represent a failure to treat uncertainty about sensory
nition in schizophrenia patients using both an objective measure information when there is a mismatch between expectancy and
(reaction time and accuracy) and a subjective measure (self- experience. Thus, instead of updating the top-down expectations
report of daily self-face recognition difculties). We hypothesized with new learning, predictor error reinforces the delusional beliefs.
that patients suffering from schizophrenia would be overall slower On the other hand, studies have suggested an unbalance between
than healthy controls, but not necessarily less accurate. Instead, bottom-up and top-down processing under perceptual ambiguity,
only under perceptual ambiguity patients suffering from schizo- which might be associated with hallucinations (Aleman et al.,
phrenia would perform worse than controls. We also expected that 2003) and delusions (Corlett et al., 2010). Here, we argue that
schizophrenia patients would score higher than healthy controls similar processes could be implicated in visual perceptual ambi-
suggesting more everyday life difculties in recognizing their own guity during self-face processing. We suggest two factors could
face. contribute to difculties in processing ambiguity of one's own face:
In accordance with our hypothesis, our results showed that (1) an inability to deal with ambiguity of the sensory information
patients with schizophrenia performed similarly to healthy con- when facing a mismatch between the internal representation of
trols when recognizing faces at 100% and 80% in the morphing one's own face (and the famous face's representation) and the
continuum. Nevertheless, they performed worse when the image displayed (60% self and 40% famous); and (2) an increased
perceptual ambiguity of the self-face increased. More specically, inuence of top-down expectations (at the cost of bottom-up fac-
when viewing a face containing 60% of their own face and 40% of a tors) on decision making.
famous face, they tended to answer that it was an unknown Regarding the controls task, patients with schizophrenia were
face, rather than their own face, more frequently than healthy slightly slower than healthy controls. Nevertheless, no statistically
controls. This result was specic to the self-famous morphing signicant differences were found between groups. This result
continuum. Although they performed slightly worse during the could be either due to a specic slow-down during face processing
self-unknown morphing continuum, differences did not reach or due to the fact that the control task was a two-option forced
statistical signicance. It is possible that patients with schizo- choice task, while that face processing task was a three-option
phrenia compensate their difculty in the self-unknown morphing forced choice task. It is well known that reaction time rises with
continuum by increasing their reaction time. Concerning reaction the number of different possible stimuli (Nickerson, 1972). There-
time measures, schizophrenia patients were overall slower than fore, it is difcult to draw concrete conclusions from this result.
healthy controls, in particular, for the self-unknown and famous- Previously, Laroi et al. (2007) found that individuals with higher
unknown morphing continuums. Nevertheless, a similar pattern levels of schizotypal traits reported difculties in daily self-face
of response speed was observed. Both groups were faster when recognition such as confounding their own face with someone
responding to faces containing more than 80% of the identity else. Contrary to this study, we found no signicant differences
compared to 60%. Therefore, our results suggest that patients with between schizophrenia patients and non-clinical controls.
 C. Bortolon et al. / J. Behav. Ther. & Exp. Psychiat. 50 (2016) 215e222
Nevertheless, both groups reported different patterns of daily self-
face recognition difculties. Healthy controls reported more expe-
riences in which they recognized their own face but found it weird
(unusual perception). Patients, conversely, reported more
frequently to judge their own face as being an unfamiliar person
(recognition failure) compared to other daily difculties in self-face
recognition, but not compared to healthy participants.
We also investigated the correlation between subjective self-
report measures and objective measures and between the self-
face recognition measures and patients symptomatology including
positive and negative symptoms, and insight. We found no signi-
cant correlations between self-report measures (subjective mea-
sure) and the reaction time/accuracy (objective measure)
suggesting a dissociation between experimental task and daily life.
Important differences between laboratory settings and self-face
recognition in daily life might explain this lack of association.
First, most people rarely see their own face displayed in a neutral
expression and without hair or ears. Individuals see their own face
(and other too) with different expressions and different haircuts.
Second, in the laboratory settings several factors are controlled for
such as the light, the angle and the duration of presentation. In daily
life, however, all these aspects tend to vary, which can inuence the
way we perceive our own face. As suggested by Burton, Jenkins, and
Schweinberger (2011) the variability in the world (e.g. ambient
lighting conditions) is part of one's representation of the face and,
thus, is an important factor to be considered when evaluating face
recognition. Third, the self-face recognition questionnaire evaluated
different kinds of difculties including failures to recognition,
misidentication and abnormal perceptions in different context
(mirror, windows, photos and video). Conversely, laboratory set-
tings evaluate only self-face recognition using photographs and do
not take into account the different kinds of difculties individuals
can experience while seeing their own face. Finally, since ques-
tionnaires rely on memory and not on real life experience, this lack
of correlation might also be due the different mechanisms impli-
cated when responding to a questionnaire and actually trying to
recognize one's own face. Thus, this result highlight the importance
of taking into account different measures of self-face processing. It
seems that each measure provides unique information about the
different mechanism implicated in self-face recognition. Further
studies investigating daily difculties in self-face and face pro-
cessing should employ experience-sampling procedures that cap-
ture the representation of experience as it occurs.
Regarding the correlation analysis, we found that subjective, but
not objective, measures of self-recognition were associated with
positive symptomatology (i.e. hallucinations and disorganization).
In agreement with Laroi et al. (2007), results revealed that feelings
of strangeness when seeing one's own face were associated with
both hallucinations and disorganization. In their study, Laroi et al.
(2007) suggested that these correlations could be understood in
light of the ipseity-disturbance model proposed by Sass and Parnas
(2003); Sass and Parnas (2003) proposed that positive symptoms
could be understood in association with disturbances in the mini-
mal self. In this context, both disorganized symptoms and halluci-
nations appear in a context of hyperreexivity and diminished self-
affection.
Conversely, no signicant correlations were found between self-
face recognition and insight, contrary to the results of Heinisch
et al. (2013). Different hypothesis might explain the lack of asso-
ciation between self-face recognition and insight. First, Heinisch
et al. (2013) calculated a specic score of reaction time, which
implicate only the differentiation between self and famous faces,
but not self and unknown faces. In the present study we just used
the average of the self-face corrected answers and the self-face
reaction time. Second, insight was measured with the unique
221
item G12 of the PANSS. It is well known that insight is a multidi-
mensional factor; therefore the use of a specic and multidimen-
sional measure of patients' clinical insight e.g. Scale to assess
Unawareness of Mental Disorder (Amador et al., 1993) is needed.
Third, it is possible that the different direct and indirect measures of
self-awareness used in the present study are independent in
agreement with previous studies showing that awareness is a
multiply fractionated factor that can be determined by different
variables (Gilleen, Greenwood, & David, 2011).
4.1. Limitations
The present study has some limitations that should be consid-
ered while interpreting the results. First, the small sample size does
not allow us to generalize our results, in particularly the correlation
analysis. Second, we did not use a proper measure of insight.
Aforementioned, multidimensional measure of patients' clinical
insight would be indicated in this case. Third, it is possible that the
relatively mild severity of psychopathology (e.g. positive symp-
toms) in our patient sample precluded signicant differences be-
tween groups, in particular, on subjective measures. Forth, a more
naturalistic design using mirrors for self-face recognition and
photographs including the person's body and background envi-
ronment could shed more light on face processing in schizophrenia.
Fifth, unfortunately we did not measure schizotypal personality
traits in our controls group, which could help us to better under-
stand our results. Although several studies provide consistent evi-
dence in favor of the continuum hypothesis, there are some
differences between psychotic-like experiences in general popu-
lation and the psychotic experiences in patients with schizophrenia
(Waters, Allen, et al, 2012; Waters, Woodward, Allen, Aleman, &
Sommer, 2012).
5. Conclusions
In sum, several conclusions seem to emerge from our results.
First, individuals with schizophrenia have a preserved ability to
recognize their own face under lower levels of perceptual ambi-
guity. We highlighted some factors that could contribute to this
difculty in dealing with the mismatch between the mental rep-
resentation of their own face and the image displayed (containing
60% self and 40% famous): (1) hesitation and impaired decision
making under ambiguity; (2) abnormal prediction-error hypothe-
sis; and (3) abnormal inuence of top-down expectations. Second,
we found that patients with schizophrenia recruited for this study
did not report more daily life difculties in self-face recognitions
compared to healthy controls. Nevertheless, unusual perceptions
while recognizing their own face were signicantly correlated with
hallucinations and disorganization. Third, no signicant correla-
tions were found between subjective and objective measures
which could be attributed to the different processes underlying
both measures. Finally, no signicant correlations were found be-
tween self-face recognition measures, which could be explained by
the different measures employed in our study and those used by
Heinisch et al. Thus, further studies should take into account: daily
live experience, the fact that recognition of oneself is supported by
the integration of multiple sources of knowledge including face,
behavior, voice, body, and the associated feelings.
Conict of interest
We wish to conrm that there are no known conicts of interest
associated with this publication and there has been no signicant
nancial support for this work that could have inuenced its
outcome.
222 C. Bortolon et al. / J. Behav. Ther. & Exp. Psychiat. 50 (2016) 215e222
Acknowledgments
This experiment was supported by the European Union's Sev-
enth Framework Program (FP7 ICT 2011 Call 9) under grant
agreement nu FP7-ICT-600610. This project is promoted by CHRU of
Montpellier. We express our gratitude to the patients for partici-
pating in this research and thereby made it possible. Authors would
like to thank Charline Dellouve for her involvement in the
recruitment of participants.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.jbtep.2015.09.006.
References
Aleman, A., Bocker, K. B., Hijman, R., de Haan, E. H., & Kahn, R. S. (2003). Cognitive
basis of hallucinations in schizophrenia: role of top-down information pro-
cessing. Schizophrenia Research, 64, 175e185.
Amador, X. F., Strauss, D. H., Yale, S. A., Flaum, M. M., Endicott, J., & Gorman, J. M.
(1993). Assessment of insight in psychosis. American Journal of Psychiatry, 150,
873e879.
Burton, A. M., Jenkins, R., & Schweinberger, S. R. (2011). Mental representations of
familiar faces. British Journal of Psychology, 102, 943e958.
Butler, P. D., Tambini, A., Yovel, G., Jalbrzikowski, M., Ziwich, R., Silipo, G., et al.
(2008). What's in a face? effects of stimulus duration and inversion on face
processing in schizophrenia. Schizophr Research, 103, 283e292.
Corlett, P. R. (2015). Answering some phenomenal challenges to the prediction error
model of delusions. World Psychiatry, 14, 181e183.
Corlett, P. R., Taylor, J. R., Wang, X. J., Fletcher, P. C., & Krystal, J. H. (2010). Toward a
neurobiology of delusions. Progress in Neurobiology, 92, 345e369.
Darke, H., Peterman, J. S., Park, S., Sundram, S., & Carter, O. (2013). Are patients with
schizophrenia impaired in processing non-emotional features of human faces?
Front Psychology, 4, 529.
Fond, G., Bayard, S., Capdevielle, D., Del-Monte, J., Mimoun, N., Macgregor, A., et al.
(2013). A further evaluation of decision-making under risk and under ambiguity
in schizophrenia. European Archives of Psychiatry and Clinical Neuroscience, 263,
249e257.
de Gelder, B., Vroomen, J., de Jong, S. J., Masthoff, E. D., Trompenaars, F. J., &
Hodiamont, P. (2005). Multisensory integration of emotional faces and voices in
schizophrenics. Schizophr Research, 72, 195e203.
Gilleen, J., Greenwood, K., & David, A. S. (2011). Domains of awareness in schizo-
phrenia. Schizophrenia Bulletin, 37, 61e72.
Goghari, V. M., Macdonald, A. W., 3rd, & Sponheim, S. R. (2011). Temporal lobe
structures and facial emotion recognition in schizophrenia patients and
nonpsychotic relatives. Schizophrenia Bulletin, 37, 1281e1294.
Goldner, E. M., Hsu, L., Waraich, P., & Somers, J. M. (2002). Prevalence and incidence
studies of schizophrenic disorders: a systematic review of the literature. Ca-
nadian Journal of Psychiatry, 47, 833e843.
Heinisch, C., Wiens, S., Grundl, M, Juckel, G., & Brune, M. (2013). Self-face recog-
nition in schizophrenia is related to insight. European Archives of Psychiatry
Clinical Neuroscience, 263, 655e662.
Irani, F., Platek, S. M., Panyavin, I. S., Calkins, M. E., Kohler, C., Siegel, S. J., et al.
(2006). Self-face recognition and theory of mind in patients with schizophrenia
and rst-degree relatives. Schizophrenia Research, 88, 151e160.
Jia, H., Yang, J., Zhu, H., Liu, J., & Barnaby, N. (2013). Self-face recognition in the ultra-
high risk for psychosis population. Early Intervention in Psychiatry, 9(2),
126e132.
Kay, S. R., Fiszbein, A., & Opler, L. A. (1987). The positive and negative syndrome
scale (PANSS) for schizophrenia. Schizophrenia Bulletin, 13, 261e276.
Kircher, T. T., Seiferth, N. Y., Plewnia, C., Baar, S., & Schwabe, R. (2007). Self-face
recognition in schizophrenia. Schizophrenia Research, 94, 264e272.
Laroi, F., D'Argembeau, A., Bredart, S., & van der Linden, M. (2007). Face recognition
failures in schizotypy. Cognitive Neuropsychiatry, 12, 554e571.
Lee, J., Kwon, J. S., Shin, Y. W, Lee, K. J., & Park, S. (2007). Visual self-recognition in
patients with schizophrenia. Schizophrenia Research, 94, 215e220.
Mackinnon, A., & Mulligan, R. (2005). The estimation of premorbid intelligence
levels in French speakers. Encephale, 31, 31e43.
Nickerson, R. S. (1972). Binary-classication reaction time: A review of some studies
of human information-processing capabilities. Psychonomic Monograph Sup-
plements, 4, 275e318.
Sass, L. A., & Parnas, J. (2003). Schizophrenia, consciousness, and the self. Schizo-
phrenia Bulletin, 29, 427e444.
Schatz, J. (1998). Cognitive processing efciency in schizophrenia: generalized vs
domain specic decits. Schizophrenia Research, 30, 41e49.
xSheehan, D. V., Lecrubier, Y., Sheehan, K. H., Amorim, P., Janavs, J., Weiller, E., et al.
(1998). The mini-international neuropsychiatric interview (M.I.N.I.): the
development and validation of a structured diagnostic psychiatric interview for
DSM-IV and ICD-10. Journal of Clinical Psychiatry, 59(Suppl. 20), 22e33. quiz
34e57.
Vercammen, A., & Aleman, A. (2010). Semantic expectations can induce false per-
ceptions in hallucination-prone individuals. Schizophrenia Bulletim, 36, 151e156.
Waters, F., Allen, P., Aleman, A., Fernyhough, C., Woodward, T. S., Badcock, J. C., et al.
(2012). Auditory hallucinations in schizophrenia and nonschizophrenia pop-
ulations: a review and integrated model of cognitive mechanisms. Schizo-
phrenia Bulletin, 38, 683e693.
Waters, F., Woodward, T., Allen, P., Aleman, A., & Sommer, I. (2012). Self-recognition
decits in schizophrenia patients with auditory hallucinations: a meta-analysis
of the literature. Schizophrenia Bulletin, 38, 741e750.