Content uploaded by Junichi Takahashi
Author content
All content in this area was uploaded by Junichi Takahashi on Jan 23, 2019
Content may be subject to copyright.
Effects of facial expressions on visual short-term memory in relation
to alexithymia traits
q
Junichi Takahashi
a,
⇑
, Tomohisa Hirano
b
, Jiro Gyoba
c
a
Department of Human Development, Faculty of Human Development and Culture, Fukushima University, 1 Kanayagawa, Fukushima-shi, Fukushima 960-1296, Japan
b
Department of Psychology, Faculty of Arts and Letters, Tohoku University, 27-1, Kawauchi, Aoba-ku, Sendai-shi, Miyagi 980-8576, Japan
c
Department of Psychology, Graduate School of Arts and Letters, Tohoku University, 27-1, Kawauchi, Aoba-ku, Sendai-shi, Miyagi 980-8576, Japan
article info
Article history:
Received 12 January 2015
Received in revised form 6 April 2015
Accepted 6 April 2015
Keywords:
Alexithymia
Facial expressions
Angry face
Happy face
Visual short-term memory
abstract
We examined dysfunctional memory processing of facial expressions in relation to alexithymia.
Individuals with high and low alexithymia, as measured by the Toronto Alexithymia Scale (TAS-20), par-
ticipated in a visual search task (Experiment 1A) and a change-detection task (Experiments 1B and 2), to
assess differences in their visual short-term memory (VSTM). In the visual search task, the participants
were asked to judge whether all facial expressions (angry and happy faces) in the search display were
the same or different. In the change-detection task, they had to decide whether all facial expressions
changed between successive two displays. We found individual differences only in the change-detection
task. Individuals with high alexithymia showed lower sensitivity for the happy faces compared to the
angry faces, while individuals with low alexithymia showed sufficient recognition for both facial
expressions. Experiment 2 examined whether individual differences were observed during early storage
or later retrieval stage of the VSTM process using a single-probe paradigm. We found no effect of
single-probe, indicating that individual differences occurred at the storage stage. The present results pro-
vide new evidence that individuals with high alexithymia show specific impairment in VSTM processes
(especially the storage stage) related to happy but not to angry faces.
Ó2015 Elsevier Ltd. All rights reserved.
1. Introduction
Alexithymia is an individual characteristic that is known as a
reduced ability to identify and describe one’s own feelings and a
reduced capacity to engage in fantasy and imagery (Bagby,
Parker, & Taylor, 1994a; Bagby, Taylor, & Parker, 1994b; Sifneos,
1973). These difficulties relate to a variety of behavioral disorders,
for example, individuals with alexithymia show deficits in com-
municating emotions to others. Although alexithymia is not
included as a diagnostic category in the DSM-IV-TR (American
Psychiatric Association., 2000) or DSM-5 (American Psychiatric
Association, 2013), it may overlap with various psychiatric and
developmental disorders, such as autism or Asperger’s syndrome
(Fitzgerald & Bellgrove, 2006; Szatmari et al., 2008).
Human beings have developed communication through feelings
to maintain the social group. In social communication, we have to
guess others’ feelings. Specifically, the recognition of others’ facial
expressions plays an important role in social communication. In
order to adapt to the surroundings, there is a great need to retain
the identity of angry or happy faces and rapidly and effectively
decide whether to approach or avoid these people. However, as
described above, individuals with alexithymia generally show
impairments to identify and describe feelings. In these situations,
we speculate that individuals with alexithymia may have difficulty
in communication and experience maladaptation in a social group.
Thus, it is necessary to examine the cognitive process of facial
expressions in individuals with alexithymia so they can deal with
angry or happy faces rapidly and effectively.
Recent studies examined differences in emotional property in
alexithymia individuals. The researchers used an experimental
method, in which they examined the relationship between alex-
ithymia traits measured by the 20-item Toronto Alexithymia
Scale (TAS-20; Bagby et al., 1994a, 1994b) and emotional tasks
using facial expressions and emotional words as experimental
stimuli (Grynberg, Vermeulen, & Luminet, 2014; McDonald &
Prkachin, 1990; Pandey & Mandal, 1997; Parker, Taylor, & Bagby,
1993; Prkachin, Casey, & Prkachin, 2009; Vermeulen & Luminet,
2009; Vermeulen, Luminet, & Corneille, 2006). Other studies have
examined the physiological basis of the relationship between the
http://dx.doi.org/10.1016/j.paid.2015.04.010
0191-8869/Ó2015 Elsevier Ltd. All rights reserved.
q
This work was supported by a Japan Society for the Promotion of Science (JSPS)
Grant-in-Aid for Scientific Research to J.G. (Grant No. 24300087).
⇑
Corresponding author. Tel./fax: +81 24 548 8171.
E-mail address: j-takahashi@educ.fukushima-u.ac.jp (J. Takahashi).
Personality and Individual Differences 83 (2015) 128–135
Contents lists available at ScienceDirect
Personality and Individual Differences
journal homepage: www.elsevier.com/locate/paid
TAS-20 and emotional tasks, using event-related potentials (ERP;
Franz, Schaefer, Schneider, Sitte, & Bachor, 2004; Vermeulen,
Luminet, de Sousa, & Campanella, 2008), functional magnetic reso-
nance imaging (fMRI: Berthoz et al., 2002; Mantani, Okamoto,
Shirao, Okada, & Yamawaki, 2005), and positron emission tomogra-
phy (PET: Kano et al., 2003). Some of these studies have provided
evidence of memory deficits in individuals with alexithymia.
Using a behavioral approach, for example, Prkachin et al. (2009)
examined dysfunctional memory processing of facial expressions
in individuals with alexithymia. They used 90 different faces, each
displaying one of six basic facial emotions (Ekman & Friesen, 1975)
on video recordings. After presentation of the memory stimuli, test
faces were presented, and the participants were asked to judge
whether the presented faces in the test phase were the same as
in the original series. The authors compared high- and low-alex-
ithymia individuals’ facial emotion sensitivity and found that
individuals with high alexithymia showed less sensitivity to anger,
fear, and sadness conditions than individuals with low alex-
ithymia. In contrast, there were no significant differences between
high- and low-alexithymia individuals in sensitivity to happy faces.
Using a physiological approach, for example, Vermeulen et al.
(2008) showed dysfunctional processing of facial expressions in
individuals with alexithymia in terms of oddball task with ERP
measurements. In the task, they examined the categorical percep-
tions of facial expressions. Results showed that latencies of N2b
and P3a, reflecting the degree of attention devoted to specific infor-
mation, were delayed among individuals with alexithymia com-
pared with individuals without alexithymia. In this study, they
used only negative (angry and disgusted) faces and no positive
(happy) faces. These findings suggest that individuals with high
alexithymia showed weaker attention levels to deal with faces
expressing anger and disgust compared to individuals with low
alexithymia.
To the best of our knowledge, basic research of memory for
facial expressions in alexithymic individuals is quite limited and
are few studies examining memory processes of facial expressions
in relation to alexithymia. For example, DiStefano and Koven
(2012) showed dysfunctional processing of faces in individuals
with alexithymia in terms of visual immediate and delayed memo-
ries. In their task, color photographs with human faces (neutral
expressions) were presented. Experimental settings consisted of
an immediate memory condition (participants responded immedi-
ately after the series of test faces were presented) and delayed
memory condition (delayed presentation of series of the test faces).
The results showed that individuals with alexithymia showed
lower memory performance in both conditions. However, this
study did not manipulate facial expression. Pandey and Mandal
(1997) showed no differences in emotional matching between
individuals with and without alexithymia. They used photographic
faces (happiness, sadness, fear, anger, surprise, disgust, and neu-
tral) and the participants were asked to match the test pho-
tographs with the target. The results showed that there was no
difference in matching performance between the two groups. As
described above, Prkachin et al. (2009) showed that individuals
with high alexithymia showed less sensitivity to anger, fear, and
sadness conditions than individuals with low alexithymia, but this
study did not strictly manipulate time properties of task.
Considering these studies, heterogeneities exist when examining
visual short-term memory (VSTM) processes of facial expressions
in relation to alexithymia, such as memory properties (including
time properties), task (response mode), and stimulus type. To fill
this gap, we propose the following adjustments: First, we consider
the fact that the findings on memory deficits differ between these
previous studies. In other words, it is not obvious what kind of
memory impairment was addressed in each of these studies (see
also DiStefano & Koven, 2012). Second, in relation to the task, many
researchers used free-recall, recognition, or same-different tasks to
assess memory. These tasks could not strictly control the memory
processing times; thus, we found that many previous studies
examined long-term memory processes. Third, many previous
studies used photographic faces as experimental stimuli. Since
photographs faces include low-level artifacts, such as contrast
(Purcell, Stewart, & Skov, 1996), it is not clear whether the alex-
ithymic individuals’ deficiencies to memorize facial expressions
were caused by the facial expressions themselves (i.e., anger and
happiness) or by other low-level perceptual artifacts.
The purpose of the present study was therefore to examine dys-
functions of facial expression memory processes in individuals
with alexithymia, by conforming to strict experimental techniques
of psychophysics. Considering the first issue, basic research into
facial expression among individuals without alexithymia (i.e.,
individuals with typical development and without psychiatric dis-
orders) has confirmed the threat-advantage (negative) hypothesis
in a visual search task (e.g., Öhman, Lundqvist, & Esteves, 2001)
and in a change-detection task (e.g., Jackson, Wu, Linden, &
Raymond, 2009; however, a positive advantage hypothesis was
proposed by D’Argembeau and Van der Linden, 2007). The
threat-advantage hypothesis proposes that participants are faster
and more accurate in detecting and memorizing angry (or sad)
faces among a crowd of happy faces than vice versa. This hypothe-
sis seems to have higher repeatability and robustness in the field of
basic facial expression memory. Thus, in the present study, we
address dysfunctional processing of facial expressions in individu-
als with alexithymia in terms of this hypothesis. To examine the
second issue, we focused on VSTM, which is an early component
of the visual memory process that plays a role in the temporary
storage of object representation. The VSTM process consists of 3
stages: encoding, storage, and retrieval (e.g., Xu & Chun, 2006). A
visual search task and change-detection task are generally useful
in examining the VSTM process. Since many previous studies have
used schematic faces as experimental stimuli (e.g., Barratt &
Bundesen, 2012; Eastwood, Smilek, & Merikle, 2001; Eastwood,
Smilek, & Merikle, 2003), we also used schematic faces.
In the present study, we assessed TAS-20 data from a broader
sample and selected participants from the highest and lowest seg-
ments of the TAS-20 distribution (High TAS and Low TAS groups,
respectively) for the main experiments. In Experiment 1, we
assessed whether individual differences were observed in the
encoding or storage stages of the VSTM process by utilizing visual
search (Experiment 1A) and change-detection (Experiment 1B)
tasks, respectively. In Experiment 2, we further examined whether
individual differences were observed in storage or retrieval (deci-
sion making) stages of the VSTM process by establishing a single-
probe paradigm (Wheeler & Treisman, 2002) in the change-detec-
tion task. Considering the previous findings (Prkachin et al., 2009),
we predict that our High TAS group will show lower performance
for the angry and happy faces in the visual search and change-de-
tection tasks.
2. Experiment 1
2.1. Method
2.1.1. Participants
One hundred and thirty-seven Tohoku University students (72
men and 65 women; mean age = 21.20, SD = 2.48) completed the
TAS-20 assessment (mean TAS = 53.30, SD = 9.57). Thirty people
from the higher and lower portions of the TAS distribution were
invited to participate in a measurement of visual ability using
Raven’s Standard Progressive Matrices (SPM; Raven, Raven, &
Court, 1998) and to complete the visual search (Experiment 1A)
J. Takahashi et al. / Personality and Individual Differences 83 (2015) 128–135 129
and change-detection tasks (Experiment 1B). On the basis of
Taylor, Bagby, and Parker’s (1997) study, cutoff scores of 61 or
more and 51 or less were divided as members of either the High
TAS group (n= 15, 4 men and 11 women; mean age = 20.01,
SD = 1.79; mean TAS = 66.90, SD = 4.22) or the Low TAS group
(n= 15, 7 men and 8 women; mean age = 21.40, SD = 1.45; mean
TAS = 45.30, SD = 5.32), respectively.
1
The characteristics of the
High and Low TAS groups are described in Table 1. All participants
had normal or corrected-to-normal visual acuity. Informed consent
was obtained from each participant prior to his or her participation.
The ethics committee of the Graduate School of Arts and Letters,
Tohoku University, approved the study protocol.
2.1.2. Apparatus
Both Experiments 1A and 1B were programmed in Matlab
(Mathworks, Inc.) with the Cogent graphics packages and pre-
sented on a CRT monitor (Trinitron GDM-FW900, SONY, 24 inch)
with a 60-Hz refresh rate and a resolution of 1024 768 pixels
that was controlled by a PC (Precision 390, DELL). A chin rest
was used to set the participants’ head position. The viewing dis-
tance was 60 cm.
2.1.3. Stimuli
Figure 1(a) shows the face stimuli used in Experiments 1A and
1B. These faces were composed of a white circle and line (37.0 cd/
m
2
) against a black background (0.47 cd/m
2
). The diameters of the
faces (two angry and two happy faces) were approximately 4 [hori-
zontal] 4 [vertical] degrees. We created the faces with reference
to previous researches (e.g., Barratt & Bundesen, 2012; Eastwood
et al., 2001; Eastwood et al., 2003). We preliminarily confirmed
that the participants were able to accurately interpret the emo-
tions of each face. We also presented a red fixation circle
(12.0 cd/m
2
) at the center of the display.
2.1.4. Procedure
All 30 participants completed the SPM, the visual search task
(Experiment 1A), and the change-detection task (Experiment 1B).
The SPM was completed first; and then, half the participants com-
pleted Experiments 1A and 1B in this order, while the other half
completed them in reverse order. In both Experiments 1A and
1B, the anger condition meant that one angry face was presented
as the target stimulus with the happy face as a distractor, whereas
the happiness condition consisted of the reverse.
2.1.5. Experiment 1A
Figure 1(b) illustrates examples of the visual search task used in
Experiment 1A. In each trial, after the fixation point was presented
for 1000 ms at the center of the CRT monitor, the search display
including 4 or 8 various facial expressions was shown until the par-
ticipants responded. The participants were asked to judge whether
all facial expressions were the same or different (two-forced choice
method). Accuracy was emphasized in addition to response time.
The inter-trial interval (ITI) was 1000 ms.
The experiment comprised 144 trials (condition of changes in
facial expression [2; same and different] facial expression [2;
anger and happiness] set size [2; 4 and 8] 18 trials).
Participants completed 20 practice trials before the main session.
The order of conditions was randomized for each participant.
2.1.6. Experiment 1B
Figure 1(c) depicts examples of the change-detection task used
in Experiment 1B. On each trial, after the fixation point was pre-
sented for 1000 ms at the center of the CRT monitor, a memory dis-
play including 4 or 8 various facial expressions was shown for
2000 ms. Following a 100 ms mask display, a 2000 ms blank inter-
val was presented. Subsequently, a test display (either identical to
or different from the memory display) including 4 or 8 various
facial expressions was presented for 900 ms. Half the stimuli pre-
sented angry faces, and the other half showed happy faces. A blank
display was presented until the participants responded. The par-
ticipants were asked to judge whether the facial expressions on
all the stimuli in the test display were the same or different from
those in the memory display (two-alternative forced choice
method). Accuracy rather than speed was emphasized. The ITI
was 1000 ms.
The experiment comprised 144 trials (condition of changes in
facial expression [2; same and different] facial expression [2;
anger and happiness] set size [2; 4 and 8] 18 trials).
Participants completed 20 practice trials before the main session.
The order of the conditions was randomized for each participant.
2.1.7. Measurement of alexithymia traits
The TAS-20 (Bagby et al., 1994a, 1994b) is the most widely used
measure of the degrees of alexithymia traits. We used the Japanese
version of the TAS-20 (Fukunishi, Nakagawa, Nakamura, Kikuchi, &
Takubo, 1997). The reliability and validity of the English version
have been shown to generalize well to the Japanese version. The
TAS-20 comprises 20 items referring to various situations in rela-
tion to emotional processing that participants are asked to rate
on a 5-point, Likert-type scale (1 = not at all to 5 = very much). For
example, ‘‘I am often confused about which emotion I am feeling’’
and ‘‘It is difficult for me to find the right words for my feelings’’
(for more detail, see Bagby et al., 1994a, 1994b). The TAS-20 is
composed of 3 subscales: difficulty identifying feelings (7 items),
difficulty describing feelings (5 items), and externally oriented
thinking (8 items). A high TAS-20 score represents a high level of
alexithymia.
2.1.8. Measurement of nonverbal abilities
Participants completed the SPM (Raven et al., 1998) to test for
differences in nonverbal abilities between the High and Low TAS
groups. In order to confirm that individual differences in facial
recognition were independent of nonverbal abilities, we first
conducted the SPM to control for nonverbal abilities.
2.2. Results
The significance level was set at .05 (p< .05) in all analyses.
We first tested for differences in nonverbal abilities (repre-
sented by the sum of the SPM scores and response times) between
the High and Low TAS groups. The ttest showed no significant
Table 1
Characteristics of the High and Low TAS groups who participated in Experiments 1A,
1B, and 2.
Experiments 1A and 1B Experiment 2
High TAS
(n= 15)
Low TAS
(n = 15)
High TAS
(n=8)
Low TAS
(n=8)
Male:
Female
4: 11 7: 8 3: 5 3: 5
Age-years
(SD)
20.01 (1.79) 21.40 (1.45) 19.75 (1.16) 21.34 (1.60)
TAS score
(SD)
66.9 (4.22) 45.30 (5.32) 67.8 (6.09) 44.60 (2.92)
1
One might argue that the cutoff score of 61 or more might be arbitrary to define
individuals with alexithymia. Parker, Taylor, and Bagby (2003) collected TAS-20 data
from a large sample (N= 1933) and showed that 1.5 SD above the mean score was
close to the cutoff score of 61 or more. In the present study, we adopted this criterion.
However, it may be necessary to re-consider the experimental method in terms of the
examination of the TAS-20 as a continuous variable (see also Section 4).
130 J. Takahashi et al. / Personality and Individual Differences 83 (2015) 128–135
differences in nonverbal abilities between the two groups [scores:
t(28) = 2.01, p= .054; response time: t(28) = 0.50, p= .62].
2.2.1. Experiment 1A
In the visual search task, the proportion of correct responses
within the target-present condition for all participants was
approximately 90%. The accuracy for facial expressions of anger
with set sizes of 4 and 8 was 93.3% and 90.4%, respectively, for
the High TAS group and 93.0% and 86.7%, respectively, for the
Low TAS group. The accuracy for facial expressions of happiness
with set sizes of 4 and 8 was 93.7% and 90.0%, respectively, for
the High TAS group and 93.7% and 91.1%, respectively, for the
Low TAS group. These results indicate a ceiling effect, and there
was no trade-off between accuracy and response times.
Considering this finding, we used response time as our measure-
ment index.
Figure 2(a) represents the response times (target-present trials).
We conducted a three-way repeated measures ANOVA with the
factors of TAS-20 score (2; High and Low TAS) as a between-par-
ticipants factor and facial expression (2; anger and happiness)
and set size (2; 4 and 8) as within-participants factors. The results
showed main effects of facial expression and set size [facial expres-
sion: F(1,28) = 5.62, p< .05,
g
p
2
= .17; set size: F(1, 28) = 35.74,
p< .05,
g
p
2
= .56]. Response times for the angry faces were shorter
than for the happy faces. Moreover, response times for the smaller
set size were shorter than for the larger set size. However, there
was no significant main effect of the TAS-20 score [F(1, 28) = 2.05,
p= .16,
g
p
2
= .07] and no significant interactions between these fac-
tors [F(1,28) = 1.17, p= .29,
g
p
2
= .04].
2.2.2. Experiment 1B
Based on the proportions of correct answers, we calculated the
sensitivity (d-prime) scores for the detection of changes in facial
expressions in the change-detection task using the following for-
mula: d-prime = Z [H] Z [FA] (Green & Swets, 1966). When
changes in facial expressions were correctly detected, it was
regarded as a ‘‘hit’’ (H) for the ‘‘same’’ trials and as a ‘‘false alarm’’
(FA) for the ‘‘different’’ trials. A higher d-prime value indicates a
higher sensitivity for detecting change.
Figure 2(b) shows the calculated d-prime scores. We conducted
a three-way repeated measures ANOVA with the TAS-20 score (2;
High and Low TAS) as a between-participants factor and facial
expression (2; anger and happiness) and set size (2; 4 and 8) as
within-participants factors. There was a main effect of set sizes
[F(1, 28) = 142.84, p< .001,
g
p
2
= .84], showing higher d-prime
scores for the smaller set size compared to the larger one.
Moreover, there was a significant three-way interaction between
all factors [F(1, 28) = 9.18, p< .01,
g
p
2
= .25]. The simple two-way
(a)
(b) (c)
Fig. 1. Face stimuli used in this study and the stimulus sequences used in Experiments 1A and 1B. (a): We created these face stimuli with reference to previous researches
(e.g., Barratt & Bundesen, 2012; Eastwood et al., 2001, 2003). We used two types of angry and happy faces in Experiments 1A and 1B. In Experiment 2, we adopted neutral
faces in addition to the angry and happy faces. (b): Stimulus sequences used in the visual search task (Experiment 1A). This example shows the different trial with set size 8, in
which the correct response was made if the participants could detect the angry face in the lower right corner of the display. (c): Stimulus sequences used in the change-
detection task (Experiment 1B). This example shows the different trial with set size 8. If the participants could detect the angry faceat the top of the display, this wascounted
as a correct response.
J. Takahashi et al. / Personality and Individual Differences 83 (2015) 128–135 131
interactions between TAS-20 score and facial expression within the
larger set size [F(1, 56) = 9.41, p< .005,
g
p
2
= .14] and between facial
expression and set size in the High TAS group [F(1, 28) = 7.44,
p< .05,
g
p
2
= .21] were both significant. A significant simple main
effect of TAS-20 score with the angry face condition in set size 8
was observed [F(1, 112) = 4.44, p< .05,
g
p
2
= .04], indicating that
d-prime scores for the High TAS group were lower than for the
Low TAS group in the happiness condition for the larger set size.
In addition, we observed a significant simple main effect of facial
expression of the High TAS group in set size 8 [F(1, 56) = 9.03,
p< .005,
g
p
2
= .14], suggesting that d-prime scores for happy faces
of the High TAS group in the larger set size were lower than for
the angry faces. Moreover, there were significant simple main
effects of set size for all conditions [High TAS with angry faces:
F(1, 56) = 56.21, p< .001,
g
p
2
= .50; High TAS with happy
faces: F(1, 56) = 84.66, p< .001,
g
p
2
= .60; Low TAS with angry
faces: F(1, 56) = 67.14, p< .001,
g
p
2
= .55; Low TAS with happy
faces: F(1, 56) = 52.16, p< .001,
g
p
2
= .48]. These results showed that
d-prime scores for the smaller set size were higher than those of
both the High and Low TAS groups in both the anger and happiness
conditions.
2.3. Discussion
In Experiment 1, we examined the differences in response times
and sensitivity in relation to alexithymia traits by using visual
search (Experiment 1A) and change-detection (Experiment 1B)
tasks. We found that alexithymia affected the change-detection
task but not the visual search task. Concerning the change-detec-
tion task, the High TAS group showed lower sensitivity to happy
faces compared to angry ones, whereas the Low TAS group showed
equal accuracy to both angry and happy faces. The Low TAS group
seemed to be more sensitive to the happy faces than the High TAS
group. In the present experimental settings, the visual search and
change-detection tasks reflected the encoding and storage stages
of the VSTM process, respectively. Since the differences related to
TAS-20 were observed in the change-detection task but not in
the visual search task, these results suggest that individual differ-
ences occurred at the temporal storage level but not at the encod-
ing level. Specifically, the High TAS group showed impairments in
memorizing only happy faces, whereas the Low TAS group showed
no impairments in processing facial expressions.
3. Experiment 2
In Experiment 1, we showed individual differences in the stor-
age stage of memory processing with the change-detection task,
but there were no differences in the encoding stage with the visual
search task. Since the VSTM process includes encoding, storage,
and retrieval stages, there was a further possibility that individual
differences might be observed in the retrieval stage. To examine
this, we adopted the single-probe paradigm (Wheeler &
Treisman, 2002). In the single-probe condition, only one stimulus
is presented for the test display, whereas all stimuli are presented
for the test display in the whole-probe condition, similar to
Experiment 1B. By using this paradigm, Wheeler and Treisman
(2002) showed that memory performance improved in the sin-
gle-probe condition compared to the whole-probe condition (sin-
gle-probe advantage). This result could be interpreted as a
reduced interference or cognitive load in the memory retrieval of
the single-probe condition. In this way, these results might be
explained by the retrieval stage of the VSTM process. In contrast,
if there were no differences in memory performance between the
single- and whole-probe conditions, the results might indicate
something that is occurring at the storage stage of the VSTM
process.
Moreover, we adopted neutral (non-expressive) facial stimuli in
Experiment 2 (see Fig. 1(a)). In Experiment 1, because the partici-
pants compared expressive facial stimuli (i.e., angry and happy
faces), the effects of expression could not be assessed in detail.
By using neutral facial stimuli as a baseline in Experiment 2, we
can make a more precise examination.
In Experiment 2, we used the single-probe paradigm in the
change-detection task to examine the above possibility. If sensitiv-
ity in the single-probe condition was higher than in the whole-
probe condition, we could interpret this to mean that the individ-
ual differences might be explained at the retrieval stage of the
VSTM process. In contrast, if there were no differences found in
sensitivity between the single- and whole-probe conditions, the
individual differences might best be explained by the storage stage
of the VSTM process.
3.1. Method
3.1.1. Participants
Sixteen people, who did not participate in Experiment 1 but
were drawn from the previous participant pool, participated in
Experiment 2. We used the same cutoff scores (Taylor et al.,
1997) of 61 or more and 51 or less that determined whether
individuals were placed in either the High TAS (n= 8, 3 men and
5 women; mean age = 19.75, SD = 1.16; mean TAS = 67.8,
SD = 6.09) or Low TAS (n= 8, 3 men and 5 women; mean
age = 21.34, SD = 1.60; mean TAS = 44.60, SD = 2.92) groups,
respectively. Table 1 describes the characteristics of the High and
(a) (b)
Fig. 2. Response time of the visual search task (a) and sensitivity (d-prime) of the change-detection task (b) for the High and Low TAS groups (each n= 15). Error bars denote
the standard error (SE) of the mean.
132 J. Takahashi et al. / Personality and Individual Differences 83 (2015) 128–135
Low TAS groups. All the participants had normal or corrected-to-
normal visual acuity. Informed consent was obtained from each
participant prior to his or her participation. The ethics committee
of the Graduate School of Arts and Letters, Tohoku University,
approved the study protocol.
3.1.2. Apparatus
We used the same experimental settings as were used in
Experiment 1.
3.1.3. Stimuli
Figure 1(a) shows the facial stimuli used in Experiment 2. We
added neutral facial stimuli in addition to angry and happy faces.
The size and brightness of the facial stimuli used in Experiment 2
were the same as what was used in Experiment 1.
3.1.4. Procedure
All 16 participants completed the change-detection task.
2
Figure 3(a) depicts examples of the change-detection task used in
Experiment 2. The changes in the experimental settings, compared
with Experiment 1B, were that we used a single- and whole-probe
paradigm in the test display, we presented experimental facial
expressions (i.e., angry or happy faces) simultaneously with the neu-
tral faces, and the number of presented stimuli was fixed at only one
set size of 8.
3
The experiment comprised 144 trials (condition of change in
facial expression [2; same and different] facial expression [2;
anger and happiness] probe [2; single and whole] 18 trials).
Participants completed 20 practice trials before the main session.
The order of the conditions was randomized for each participant.
3.2. Results
Figure 3 shows the calculated d-prime scores. We conducted a
three-way repeated measures ANOVA with the TAS-20 score (2;
High and Low TAS) as between-participants factor and facial
expression (2; anger and happiness) and probe (2; single and
whole) as the within-participants factors. We did not observe
any main effect [TAS-20 score: F(1, 14) = 1.26, p= .28,
g
p
2
= .08;
facial expression: F(1, 14) = 0.02, p= .89,
g
p
2
= .001 probe: F(1,
14) = 0.004, p= .95,
g
p
2
= .0002]. In addition, there were no interac-
tions between TAS-20 score and facial expression [F(1, 14) = 2.21,
p= .17,
g
p
2
= .13], between TAS-20 score and probe [F(1,
14) = 2.04, p= .18,
g
p
2
= .13], or between TAS-20 score, facial
expression, and probe [F(1, 14) = 0.015, p= .90,
g
p
2
= .001].
3.3. Discussion
In Experiment 2, we further examined the effect of alexithymia
on memory processing of facial expressions via the change-detec-
tion task. Specifically, we adopted the single-probe paradigm in the
change-detection task and tested whether the effects of alex-
ithymia were observed in the retrieval or storage stage of the
VSTM process. We did not observe any significant differences
between the single-probe and whole-probe conditions. This
indicates that the effects of alexithymia might be observed in the
storage stage of the VSTM process but not in the retrieval stage.
4. General discussion
The present study assessed the differences in effects of facial
expressions (anger and happiness) on VSTM based on the partici-
pants’ TAS-20 scores (Bagby et al., 1994a, 1994b). Specifically, we
focused on VSTM processes, such as encoding, storage, and
(a) (b)
Fig. 3. Stimulus sequences and results in the change-detection task (Experiment 2). (a): This example shows the same trial within the single-probe condition. In this trial,
participants responded correctly if they could judge that the facial expression (the angry face) in the test display was the same as the facial expression in the memory display.
(b): Sensitivity (d-prime) of Experiment 2 for the High and Low TAS groups (each n= 8). Error bars denote the standard error (SE) of the mean.
2
We did not conduct the SPM in Experiment 2, as we showed no difference in SPM
score between people with high- and low-alexithymia in Experiment 1. A previous
study (Mikolajczak & Luminet, 2006, their unpublished data) also indicated no
differences in SPM score between people with and without alexithymia. Considering
these findings, we presumed that the nonverbal abilities between high- and low-
alexithymia individuals were similar.
3
In Experiment 1B, we showed individual differences in sensitivity between high-
and low-alexithymia only for set size of 8. Thus, in Experiment 2, we used a set size of
8 only, and manipulated whole- and single-probe in the change-detection task
instead.
J. Takahashi et al. / Personality and Individual Differences 83 (2015) 128–135 133
retrieval, and we examined at which stage we could observe
individual differences related to alexithymia. High and Low TAS
groups participated in the visual search task (Experiment 1A) and
the change-detection task (Experiments 1B), which are related to
the encoding and storage stages of the VSTM process, respectively.
In Experiments 1A and 1B, we found individual differences in the
change-detection task (Experiment 1B) but not in the visual search
task (Experiment 1A). In Experiment 1B, the High TAS group
showed less sensitivity to the happy faces compared to the angry
faces, while the sensitivity of the Low TAS group was similar
across facial expressions. Experiment 2 examined whether individ-
ual differences were observed at the storage or retrieval stages of
the VSTM process by using a single-probe paradigm (Wheeler &
Treisman, 2002). There was no single-probe effect on sensitivity
to facial expressions, suggesting that individual differences might
be observed at the storage stage of the VSTM process. We found
that alexithymia’s effects on recognition of facial expressions were
seen at the storage stage of the VSTM process, during which the
High TAS group exhibited dysfunctional processing for the happy
faces but not the angry faces.
Considering the results of the SPM (Raven et al., 1998), which
compared the nonverbal abilities of the High and Low TAS groups,
we found that their abilities were similar. Thus, the results we
describe below were likely associated with alexithymia and not
caused by differences in nonverbal abilities.
In the visual search task (Experiment 1A), we observed a threat
advantage (Öhman et al., 2001). The response times to the angry
faces were significantly shorter than to the happy faces. This result
is consistent with previous studies (Barratt & Bundesen, 2012;
Eastwood et al., 2001; Eastwood et al., 2003) that also used sche-
matic faces as experimental stimuli. Thus, we conclude that with
our schematic face stimuli, we could sufficiently replicate previous
findings. However, we observed no differences in response times
between the two groups. In the change-detection task
(Experiment 1B), we found differences in sensitivity between the
two groups. The Low TAS group showed similar identification accu-
racy to both angry and happy faces. The High TAS group showed
significantly less sensitivity to the happy faces than they did to
the angry faces. In addition, the High TAS group’s sensitivity to
happiness was significantly lower than the Low TAS group,
although there were no differences in sensitivity to angry faces
between the two groups. This indicates that the High TAS group
had a specific impairment in memorizing the happy faces.
Considering the results of Experiments 1A and 1B, individual dif-
ferences were observed in the change-detection task but not the
visual search task. The visual search task focused on detecting
facial stimuli, whereas the change-detection task required tempor-
ary memory storage. Thus, individual differences were observed at
the storage stage of the VSTM process.
This possibility was further examined in Experiment 2. We
tested at which stage of the VSTM process (i.e., storage or retrieval)
our present effects could be located. To examine this, we used the
single-probe paradigm (Wheeler & Treisman, 2002), which com-
pares sensitivity between the single-probe condition with rela-
tively low cognitive load of memory retrieval and the whole-
probe condition with relatively high cognitive load of memory
retrieval. If the single-probe effect were found, we could conclude
that the retrieval stage of the VSTM process might be largely
responsible for our present results. However, the sensitivity of
the single-probe condition was similar to that of the whole-probe
condition in Experiment 2. Thus, we assume that our present dif-
ferences represent processes happening at the storage stage. The
difficulty of the High TAS group in processing the happy faces
occurred at the temporary storage stage of the VSTM process.
Several previous studies have indicated that individuals with
alexithymia tend to be less accurate in processing angry faces
(Vermeulen et al., 2006; Vermeulen et al., 2008) and other negative
faces, such as afraid and sad faces (Prkachin et al., 2009), whereas
there are no accuracy differences between individuals with and
without alexithymia in processing of happy faces. Based on these
previous studies, we predicted that our High TAS group might
show lower performance for the angry but not for the happy faces
in the visual search and change-detection tasks. However, our High
TAS group showed impairment only in the VSTM for the happy
faces. We explain potential reasons for this disagreement
between our results and those of previous studies in the following
paragraphs.
Vermeulen et al. (2006) used an affective priming paradigm and
showed a decrease in automatic processing of angry faces on
evaluative judgments. Later, Vermeulen et al. (2008) used an odd-
ball task with ERP measurements to examine the categorical per-
ception of facial expressions and to show that latencies of N2b
and P3a, which may reflect the degree of attention devoted to
specific information, were delayed in highly alexithymic individu-
als but not in individuals with low alexithymia. However, they
used only negative faces, such as angry and disgusted faces, and
no positive faces, such as happy faces. Prkachin et al. (2009) used
a memory task with a relatively long delay and showed that
individuals with high alexithymia showed lower sensitivity to
angry, afraid, or sad faces compared to individuals with low alex-
ithymia. These studies examined either levels of evaluative judg-
ment or long-term memory. On the other hand, our study used
the visual search and change-detection tasks to examine VSTM
levels. We assumed that this would create different results
between our study and those of previous studies.
Moreover, a discussion of memory processing levels may be
applicable to our speculations regarding the threat advantage
hypothesis. In the threat advantage hypothesis, angry faces are
particularly effective at capturing attention that enhances the
encoding and/or the temporary storage of angry faces (e.g.,
Eastwood et al., 2003). In addition, Jackson et al. (2009) discussed
the possibility that angry faces may be stored in VSTM. Considering
the nature of response adaptation, people have a greater need to
immediately identify an angry person in their online memories
(i.e., VSTM) in order to decide whether to approach or avoid that
person. Thus, it is possible that threat situations such as angry
faces may capture effective attention through VSTM. In our visual
search task, response times to the angry faces were shorter than
to the happy faces and there were no individual differences. This
suggests that both groups deployed sufficient attentional capacity
for the angry faces, thus confirming the threat advantage. In our
change-detection task, sensitivity to the angry faces was similar
between two groups because they both effectively attended to
the angry faces. Although the Low TAS group effectively attended
to the happy faces in addition to the angry faces, the High TAS
group could not afford the capacity to deal with the happy faces
because of their dysfunction in processing them. In order to suc-
cessfully adapt to potentially threatening social situations, there
is greater need to retain and rapidly identify angry faces rather
than happy faces to effectively decide whether to approach or
avoid these people. Considering these situations, it may not be
necessary to detect and memorize happy faces. These speculations
must be considered carefully when planning future research on
this issue. Given our present results, it would be premature to
make any strong assumptions concerning the exact mechanisms
involved in emotional facial processing of alexithymic individuals.
Considering the present results, we further attempted to infer
day-to-day functioning of people with alexithymia. We found that
individuals with alexithymia showed impairment in storage of
happy faces but not in retrieval, suggesting that alexithymia may
often fail to maintain recovery of information that was successfully
stored. This characteristic may decrease the quality of their social
134 J. Takahashi et al. / Personality and Individual Differences 83 (2015) 128–135
communication and the chance to form close human relationship
in social situations.
The present study has some limitations: First, we used the fac-
tor of alexithymia as a nominal variable, adopting the experimen-
tal method used by previous studies (i.e., cutoff score of 61 or more
to define individuals with alexithymia). However, we should con-
sider treating the TAS-20 score as a continuous variable.
Regarding other disorders such as autism, a recent study has sug-
gested a cognitive and behavioral continuity between groups with
typical development, Asperger’s syndrome, and autism, which are
called ‘‘spectrum’’ and measured by the autism-spectrum quotient
(Baron-Cohen, Wheelwright, Skinner, Martin, & Clubley, 2001).
Thus, future research will have to confirm the repeatability of
our present results by treating the factor of alexithymia as a con-
tinuous variable, that is, using regression analysis or other sta-
tistical methods (see also a review article, Grynberg et al., 2012).
Second, our present study used schematic faces as experimental
stimuli because we wanted to remove low-level perceptual arti-
facts, which might be included in photographic faces (Purcell
et al., 1996). By using these stimuli, we could extract the factor
of alexithymia in relation to recognition of facial expression. It is
natural to assume that socio-affective intensity (or reality) is stron-
ger for photographic faces than for schematic faces. Future
research should re-examine our results by using photographic
faces.
In conclusion, we showed that the High TAS group had less
sensitivity for processing of happy faces during the storage stage
of the VSTM process, but showed similar sensitivity as the Low
TAS group in the VSTM of angry faces. Based on the threat advan-
tage hypothesis, we assume that the High TAS group could effec-
tively memorize angry (threatening) faces, while they were less
efficient in memorizing happy faces due to impairments in pro-
cessing. From these accounts, our present study suggests that the
High TAS group shows specific dysfunctions in the processing of
happy faces within the storage stage of the VSTM process.
Conflicts of interest
There are no conflicts of interest.
References
American Psychiatric Association. (2000). Diagnostic and statistical manual of mental
disorders (4th ed., text revision). Washington, DC: APA.
American Psychiatric Association (2013). Diagnostic and statistical manual of mental
disorders (5th ed.). Washington, DC: APA.
Bagby, R. M., Parker, J. D., & Taylor, G. J. (1994a). The twenty-item Toronto
Alexithymia Scale-I. Item selection and cross-validation of the factor structure.
Journal of Psychosomatic Research, 38, 23–32.
Bagby, R. M., Taylor, G. J., & Parker, J. D. (1994b). The twenty-item Toronto
Alexithymia Scale-II. Convergent, discriminant, and concurrent validity. Journal
of Psychosomatic Research, 38, 33–40.
Baron-Cohen, S., Wheelwright, S., Skinner, R., Martin, J., & Clubley, E. (2001). The
Autism-spectrum Quotient (AQ): evidence from Asperger Syndrome/high-
functioning Autism, males and females, scientists and mathematicians.
Journal of Autism and Developmental Disorders, 31, 5–17.
Barratt, D., & Bundesen, C. (2012). Attentional capture by emotional faces is
contingent on attentional control settings. Cognition and Emotion, 26,
1223–1237.
Berthoz, S., Artiges, E., Van de Moortele, P.-F., Poline, J.-B., Rouquette, S., Consoil, S.
M., et al. (2002). Effect of impaired recognition and expression of emotions on
frontcingulate cortices: an fMRI study of men with alexithymia. The American
Journal of Psychiatry, 159, 961–967.
D’Argembeau, A., & Van der Linden, M. (2007). Facial expressions of emotion
influence memory for facial identity in an automatic way. Emotion, 7, 507–515.
DiStefano, R. A., & Koven, N. S. (2012). Dysfunctional emotion processing may
explain visual memory deficits in alexithymia. Personality and Individual
Differences, 52, 611–615.
Eastwood, J. D., Smilek, D., & Merikle, P. M. (2001). Differential attentional guidance
by unattended faces expressing positive and negative emotion. Perception &
Psychophysics, 63, 1004–1013.
Eastwood, J. D., Smilek, D., & Merikle, P. M. (2003). Negative facial expression
captures attention and disrupts performance. Perception & Psychophysics, 65,
352–358.
Ekman, P., & Friesen, W. V. (1975). Unmasking the face: A guide to recognizing the
emotions from facial cues. Englewood Cliffs, NJ: Prentice Hall.
Fitzgerald, M., & Bellgrove, M. A. (2006). The overlap between Alexithymia and
Asperger’s syndrome. Journal of Autism and Developmental Disorders, 36,
573–576.
Franz, M., Schaefer, R., Schneider, C., Sitte, W., & Bachor, J. (2004). Visual event-
related potentials in subjects with alexithymia: Modified processing of
emotional aversive information? The American Journal of Psychiatry, 161,
728–735.
Fukunishi, I., Nakagawa, T., Nakamura, H., Kikuchi, M., & Takubo, M. (1997). Is
alexithymia a culture-bound construct? Validity and reliability of the Japanese
versions of the 20-item Toronto Alexithymia Scale and modified Beth Israel
Hospital Psychosomatic Questionnaire. Psychological Reports, 80, 787–799.
Green, D., & Swets, J. A. (1966). Signal detection theory and psychophysics. New York:
John Wiley and Sons.
Grynberg, D., Chang, B., Corneille, O., Maurage, P., Vermeulen, N., Berthoz, S., et al.
(2012). Alexithymia and the processing of emotional facial expressions (EFEs):
Systematic review, unanswered questions and further perspectives. PLoS One, 7,
1–20.
Grynberg, D., Vermeulen, N., & Luminet, O. (2014). Amplification of attentional blink
by distress-related facial expressions: Relationships with alexithymia and
affectivity. International Journal of Psychology, 49, 371–380.
Jackson, M. C., Wu, C.-Y., Linden, D. E. J., & Raymond, J. E. (2009). Enhanced visual
short-term memory for angry faces. Journal of Experimental Psychology: Human
Perception and Performance, 35, 363–374.
Kano, M., Fukudo, S., Gyoba, J., Kamachi, M., Tagawa, M., Mochizuki, H., et al. (2003).
Specific brain processing of facial expressions in people with alexithymia: An
H215O-PET study. Brain, 126, 1474–1484.
Mantani, T., Okamoto, Y., Shirao, N., Okada, G., & Yamawaki, S. (2005). Reduced
activation of posterior cingulate cortex during imagery in subjects with high
degrees of alexithymia: A functional magnetic resonance imaging study.
Biological Psychiatry, 57, 982–990.
McDonald, P. W., & Prkachin, K. M. (1990). The expression and perception of facial
emotion in alexithymia: A pilot study. Psychosomatic Medicine, 52, 199–210.
Mikolajczak, M., & Luminet, O. (2006). Is alexithymia affected by situational stress
or is it a stable trait related to emotion regulation? Personality and Individual
Differences, 40, 1399–1408.
Öhman, A., Lundqvist, D., & Esteves, F. (2001). The face in the crowd revisited: A
threat advantage with schematic stimuli. Journal of Personality and Social
Psychology, 80, 381–396.
Pandey, R., & Mandal, M. K. (1997). Processing of facial expressions of emotional and
alexithymia. British Journal of Clinical Psychology, 36, 631–633.
Parker, J. D. A., Taylor, G. J., & Bagby, R. M. (1993). Alexithymia and the recognition
of facial expressions of emotion. Psychotherapy and Psychosomatics, 59,
197–202.
Parker, J. D. A., Taylor, G. J., & Bagby, R. M. (2003). The 20-item Toronto Alexithymia
Scale III. Reliability and factorial validity in a community population. Journal of
Psychosomatic Research, 55, 269–275.
Prkachin, G. C., Casey, C., & Prkachin, K. M. (2009). Alexithymia and perception of
facial expressions of emotion. Personality and Individual Differences, 46,
412–417.
Purcell, D. G., Stewart, A. L., & Skov, R. B. (1996). It takes a confounded face to pop
out of a crowd. Perception, 25, 1091–1108.
Raven, J., Raven, J. C., & Court, J. H. (1998). Raven manual: Section 3. Standard
progressive matrices. Oxford, England: Oxford Psychologists Press.
Sifneos, P. E. (1973). The prevalence of ‘‘alexithymic’’ characteristics in
psychosomatic patients. Psychotherapy and psychosomatics, 22, 255–262.
Szatmari, P., Georgiades, S., Duku, E., Zwaigenbaum, L., Goldberg, J., & Bennett, T.
(2008). Alexithymia in parents of children with autism spectrum disorder.
Journal of Autism and Developmental Disorders, 38, 1859–1865.
Taylor, G. J., Bagby, R. M., & Parker, J. D. A. (1997). Disorders of affect regulation.
Alexithymia in medical and psychiatric illness. Cambridge, UK: Cambridge
University Press.
Vermeulen, N., & Luminet, O. (2009). Alexithymia factors and memory
performances for neutral and emotional words. Personality and Individual
Differences, 47, 305–309.
Vermeulen, N., Luminet, O., & Corneille, O. (2006). Alexithymia and the automatic
processing of affective information: Evidence from the affective priming
paradigm. Cognition and Emotion, 20, 64–91.
Vermeulen, N., Luminet, O., de Sousa, M. C., & Campanella, S. (2008). Categorical
perception of anger is disrupted in alexithymia: Evidence from a visual ERP
study. Cognition and Emotion, 22, 1052–1067.
Wheeler, M. E., & Treisman, A. M. (2002). Binding in short-term visual memory.
Journal of Experimental Psychology: General, 131, 48–64.
Xu, Y., & Chun, M. M. (2006). Dissociable neural mechanisms supporting visual
short-term memory for objects. Nature, 440, 91–95.
J. Takahashi et al. / Personality and Individual Differences 83 (2015) 128–135 135
