Avatar Assistant: Improving Social Skills in Students with an ASD
Through a Computer-Based Intervention
Ingrid Maria Hopkins•Michael W. Gower•
Trista A. Perez•Dana S. Smith•Franklin R. Amthor•
F. Casey Wimsatt•Fred J. Biasini
Published online: 2 February 2011
? Springer Science+Business Media, LLC 2011
computer-based social skills training program for children
with Autism Spectrum Disorders (ASD). This randomized
controlled study (N = 49) indicates that providing children
with low-functioning autism (LFA) and high functioning
discriminating facial expressions and recognizing faces and
emotions in FaceSay’s structured environment with inter-
active, realistic avatar assistants improved their social skills
abilities. The children with LFA demonstrated improve-
ments in two areas of the intervention: emotion recognition
improvements in all three areas: facial recognition, emotion
recognition, and social interactions. These findings, partic-
ularly the measured improvements to social interactions in a
natural environment, are encouraging.
This study assessed the efficacy of FaceSay, a
Facial recognition ? Social interactions ? Generalization
Autism ? Intervention ? Emotion recognition ?
The development of social competence is an important goal
for all children. However, some children are at risk for
challenges in social competence due to deficits in social
skills (Gresham 1981). Children with Autism Spectrum
Disorders (ASD) are particularly affected by this impair-
ment as evidenced by their difficulties in reciprocal social
interaction skills (American Psychiatric Association 1994).
Individuals with ASD display marked impairments in the
use and interpretation of nonverbal behaviors, such as eye-
to-eye gaze, facial expressions, body posture, and gestures
to regulate social interaction (Hobson 1986). In addition,
individuals with ASD often fail to monitor the effect of
their conversations or behaviors on other people. For
example, they frequently monopolize conversations or
walk away while others are trying to interact with them
(Bailey 2001). This may be the result of an inability to
interpret nonverbal communications provided in facial
expressions and body posture.
The ability to recognize emotions in others is a crucial
component of social development (Hobson 1986). Impair-
ment in this skill severely reduces one’s ability to partici-
pate in or interpret social interactions. In order to establish
the extent to which emotion recognition skills develop in
children with developmental disabilities, previous research
has tended to focus on children with ASD for whom
impaired social development is seen as a key feature.
Differences have been found between children with ASD
and typically developing children. For example, Baron-
Cohen et al. (1997) developed a theory of mind task where
the participants looked at photos of either an entire face or
just a region around the eyes to determine emotions. The
researchers found that individuals with autism performed
significantly worse on the task than age and IQ matched
participants without autism. In addition, individuals with
autism had marked deficits in their performance in the
eyes-only condition. These results support the idea that
emotion recognition deficits are key characteristics in
understanding the social deficits of individuals with ASD.
I. M. Hopkins (&) ? M. W. Gower ? T. A. Perez ?
D. S. Smith ? F. R. Amthor ? F. J. Biasini
Department of Psychology, University of Alabama
at Birmingham, 1300 University Blvd., CH 328,
Birmingham, AL 35924-1170, USA
F. Casey Wimsatt
Symbionica, LLC, San Jose, CA, USA
J Autism Dev Disord (2011) 41:1543–1555
Emotion processing deficits may not be specific to
emotions, but may reflect more general information pro-
cessing deficits. For example, difficulties in processing
emotional expressions in ASD may stem from more gen-
eral difficulties in processing faces (Boucher and Lewis
1991) or still more general perceptual difficulties related to
the processing of relational information (Shah and Frith
1993). Davies et al. (1994) performed two experiments
designed to tease apart these various possibilities. Their
results support the conclusion that emotion processing and
face processing deficits in ASD can be attributed to more
general impairment in processing of relational information.
Participants in their study showed similar deficits in pro-
cessing emotional, facial, and nonfacial stimuli.
Faces are remarkably homogenous as a class of visual
stimuli in that they share a highly similar structure, always
consisting of the same set of parts (e.g. eyes, nose, and
mouth) in the same basic configuration (e.g. nose centered
below the eyes and above the mouth). Yet, despite this
basic similarity, most people can easily recognize and
discriminate among hundreds of faces. This ease with
which humans are able to distinguish between faces has
been contributed to a holistic perceptual and encoding
process (Bartlett and Searcy 1993; Mundy 2003; Rhodes
Individuals with ASD, however, appear to process faces
by relying more on local facial features than on holistic
facial configuration (i.e. the relationship between the dif-
ferent parts). For example, persons with autism show much
less of an inversion effect (i.e. their performance was not
impaired for recognition of upside down faces versus
upright faces) compared to controls when presented with
inverted faces (Langdell 1978). The lack of inversion effect
is taken as a sign of a local rather than a configural or
global processing of faces, since local information is most
relevant in identifying inverted faces.
People with ASD can discriminate between faces
(Ozonoff et al. 1990). However, as the demands are
increased, or elements of emotion are included, perfor-
mance is impaired for individuals with ASD (Davies et al.
1994). Joseph and Tanaka (2003) found evidence of
holistic strategy use by children with autism, but, unex-
pectedly, only when face recognition depended on the
mouth. Marked deficiencies in recognition occur when
identification is dependent on the eyes. Hopkins and Bia-
sini (2005) also found that children with autism recognized
schematic face stimuli by relying on lower facial features
only. The same conclusion was drawn when using photo-
graphs of faces. The Joseph and Tanaka (2003) finding is
consistent with Hopkins and Biasini (2005) study: children
with autism focus on the mouth when attending to faces
and are poor at recognition involving eyes.
The Current Study
Recent population statistics suggest that autism is the
fastest growing developmental disability in the United
States, with an estimated annual cost for diagnosis, and
treatment of $90 billion (Autism Society of America 2003).
This growth, combined with the likely long-term implica-
tions, necessitates an immediate effort to improve social
skills among children with ASD. Any technology that
could teach individuals with ASD necessary social skills
would not only be invaluable for the individuals affected,
but would also result in financial savings. Such a technol-
ogy appears to exist in the rudimentary form as avatars,
computer embodied virtual people that have a knowledge
base and the ability to converse with humans in natural
Computers are promising teaching instruments for
children with ASD (e.g. Chen and Bernard-Opitz 1993;
Colby 1973; Faja et al. 2008; Panyan 1984). Multisensory
interactions, controlled and structured environments, use of
multilevel interactive functions, and the ability to individ-
ualize instruction are some of the features that can assist
children with ASD when working with computers. Those
functions have been found to be successful for various
computer-based interventions (e.g. Bernard-Opitz 1989;
Panyan 1984; Yamamoto and Miya 1999). For instance,
Chen and Bernard-Opitz (1993) found that 3 of 4 students
with autism exhibited higher motivation to learn and fewer
disruptive behaviors using computer-based instruction
compared to more traditional personal instruction. Heiman
et al. (1995) found that an interactive environment pro-
vided by a computer enhanced the reading and writing
skills of children with ASD. ‘‘Baldi’’ was developed as a
three-dimensional computer-animated talking head (ava-
tar). It provides realistic visible speech that is almost as
accurate as a natural speaker. Computer-based training
programs using Baldi to carry out language tutoring for
children have been found to facilitate language learning in
children with ASD (Bosseler and Massaro 2003). Finally,
Faja et al. (2008) found that computer-based face training
can affect processing of faces. Faja and colleagues found
that 8 h of computerized face training resulted in sensi-
tivity to holistic processing. However, no difference was
found in face recognition, possibly because the face
training was two-dimensional.
Several published studies have evaluated the efficacy of
computer or DVD video based interventions to enhance the
social skills abilities of children with ASD (Bernard-Opitz
et al. 2001; Silver and Oakes 2001; Bo ¨lte et al. 2002; Golan
and Baron-Cohen 2006; LaCava et al. 2007; Golan et al.
1544 J Autism Dev Disord (2011) 41:1543–1555
2010; Tanaka et al. 2010; Whalen et al. 2010; Lacava
et al. 2010). These technology-based social skills inter-
ventions all leverage the affinity of students with an ASD
for the predictable and animated environment of a com-
puter game or video (Goldsmith and LeBlanc 2004).
These interventions typically do not offer the individual
an opportunity for more real life like interactions. Unlike
these interventions, FaceSay uses interactive, realistic
Generalization, the ability to transfer a learned behavior
acquired during a training activity to another similar or
related activity or situation, is considered a difficult task for
children with an ASD (Koegel et al. 2001). Some of the
studies with children with an ASD found no generalization
across tasks (Bo ¨lte et al. 2002; Silver and Oakes 2001) or
only near generalization (Golan and Baron-Cohen 2006).
Other studies have measured distant generalization from
one setting to the next (Bernard-Opitz et al. 2001; Golan
et al. 2010). While other studies mention anecdotal reports
of generalization to real settings (Golan and Baron-Cohen
2006; LaCava, et al. 2007; Golan et al. 2010; Whalen et al.
2010; Lacava et al. 2010) or limited generalization to real
settings in unpublished laboratory findings (Bernard-Opitz
et al. 2001). None of these studies have shown social skills
generalization to a real setting and several call for further
research to investigate it (Golan and Baron-Cohen 2006;
LaCava et al. 2007; Golan et al. 2010; Tanaka et al. 2010;
Whalen et al. 2010).
The challenge of generalization to real settings has also
been reported for non-computer based social skills inter-
ventions. A review of studies of social skills group inter-
ventions, for example, concludes, ‘‘there is evidence that
skills may be displayed in laboratory/clinic settings, but not
necessarily applied in the child’s daily life at school or
home. Generalization and flexible skill use in natural
environments continues to be a challenge.’’ (Williams
White et al. 2007).
The overall purpose of the current study was to evaluate
FaceSay, a new computer based social skills intervention
designed and created especially for this study by Symbio-
nica, LLC, and its use of interactive, realistic avatar
assistants for social skills learning in children with ASD,
including FaceSays’s potential for improving the children’s
social interactions in natural environments. Previous
computer based approaches to teaching face and emotion
recognition skills to persons with an ASD were mostly
static response based software (e.g., Tanaka et al. 2003).
FaceSay, however, uses an interactive approach with
computer animated avatars, both humans and animals, to
create a more life-like software program to teach face and
emotion recognition skills. FaceSay also includes an
appealing predictability (Goldsmith and LeBlanc 2004)
and restricted field of focus (Corbett and Abdullah 2005)
made possible through computer technology using inter-
active video-realistic avatars.
FaceSay tries to incorporate activities to address known
challenges for persons with an ASD. For example, in the
first of the three games the goal was to improve joint
attention skills by creating an interactive problem solving
task focusing on tracking the eyes of the avatar to respond
to the avatar’s request. The other two games are based, in
part, on the idea that persons with an ASD have weak
central coherence (Frith 1989). One of the goals was to
increase their abililty to be more global lookers when
interpreting expressions of emotions and discriminating
faces. These games are described in more detail in the
Procedures section of this paper.
Aims and Predictions
effect of FaceSay on children’s emotion and facial recog-
nition skill development. It was predicted that all children
who participated in the training program would attain
improved emotion recognition skills following the inter-
vention (Hypothesis 1). Also, given the encouraging results
of a computer-based social skills training (e.g. Bernard-
Opitz et al. 2001), it was predicted that children with ASD
who participated in the training would have improved
facial recognition skills following the training program
The primary aim of this study was to examine the
intervention on social behaviors in the natural environment.
It was predicted that children with ASD who received the
intervention would demonstrate improvements in observed
and reported social skills following the intervention
(Hypothesis 3). This prediction was based on the premise
that the avatar based games in FaceSay would improve the
face and emotion recognition skills of the participants
leading to more appropriate and less dysfunctional social
interactions related to improvements in the recognition of
nonverbal facial communication. The improved face and
emotion recognition skills should create a greater aware-
ness of the social value of facial features, particularly the
eyes, thereby increasing interest in and decreasing social
confusion in real world social interactions.
The study also investigated the impact of the
The protocol for the current study was approved by the
institutional review board of the University of Alabama at
J Autism Dev Disord (2011) 41:1543–15551545
Birmingham. Participants were recruited from several
sources in central Alabama. All parents gave written
informed consent after study procedures were fully
explained. Formal child assent was obtained from children
whose mental age was above seven years. Before the study
began, a meeting with each parent and child was held to
give an overview of the study’s procedures, explain the
importance of regular attendance, and to answer any
questions the parent or child had.
Fifty-one children were initially enrolled in the study.
However, two children were excluded from the study due
to low attendance rates (one student moved and one student
was hospitalized for an extended period). Forty-nine chil-
dren with LFA or HFA completed the project. All the
children had previously received a diagnosis of an ASD
according to the criteria specified by the Diagnostic and
Statistical Manual of Mental Disorders (DSM-IV; Ameri-
can Psychiatric Association 1994) by a licensed commu-
nity professional. To insure that all the participants met this
criteria the investigators confirmed the ASD diagnosis with
an administration of the CARS (M = 37.14). Due to the
variability in the children’s functioning, they were grouped
into High Functioning Autism (HFA; KBIT greater than
70; n = 24) and Low Functioning Autism (LFA; KBIT less
than 70; n = 25). Children who attended greater than 83%
of the sessions were included in the analyses.
Children with a mental age between 6 and 10 years were
recruited from several sources located in a large geographic
area in a southeastern state in the USA. Among the chil-
dren in the sample, 5 were girls and 44 were boys, ranging
in age from 6 years, 3 month to 15 years, 1 month
(M = 10.17). Possible demographic differences between
the intervention and control groups of children with LFA
and HFA were tested using ANOVA or Chi square. No
significant differences were found between the groups.
Tables 1 and 2 provide a listing of the demographic char-
acteristics for all of the children.
Design and Instruments
The current study involved a 2 (Training) 9 2 (Group) 9 2
(Time) mixed factorial design. The within factor was Time,
which had two levels (baseline and post-intervention). The
between factors were Group, and it had two levels (LFA and
HFA) and Training, and it had two levels (Avatar Training
Kaufman Brief Intelligence Test
Cognitive functioning for all groups was obtained using the
Kaufman Brief Intelligence Test, Second Edition (KBIT,
II) (Kaufman and Kaufman 1990). The KBIT, II assesses
general cognitive abilities and generates verbal, non-ver-
bal, and composite domain scores. The KBIT, II correlates
highlywith the Wechsler
(r = 0.84). The KBIT, II has demonstrated an internal
consistency coefficient of 0.92 and test–retest reliability
coefficients of 0.90. (Kaufman and Kaufman 1990).
Childhood Autism Rating Scale
The Childhood Autism Rating Scale (CARS) was used to
reconfirm the previous diagnosis of ASD. The CARS is a
widely used instrument developed to distinguish children
with autism from those with other developmental disabil-
ities or normal functioning (Schopler et al. 1988). The
CARS has been shown to have high reliability (0.81) and
internal consistency (0.94) (Perry et al. 2005). Criterion-
related validity has been determined by correlating CARS
diagnoses to diagnoses made independently by child psy-
chologists and psychiatrists (r = 0.80). The CARS has also
been shown to have 100% predictive accuracy when dis-
tinguishing between groups of children with autism and
children with mental retardation (Teal and Wiebe 1986).
Both photographs and schematic drawings were used to
measure children’s ability to recognize emotional expres-
sions before and after the training. The photographs con-
sisted of six black and white pictures that illustrate a
woman with six different emotional expressions (anger,
disgust, fear, happiness, sadness, and surprise). They were
selected from Ekman and Friesen’s (1975) photos of faces
of emotional expressions. The photographs are contained in
Unmasking the Face (Ekman and Friesen 1975) and have
been used in previous studies (e.g. Hopkins and Biasini
2005; Sullivan 1996). The schematic drawings were
essentially versions of a happy or smiley face that were
designed to depict the six emotions and were validated by
groups of undergraduate students. The drawing had to be
correctly identified by at least 95% of the undergraduate
students to be included in the array. The photo measure has
been found to have strong reliability (0.89–0.91) and
validity (0.71–0.86; Ekman and Friesen 1975).
The children in the current study were presented with
the arrays of photographs and schematic drawings of
emotions. Each array consisted of six faces. As they were
presented the children heard a label (angry, disgusted,
scared, happy, sad, or surprised) and were asked to match
the emotion label by touching the appropriate photo or
drawing. Presentation of the photo or schematic drawing
sets was counterbalanced. The order of the six photo-
graphs anddrawings withinthearrays werealso
1546J Autism Dev Disord (2011) 41:1543–1555
counterbalanced. The child’s emotion recognition score
was the total correct responses across the 12 trials of
photos and drawings.
To measure the children’s facial recognition skills, the
Benton Facial Recognition Test (Short Form) was admin-
istrated (Benton 1980). The Benton test is a standardized
27-item task for assessing the ability to identify and dis-
criminate photographs of unfamiliar human faces. The test
has been normed for children and adults. The internal-
consistency reliability has been found to be 0.71, with a
test–retest reliability of 0.66. Male and female faces are
used, and the faces are closely cropped so that no clothing
and little hair are visible.
The children completed the Benton test before and after
the training. The faces are centered within a black back-
ground, and the entire image is 6.5 cm by 6.5 cm. For the
first six items, only one of the six test faces displays the
target individual, and the target image and the test image
are identical. In the next seven items, three of the test faces
match the target face, and the poses for the test images are
different from the target image. No time limits are placed
on individual items or the test as a whole. On each item, the
children were presented with a target face above six test
faces, and they were asked to indicate which of the six
images matched the target face. For the purpose of this
study, the task was slightly modified in that the target face
was presented as a card instead of as a photograph in the
stimulus booklet. The children were asked to match the
card with the correct test image in the booklet.
Social Skills Rating System
The Social Skills Rating System (SSRS) (Gresham and
Elliott 1990) was administered as both a pre- and posttest
measure. It is a standardized, norm-referenced 38-item
parent-report questionnaire that measures a wide range of
social skills, including the broad domains of cooperation,
assertion, responsibility, and self-control. The SSRS has
been found to have an internal-consistency of 0.87-0.90
and test re-test reliability of 0.87 (Gresham and Elliott
1990). Children with ASD have been shown to have
impairments on the SSRS when compared to the normative
sample (Gresham, 1981). The total score on the 38-items
yields a Social Skills standard score, which has demon-
strated a test–retest reliability coefficient of 0.87 (Gresham
and Elliott 1990). The parents were blind to their child’s
group assignment (i.e. training or control).
Social Skills Observation
To measure children’s social skills, an observation of
each child was conducted during recess or free time. At
baseline, and following the intervention, each child was
observed for two 5-min assessment periods at a random
time during their school recess by two research assistants.
The child was observed interacting with peers on two
separate days. The two research assistants were blind to
Table 1 Frequency (percentage) for particpants’ demographic charactersitics
Variable LFA training (N = 11) LFA control (N = 14) HFA training (N = 13) HFA control (N = 11) Total (N = 49)
Male10 (90.9) 13 (92.9)12 (92.3)9 (81.8) 44 (89.8)
Female1 (9.1) 1 (7.1)1 (7.7) 2 (18.2) 5 (10.2)
African American 3 (27.3) 4 (28.6) 3 (23.1)3 (27.3) 13 (26.5)
Caucasian8 (72.7) 10 (71.4)10 (76.9)7 (63.3) 35 (71.4)
Other0 (0) 0 (0) 0 (0) 1 (9.1)1 (2.0)
Table 2 Means (standard deviations) of age, CARS, and IQ scores for all groups
Variable LFA training (N = 11) LFA control (N = 14) HFA training (N = 13) HFA control (N = 11)Total (N = 49)
Age10.31 (3.31) 10.57 (3.2)10.05 (2.30)9.85 (2.87)10.17 (3.02)
CARS38.64 (3.93) 38.92 (5.79)36.00 (5.26)35.00 (5.22)37.14 (5.24)
Verbal52.09 (16.68)50.00 (15.46)92. 05 (18.63)93.09 (21.91)74.51 (25.59)
Nonverbal 59.00 (23.54)58.38 (19.03)91.76 (20.98)93.81 (26.05)78.86 (28.94)
Composite55.09 (20.91)54.79 (16.41) 91.88 (19.54)93.00 (25.47) 75.71 (27.34)
J Autism Dev Disord (2011) 41:1543–15551547
the group status of the participants in the study. Prior to
collecting data, the two research assistants spent a 2 hour
practice session scoring children’s social interactions until
90 percent agreement had been reached consistently
between the assistants and the investigator. Inter-rater
reliability was established for 100% of the coding data.
They then observed the social interactions and coded the
interactions on a rating system that has been developed
and used in previous studies (Hauck et al. 1995). The
items on the rating scale assess specific social skills and
produce scores for three social skills factors: positive
social interaction, low-level social interaction, and nega-
tive social interaction. Positive social interactions include
activities that exhibit verbal and nonverbal social behav-
iors that lead to an effective social process with peers.
These behaviors serve to start or maintain social inter-
actions. Low-level interactions include behaviors that
indicate social intention but with minimal social enact-
ment such as close proximity to other children without
initiating a positive social interaction. Negative social
interactions include unpleasant social behaviors that
operate to stop or decrease the likelihood of an adequate
social interaction. Each of the behaviors was coded as
present, not present, or not applicable to the situation.
The observers maintained close proximity to the chil-
dren during recess, whether in the gymnasium or outdoors;
however, they did not interact with the children and
politely rejected any overtures made towards them. Chil-
dren were told by their teachers that that the observers were
interested in learning about their play habits. Reliability for
the two observers across the 2 social skills factors was
established using Cohen’s kappa (Cohen 1960). Cohen’s
kappas for the two raters were 0.95 for positive social
interaction, 0.74 for low-level social interaction, and 0.86
for negative social interaction. According to Fleiss (1981),
kappa levels of 0.40– 0.60 are fair, 0.60–0.75 are good, and
0.75 and above are excellent. Thus, these kappa levels are
considered good to excellent.
All procedures occurred at the child’s school or after-
school facility. At the baseline assessment session, parents
of the children completed a demographic information form
and the Social Skills Rating System. The children com-
pleted the Emotion Recognition test, the Benton Facial
Recognition test, and the Kaufman Brief Intelligence Scale.
The participants were then observed interacting with other
children during recess. The research assistants completed
the CARS and the Social Skills Observation. The children
were randomly assigned to the training group or the control
group. All post-test measures were completed within
2 weeks after the final intervention session.
Fourteen participants with LFA and 11 children with HFA
were asked to use Tux Paint, open source drawing software
for children (www.tuxpaint.org), at the school with the
assistance of one or two investigator(s), twice a week for
approximately 10–25 min per session over a period of
6 weeks, a total of 12 sessions. The curriculum of the art
software (i.e. teaching painting, drawing, and coloring) was
not associated with the aims of the intervention software
(i.e. joint attention, face processing, and eye gaze). This
activity was selected to control for the amount of time
spent playing on the computer during the study as well as
social interactions with the experimenters.
Eleven participants with LFA and 13 participants with
HFA were asked to use the FaceSay software (Symbionica,
LLC, San Jose, CA) at the school with the assistance of one
or two investigators, twice a week for approximately
10–25 min per session over a period of 6 weeks, a total of
12 sessions. The decision to limit the intervention to 12
sessions was determined based on the limited time in the
school setting that was available during one semester. In
order to complete baseline measures, the intervention, and
post-intervention measures, 12 intervention sessions was
the maximum number that would allow for the completion
of the study during one school term thus minimizing
attrition and lapses in study protocol.
Before beginning the computer sessions, training ses-
sions were conducted to introduce the children to the
computer. The students learned to sit at the computer, to
listen and respond to the software, and use the mouse or to
touch the screen. Each student had the option to respond
with either an external mouse (Logitech, M-CAA42, Fre-
mont, CA) or a touch screen (KEYTEC Magic Touch,
Richardson, TX). If the students stayed in their seat and
appropriately used the mouse or touch screen their
behavior was reinforced with praise or snack foods
depending on the type of reinforcer that was recommended
by their teacher. The computer-based intervention and
control sessions began following the two training sessions
and continued for six weeks.
FaceSay is a colorful program that contains three different
games with realistic avatars designed to teach children
specific social skills. The avatars were animated photos of
real persons that could interact with the children by
drawing on a pre-programmed knowledge base. The
overall goal of the games is to promote awareness of the
1548 J Autism Dev Disord (2011) 41:1543–1555
movements and features of the face, particularly the area
around the eyes. Interactive features of the software pro-
vide opportunities for children to respond to social situa-
tions. Targeted social skills included teaching specific
social skills for responding to joint attention, particularly
eye gaze, recognizing facial expressions and recognizing
faces. The children were asked to attend to and interact
with a computer animated avatar that initiated an interac-
tion with the child and asked them to complete certain
activities that involved following an eye gaze, completing a
face puzzle, and matching and manipulating facial
expressions. For example joint attention was taught by
instructing the children to follow the avatar’s eyes to
determine what face or object the avatar was attending to.
The tasks increased in difficulty to assure that participants
of various levels could be successful as well as challenged
by the tasks. For example in the joint attention task there
are initially 4 response choices and as the children are
successful the response choices increase to 12. In the face
puzzle the children initially have 3 options to complete the
face and the number of choices is increased to 6 as they
progress through the game.
For the purpose of the current study, three games from
FaceSay were used to teach specific social skills. The
‘‘Amazing Gazing’’ game is designed to teach children to
attend to eye gaze, respond to joint attention and to
understand that eye gaze can convey intent (see Fig. 1). In
the game, the avatar is surrounded by an array of objects,
numbers, or faces. The child is asked to touch the object,
number, or face at which the avatar is gazing. If the child is
correct, the animal ‘‘coach’’ avatar at the bottom right of
the screen praises the child using the child’s name (e.g.
‘‘Good job, Johnny!’’).
A second game, ‘‘Band Aid Clinic,’’ is designed to teach
holistic facial processing and face recognition. The child is
asked to select the appropriate face ‘‘band aid’’ that would
fit over the distorted portion of the avatar’s face. The
possible matches increase in number and similarity as the
games progresses. Once reconstructed, the face ‘‘comes
alive’’ and expresses gratitude for fixing the face. (see
The third game, ‘‘Follow the Leader’’, is designed to
teach children to attend to movements in the area around
the eyes to improve their ability to discriminate facial
expressions. The game specifically emphasizes how subtle
changes in eye information, often ignored by children
with ASD, can alter the perception of the facial
In the first level, the child is asked to identify identical
facial expressions of emotions by selecting ‘‘Yes’’ for same
and ‘‘No’’ for different expressions. In an advanced level,
the child is asked to change the expression of the avatar’s
twin to match the avatar’s expression moving the eyes up
or down. This was accomplished with the touch screen or
mouse by touching or clicking a button to expand or con-
tract the eyes (see Fig. 3).
Fig. 1 Screenshots from the
‘‘Amazing Gazing’’ game
Fig. 2 Screenshots from the ‘‘Band Aid Clinic’’ game
J Autism Dev Disord (2011) 41:1543–15551549
or HFA who received the FaceSay intervention would
demonstrate increased emotion recognition skills following
the intervention as measured by their responses to the six
Ekman & Friesen photographs and the six drawings of
facial expressions. To analyze the impact FaceSay had on
emotion recognition skills, separate ANCOVAs were run
for the LFA group and the HFA group. For these analyses
the independent variable was the group (training or con-
trol). The dependent variable was the emotion recognition
post-test score, and the covariates were the emotion pre-test
score and KBIT score. Using the pre-test score and IQ
score as covariates allowed an adjustment for a difference
between the groups prior to treatment. In addition, the
resulting analysis was capable of reflecting a change in
emotion recognition skills that takes into account the
cognitive functioning of each group prior to treatment.
It was hypothesized that children with LFA
The first analysis compared the change in emotion rec-
ognition skills for the children with LFA who received
training with that of the children with LFA who did not
receive the training. There was a significant difference in
total emotion recognition skills (photos and drawings), F(1,
21) = 4.52, p\0.05 (adjusted Ms 6.53 and 5.23, respec-
tively). Also, there was a significant difference in emotion
recognition skills using only photographs as stimuli, F(1,
21) = 4.56, p\0.05 (adjusted Ms: 3.59 and 2.79,
respectively). However, there was no significant difference
in emotion recognition skills using only drawings as
stimuli, F(1, 21) = 1.64, p[0.05; (adjusted Ms: 2.95, and
The second analysis compared the children with HFA
who received training with that of the children with HFA
who did not receive the training. There was a significant
difference in total emotion recognition skills (photos and
drawings), F(1, 20) = 29.31, p\0.001 (adjusted Ms 8.70
and 6.79, respectively). Also, there were significant dif-
ferences in emotion recognition skills using only photo-
graphs as stimuli, F(1, 20) = 24.52, p\0.001 (adjusted
Ms 4.58 and 3.66, respectively), and in emotion recognition
skills using only drawings as stimuli, F(1, 20) = 15.48,
p\0.01; (adjusted Ms: 4.11, and 3.14, respectively). The
results of these two analyses indicate that there was an
overall change in emotion recognition skills using photo-
graphs as stimuli after training for the children who
received the intervention. However, only the children with
HFA who received the training improved in their ability to
interpret drawings of emotions following the intervention.
with LFA or HFA who participated in the FaceSay inter-
vention would demonstrate improved facial recognition
skills after the intervention. This was measured using the
Benton Facial Recognition Test. To analyze the impact
FaceSay had on facial recognition skills, separate ANCO-
VAs were run for the LFA group and the HFA group. The
Benton pre-test scores and the KBIT score were used as the
covariates and the Benton post-test scores as the dependent
variable. The independent variable was the group (training
or control). The results for children with LFA indicated
that there was no significant difference in their perfor-
mance on the Benton-Short Form F(1, 21) = 0.67,
p[0.05 (adjusted Ms: 14.48 and 12.84, respectively). For
children with HFA, ANCOVA results showed that the
training group and the control group differed significantly
on the Benton-Short Form F(1, 20) = 10.86, p\0.01
(adjusted Ms: 18.41 and 15.42, respectively), with the
training group having significantly higher post-test scores.
The results of these two analyses indicate that there was an
overall change in facial recognition skills for the children
with HFA who received the intervention. The analyses for
Hypotheses 1 and 2 are summarized in Table 3.
The second hypothesis was that children
or HFA who participated in the FaceSay intervention
would demonstrate improved social interaction skills in a
natural environment after the training program. To analyze
the impact FaceSay had on social interaction skills, sepa-
rate ANCOVAs were run for the LFA group and the HFA
group for their scores on the SSRS and the Social Skills
Observation. The SSRS or the Social Skills Observation
pre-test scores and the KBIT score were used as the
covariates and the SSRS or the Social Skills Observation
post-test scores as the dependent variable. The independent
variable was the group (training or control).
It was hypothesized that children with LFA
The first analyses compared the children who received
training with that of the children who did not receive the
Fig. 3 Screenshot from the ‘‘Follow the Leader’’ game
1550J Autism Dev Disord (2011) 41:1543–1555
training on their parents’ reported social skills on the
SSRS. For children with LFA, there was a significant dif-
ference in their scores on the SSRS, F(1, 21) = 14.42,
p\0.01 (adjusted Ms 64.99 and 58.51, respectively).
Also, there were significant differences in Assertion, F(1,
21) = 5.89 p\0.05 (adjusted Ms 8.06 and 5.92, respec-
tively) and Self-control, F(1, 21) = 6.00 p\0.05 (adjus-
ted Ms 7.57 and 4.91, respectively). However, there was no
significant difference in Cooperation, F(1, 21) = 2.74,
p[0.05 or in Responsibility, F(1, 21) = 2.22 p\0.05
(adjusted Ms 5.36 and 4.09, respectively).
For children with HFA who received training compared
with that of the children with HFA who did not receive the
training, a trend emerged in their parents’ reported social
skills on the SSRS, F(1, 20) = 4.36, p = 0.05 (adjusted Ms
67.77 and 62.27, respectively). However, there were no
significant differences in Cooperation, F(1, 20) = 4.04,
p[0.05, Assertion, F(1, 20) = 0.26 p[0.05 Responsi-
bility, F(1, 20) = 0.31 p[0.05, or Self-control, F(1,
20) = 0.84 p[0.05.
The second set of analyses compared the children who
received the training to that of the controls on their
observed social interactions. For children with LFA, there
was a significant difference in their total scores on the
Social Skills Observation, F(1, 21) = 5.05, p\0.05
(adjusted Ms 9.60 and 11.05, respectively), with the
training group having significantly lower post-test scores
(i.e. less inappropriate social interactions). Also, there was
significant difference in their Negative Interactions, F(1,
21) = 5.52 p\0.05 (adjusted Ms 0.67 and 1.69, respec-
tively), with the training group having significantly lower
post-test scores (i.e. less negative interactions). However,
there were no significant differences in their Positive
Interactions, F(1, 21) = 0.76, p[0.05 or in their Low-
level interactions, F(1, 21) = 0.13, p[0.05.
For children with HFA who received training compared
with that of the children with HFA who did not receive the
training, there was a significant difference in their total
scores on the Social Skills Observation, F(1, 20) = 13.61,
p\0.001 (adjusted Ms 7.54 and 10.46, respectively), with
the training group having significantly lower post-test
scores (i.e. less inappropriate social interactions). Also,
there were significant a difference in their Positive Inter-
actions, F(1, 20) = 11.49, p\0.01 (adjusted Ms 5.93 and
7.75, respectively), with the training group having signifi-
cantly lower post-test scores (i.e. more positive interac-
tions). However, there were no significant differences in
their Negative Interactions, F(1, 20) = 2.72, p[0.05 or in
their Low-level interactions, F(1, 20) = 0.42, p[0.05.
The analyses for Hypothesis 3 are summarized in
Table 4. The Cohen’s d effect size (comparing control to
training post measures, post-hoc) for the SSRS Composite
are 1.01 (LFA) and 0.29 (HFA) and for the SSO Total 0.81
(LFA) and 1.34 (HFA).
The purpose of this study was to examine the effects of a
new interactive software intervention for social skills for
children with an ASD, including in natural environments.
The results of this study provide general support for the
promise of using computer-based interactive games for
enhancing specific social skills. In particular, this study
suggests that providing children with ASD opportunities to
practice eye gaze, expression matching and face recogni-
tion in FaceSay’s controlled, structured, and interactive
environment with realistic avatar assistants improved
their social skill. The children with LFA demonstrated
improvement in two areas of the intervention: emotion
recognition and social interactions. The children with HFA
demonstrated improvements in all three areas of the
intervention: facial recognition, emotion recognition, and
social interaction in natural environments.
First, the ability to recognize unfamiliar faces improved
for children with HFA following the intervention, but not
for children with LFA. This difference could be an indi-
cation of individual differences such as intellectual func-
tions. The children with LFA had significantly lower
cognitive functioning than the sample of children with
Table 3 Means (standard deviations) of measures of emotional and facial recognition pre and post intervention for all groups
Measure LFA control (N = 1 4) LFA training (N = 11) HFA control (N = 11)HFA training (N = 13)
Pictures2.91 (1.38)2.91 (1.51)2.71 (1.49)3.50 (1.22)3.61 (1.19)3.31 (1.18) 4.54 (1.63)5.00 (1.00)
Drawings2.36 (0.67)2.27 (1.10)2.71 (1.38) 3.07 (1.21)2.69 (0.95)2.77 (1.24)3.64 (1.91)4.54 (1.37)
Total 5.27 (1.95) 5.18 (2.44)5.43 (2.59)6.57 (2.28) 6.31 (1.97)6.08 (2.33)8.00 (3.13) 9.54 (2.34)
Short form 11.18 (4.87)12.64 (4.20)12.79 (4.19) 14.64 (4.96)13.23 (3.37) 14.54 (3.26)16.27 (5.91)19.45 (4.27)
Long form28.18 (4.21)29.36 (4.24)29.86 (4.72)32.64 (6.81)29.31 (5.59)31.23 (5.79)36.00 (7.03)40.64 (6.67)
J Autism Dev Disord (2011) 41:1543–1555 1551
HFA. Thus, it is possible that the children with LFA did not
completely understand the concepts or directions in the
games, and therefore, did not fully benefit from the
Second, after the intervention, the children with LFA
and HFA demonstrated improvement in their ability to
recognize emotions. More specifically, children with LFA
improved in their ability to recognize emotions when
provided with photographs, whereas children with HFA
improved in emotion recognition when provided with
photographs or drawings. As the computer games only
included photographs of adults and children, it is interest-
ing to note that children with HFA generalized their
improved emotion recognition skills to schematics of faces.
It is also interesting to note that both groups improved their
skills to label the emotions even though the computer
games never mention emotion labels.
These results support the findings from other studies that
emotion recognition and facial recognition abilities can be
improved in laboratory settings using computer based
training (Goldsmith and LeBlanc 2004). However, a major
question raised by the current study, in line with the main
difficulty encountered in other intervention programs for
this population, is whether children’s improvements were
transferred into the child’s more global social competence
with peers and family in real settings. Results of this study
demonstrate that when students with LFA and HFA were
provided with an opportunity to learn and practice specific
social skills in FaceSay’s controlled environment that
simulated a natural setting, their social interactions
improved in natural environments.
Children showed improvements in their social interac-
tions with peers and family members. More specifically,
the children with HFA demonstrated growth in their posi-
tive social interaction behaviors (e.g. more likely to share
experiences or an object with a peer). For children with
LFA, significant decreases in their negative behaviors were
demonstrated following the intervention (e.g. the children
were less likely to avoid social overtures made towards
them by a peer). Their parents also reported an improve-
ment in their ability to be more assertive (e.g. initiates
more social activities with peers), to be more responsible
(e.g. acknowledges compliments), and to have more self-
control (e.g. cooperates with family members without
being told to do so).
Previous research has highlighted the role of exposure to
faces in the development of face processing and its
underlying mechanisms (Gauthier et al. 2000). It is possi-
ble that children with ASD avoid looking at faces during a
period through which typically developing children acquire
face processing skills, thus, they do not acquire such skills
at a typically developing pace. The lower levels of expo-
sure to faces that children with ASD experience relative to
other children may inhibit the acquisition and development
of basic face processing mechanisms. The current study
indicates that with increased exposure to faces and practice
recognizing key features of faces, it is possible to enhance
children with HFA’s face processing skills.
The purpose of the current study was to examine the effects
of a computer interactive intervention program on specific
social skills of children who have ASD. This study indi-
cates that practicing simulated activities on the computer
enhances facial and emotion recognition abilities. The
results provide support for the effectiveness of using a
computer-based interactive simulation program as a vehi-
cle for enhancing observed and reported social skills. These
results are consistent with previous results on the
Table 4 Means (standard deviations) of measures of social skills pre and post intervention for all groups
MeasureLFA control (N = 11) LFA training (N = 14) HFA control (N = 13) HFA training (N = 11)
Pre PostPrePost Pre PostPrePost
Cooperation 6.09 (2.91) 6.00 (3.41)6.43 (2.24)8.14 (2.07)8.38 (1.75)7.08 (2.87) 6.36 (3.07) 6.73 (2.24)
Assertion 7.27 (2.33)6.09 (1.97) 6.93 (2.97)7.93 (2.62)8.85 (1.86) 7.77 (2.68)7.00 (2.45)7.82 (2.63)
Responsibility4.09 (3.01)4.09 (2.84) 4.07 (3.19) 5.36 (2.76)6.69 (2.75) 6.15 (3.05) 4.82 (3.49)4.82 (3.37)
Self-Control 5.91 (2.51) 4.09 (2.34)5.71 (2.73)7.57 (2.56)7.00 (1.96) 6.31 (3.17)4.64 (2.29) 6.54 (1.50)
Composite 60.91 (8.39)59.45 (9.49)61.71 (8.62)68.64 (9.93) 68.38 (7.13)66.78 (10.68) 61.55 (9.43) 63.90 (8.84)
Positive 7.36 (1.63)7.82 (1.25) 7.71 (1.58)7.46 (1.31)6.81 (2.17)7.54 (1.19) 6.86 (1.80)6.18 (2.12)
Negative 0.95 (0.91)1.68 (1.23)0.89 (1.21) 0.68 (0.79)0.69 (0.72) 1.35 (0.99)0.45 (0.65) 0.59 (0.92)
Low-Level1.82 (1.08) 1.45 (0.93)1.39 (0.92) 1.68 (0.75)1.35 (0.80)1.42 (0.81) 1.36 (0.84) 1.05 (0.82)
Total10.14 (2.17)10.95 (1.42) 9.86 (2.48) 9.68 (1.71) 8.85 (2.58)10.31 (0.97)8.68 (2.11)7.72 (2.56)
1552 J Autism Dev Disord (2011) 41:1543–1555
effectiveness of computer-based interventions for children
with ASD (e.g. Bernard-Optiz et al. 2001; Chen and
Bernard-Optiz 1993; Panyan 1984; Yamamoto and Miya
One of the most important factors in the success of this
study was that all participants who received the interven-
tion easily adapted to the computer interactive program.
They always wanted to go to work on the computer
(although a few wanted to finish an activity first), and they
typically participated in the games until the activities were
terminated. Although not measured, anecdotal observations
of the children who received the intervention indicated that
the children with ASD enjoyed the programs. They fre-
quently asked to play the games, and became upset if there
was an interruption of the sessions due to field trips or
holidays. The students also provided themselves with ver-
bal praise such as ‘‘Good job’’ or prompted the experi-
menter to say ‘‘Good job.’’ Also, observations of the
participants indicate that some of the students increased
their computer skills following the study. For example, one
participant who previously had no experience with com-
puters learned to navigate a mouse, turn the program on
and off, and log off the computer. An objective measure of
treatment acceptability for the computer based activity and
a valid computer-skills measure would serve as useful
collateral data in future studies.
FaceSay, a computer program with realistic avatar
assistants, appears to be a promising strategy for teaching
specific social skills for children with ASD. Multidisci-
plinary approaches, involving educational specialists,
psychologists, software designers, as well as parents and
their child with ASD, could be useful. Although real-
life practice remains the most important part of social
skills training, computer-based simulations might be a
non-threatening starting point for individuals with ASD,
contributing to the facilitation of better social and com-
A few limitations to this study design are important to note.
It is not clear how length of treatment is related to the
effectiveness of FaceSay. While the present study imple-
mented training for 6 weeks, fewer sessions may have been
just as beneficial. Conversely, longer term treatment might
impart more improvement than found in this investigation.
This issue is important to clarify when resources are lim-
ited. Another question unaddressed by the present study is
the duration of improvement. This study focused on short-
term follow-up. It is not yet known whether these
improvements continue as children mature. Are there
continued gains 6 months and a year after cessation of the
training program? Although the intention of this training
model is to provide children with lifelong tools to assist the
child, is this goal successfully attained? Do children con-
tinue to make gains after formal involvement ends?
In addition, this study did not attempt to directly com-
pare the computer-based program with other treatment
models. What is needed in future studies are head-to-head
comparisons of different programs for treating ASD, in
which variables such as number of hours of intervention
and parent involvement are tightly controlled, while
teaching models are varied (e.g. interactive computer-
based teachingvs. two-dimensional
The measure of social skills interactions is also subject
to a few methodological limitations. First, although the
current study involved a social interaction observation to
evaluate if the emotion and face recognition skills taught
during the intervention resulted in improved social inter-
actions with others, the children were evaluated during
recess at school with familiar peers. Thus, it is difficult to
say whether the social skills learned in the intervention
generalize across other real settings and to unfamiliar age
peers. Second, the items on the social skills observation
rating scale did not address the frequency or duration of the
social skills, but rather focused on the presence or absence
of each skill during the 5-min period.
Finally, the groups of children with LFA and HFA were
from a sample of children with developmental disabilities
coming to a specialized school or after-school center, often
because parents wanted the children to receive therapy, or
because their problems were particularly challenging. In
this sense, it seems likely that the sample may have
included children who are more severely impaired than the
general population of children with ASD. The outcomes of
this study might therefore be limited in terms of general-
izability to the population of individuals with ASD as a
whole. Thus, there is a need to replicate the findings from
this study with a different sample, and ideally, from a
population-based sample, where more heterogeneity is
likely to be found.
The neurobiological basis for face processing difficulties in
ASD has become a topic of recent interest (Sasson 2006). It
remains unclear to what extent the abnormal processing of
social and emotional information in individuals with ASD
could be due to a dysfunction in the amygdala, the fusiform
gyrus, a lack of developmentally appropriate experience
with human faces, or a combination of these factors (Sas-
son 2006). The theory that amygdala pathology could
contribute to some of the neuropsychological impairments
in social and emotional processing seen in ASD (Baron-
Cohen et al. 2002) is supported by the finding that
J Autism Dev Disord (2011) 41:1543–15551553
individuals with damage to the amygdala also show
abnormal emotional and social processing. In particular,
several studies have found that the amygdala is important
for recognition of certain emotions and that it is important
for making complex social judgments from faces (Adolph
et al. 1994). In addition, persons with an ASD have been
found to have less activation of the fusiform gyrus, a face
processing area of the cortex, when viewing novel faces
(Schultz et al. 2000).
The results of the current study call for neuroimaging
studies to examine possible changes in the functioning of
brain areas (e.g. in the amygdala, fusiform gyrus, or pre-
frontal cortex), and gaze tracking studies to objectively
measure attention to the area around the eyes following the
use of FaceSay. Such studies would throw light on whether
the observed facial processing changes reported here are
arising from changes in those neural regions that are typ-
ically recruited by the brain in typically developing indi-
viduals, or if they are due to compensatory strategies by
other neural regions.
This investigation focused on facial and emotion rec-
ognition skills, as well as social interaction improvements
in children using FaceSay. A number of additional outcome
variables could be explored in future studies, including
objective measures of eye gaze and child satisfaction to the
games. Also, future research should clarify which variables
predict most successful utilization of FaceSay. The current
study found that children with LFA or HFA who had higher
initial IQ and lower autism symptomology benefited most
from the intervention. The effects of other child variables,
such as visual spatial ability, and behavior problems,
should also be explored. In addition, parent and family
variables, such as stress and depression levels, and socio-
economic status may predict treatment outcomes. Since
services are, at least in most areas of the country, a limited
resource available to only a subset of children, determi-
nation of who benefit most from the training model is
critical. Future research should also investigate the gener-
alization from computer-based programs that explore sit-
uations in the classroom, at home, and in the community.
The use of such programs and the development of addi-
tional strategies could enhance our knowledge of instruc-
tional strategies for enhancing the social skills of children
with ASD. Additional computer-based programs that
enhance the social skills of children who have other
developmental disabilities, such as Attention-Deficit/
Hyperactivity Disorder, could also further our under-
standing and could provide additional opportunities to
expand on understanding of theory of mind and impair-
ments in social skills.
Civitan International. The author thanks the Autism Lab at UAB,
This study was funded in part by a grant from
Dr. Franklin R. Amthor as well as Casey Wimsatt, Symbionica, LLC.
This paper is adapted from the author’s dissertation.
Adolph, R., Tranel, D., Damasio, H., & Damasio, A. R. (1994).
Impaired recognition of emotion in facial expressions following
bilateral damage to the human amygdala. Nature, 372, 669–672.
American Psychiatric Association. (1994). Diagnostic and statistical
manual of mental disorders (4th ed.). DC: Washington.
Autism Society of America. (2003). New autism facts. Advocate,
Bailey, K. J. (2001). Social competence of children with autism
classified as best-outcome following behavior analytic treatment.
Dissertation Abstracts International: Section B the Sciences and
Engineering, 61(12-B), 6696.
Baron-Cohen, S., Ring, H. A., Bullmore, E. T., Wheelwright, S.,
Ashwin, A., & Williams, P. (2002). Social intelligence in the
normal and autistic brain: An fMRI study. European Journal of
Neuroscience, 11, 1891–1898.
Baron-Cohen, S., Wheelwright, S., & Jolliffe, T. (1997). Is there a
‘‘language of the eyes’’? Evidence from normal adults and adults
with Autism or Asperger syndrome. Visual Cognition, 4,
Bartlett, J. C., & Searcy, J. (1993). Inversion and configuration of
faces. Cognitive Psychology, 25, 281–316.
Benton, A. L. (1980). The neuropsychology of facial recognition.
American Psychologist, 35, 176–186.
Bernard-Opitz, V. (1989). Computer assisted instruction for autistic
children. Journal of Child and Adolescent Psychiatry, 17,
Bernard-Opitz, V., Sriram, N., & Nakhoda-Sapuan, S. (2001).
Enhancing social problem solving in children with autism and
normal children through computer-assisted instruction. Journal
of Autism and Developmental Disorders, 31, 377–398.
Bo ¨lte, S., Feineis-Matthews, S., Leber, S., Dierks, T., Hubl, D., &
Poustka, F. (2002). The development and evaluation of a
computer-based program to test and to teach the recognition of
facial affect. International Journal of Circumpolar Health,
61(Suppl 2), 61–68.
Bosseler, A., & Massaro, D. W. (2003). Development and evaluation
of a computer-animated tutor for vocabulary and language
learning in children with autism. Journal of Autism and
Developmental Disorders, 33, 653–673.
Boucher, J., & Lewis, V. (1991). Unfamiliar face recognition in
relatively able autistic children. Journal of Child Psychology and
Psychiatry, 33, 843–859.
Chen, S. H., & Bernard-Opitz, V. (1993). Comparison of personal and
computer-assisted instruction for children with autism. Mental
Retardation, 31, 368–376.
Cohen, J. (1960). A coefficient of agreement for nominal scales.
Educational and Psychological Measurement, 20, 37–46.
Colby, K. (1973). The rationale for computer-based treatment of
language difficulties in non-speaking autistic children. Journal of
Autism and Childhood Schizophrenia, 3, 254–260.
Corbett, B. A., & Abdullah, M. (2005). Video Modeling: Why does it
work for children with autism? Journal of Early and Intensive
Behavior Intervention, 2(1), 2–8.
Davies, S., Bishop, D., Manstead, A. S., & Tantam, M. (1994). Face
perception in children with autism and Asperger’s Syndrome.
Journal of Child Psychology and Psychiatry, 35, 1033–1058.
Ekman, P., & Friesen, W. V. (1975). Unmasking the face. Englewood
Cliffs, NJ: Prentice Hall.
1554J Autism Dev Disord (2011) 41:1543–1555
Faja, S., Aylward, E., Bernier, R., & Dawson, G. (2008). Becoming a
face expert: A computerize face-training program for high-
functioning individuals with autism spectrum disorders. Devel-
opmental Neuropsychology, 33, 1–24.
Fleiss, J. L. (1981). Statistical methods for rates and proportions.
New York: John Wiley & Sons, Inc.
Frith, U. (1989). Autism: Explaining the enigma. Oxford: Blackwell.
Gauthier, I., Skudlarski, P., Gore, J. C., & Anderson, A. W. (2000).
Expertise for cars and birds recruits brain areas involved in face
recognition. Nature Neuroscience, 3, 191–197.
Golan, O., Ashwin, A., Granader, Y., McClintock, S., Day, K.,
Leggett, V., et al. (2010). Enhancing emotion recognition in
children with autism spectrum conditions: An intervention using
animated vehicles with real emotional faces. Journal of Autism
and Developmental Disorders, 40, 269–279.
Golan, O., & Baron-Cohen, S. (2006). Systemizing empathy:
Teaching adults with Asperger syndrome or high-functioning
autism to recognize complex emotions using interactive multi-
media. Development and Psychopathology, 18(2), 591–617.
Goldsmith, T. R., & LeBlanc, L. A. (2004). Use of technology in
interventions for children with autism. Journal of Early and
Intensive Behavior Intervention, 1(2), 166–178.
Gresham, F. M. (1981). Assessment of children’s social skills.
Journal of School Psychology, 19, 120–134.
Gresham, F. M., & Elliott, S. N. (1990). Social skills rating system
manual. Circle Pines, MN: American Guidance Service.
Hauck, M., Fein, D., Waterhouse, L., & Feinstein, C. (1995). Social
initiations by autistic children to adults and other children.
Journal of Autism and Developmental Disorders, 24, 129–154.
Heiman, M., Nelson, K. E., Tjus, T., & Gillberg, C. (1995). Increasing
reading and communication skills in children with autism
through an interactive multimedia computer program. Journal
of Autism and Developmental Disorders, 25, 459–480.
Hobson, R. P. (1986). The autistic child’s appraisal of expressions of
emotion. Journal of Child Psychology and Psychiatry, 27,
Hopkins, M., & Biasini, F. (2005). Children with fragile x syndrome
and children with autism show distinct impairments in emotional
understanding. Presented at the American psychological society
conference, New York, NY.
Joseph, R. M., & Tanaka, J. (2003). Holistic and part-based face
recognition in children with autism. Journal of Child Psychology
and Psychiatry, 44, 529–542.
Kaufman, A. S., & Kaufman, N. L. (1990). Kaufman brief intelligence
test. Circle Pines, MN: American Guidance Service.
Koegel, R. L., Koegel, L. K., & McNerney, E. K. (2001). Pivotal
areas in intervention for autism. Journal of Clinical Child
Psychology, 27, 233–243.
LaCava, P., Golan, O., Baron-Cohen, S., & Smith Myles, B. (2007).
Using assistive technology to teach emotion recognition to
students with Asperger syndrome: A pilot study. Remedial and
Special Education, 28, 174–181.
LaCava, P., Rankin, A., Mahlios, E., Cook, K., & Simpson, R. (2010).
A single case design evaluation of a software and tutor
intervention addressing emotion recognition and social interac-
tion in four boys with ASD. Autism, 14, 161–178.
Langdell, T. (1978). Recognition of faces: An approach to the study
of autism. Journal of Child Psychology and Psychiatry, 19,
Mundy, P. (2003). Annotation: The neural basis of social impairments
in autism: The role of the dorsal medial-frontal cortex and
anterior cingulated system. Journal of Child Psychology and
Psychiatry, 44, 793–809.
Ozonoff, S., Pennington, B., & Rogers, S. (1990). Are there specific
emotion perception deficits in young autistic children? Journal
of Child Psychology and Psychiatry, 31, 342–361.
Panyan, M. V. (1984). Computer technology for autistic students.
Journal of Autism and Developmental Disorders, 14, 375–382.
Perry, A., Condillac, R., Freeman, N. L., Dunn-Geier, J., & Belair, J.
(2005). Multi-site study of the Childhood Autism Rating Scale
(CARS) in five clinical groups of young children. Journal of
Autism and Developmental Disorders, 35, 625–634.
Rhodes, G. (1988). Looking at face: First-order and second order
features as determinants of facial appearance. Perception, 17,
Sasson, N. J. (2006). The development of face processing in autism.
Journal of Autism and Developmental Disorders, 36, 381–394.
Schopler, E., Reichler, R. J., & Renner, B. R. (1988). The childhood
autism rating scale (CARS). Los Angeles: Western Psycholog-
Schultz, R. T., Gauthier, I., Klin, A., Fulbright, R. K., Anderson, A.
W., Volkmar, F., et al. (2000). Abnormal ventrical temporal
cortical activity during face discrimination among individual
with autism and Asperger syndrome. Archives of General
Psychiatry, 57, 331–340.
Shah, A., & Frith, U. (1993). Why do autistic individuals show
superior performance on the block design task? Journal of Child
Psychology and Psychiatry, 34, 1351–1364.
Silver, M., & Oakes, P. (2001). Evaluation of a new computer
intervention to teach people with autism or Asperger syndrome
to recognize and predict emotions in others. Autism, 5, 299–316.
Sullivan, L. A. (1996). Facial recognition of emotions by children and
adults. Dissertation Abstracts International: Section B The
Sciences and Engineering, 57(11-B), 7253.
Tanaka, J. W., Lincoln, S., & Hegg, L. (2003). A framework for the
study and treatment of face processing deficits in autism. In G.
Schwarzer & H. Leder (Eds.), The development of face
processing (pp. 101–119). Ashland, OH: Hogrefe & Huber
Tanaka, J. W., Wolf, J. M., Klaiman, C., Koenig, K., Cockburn, J.,
Herlihy, L., et al. (2010). Using computerized games to teach
face recognition skills to children with autism spectrum disorder:
the Let’s Face It! program. Journal of Child Psychology and
Psychiatry, 51(8), 944–952.
Teal, M. B., & Wiebe, M. J. (1986). A validity analysis of selected
instruments used to assess autism. Journal of Autism and
Developmental Disorders, 16, 485–494.
Whalen, C., Moss, D., Ilan, A., Vaupel, M., Fielding, P., Macdonald,
K., et al. (2010). Efficacy of TeachTown: Basics computer-
assisted intervention for the Intesntive Comprehensive Autism
Program in Los Angeles Unified School District. Autism, 14,
Williams White, S., Keonig, K., & Scahill, L. (2007). Social skills
development in children with autism spectrum disorders: A
review of the intervention research. Journal of Autism and
Developmental Disorders, 37(10), 1858–1868.
Yamamoto, J., & Miya, T. (1999). Acquisition and transfer of
sentence construction in autistic students: Analysis of computer-
based teaching. Research in Developmental Disabilities, 20,
J Autism Dev Disord (2011) 41:1543–1555 1555