ng
or
ieb
rs, I
Article history:
Received 10 May 2010
Received in revised form
10 December 2010
Accepted 11 January 2011
Keywords:
the ages of 2 and 3 years, but early signs may be detectable by 12
months of age (Osterling and Dawson, 1994). Early recognition of
the disorder is an important challenge since early treatment is
more effective in helping children develop adaptation skills thus
initial study, Anderson et al. (2006) investigated pupil responses to
face and non-face stimuli in ASD children and found that the ASD
group showed pupil constriction in response to children’s faces,
whereas the control groups showed pupil dilation. In 2009, they
analyzed tonic (baseline) pupil responses obtained from the
previouslymentioned investigation (Anderson and Colombo, 2009)
and found that the ASD group exhibited a greater baseline pupil
size than age-matched controls. Fan et al. (2009) did not observe
any differences in resting pupil size in children with and without
* Corresponding author. Inserm U 930, Bâtiment B1A, CHRU Bretonneau, 2,
boulevard Tonnellé, 37044 Tours Cedex 9, France. Tel.: þ33 2 4747 88 40; fax: þ33 2
47 47 38 46.
Contents lists availab
Journal of Psych
lse
Journal of Psychiatric Research 45 (2011) 1077e1082E-mail address: joelle.martineau@univ-tours.fr (J. Martineau).1. Introduction
Autism spectrum disorder (ASD) is a heterogeneous neuro-
developmental disorder affecting between 1 in 165 (Fombonne,
2003) or even 1 in 90 children (King and Bearman, 2009). It is
four times more prevalent in boys than in girls. The disorder was
first described by Kanner (1943) who isolated three core features:
impaired reciprocal social interactions, abnormal development and
use of language; repetitive and ritualized behavior and a narrow
range of interests (APA, 2000). ASD is usually diagnosed between
allowing them to attain a better level of integration into society. In
addition, it can reduce the intensity of the condition and the level of
handicap (Robel, 2009).
Previous studies on the pupil responses of children with autism
have reported conflicting results. The first study was performed
by Rubin (1961) who reported smaller pupil sizes in children
with autism during pupil dilation. In a habituation paradigm,
van Engeland et al. (1991) used geometric figures of differing
complexity and found no significant differences in baseline pupil
size or constriction amplitude in children with ASD. In anotherAutism
Eye-tracking
Pupil size
Pupil responses
Discriminant analysis0022-3956/$ e see front matter
2011 Elsevier Ltd.
doi:10.1016/j.jpsychires.2011.01.008The purpose of this study was to determine whether baseline pupil size and pupil responses during
visual scanning with eye-tracking technology could discriminate children with Autism Spectrum
Disorder (ASD) from mental age-matched and chronological age-matched controls. To this end, we used
stimuli consisting in still color photographs presented centrally to the participant’s midline on a stimulus
monitor. Each child was presented with a series of neutral faces, virtual faces (avatars) and different
objects, separated by black slides. We recorded the mean pupil size and pupil size changes over time in
each of the three categories of stimuli and during exposure to the black slides. Fifty-seven children
participated in study (19 ASD, mean age 118 months; 19 mental age-matched controls, mean age 87
months; and 19 chronological age-matched controls, mean age 118 months). We compared the baseline
pupil size and pupil responses during visual scanning among the three diagnostic groups. During the
presentation of slides, the mean pupil size in the ASD group was clearly smaller than in the MA-matched
and CA-matched groups. Discriminate analysis of pupil size during the presentation of black slides and
slides with visual stimuli successfully predicted group membership in 72% of the participants. Group
membership was correctly classified in 89% of the participants in the ASD group, in 63% in the
MA-matched group and in 63% in the CA-matched group. These potential biomarkers may contribute to
our understanding of the differences in neurological development in the brain in autism and could prove
useful as indicators of ASD.

2011 Elsevier Ltd. All rights reserved.a r t i c l e i n f o a b s t r a c tCan pupil size and pupil responses duri
to the diagnosis of autism spectrum dis
Joëlle Martineau a,*, Nadia Hernandez a, Lorraine H
Frédérique Bonnet-Brilhault a
aUMR_S Inserm U 930, CNRS ERL 3106, Université François Rabelais de Tours, CHU Tou
bUniversity François Rabelais de Tours, 37000 Tours, France
journal homepage: www.eAll rights reserved.visual scanning contribute
der in children?
el b, Laetitia Roché a, Aude Metzger b,
FR 135 Imagerie Fonctionnelle, 37000 Tours, France
le at ScienceDirect
iatric Research
vier.com/locate/psychires

ASD, but participants with ASD showed a significantly longer PLR
(pupillary light reflex) latency, smaller constriction amplitude and
slower constriction velocity than children who had normal devel-
opment. All these results suggest abnormal functioning in the
autonomic nervous system (ANS). Reduced duration and quality of
sleep (Daoust et al., 2004; Williams et al., 2004), heightened
autonomic responses at rest such as increased skin conductance
(Hirstein et al., 2001; Zahn et al., 1987), a decrease in skin
conductance during emotional judgment (Hubert et al., 2009),
increased heart-rate (Hirstein et al., 2001; Ming et al., 2005) and
increased arterial blood pressure (Ming et al., 2005) and an increase
in respiratory rate (Zahn et al., 1987) have been reported in ASD
cognitive development and were progressing in age-appropriate
education programs. The MA group was matched with the ASD
group for mental age (t ¼ 0.53, df ¼ 1, n.s.) and the CA group was
matched with the ASD group for chronological age (t ¼ 0.03, df ¼ 1,
n.s.). None of the participants were taking any prescribed or over-
the-counter medication.
The University Hospital Center of Tours utilizes a standard data
set for recording information which includes the prenatal and
perinatal periods, development, language acquisition, behavior,
neurological development, diet and data concerning past medical
and family history. We obtained a detailed history of the onset of
the autistic symptoms including the age of onset of each symptom,
J. Martineau et al. / Journal of Psychiatric Research 45 (2011) 1077e10821078patients compared to controls and by consistent with the hypoth-
esis of an altered balance in the inhibitory and excitatory activity
within the sympathetic and parasympathetic divisions of the ANS.
The purpose of this study was to determine whether baseline pupil
size and pupil responses during visual scanning could discriminate
between children with ASD and controls. We chose to measure
pupil diameter because pupil size assessment has long been used to
evaluate neurological functioning, alertness and cognitive func-
tioning (Loewenfeld, 1999). Pupil responses have also been asso-
ciated with attention and information processing (Hess, 1975). In
addition, these measurements are more reliable and more sensitive
than other autonomic measurements (Beatty and Lucero-Wagoner,
2000 for review). Since previous reports showed inconsistent
results in resting pupil responses and task-evoked pupil responses,
we chose to analyze baseline pupil size and pupil responses to the
faces, objects and avatars used in a previous study (Hernandez
et al., 2009).
2. Methods
2.1. Participants
The study sample consisted in 19 patients who had been diag-
nosed with ASD (American Psychiatric Association, 2000) at the
University Hospital Center of Tours (France). Inclusion criteria
included normal vision or vision correctable to 10/10 or better, no
past history of any eye disorder and the ability to comply with the
test procedure. The patients’ mean age was 118 months, chrono-
logical age (CA) between 41 and 181 months, mental age (MA)
between 35 and 136months and the gender ratio was 16 males to 3
females.
We also recruited thirty-eight normally developing children
from the local community as controls and divided them into two
groups: a chronological age-matched group (19 children, mean age
116 months, range 44e178 months, 8 females and 11 males) and
a mental age-matched group (19 children, mean age ¼ 87 months,
range 41e136 months, 7 females and 12 males). All controls had
normal vision or vision correctable to 10/10 or better and none had
a past history of any eye disorder. They all had undergone normalFig. 1. The three types of stimuli used in thewhether language acquisition was delayed or lost, gross and fine
motor activity and social interactions.
The diagnosis of autism was made by the specialists in the
Autism Center using a center-based version of the ADI scoring
protocol. The ADI-R confirmed the previous diagnosis in every case.
All 19 subjects had the essential autism phenotype.
Mental Ages (MA), based on a neuropsychological evaluation,
were available for all 19 ASD participants. The mental ages of eight
of the participants were obtained using the BruneteLezine scale,
the French version of the Gesell scale (Brunet and Lézine, 1976) and
in the remaining 11 children, by using the EDEI (Echelles différ-
entielles d’efficiences intellectuelles, Perron-Borelli, 1996) The
average mental age in the 19 participants was 85 months (range
35e136).
2.2. Materials and procedure
The eye-tracking procedures and some portions of the experi-
mental paradigm and stimuli have already been described else-
where (Hernandez et al., 2009). Briefly, we made recordings with
an eye-tracking system consisting in a computer equipped with
two cameras which allowed us to film the eyes of a subject while
he/she was looking at images on the computer screen. Gaze
directionwasmonitored by software supporting the camera system
which was able to capture pupil-illumination of the through
infrared diodes, thus eliminating constraints on the subject, who
remained free to move. We measured head position and orienta-
tion, eye direction and eye-tracking in real time by using the
FaceLAB monitoring system. Gaze Tracker software then
provided measurement and real-time analysis of the duration of
fixation on various parts of the image and of eye pathways between
different parts.
The study subjects were shown the stimuli (still color photo-
graphs) for 4 s centrally to the participant’s midline on the stimulus
monitor. Each child was presented with a series of ten neutral faces,
a series of ten virtual faces (avatars) and a series of objects (Fig. 1).
Stimulus validation has been described elsewhere (Hernandez et al.,
2009). The stimuli were presented on a beige background randomly
to the subjects using a constant (7.5e9.8 lx) approximately averagestudy: neutral faces, avatars and objects.