Neural correlates of eye gaze processing in the infant broader autism phenotype.

Abstract

Studies of infant siblings of children diagnosed with autism have allowed for a prospective approach to study the emergence of autism in infancy and revealed early behavioral characteristics of the broader autism phenotype. In view of previous findings of atypical eye gaze processing in children and adults with autism, the aim of this study was to examine the early autism phenotype in infant siblings of children diagnosed with autism spectrum disorder (sib-ASD), focusing on the neural correlates of direct compared with averted gaze.
A group of 19 sib-ASD was compared with 17 control infants with no family history of ASD (mean age=10 months) on their response to direct versus averted gaze in static stimuli.
Relative to the control group, the sib-ASD group showed prolonged latency of the occipital P400 event-related potentials component in response to direct gaze, but they did not differ in earlier components. Similarly, time-frequency analysis of high-frequency oscillatory activity in the gamma band showed group differences in response to direct gaze, where induced gamma activity was late and less persistent over the right temporal region in the sib-ASD group.
This study suggests that a broader autism phenotype, which includes an atypical response to direct gaze, is manifest early in infancy.

Full text

Page 1

Neural Correlates of Eye Gaze Processing in the Infant
Broader Autism Phenotype
Mayada Elsabbagh, Agnes Volein, Gergely Csibra, Karla Holmboe, Holly Garwood, Leslie Tucker,
Sanya Krljes, Simon Baron-Cohen, Patrick Bolton, Tony Charman, Gillian Baird, and Mark H. Johnson
Background: Studiesofinfantsiblingsofchildrendiagnosedwithautismhaveallowedforaprospectiveapproachtostudytheemergence
of autism in infancy and revealed early behavioral characteristics of the broader autism phenotype. In view of previous findings of atypical
eyegazeprocessinginchildrenandadultswithautism,theaimofthisstudywastoexaminetheearlyautismphenotypeininfantsiblingsof
children diagnosed with autism spectrum disorder (sib-ASD), focusing on the neural correlates of direct compared with averted gaze.
Methods: A group of 19 sib-ASD was compared with 17 control infants with no family history of ASD (mean age ? 10 months) on their
response to direct versus averted gaze in static stimuli.
Results: Relative to the control group, the sib-ASD group showed prolonged latency of the occipital P400 event-related potentials
component in response to direct gaze, but they did not differ in earlier components. Similarly, time-frequency analysis of high-frequency
oscillatoryactivityinthegammabandshowedgroupdifferencesinresponsetodirectgaze,whereinducedgammaactivitywaslateandless
persistent over the right temporal region in the sib-ASD group.
Conclusion: This study suggests that a broader autism phenotype, which includes an atypical response to direct gaze, is manifest early in
infancy.
Key Words: Autism, broader autism phenotype, event-related po-
tential, eye gaze processing, infancy
A
observed in this population. Behavioral studies with children and
adults with autism have demonstrated that the use of gaze cues in
social contexts such as joint attention (1–3) or in inferring mental
states (4) is an area of difficulty. This does not, however, imply a
complete lack of sensitivity to gaze in this clinical group. Individuals
with autism are able to estimate gaze direction (5), and they show
reflexive orienting toward objects cued by another’s gaze (6,7);
however, they are atypical in the context of mutual gaze, that is,
when the perceiver makes eye contact (8).
Electrophysiological and neuroimaging studies have docu-
mented atypical neural correlates of gaze processing in autism.
Using event-related potential (ERP) recording, passive viewing of
faces with direct gaze elicited larger occipitoparietal negativity
than averted gaze in 4 to 7 year olds with autism, a pattern not
seen in typically developing children of the same age (9). In
contrast to this study using passive viewing, actively detecting
direct versus averted gaze targets elicited an occipitotemporal
negativity in children with and without autism. However, the
response was stronger and predominantly right lateralized for
typically developing children, whereas children with autism
prominent aspect of the autism phenotype is a qualitatively
unusual pattern of eye contact, which may reflect the more
widespread deficits in communication and social interaction
showed a bilaterally distributed response (8). Converging evi-
dence for atypical gaze processing comes from functional mag-
netic resonance imaging (fMRI) studies. In typical individuals,
eye gaze is processed in specialized regions including the
superior temporal sulcus and the amygdala (10). Furthermore,
brain activation patterns are modulated by the referential nature
of eye gaze, that is, when gaze shifts are either congruent or
incongruent with the appearance of peripheral visual targets.
Individuals with autism show brain activity in similar regions, but
the modulation of the response as a function of the referential
context of eye gaze is reduced (11).
Processing of eye gaze in autism has also been tied to face
processing. Although there is some consensus that the neural
correlates of face processing in autism are atypical (12), recent
evidence suggests that brain activation patterns in response to
faces correlate strongly with atypical behavioral scanning pat-
terns in autism. More specifically, a reduction in the typical
fusiform and amygdala response to faces seen in autism corre-
lates with reduced fixation to the eye region (13–15).
For genetic reasons (16–18), some of the atypical responses
to socially relevant stimuli including gaze found in individuals
with autism are also seen in their first-degree relatives who do
not have a diagnosis. This broader autism phenotype (BAP)
includes several cognitive and neural characteristics (13,19,20).
For instance, parents of children on the autistic spectrum show
atypical brain activity in response to eyes (21). Also, adult
siblings of individuals with autism, who do not themselves have
a diagnosis, show diminished fusiform activation correlated with
reduced gaze fixation, similar to that seen in their affected
siblings (14). Even at the neuroanatomic level, amygdala volume
in siblings has been found to be significantly reduced (14).
The significance of these findings in formulating a develop-
mental model of autism has been recognized. The reduced
sensitivity to gaze in the social context seems to suggest that the
early developmental course in autism contrasts with that in
typical development. Infants display very early sensitivity to gaze
(22,23), which develops rapidly in the first few years in the form
of social referencing, joint attention, and communication. In
From the Centre for Brain and Cognitive Development (ME, AV, GC, KH, HG,
LT, SK), Guy’s and St Thomas’ NHS Trust (GB), and Centre for Brain &
Cognitive Birkbeck, University of London (MHJ); Autism Research Cen-
tre, University of Cambridge (SB-C); Institute of Psychiatry (PB), King’s
College, London; and Institute of Child Health (TC), London, United
Kingdom.
Address reprint requests to Mayada Elsabbagh, Ph.D., Centre for Brain and
Cognitive Development, Birkbeck, University of London, Henry Wellcome
Building, London WC1E 7HX, United Kingdom; E-mail: m.elsabbagh@
bbk.ac.uk.
ReceivedOctober3,2007;revisedSeptember9,2008;acceptedSeptember
10, 2008.
BIOL PSYCHIATRY 2009;65:31–38
© 2009 Society of Biological Psychiatry
0006-3223/09/$36.00
doi:10.1016/j.biopsych.2008.09.034

Page 2

autism, it has been suggested that a form of neurodevelopmental
deficit in brain networks subserving social cognition leads to
decreased attention to, or interest in, the social world (12,19,24).
This early lack of attention to or interest in social stimuli,
including gaze, may interfere with the emergence of critical
developmental milestones relevant for social cognition such as
shared attention. These cascading influences may eventually
preclude the typical development of sociocommunicative skills.
Compelling as they are, such developmental accounts face
serious empirical challenges. Because a confirmed and reliable
diagnosis of autism is usually only made from approximately 3
years of age (25), our knowledge of the early neural, behavioral,
and cognitive profile in autism is limited. Retrospective studies
indicate that subtle deficits are present before the typical age of
diagnosis in autism. This evidence relates to early differentiation
between social and nonsocial stimuli, which continues into
adulthood. Home videos of the first 2 years of life of infants later
diagnosed with autism show less orienting toward social stimuli
from as early as 9 months (26–29) or even younger (30)
compared with infants later diagnosed with developmental de-
lay. However, these findings remain limited as far as controlled
experimental assessment. Furthermore, it is unclear whether
these early differences in social orienting in general, or in eye
gaze processing in particular, found in infants later diagnosed
with autism extend to the BAP.
An emerging area of research focusing on infants at genetic
high risk for autism has begun to address the emergent nature of
autism symptoms more directly. Research on infant siblings of
children diagnosed with autism spectrum disorders (ASD; here-
after, “infant siblings”) offers this opportunity because the recur-
rence rate of ASD is significantly elevated above the general
population (31) and has been even higher in recent studies
(32,33). Research on infant siblings may clarify why autism
emerges in some cases and not in others, as well as help to
explain variations associated with the autism phenotype.
Some studies have followed up infant siblings from 4 or 6
months of age to the stage when some received a formal
diagnosis. These studies have confirmed that the expression of
autism in the first year is often subtle, at least as far as overt
symptoms or delays in the expected developmental milestones
(32–35). Furthermore, several studies have documented differ-
ences in groups of siblings of children with autism, relative to
matched groups of infants who do not have affected siblings.
One study examined gaze fixation in infant siblings as young as
4–6 months and reported that they looked more to the mother’s
mouth relative to her eyes (36). Other studies did not find
differences in visual fixation but reported subtle differences in
affect (37,38). Slightly older infant siblings (? 14 months) show
somewhat clearer differences in a number of cognitive and motor
measures as well as sociocommunicative measures (39–41). For
instance, in a study examining infant’s response to different
combinations of joint attention cues (e.g., eye gaze or eye gaze
and head movement), siblings showed less responding to joint
attention cues relative to a low-risk control group (41). Although
follow-up data are essential in understanding the functional
impact of such early behavioral patterns on later development
(e.g., 42), these findings nonetheless suggest that early atypical
gaze behavior may form part of the BAP (38).
Despite the behavioral results just described, as yet nothing is
known about the early neural basis of gaze processing in the
BAP, including in infant siblings at risk for autism. Thus, the aim
of this study was to explore differences in response to gaze in
infant siblings as a group using electrophysiological measures.
We examined the neurophysiological correlates of gaze process-
ing in infant siblings while they viewed photographs of females
displaying direct or averted gaze. The stimuli and paradigm we
employed have previously been used with typically developing
infants (43,44) as well as young children diagnosed with autism
(9). We used two established techniques for analysis of brain
activity during this task. First, we measured ERPs that are phase-
locked to stimulus onset. ERP relies on contrasting average neuro-
physiological activity in response to direct gaze relative to averted
gaze in infant siblings relative to the control group. Second, we
analyzed high-frequency oscillatory activity in the gamma band
(20–60 Hz), which is thought to reflect synchronization of brain
activity in response to the task (44–46). Activity in the gamma
band can be evoked, that is, time locked to the eliciting stimulus,
or induced, that is, it jitters in its latency from one trial to the next
and cannot be detected after the averaging process used in ERP
analysis. Detecting the latter form of brain activity relies on
observed variation in spectral power over time (time-frequency
analysis; TFA). Hence, the two techniques, ERP and TFA, provide
converging sources of evidence, each measuring a different form
of neural activity associated with the eliciting stimuli.
Previous studies employing the same paradigm have estab-
lished differentiation of response to direct versus averted gaze in
typically developing infants using both methods (43,44). On the
basis of previous work suggesting that older children and adults
with autism as well as their adult siblings exhibit atypical neural
correlates of eye gaze, we hypothesized that a similar pattern
would be observed in a group of infant siblings. More specifi-
cally, we expected atypical differentiation of direct as opposed to
averted gaze to be reflected in ERP and electroencephalogram
(EEG) components sensitive to face and gaze processing in
infancy. Based on previous findings from autism, we anticipated
that early visual processing of eye gaze in the occipital region
would not differentiate high-risk from low-risk infants but that
differences might be observed in later components in that region,
particularly those sensitive to attentional modulation and top-
down visual processing (47) or components that are sensitive to
the detection of mutual eye gaze and sensitive to its referential
context found over central and right anterior regions (44).
Because infant waveforms may differ substantially from those
found in children and adults, no specific predictions were made
in relation to whether any differences would be reflected in terms
of amplitude or latency between the groups.
Methods and Materials
Participants
Sixty-two infants were recruited for this study, including 31
infant siblings of children with ASD (sib-ASD; 17 male, mean
age ? 10:1 months, SD ? 1.5) and 31 infants who have no family
history with ASD (control; 18 male, mean age ? 10:1 months,
SD ? 1.6). Allinfantsexceptonefromthesib-ASDgroupwereborn
full term. Informed consent was obtained from the parents. All
infants in the sib-ASD group had an older brother or sister who had
received a clinical diagnosis of an autism spectrum disorder from a
U.K. clinician based on ICD-10 criteria (48). All but six infants from
the control group had at least one or more siblings. The infants in
the sib-ASD group fell within the average range of functioning as
verified by the Mullen Scales for Early Development (mean ? 104,
SD ? 9.6). Mullen scores were not available for the control group.
Apparatus and Stimuli
The infants sat on their parents’ laps at 60-cm distance from a
40 ? 29 cm computer screen. The infants’ gaze during stimulus
32 BIOL PSYCHIATRY 2009;65:31–38
M. Elsabbagh et al.
www.sobp.org/journal

Page 3

presentation was recorded using a video camera. Each trial
began with a static colorful fixation stimulus followed by a color
image of a female face. In these images, gaze was either direct or
averted. Four faces were presented in a pseudorandom order.
The faces were aligned with the center of the screen so that the
eyes appeared at a location where the fixation stimuli had been
presented. The faces subtended 21.3 ? 13.9 degrees of visual
angle, and one eye subtended to about 1.6 ? 2.7 degrees.
Fixation stimuli subtended approximately 1.6 ? 1.6 degrees and
were presented for a variable duration of 800–1200 msec. The
face stimuli were presented for 1000 msec followed by a
500-msec interval without a visual stimulus.
Procedure
A 63-channel Geodesic Sensor Net was mounted on the infants’
heads while they sat on their parents’ laps in front of the stimulus
screen. When the baby was attending toward the screen, trials were
presented continuously for as long as the infant attended. The
brain electrical activity was measured simultaneously. The refer-
ence electrode was the vertex (Cz in the conventional 10/20
system). The electrical potential was amplified with .1–100 Hz
bandpass, digitized at 250-Hz sampling rate.
Behavioral and Electrophysiological Data Analysis. First,
participants’ overall behavior was initially coded from the video-
tape, including any gaze shifts during trial presentation (see
Supplement 1 for further details regarding gaze shifting during
the task). For electrophysiological data analysis, trials were
included only if the infants were fixating the center of the screen
at target onset, without any gaze shifts, blinking, or head
movements at any time during the 800-msec segment following
the onset of the face. Furthermore, data from individual sensors
were excluded if they contained artifacts created by movement
or poor contact. The entire trial was excluded if data from more
than 12 sensors were removed or if the trial contained blinks or
other artifacts, and missing data for trials with 12 or fewer bad
channels, irrespective of their location, were interpolated. Infants
with less than 10 artifact-free trials in any condition were
excluded. Data were then referenced to the average.
The same EEG data were further analyzed using two tech-
niques, ERP and TFA. For the ERP analysis, the 1000-msec
segments were baseline corrected, with the baseline a 200-msec
segment before stimulus onset. Individual participant averages
were computed for each trial type. The same ERP analysis was
conducted over the filtered raw data (using a 30-Hz low-pass
filter) to check whether the same results held after filtering.
For the TFA, a Morlet wavelet transformation (45,46) was
applied using 1-Hz intervals for oscillatory activity in the gamma
range. To avoid distortion caused by this transformation, 100-
msec segments at the beginning and end of each trial were
removed so that the resulting segment length was 800-msec long.
For analysis focusing on condition differences between the
groups, the Morlet transformation was applied with baseline
correction either to single EEG trials to compute induced oscil-
latory gamma activity or the infants’ average ERPs to compute
evoked activity. Another analysis was conducted to assess any
overall group differences during a baseline period, which was
100 msec before stimulus onset. This was done by applying the
Morlet transformation to single uncorrected baseline EEG seg-
ments in each trial.
Regions of Interest and Statistical Analysis. Four channel
groups (Figure 1) were selected on the basis of visual inspection
of grand averages for both groups. Furthermore, the selection of
these regions was based on previous research (reviewed in 24)
showing that face- and gaze-sensitive ERP and EEG components
are found over occipital, central, and right temporal channel
groups. These were a posterior group, a right temporal group, a
left temporal group, and an anterior central group. For compo-
nents of interest, statistical models focused on the interaction of
Group ? Condition separately for each channel group. Only
when this was significant were effects explored further.
Results
Twelve infants from the sib-ASD group and 14 from the
control group were excluded because of excessive artifacts or for
completing too few trials because of fussiness or fatigue. The
final sample consisted of 19 sib-ASD and 17 control infants. The
groups did not differ on a number of baseline measures,
including the total number of trials or the number of valid trials
retained for analysis (Table 1).
Figure 1. Regions selected for analysis: anterior central, left and right tem-
poral, and posterior. VREF, vertex reference sensor.
Table 1. Characteristics of Participants Included in the Analysis and Their
Behavior During the Task
Group
ControlSib-ASD
N
Male:Female
Age (SD)
Total
Number of trials (SD)
Valid trials (SD)
Direct Gaze
Number of trials (SD)
Valid trials (SD)
Averted Gaze
Number of trials (SD)
Valid trials (SD)
17
10:7
19
10:9
297.64 (55.05) 291.94 (39.07)
119.42 (49.49)
44.36 (25.96)
131.68 (38.76)
43.21 (22.42)
66.01 (28.81)
22.11 (12.94)
65.79 (19.51)
21.42 (11.25)
53.42 (22.92)
22.26 (13.27)
65.89 (19.26)
21.79 (11.43)
sib-ASD, siblings of children diagnosed with autism spectrum disorder.
M. Elsabbagh et al.
BIOL PSYCHIATRY 2009;65:31–38 33
www.sobp.org/journal

Page 4

ERP Analysis of Gaze Effects
The ERP effects were observed in the posterior channel group
but not in any of the other channel groups. The corresponding
waveform is displayed in Figure 2A. Mean amplitude and latency
of three components sensitive to face and gaze processing in
infancy (49–52) were averaged for each of the channel groups.
The components analyzed were the P1 around 100–199 msec,
N290 around 200–319 msec, and P400 around 320–539 ms. The
ERP effects were assessed using analysis of variance (ANOVA)
with Condition (Direct, Averted) as a within-subject factor and
Group (sib-ASD, control) as a between-subject factor for the
amplitude and latency of each ERP component.
There were no significant main effects or interactions for
the amplitude of any of the components. On the other hand,
there was a main effect of Condition on the latency of the P1
[F(1,34) ? 6.1, p ? .01] and N290 [F(1,34) ? 12.0, p ? .001] in
the posterior channel group; in both groups, the response to
Direct gaze was faster than to Averted gaze. Furthermore, the
interaction of Condition and Group was significant for the
latency of the P400 [F(1,34) ? 6.4, p ? .01] over the posterior
region. Figure 2B shows the difference in latency between the
two conditions for each group, and Figure 2C is a scatter plot
of individual data points (see Supplement 1 for examples of
waveforms for individual infants). Further analysis revealed
that the sib-ASD group showed a slower P400 response to
Direct gaze [F(1,34) ? 10.1, p ? .003], but the P400 latency in
the two groups did not differ for Averted gaze [F(1,34) ? 1.8,
p ? .81]. Within-group comparisons indicated that whereas
the control group showed no latency difference between
Direct and Averted gaze (p ? .1), the difference between the
two conditions in the sib-ASD group approached significance
(p ? .05). The sib-ASD group tended to respond faster to the
Averted relative to the Direct gaze condition. Analysis of peak
amplitude for the three components revealed no significant
interactions between the two groups. Reanalysis of the raw
data after applying a 30-Hz low-pass filter had negligible
effects on the level of significance of any of the results
reported here, indicating that potential noise from high fre-
quencies had little impact on the findings.
Figure 2. (A) The event-elated potential wave form for posterior channel groups, (B) the differences in latency of the response to direct relative to averted
gazeforthethreeposteriorcomponents,and(C)scatterplotofindividualdatapoints.AsignificantinteractionbetweenGroupandConditionwasobserved
for the P400 but not for earlier components. sib-ASD, siblings of children diagnosed with autism spectrum disorder.
34 BIOL PSYCHIATRY 2009;65:31–38
M. Elsabbagh et al.
www.sobp.org/journal

Page 5

TFA Analysis of Gaze Effects
Evoked gamma activity in the lower-frequency band
(20–30 Hz) and induced activity in the higher-frequency band
(30–50 Hz) were assessed by calculating mean amplitude of
spectral power over 100-msec segments over the selected
channel groups. ANOVA was used to assess the relationship
between evoked and induced gamma and Condition (two
levels) ? Time (seven levels) ? Group (two levels). Evoked
gamma activity was observed in the posterior region, but none
of the main effects or interactions were significant. In contrast,
clear condition differences were observed for induced gamma
activity (30–50 Hz) in the right temporal region, where there
was a significant main effect of Condition [F(1,34) ? 5.8, p ?
.02] and a two-way interaction between Time and Group
[F(6,29) ? 3.1, p ? .01]. Furthermore, the three-way interac-
tion between Condition, Time, and Group approached signif-
icance [F(6,29) ? 2.3, p ? .06). The time-frequency plot for
this region is shown in Figure 3. The effects observed did not
extend above 50 Hz and could thus not have been due to
muscular activity. Subsequent analyses were conducted sepa-
rately for each time segment, and pairwise t tests were used to
assess differences between the two conditions within each
group. As Figure 3 illustrates, the control group showed clear
differentiation of the Direct relative to the Averted gaze condition
around 200 msec, and induced gamma effect persists during later
time periods. Pairwise comparisons for the control group
showed that the response to Direct gaze was significantly
different from that to Averted gaze within the following time
bins: 100–200, 200–300, 300–400, and 600–700 msec (all p ?
.05). However, the differentiation of the two conditions in the
sib-ASD group began approximately 200 msec later than the
control group and was less persistent over time (p ? .05 only in
the 300–400 msec time bin; p ? .06 in the 500–600 msec time
bin). Consistent with the ERP results, time-frequency analysis
also indicated that the two groups differed mainly in their
processing of the direct gaze condition (Figure 3).
Analysis of Baseline Gamma Activity
In view of these results, we also explored potential group
differences during the baseline interval of 100 msec before
stimulus onset while the infants were viewing different, ran-
domly presented, color cartoons. This analysis focused on abso-
lute differences in gamma activity between the two groups,
irrespective of condition, and was conducted without baseline
correction of the data. Oscillatory activity in the lower (20–30
Hz) and higher (30–50 Hz) frequency band during the baseline
period was assessed over single trials and averaged over all scalp
electrodes except peripheral ones (channels 10, 23, 26, 35, 51,
59). There was a near significant effect of Group in both
frequency ranges (20–30 Hz, p ? .06; 30–50 Hz, p ? .05).
Separate ANOVAs for each of the four regions showed a
significant main effect of Group for the central anterior region
(20–40 Hz, p ? .04; 30–50 Hz, p ? .03) and right temporal region
(20–40 Hz, p ? .02; 30–50 Hz, p ? .01) but not for the two other
regions. Hence, irrespective of the condition, the sib-ASD group
showed increased gamma oscillatory activity relative to the
control group in central and right temporal regions. This is
despite the fact that the central anterior region showed no group
differences in relation to the eliciting gaze stimuli.
Discussion
Developmental models of autism have focused on under-
standing the precursors of the observed social difficulties includ-
Figure3.Inducedgammaactivityinthegammarange(30–50Hz)overtherighttemporalregionforthetwogroups.DifferentiationoftheDirectrelativeto
theAvertedgazeconditionbeganatapproximately200msec,andinducedgammaeffectpersistsduringlatertimeperiods(p?.05forpairwisecomparisons
infollowingtimebins:100–200,200–300,300–400,600–700msec).However,thedifferentiationofthetwoconditionsinthesiblingsofchildrendiagnosed
withautismspectrumdisorder(sib-ASD)groupbeganapproximately200mseclaterthanthecontrolgroupandwaslesspersistentovertime(p?.05onlyin
the 300- to 400-msec time bin; p ? .06 in the 500- to 600-msec time bin).
M. Elsabbagh et al.
BIOL PSYCHIATRY 2009;65:31–38 35
www.sobp.org/journal

Page 6

ing orienting to social stimuli and events, joint attention, imita-
tion, and social interactions. Although various models propose
alternative explanations of the origins of these difficulties
(12,53,54), all agree that differences in attention to, or preference
for, socially relevant information has an important contribution.
Our findings are consistent with results indicating atypical gaze
processing in children and adults with autism (8,9,11) as well as
in unaffected siblings of individuals with autism (14) and offer
some insights into how the developmental process is altered in
infant siblings of children with autism relative to infants who
have no family history of autism.
In this study, siblings of children diagnosed with autism were
indistinguishable from the control group in their response to
direct relative to averted gaze in early posterior ERP components.
In contrast, the later P400 component differentiated the two
groups, with the sib-ASD group showing prolonged latency in
responding to direct gaze. This P400 ERP component is sensitive
to face processing in infants and, together with the N290, is
thought to be a precursor to the face-sensitive adult N170 (51).
This component is also sensitive to attentional modulation, or
top-down visual processing in infants (47). Consistent results
were found using TFA of high-frequency oscillatory activity in
the gamma band. The sib-ASD and control groups did not differ
in early phase-locked evoked gamma activity, but they differed
in induced gamma activity. Whereas the control group showed a
clearly differentiated and temporally persistent induced response
to direct relative to averted gaze, the response in the sib-ASD was
delayed and less persistent.
These converging results suggest that early visual processing
of eye gaze is not atypical in infant siblings at-risk for autism,
because both ERP analysis and TFA showed that early ERP
components and phase-locked gamma activity were similar in
the two groups. However, the two groups differed mainly in their
response to the direct gaze condition in later occipital ERP
components and in right temporal induced gamma activity. The
latter components are thought to reflect neural processes specific
to the detection of mutual eye gaze and sensitive to its referential
context (44). These findings complement recent behavioral
evidence indicating that infant siblings are less able to integrate
social cues, including gaze, relevant in the context of joint
attention (41).
In addition to these differences in treatment of direct relative
to averted gaze, the sib-ASD group showed increased baseline
oscillatory activity in the gamma band relative to the control
group in central and right temporal regions, but not in other
brain regions. A recent study that reported increased gamma
band oscillatory activity in children diagnosed with autism (55)
has suggested this may reflect atypical patterns of cortical
connectivity (55,56). Excess gamma oscillations have also been
tied to several psychiatric disorders, and there are a variety of
interpretations of the source and functional relevance of this
pattern (for a review, see 57). The extent to which these baseline
differences relate to eye gaze processing is less clear. In our
study, the increase in high-frequency oscillatory activity was
observed in the right temporal region where the two groups
showed differences in their treatment of direct relative to averted
gaze. Yet a similar increase was also seen in the anterior central
region, where the two groups did not differ in their response to
the two conditions. It is still possible that such baseline differ-
ences may indirectly affect processing of eye gaze. For instance,
anterior central components were found to be sensitive to the
referential nature of eye gaze in typical infants in a task in
which the infant uses eye gaze cues to predict the appearance
of peripheral targets (8), as well as in the context of joint
attention (58).
It is generally acknowledged that electrophysiological meth-
ods in infancy are challenging and confer some limitations
including relatively small samples, reduction in the number of
valid trials, and individual variability in responses. To what
extent do these limitations relate to the key results and interpre-
tation in our study? Given that this is one of the first reports using
neurophysiological methods in at-risk infants, we opted for
conservative procedures based on previous studies using the
same task and analysis methods. Despite their general limita-
tions, these procedures allow for comparison between our
results and previous findings in low-risk, typically developing
infants. Confidence in our results comes from our findings
indicating that significant differences between the groups were
specific to certain neural components and conditions but not to
others. The use of two relatively independent analysis tech-
niques also yielded converging results.
A more general challenge to this area of research also relates
to limitations in the interpretation of EEG and ERP components—
not only in infants but also in adults. For instance, it has recently
been suggested that some induced gamma activity effects re-
ported in adults may be due to small involuntary saccades
time-locked to the onset of a stimulus presentation (59). These
small involuntary eye movements are associated with “spike
potentials” that may contaminate high-frequency EEG (such as
within the gamma range) in certain paradigms. We do not
believe that our induced gamma effects can be explained in this
way because infants under 12 months are known not to have
recordable spike potentials (60). Further, our gamma effects are
not broadband (characteristic of the saccade-related spike po-
tential) but are restricted in frequency range. Finally, removing
the infants who showed measurable saccades during stimulus
presentation did not significantly change our main results (see
Supplement 1), and it seems unlikely that viewing a face with
direct gaze would elicit less foveation (more involuntary sac-
cades) than a face with averted gaze (which may induce gaze
following). Hence, studies with larger samples and with methods
specifically targeting individual correlations with behavioral pat-
terns including saccadic movements during the task (see also
Supplement 1) will be important future directions for this line of
work.
Taken together, our results suggest that the broader autism
phenotype, including an atypical response to eye gaze, is
manifest early in infancy. Although we found no evidence of
differences in the infant BAP in terms of early visual processing
of eye gaze, the neural patterns associated with later attentional
modulation and those sensitive to referential aspects of eye gaze
are comparable to those reported in children and adults diag-
nosed with autism. This early atypical response to eye gaze is
likely to combine with other risk factors, resulting in a later
diagnosis for some individuals (61). Differences also encompass
an increase in baseline oscillatory activity, likely to reflect
atypical neural connectivity, as well as atypical mechanisms for
processing of eye gaze within the first year of life.
This research was funded by the Medical Research Council
(Grant No. G9715587) and Autism Speaks (Grant Nos. 1292/MJ/
01-201-006-065-00-00 and 06/MRE02/73). We wish to extend
our sincere thanks to the families who took part in our research.
We also thank Tobias Grossman for assistance in data analysis
and Evelyne Mercure, Teodora Gliga, Atsushi Senju, and Teresa
Farroni for helpful advice and discussion.
36 BIOL PSYCHIATRY 2009;65:31–38
M. Elsabbagh et al.
www.sobp.org/journal

Page 7

None of the authors reported biomedical financial interests or
potential conflicts of interest.
Supplementary material cited in this article is available
online.
1. Baron-Cohen S (1989): Perceptual role taking and protodeclarative
pointinginautism.BritJDevPsychol7:113–127.
2. LeekamSR,HunnisettE,MooreC(1998):Targetsandcues:Gaze-follow-
ing in children with autism. J Child Psychol Psychiatry 39:951–962.
3. Mundy P, Sigman M, Ungerer J, Sherman T (1986): Defining the social
deficitsofautism:Thecontributionofnon-verbalcommunicationmea-
sures.JChildPsycholPsychiatry27:657–669.
4. Baron-CohenS,BaldwinDA,CrowsonM(1997):Dochildrenwithautism
use the speaker’s direction of gaze strategy to crack the code of lan-
guage? Child Dev 68:48–57.
5. Leekam S, Baron-Cohen S, Brown S, Perrett D, Milders M (1997): Eye-
direction detection: A dissociation between geometric and joint-atten-
tion skills in autism. British J Dev Psychol 15:77–95.
6. Chawarska K, Klin A, Volkmar F (2003): Automatic attention cueing
through eye movement in 2-year-old children with autism. Child Dev
74:1108–1122.
7. Senju A, Tojo Y, Dairoku H, Hasegawa T (2004): Reflexive orienting in
responsetoeyegazeandanarrowinchildrenwithandwithoutautism.
JChildPsycholPsychiatry45:445–458.
8. SenjuA,TojoY,YaguchiK,HasegawaT(2005):Deviantgazeprocessing
inchildrenwithautism:AnERPstudy.Neuropsychologia43:1297–1306.
9. GriceSJ,HalitH,FarroniT,Baron-CohenS,BoltonP,JohnsonMH(2005):
Neural correlates of eye-gaze detection in young children with autism.
Cortex 41:342–353.
10. AllisonT,PuceA,McCarthyG(2000):Socialperceptionfromvisualcues:
Role of the STS region. Trends Cogn Sci 4:267–278.
11. Pelphrey KA, Morris JP, McCarthy G (2005): Neural basis of eye gaze
processing deficits in autism. Brain 128:1038–1048.
12. SchultzRT(2005):Developmentaldeficitsinsocialperceptioninautism:
The role of the amygdala and fusiform face area. Int J Dev Neurosci
23:125–141.
13. Dalton KM, Nacewicz BM, Johnstone T, Schaefer HS, Gernsbacher MA,
GoldsmithHH,etal.(2005):Gazefixationandtheneuralcircuitryofface
processing in autism. Nat Neurosci 8:519–526.
14. Dalton KM, Nacewicz BM, Alexander AL, Davidson RJ (2007): Gaze-fixa-
tion, brain activation, and amygdala volume in unaffected siblings of
individuals with autism. Biol Psychiatry 61:512–520.
15. Baron-Cohen S, Ring HA, Wheelwright S, Bullmore ET, Brammer MJ,
Simmons A, et al. (1999): Social intelligence in the normal and autistic
brain: An fMRI study. Eur J Neurosci 11:1891–1898.
16. BaileyA,LeCouteurA,GottesmanI,BoltonP,SimonoffE,YuzdaE,etal.
(1995): Autism as a strongly genetic disorder: Evidence from a British
twin study. Psychol Med 25:63–77.
17. Muhle R, Trentacoste SV, Rapin I (2004): The genetics of autism. Pediat-
rics 113:e472–e486.
18. Newschaffer CJ, Fallin D, Lee NL (2002): Heritable and nonheritable risk
factors for autism spectrum disorders. Epidemiol Rev 24:137–153.
19. DawsonG,WebbS,SchellenbergGD,DagerS,FriedmanS,AylwardE,et
al. (2002): Defining the broader phenotype of autism: Genetic, brain,
and behavioral perspectives. Dev Psychopathol 14:581–611.
20. Baron-Cohen S, Hammer J (1997): Parents of children with Asperger
syndrome: What is the cognitive phenotype? J Cog Neurosci 9:548–
554.
21. Baron-Cohen S, Ring H, Chitnis X, Wheelwright S, Gregory L, Williams S,
etal.(2006):fMRIofparentsofchildrenwithAspergersyndrome:Apilot
study. Brain Cogn 61:122–130.
22. FarroniT,CsibraG,SimionF,JohnsonMH(2002):Eyecontactdetection
inhumansfrombirth.ProcNatlAcadSciUSA99:9602–9605.
23. BatkiA,Baron-CohenS,WheelwrightS,ConnellanJ,AhluwaliaJ(2000):
Isthereaninnategazemodule?Evidencefromhumanneonates.Infant
BehDev23:223–229.
24. Johnson MH, Griffin R, Csibra G, Halit H, Farroni T, de Haan M, et al.
(2005): The emergence of the social brain network: Evidence from typi-
cal and atypical development. Dev Psychopathol 17:599–619.
25. Cox A, Klein K, Charman T, Baird G, Baron-Cohen S, Swettenham J, et
al. (1999): Autism spectrum disorders at 20 and 42 months of age:
Stability of clinical and adi-r diagnosis. J Child Psychol Psychiatry
40:719–732.
26. Maestro S, Muratori F, Cavallaro MC, Pecini C, Cesari A, Paziente A, et
al. (2005): How young children treat objects and people: An empiri-
cal study of the first year of life in autism. Child Psychiatry Hum Dev
35:383–396.
27. MarsAE,MaukJE,DowrickPW(1998):Symptomsofpervasivedevelop-
mental disorders as observed in prediagnostic home videos of infants
and toddlers. J Pediatr 132:500–504.
28. Osterling JA, Dawson G, Munson JA (2002): Early recognition of 1-year-
old infants with autism spectrum disorder versus mental retardation.
DevPsychopathol14:239–251.
29. Werner E, Dawson G (2005): Validation of the phenomenon of autistic
regression using home videotapes. Arch Gen Psychiatry 62:889–895.
30. MaestroS,MuratoriF,CavallaroMC,PeiF,SternD,GolseB,etal.(2002):
Attentional skills during the first 6 months of age in autism spectrum
disorder.JAmAcadChildAdolescPsychiatry41:1239–1245.
31. Bolton P, Macdonald H, Pickles A, Rios P, Goode S, Crowson M, et al.
(1994): A case-control family history study of autism. J Child Psychol
Psychiatry 35:877–900.
32. Landa R, Garrett-Mayer E (2006): Development in infants with autism
spectrum disorders: A prospective study. J Child Psychol Psychiatry 47:
629–638.
33. Zwaigenbaum L, Bryson S, Rogers T, Roberts W, Brian J, Szatmari P
(2005): Behavioral manifestations of autism in the first year of life. Int J
DevNeurosci23:143–152.
34. MitchellS,BrianJ,ZwaigenbaumL,RobertsW,SzatmariP,SmithI,etal.
(2006): Early language and communication development of infants
later diagnosed with autism spectrum disorder. J Dev Behav Pediatr
27:S69–S78.
35. Sullivan M, Finelli J, Marvin A, Garrett-Mayer E, Bauman M, Landa R
(2007): Response to joint attention in toddlers at risk for autism spec-
trum disorder: A prospective study. J Autism Dev Disord 37:37–48.
36. MerinN,YoungGS,OzonoffS,RogersSJ(2007):Visualfixationpatterns
duringreciprocalsocialinteractiondistinguishasubgroupof6-month-
old infants at-risk for autism from comparison infants. J Autism Dev
Disord 37:108–121.
37. CasselTD,MessingerDS,IbanezLV,HaltiganJD,AcostaSI,BuchmanAC
(2007):Earlysocialandemotionalcommunicationintheinfantsiblings
of children with autism spectrum disorders: An examination of the
broadphenotype.JAutismDevDisord37:122–132.
38. Yirmiya N, Gamliel I, Pilowsky T, Feldman R, Baron-Cohen S, Sigman M
(2006):Thedevelopmentofsiblingsofchildrenwithautismat4and14
months: Social engagement, communication, and cognition. J Child
PsycholPsychiatry47:511–523.
39. Iverson JM, Wozniak RH (2007): Variation in vocal-motor develop-
ment in infant siblings of children with autism. J Autism Dev Disord
37:158–170.
40. Toth K, Munson J, A NM, Dawson G (2006): Early predictors of commu-
nication development in young children with autism spectrum disor-
der:Jointattention,imitation,andtoyplay.JAutismDevDisord36:993–
1005.
41. Presmanes AG, Walden TA, Stone WL, Yoder PJ (2007): Effects of differ-
ent attentional cues on responding to joint attention in younger sib-
lings of children with autism spectrum disorders. J Autism Dev Disord
37:133–144.
42. Rogers S (2007): Infants with autism symptoms: New approaches to early
detection and treatment. Presented at the International Meeting for
Autism Research, Seattle, Washington, May 3–5. Available at: http://
www.autism-insar.org/docs/IMFAR2007_Program.pdf.
43. FarroniT,JohnsonMH,CsibraG(2004):Mechanismsofeyegazepercep-
tion during infancy. J Cogn Neurosci 16:1320–1326.
44. GrossmannT,JohnsonMH,FarroniT,CsibraG(2007):Socialperception
in the infant brain: Gamma oscillatory activity in response to eye gaze.
SocCognAffectNeurosci2:284–291.
45. Csibra G, Davis G, Spratling MW, Johnson MH (2000): Gamma oscilla-
tions and object processing in the infant brain. Science 290:1582–
1585.
46. Kiebel S, Tallon-Baudry C, Friston K (2005): Parametric analysis of
oscillatory activity as measured with EEG/MEG. Hum Brain Map 26:
170–177.
M. Elsabbagh et al.
BIOL PSYCHIATRY 2009;65:31–38 37
www.sobp.org/journal

Page 8

47. Kotsoni E, Mareschal D, Csibra G, Johnson MH (2006): Common–onset
visual masking in infancy: Behavioral and electrophysiological evi-
dence.JCognNeurosci18:966–973.
48. World Health Organization. (1993): Mental Disorders: A Glossary and
Guide to Their Classification in Accordance with the 10th Revision of the
International Classification of Diseases—Research Diagnostic Criteria
(ICD-10). Geneva: Author.
49. de Haan M, Nelson CA (1999): Brain activity differentiates face and
objectprocessingin6-month-oldinfants.DevPsychol35:1113–1121.
50. de Haan M, Pascalis O, Johnson MH (2002): Specialization of neural
mechanismsunderlyingfacerecognitioninhumaninfants.JCognNeu-
rosci 14:199–209.
51. Halit H, de Haan M, Johnson MH (2003): Cortical specialisation for face
processing: Face-sensitive event-related potential components in 3-
and 12-month-old infants. Neuroimage 19:1180–1193.
52. SenjuA,JohnsonM,CsibraG(2006):Thedevelopmentandneuralbasis
of referential gaze perception. Soc Neurosci 1:220–234.
53. DawsonG,WebbSJ,WijsmanE,SchellenbergG,EstesA,MunsonJ,etal.(2005):
Neurocognitiveandelectrophysiologicalevidenceofalteredfaceprocessing
inparentsofchildrenwithautism:Implicationsforamodelofabnormaldevel-
opmentofsocialbraincircuitryinautism.DevPsychopathol17:679–697.
54. GepnerB,deGelderB,deSchonenS(1996):Faceprocessinginautistics:
Evidence for a generalised deficit? Child Neuropsychol 2:123–139.
55. Orekhova E, Stroganova T, Nygren G, Tsetlin M, Posikera I, Gillberg C, et
al. (2007): Excess of high frequency electroencephalogram oscillations
in boys with autism. Biol Psychiatry 62:1022–1029.
56. Murias M, Webb S, Greenson J, Dawson G (2007): Resting state cortical
connectivity reflected in EEG coherence in individuals with autism. Biol
Psychiatry 62:270–273.
57. Herrmann C, Demiralp T (2005): Human EEG gamma oscillations in
neuropsychiatric disorders. Clin Neurophysiol 116: 2719–2733.
58. StrianoT,ReidVM,HoehlS(2006):Neuralmechanismsofjointattention
ininfancy.EurJNeurosci 23:2819–2823.
59. Yuval-GreenbergS,TomerO,KerenAS,NelkenI,DeouellL(2008):Tran-
sient Induced gamma-band response in EEG as a manifestation of min-
iature saccades. Neuron 58:429–441.
60. Csibra G, Tucker L, Volein A, Johnson M (2000): Cortical development
andsaccadicplanning:Theontogenyofthespikepotential.Neuroreport
11:1069–1073.
61. Elsabbagh M, Johnson MH (2007): Infancy and autism: Progress, pros-
pects, and challenge. Prog Brain Res 164:355–383.
38 BIOL PSYCHIATRY 2009;65:31–38
M. Elsabbagh et al.
www.sobp.org/journal