Atypical attentional behaviour is emerging as a robust
marker of ASD in children as young as 12 months in
both retrospective (1, 2) and prospective longitudinal
studies (3). Young children as well as high-functioning
adults with ASD exhibit unusual looking behaviour not
only to people but to objects as well (4), suggesting
atypical looking behaviour is not strictly a social
problem but rather a broader problem with the strategic
control of visual attention. This atypical attentional
behaviour could contribute to many impairments that
are characteristic of ASD. For example, failure to
attribute meaning to eye gaze as a cue to direct attention
could lead to a failure to engage in joint attention, a
social activity in which two people share an experience
about a commonly attended object that is considered to
be an important precursor to language acquisition (5, 6).
An impairment of joint attention is thought to reflect
social difficulties, however it may be one of many
manifestations of impaired strategic control of visual
attention (7, 8). Klin et al. (9) suggest that from a very
young age children with ASD misdirect attention in
their environment, which would consequently impede
learning, as the acquisition of skills and knowledge
depends on how well children pay attention to their
environment (10). This impairment in strategic control
over the orientation of visual attention compromises the
ability of the child to selectively direct attention to
pertinent and relevant locations in the visual field. Thus,
we suggest that visual orienting is intact in ASD but that
the control of it is impaired. This specific impairment is
also supported by similar patterns of findings in
visuomotor coordination.
Visual Orienting
Visual attention can be directed either by focusing the
eyes, or foveating, on a specific location or by choosing to attend to a location in peripheral vision. One model
of visual attention is based on the metaphor of a
spotlight beam that is directed to a specific location, and
events within the beam are detected with enhanced
efficiency (11, 12). Within this context, Posner (11)
introduced the notion of the orienting of attention in
which the directing of attention to a given location
facilitated detection of a target at that location, but
impeded detection of a target at another location. Visual
cues are used to direct attention to the cued location
either overtly, with eye movements, or covertly, without
eye movements (11). Cues that elicit shifts of attention
automatically, or unconsciously, are considered to be
exogenous, as the shift is in response to the physical
properties of the stimulus. An example is a flash of
light, which attracts attention to the location of the flash.
Cues that elicit voluntary shifts of attention are
endogenous as the shift is in response to the symbolism
or meaning of the cue. An example is an arrow, which
directs attention to a secondary location away from the
actual arrow. Deficits in exogenous orienting would
suggest a basic problem in attending, whereas deficits in
endogenous orienting, with spared exogenous orienting,
would suggest a problem in the control of attention (13).
Exogenous Orienting among Persons with ASD
The findings from two studies of exogenous
orienting were taken as initial evidence of general
orienting deficits among persons with ASD. Casey et al.
(14) found that a group of adults with ASD were slower
overall to respond but that the facilitation effects with
stimulus onset asynchronies (SOAs) at both 100 ms and
800 ms were even larger than those of the comparison
group on an exogenous orienting task with predictive
peripheral cues. Similarly, Harris et al. (15) found that
children with ASD (mean age 7.5 years) showed a
larger facilitation effect at 1000 ms SOA, whereas
typically developing children showed a larger effect at
200 ms SOA. However, the implications of these two
studies are limited by methodological concerns
regarding the participants and the tasks. With regard to the participants, in both experiments, participants with
ASD were matched to typically developing persons
only on the basis of chronological age, and had mean
full scale IQ scores that were 45 (14) and 28 (15) points
lower. Thus, the differences in performance were likely
associated with the a priori differences in IQ, and
subsequently of developmental level(16, 17). With
regard to the tasks, none of the groups in either of the
studies showed the typical inhibition of return (IOR)
effects, in which participants are slower to detect a
target at the cued location following a long SOA,
because attention has drifted away from the cued
location in the absence of a target and which are a
hallmark of exogenous orienting. The failure to elicit an
IOR effect may have been because the maintenance of
the peripheral cue on the screen during the full duration
of the trial, before and during target onset, may have
held attention longer at the cued location than if the cue
was a brief flash (18, 19). In addition, the long duration
of the cue, coupled with its predictability that may
actually make it meaningful suggests that the tasks were
effectively endogenous rather than exogenous.Differences were not found in studies of exogenous
orienting when the problems with the initial studies
were eliminated. For example, Iarocci and Burack (20)
found that low functioning children and adolescents
with ASD (mean CA 11.6 years; mean MA 7.2 years)
and typically developing children matched on mental
age performed similarly on an exogenous orienting task
in which the peripheral cue and central fixation did not
overlap with target onset. A 50 ms peripheral cue,
followed by 150 ms blank screen, which served to elicit
disengagement, was presented before the target
appeared. Peripheral cues were non-predictive as the
ratio of congruent to incongruent cues was 1:1. There
were no significant differences between the groups on
overall reaction time (RT) and no interaction between
facilitation effects and group, as both groups
demonstrated the expected benefits of congruent cues.Similarly, Randolph et al. (21) found facilitation
effects in exogenous orienting among both a group of
high functioning adolescents with ASD and a group of
comparison participants matched on chronological age
and IQ. The duration of the peripheral cue was 30 ms,
and targets could appear in one of four locations instead
of the standard two. Peripheral cues were nonpredictive
as the ratio of congruent to incongruent cues
was 1:3. Both groups showed similar facilitation effects
at an SOA of 100 ms and IOR at an SOA of 800 ms.
Thus, the patterns of findings across studies of
exogenous orienting in which issues of group matching
and stimulus presentation are appropriately controlled
do not support a general impairment in shifting of
attention among persons with ASD.
Endogenous Orienting among Persons with ASD
In contrast to the findings on exogenous orienting
tasks, persons with ASD display consistent impairments
in shifting attention on endogenous orienting tasks. For
example, Wainwright-Sharp and Bryson (22) found
different facilitation effects in the two groups between
high functioning adults with ASD and age and IQ
matched typical adults on a Posner task in which
centrally located arrow cues remained onscreen for 100
ms or 800 ms and were predictive with a congruentincongruent
ratio of 4:1. The persons with ASD did not
show facilitation effects to rapidly presented cues when
a voluntary shift of attention was required. Regardless
of cue duration, the typical adults responded faster to
congruent than to incongruent trials and the magnitude
of this effect was the same at both cue durations,
whereas the adults with ASD only displayed facilitation
effects in the long cue duration, and the magnitude of
this effect was larger than for the typically developing
group at the same duration. Wainwright-Sharp and
Bryson (22) concluded that the participants with ASD
were impaired in either disengaging or shifting of
attention, or in the voluntary coordination of attention
and motor systems. The finding of facilitation effects at
the longer SOA on the endogenous task suggests that
the process of orienting to symbolic cues is not absent,
but merely slowed down.Burack et al. (13) suggested that the deficit exhibited
on endogenous orienting reported by Wainwright-Sharp
and Bryson (22) might indicate that persons with ASD
are slower to interpret the meaning of the symbolic cue.
Consistent with this hypothesis, Randolph et al. (21)
found no ASD-related deficits on an endogenous
orienting task in which predictive arrows (75%
congruent) appeared on screen for 280 ms or 980 ms;
these trial durations were long enough for the
participants with ASD to demonstrate facilitation
effects.
Perception versus Response Selection in Visual
Orienting
The evidence does not appear to support a general
orienting deficit among persons with ASD, but rather a
delayed orienting effect to endogenous cues. The
presence of the orienting effect at longer SOAs could
reflect a slower reading of the cue (13), but reports of
slower overall reaction times (14, 21-23) should not be
dismissed as irrelevant. It may be indicative of other
slowed responses that are not observed. For example,
Landry et al. (24) found that children with ASD were
able to read rapidly presented cues as well as typically
developing children, but were less able to execute a fast
enough response in terms of shifting visual attention and required longer trials to exhibit effects.
Accordingly, Landry et al. suggested that if a person
with an ASD were slower at executing that endogenous
shift of attention in response to the cue, the onset of the
target might disrupt the in-progress endogenous shift
and begin a new exogenous shift directly to the target.
Visuomotor Planning in Autism Spectrum Disorders
The finding that endogenous cues may not be
defective among persons with ASD when they are
superseded by the appearance of the peripheral target is
consistent with findings of impairments in the voluntary
control of motor responses, rather than of motor
impairments, on reach-to-grasp (25), visual pursuit (26),
and saccadic eye movements (7, 27). The findings with
these other visuomotor skills indicate that the problems
exhibited by persons with ASD on endogenous visual
orienting tasks reflect general impairments in strategic
goal-oriented behaviour. For example, Mari et al. (25)
reported that lower functioning children with ASD were
slower than higher functioning children with ASD or
typically developing children, though accurate, in their
performance on a reach-to-grasp task. The lower
functioning children showed less simultaneous
activation of reaching and grasping, and this delay
increased as a function of the precision needed to
perform the task. Mari et al. further reported that higher
functioning children with ASD, relative to typically
developing children, executed very fast movements, as
though once the action plan was finalized it must be
performed quickly in order to avoid any disruptive
feedback mechanisms. Concordantly, Masterton and
Biederman (28) found that children with ASD were
unable to visually guide reaching movements very
efficiently. These findings broadly suggest that children
with ASD have difficulty using external feedback to
guide behaviour, at least with respect to visuomotor
activity.Eye movements also appear to be atypical among
children with ASD. For example, in a study of visual
pursuit, Takarae et al. (26) found that children with
ASD were impaired relative to typically developing
children on both the open loop stage of visual pursuit
which entails the initiation of eye movement and is
sensory driven, and the closed loop stage which entails
the ability to sustain the movement, and is feedback
driven (26); although the impairments differed as a
function of stage. In the open loop stage, the
impairments were only found for pursuit in the right
hemifield, whereas closed loop stage impairments were
found bilaterally. For children with ASD, visual pursuit
performance was correlated with motor praxis, as
measured with the Grooved Pegboard, however for
typically developing children, visual pursuit performance was correlated with motor speed, as
measured by Finger Tapping. This suggests the presence
of multiple impairments, both in visual perception for
the right hemifield and in motor coordination that might
impact control over both eye movements and shifting
visual attention.Kemner et al. (7) speculated that poor control over
eye movements might underlie abnormalities in visual
attention among children with ASD. Based on Hermelin
and O’Conner’s reports of atypical looking behaviour
during the course of experiments, Kemner et al. (7)
measured the eye movements during a visual oddball
task among children with ASD, ADHD or dyslexia, and
typically developing children. The children were
presented frequent, rare, and novel, stimuli, and were
familiarized with the frequent and rare stimuli at the
beginning of the task. The children with ASD made
more eye movements between stimuli than all other
groups, and during the presentation of the frequent
stimuli than ADHD and typical groups. Further, unlike
the typical children and children with dyslexia, the
frequency of eye movements of children with ASD did
not differ as a function of stimulus type. Although the
children with ASD appeared to look at all stimuli as
though it were novel, the high frequency of eye
movements between stimuli suggests that they had a
generalized difficulty controlling eye movements.In a followup, Kemner et al. (29) found no differences
between children with ASD and typically developing
children in a sample with a mean age of 10 years on
smooth pursuit and saccadic eye movements. Minshew
et al. (30) also found no differences between highfunctioning
young adults with ASD and typically
developing peers on a visually guided saccade task,
finding instead that the participants with ASD made
more errors on an anti-saccade task and an oculomotor
delayed response task. However, Takarae et al. (27)
found reduced saccade gain, defined as the ratio of
saccade amplitude over target distance, with normal
saccade latencies in high-functioning adolescents/
young adults with Asperger syndrome, but not autism
(mean age 16 years), suggesting that the deficit might
be highly specific, subtle, and differ across ASD
subgroups.
CONCLUSION
If persons with ASD exhibited generalized
impairments in visual saccades, this would indicate
impairment in oculomotor control. However, the
findings of the experiments on visual saccade suggest
that oculomotor control is generally intact, and like the
findings on tasks of visual orienting discussed
previously, excludes any bottom-up explanations of
atypical visual attention behaviour. Similar conclusions were drawn by Hadjikhani et al. (31), who reported that
early sensory visual areas are normally organized in the
brains of persons with ASD. Rather, the evidence
suggests that visuomotor control falls apart when it is
goal-driven and/or feedback dependent rather than
simply sensory driven. Visual orienting differences
exhibited by persons with ASD stem from poor strategic
control over visual attention and eye movements, and
are a symptom of poor control over visuomotor
coordination in general. With the emergence of
orienting difficulties in ASD by 12 months old as well
as the specificity (3), visual orienting may provide
paediatricians with new symptoms to watch for and
neuroscience with new clues to the early neurological
manifestation of ASD.