reduced change blindness suggests enhanced attention to detail in individuals with autism

Reduced change blindness suggests enhancedattention to detail in individuals with autism

Hayley Smith and Elizabeth MilneDepartment of Psychology, Western Bank, Sheffield, UK

Background: The phenomenon of change blindness illustrates that a limited number of items withinthe visual scene are attended to at any one time. It has been suggested that individuals with autismfocus attention on less contextually relevant aspects of the visual scene, show superior perceptualdiscrimination and notice details which are often ignored by typical observers. Methods: In this studywe investigated change blindness in autism by asking participants to detect continuity errors deliber-ately introduced into a short film. Whether the continuity errors involved central/marginal or social/non-social aspects of the visual scene was varied. Thirty adolescent participants, 15 with autisticspectrum disorder (ASD) and 15 typically developing (TD) controls participated. Results: The particip-ants with ASD detected significantly more errors than the TD participants. Both groups identified moreerrors involving central rather thanmarginal aspects of the scene, although this effect was larger in the TDparticipants. There was no difference in the number of social or non-social errors detected by either groupof participants. Conclusion: In line with previous data suggesting an abnormally broad attentionalspotlight and enhanced perceptual function in individuals with ASD, the results of this study suggestenhanced awareness of the visual scene in ASD. The results of this study could reflect superior top-downcontrol of visual search in autism, enhanced perceptual function, or inefficient filtering of visualinformation in ASD. Keywords: Autism, change blindness, attention, perception.

It is well known that audiences generally remainunaware of continuity errors in motion pictures(Levin & Simons, 2000). Such failure to detectunexpected and implausible events is an example ofa phenomenon known as change blindness (Levin &Simons, 1997; Rensink, O’Regan, & Clark, 1997). Sopowerful is the effect that in an experimentalinvestigation, around 50% of participants failed tonotice when a passer-by (a confederate) who hadstopped to ask them for directions was replaced byan impostor during an interruption (Simons & Levin,1998). Empirical work has demonstrated that arange of circumstances, such as a change in cameraangle, a visual distraction, a luminance change ora saccade, can give rise to change blindness (Levin &Simons, 2000). Unless attention is directed towardsthe location of the change, the change is oftendetected only after effortful search and can beconsidered akin to serial searching in visual searchtasks. The phenomenon of change blindnesshighlights that we are aware of only a small propor-tion of visual information, and that we pay attentionto some things at the expense of others (Duncan,2006). Evidence suggests that the aspects of thevisual scene that are likely to capture attentionare generally those that are relevant to currentbehaviour, are surprising or unexpected or havesocial significance. This has been demonstrated byexperiments which find that changes to faces aredetected more easily than changes to clothes (Ro,Russell, & Lavie, 2001), and that changes to items

within the visual scene that are deemed to becontextually central, e.g. a helicopter as seenthrough the window of a flying plane, are detectedmore rapidly than changes to items that are deemedto be contextually marginal, e.g. a fence behind acouple having dinner (Rensink et al., 1997). A relatedphenomenon – inattention blindness, or the failureto detect distractor stimuli that are presented at thesame time as task-orientated stimuli – indicates thatother social stimuli such as drawings of humanbodies (Downing, Bray, Rogers, & Childs, 2004;Mack & Rock, 1998) also capture attention morethan drawings of non-social stimuli. It has beensuggested that attentional bias for social stimuli maybe underpinned by specialised cortical regions whichhave developed for processing images of socialstimuli such as the fusiform face area (Kanwisher,McDermott, & Chun, 1997), and the extra-striatebody area (Downing, Jiang, Shuman, & Kanwisher,2001).

The fact that attentional capture is driven by thecontext of the visual scene highlights the importanceof top-down attentional control in guiding attention.Attentional control is known to be atypical in indi-viduals with autistic spectrum disorder. For ex-ample, it has been suggested that individuals withautism have an inefficient attentional spotlight,which results in abnormally broad attentional focusand an inability to filter extraneous stimuli (Burack,1994). Behaviourally, individuals with autism aresaid to notice features of the environment that areoverlooked by others, and to pay attention to lesscontextually relevant information. It appears thatConflict of interest statement: No conflicts declared.

Journal of Child Psychology and Psychiatry 50:3 (2009), pp 300–306 doi:10.1111/j.1469-7610.2008.01957.x

� 2008 The AuthorsJournal compilation � 2008 Association for Child and Adolescent Mental Health.Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

social stimuli, which are highly salient and quick tocapture attention even in very young typicallydeveloping children (Striano & Reid, 2006), are notprioritised in children who have autism. This isdemonstrated by many reports of children withautism preferring to interact with inanimate,mechanical stimuli rather than peers, and by datawhich illustrate that children with autism orienttheir attention more to non-social events such as amusical jack in a box than to social events such asname calling or hands clapping (Dawson, Meltzoff,Osterling, Rinaldi, & Brown, 1998; Swettenhamet al., 1998).

In addition to abnormalities in attention focus andorienting, individuals with autism show enhancedability, compared to typically developing particip-ants, on tasks which require detailed analysis of thevisual scene, such as spot-the-difference (Teunisse,Cools, van Spaendonck, Aerts, & Berger, 2001) andvisual search (Plaisted, O’Riordan, & Baron-Cohen,1998). These data have led to the suggestion thatindividuals with autism show enhanced perceptualfunction which is characterised by superiordiscrimination and increased sensitivity to simplevisual stimuli (Mottron, Dawson, Soulieres, Hubert,& Burack, 2006.). This cumulative evidence pointsto abnormalities in the way in which attention iscaptured and distributed in autism, and a percep-tual style whereby details are more salient than thegestalt (Frith, 1989). As such, differences in the wayin which changes are detected may be predicted inautism.

Given the range of evidence which indicates thatindividuals with autism are more likely to noticesmall changes to their environment than theirtypically developing counterparts, superior changedetection may be predicted in autism. However, arecent study which investigated change blindness inautism did not find any differences in the number ofchange targets detected in individuals with autismcompared to a control group (Fletcher-Watson, Lee-kam, Turner, & Moxon, 2006). In fact, contraryto predictions, the participants with ASD were slowerto detect the changes than the control group. Theparadigm used to measure change blindness in thisstudy was the ‘flicker paradigm’ which mimics theperceptual effect of a blink or a saccade by present-ing two visual scenes (which are identical apart fromthe specific change) in sequence but interspersed bya blank screen (Rensink et al., 1997). While theflicker paradigm has proven to be very useful ininducing change blindness in the laboratory, therepetitive alternation of two very similar visualscenes does not represent a common perceptualoccurrence, and it is possible that when tested with amore ecologically valid paradigm, differences inchange detection between individuals with andwithout autism may become apparent.

In the study presented here we created a short filmto which we deliberately introduced continuity

errors. Watching a film on the TV provides a morerealistic representation of information processingthan looking at alternating static images, and alsointroduces a more dynamic element to the task. Wehypothesised that when tested with a dynamic, videoparadigm, individuals with ASD would be better ableto detect continuity errors than a typically develop-ing control group. To measure empirically whichaspects of the visual scene capture attentionin individuals with ASD, we controlled whether theerrors involved items which were central or marginaland whether the changes involved objects (non-social) or the actor (social).

Method

Participants

Thirty individuals participated in the study, 15 dia-gnosed with autistic spectrum disorder (hereafterreferred to as ASD) and 15 typically developing controls(hereafter TD). All the individuals in the ASD group hadbeen diagnosed by clinicians based on DSM-IV criteria.The specific diagnoses were Asperger’s syndrome(N = 8), pure autism (N = 3) and atypical autism/PDD-NOS (N = 4). The participants with ASD were recruitedfrom a youth club which provides a service exclusivelyfor individuals with ASD. All participants who attendedthe youth club were invited to participate, and 15individuals (approximately half of the group) volun-teered. The TD participants were recruited from a localsecondary school and were selected if their parent/guardian returned a signed consent form, and if theyhad no history of any developmental disorder. Symptomprofile of the participants in the ASD group was meas-ured using the Childhood Autism Rating Scale (Schop-ler, Reichler, & Renner, 1988). An index of non-verbalability was obtained for each participant using Raven’sProgressive Matrices (Raven, Court, & Raven, 1988).The group with ASD had lower scores than the TDgroup, t (14) = –2.32, p < .05, d = .85. However, therewas no significant difference between the mean chrono-logical ages of the two groups, t (14) < 1, p >.05, d = .28.There were no significant relationships between chro-nological age and IQ in either the ASD or TD groups andno significant relationship between chronological age,IQ and CARS scores in the ASD group (Pearson’s cor-relation coefficient, p >.05). Further participant detailsare presented in Table 1.

Table 1 Participant characteristics

ASD TD

Mean chronologicalage

14y 0m 14y 6m

Standard deviation 1y 8m 1y 11mRange 11y 8m – 17y 6m 12y 6m – 17y 8mMean RPM rawscore

31.9 40.5

Standard deviation 11.4 8.6Range 9–47 26 – 52Mean CARS score 33.3Standard deviation 3.2Range 29–38

Reduced change blindness in autistic spectrum disorder 301

� 2008 The AuthorsJournal compilation � 2008 Association for Child and Adolescent Mental Health.

Design

A film was created and divided into 20 short clips whichdocumented stages in the processes of baking buns.Specific continuity errors were introduced across cutsor pans in the camera angle in 16 (80%) of the clips.Eight of the errors involved the actor (social) and eightinvolved objects (non-social) within the scene. The clipswere also counterbalanced so that eight of the errorsinvolved items that were central to the visual scene andeight involved items that were marginal to the scene. Wedefined an error as central if the changing item wascontextually relevant within the content of the clip oroccurred to a part of the scene that attracted visualattention, and marginal if the item was contextuallyirrelevant within the context of the clip. For example,the mixing spoon which changed from being a woodenspoon to being a silver spoon in the clip which describeshow the bun mixture is stirred (clip 8) was consideredas central. However, the actors’ nails changing frombeing unpainted in one shot to being painted in the next(clip 15) was defined as being marginal. Some changesthat involved the actor, i.e., in one part of the clip she iswearing blue jeans and in the other she is wearing beigecotton trousers, were also defined as being centraldespite not being contextually relevant to the clip. Thiswas based on pilot testing of typically developingparticipants which demonstrated that these items

attracted attention in the majority of the pilot particip-ants. Table 2 presents a more detailed description ofeach of the clips and the types of continuity errors. Pilotdata collected from 21 typically developing adolescentswith a group mean age of 16 years (SD = 1 year7 months) confirmed that, in line with previous data(Rensink et al., 1997), typical observers detected moreerrors when they involved central rather than marginalfeatures, v2 (1) = 37.4, p < .01.

Procedure

Informed consent was obtained from the parents orguardians of all participants, and verbal assent wasgiven by each participant. The study received ethicalapproval from the local University ethics committeebefore commencing. Participants were tested in a quietroom either in their school (TD sample) or at the youthclub (children with ASD). In addition to the experi-menter (H.S.) who was familiar to the participants withASD but not to the typically developing group, two otherpeople were present during the experiment. In the ASDgroup this was an unfamiliar research assistant, in theTD group this was their class teacher.

Participants were told that they were going to watch afilm which contained some mistakes, and that theywere to try to spot the mistakes. Examples of the type of

Table 2 A description of each of the 16 continuity errors and the number of participants who detected the error in the clip andanswered at least one of the information questions correctly

Number ofparticipants who

correctlyidentified the

change

Number ofparticipants whoanswered at leastone information

questioncorrectly

ASD TD ASD TD

Central – ActorClip 19 – Actor’s trousers change from blue jeans to beige trousers. 12 9 – –Clip 6 – Actor’s top changes colour. 12 4 15 15Clip 12 – An impostor actor replaces the actor. 11 8 15 15Clip 7 – Actor’s cardigan disappears. 13 10 15 15Central – ObjectClip 18 – Buns become doughnuts as they are being decorated. 14 14 5 12Clip 8 – The mixing spoon changes during mixing of ingredients. 9 5 14 15Clip 11 – The bowl disappears while mixture is put in the bun cases. 8 3 14 15Clip 4 – The sugar disappears as the actor describes the ingredients. 12 9 14 14Marginal – ActorClip 15 – Actor’s nails become red as the chocolate topping is melted. 10 5 15 15Clip 9 – Actor’s ring changes as the ingredients for icing are assembled. 10 4 15 15Clip 13 – Actor’s watch disappears as she cleans the kitchen. 11 1 15 15Clip 2 – A necklace appears on the actor’s neck as she wipes the surfaces. 4 2 15 14Marginal – ObjectClip 14 – A plate changes as the actor describes the icing. 9 1 15 15Clip 3 – The brand of butter changes as the actor prepares the ingredients. 3 2 14 15Clip 17 – The empty bowl vanishes while icing is spread on the buns. 12 1 14 12Clip 5 – The flour disappears as the actor prepares the equipment. 15 2 14 15No ChangeClip 1 15 14 11 15Clip 10 15 14 14 15Clip 16 15 14 13 15Clip 20 15 15 14 9

Note: The information questions were not accurately recorded for question 19 due to experimenter error.An incorrect response in the no change clips is a false positive.

302 Hayley Smith and Elizabeth Milne


continuity errors that might occur, for example that theactor’s hair might be in a different style, or that objectson the work surfaces might change, were given so thatthe participants understood what was meant by mis-takes within the clips. They were also told that not allclips contained mistakes.

After watching each clip, the film was paused andparticipants were asked whether they had noticed anymistakes or if anything unusual had happened, and ifso, to describe what had happened. Following thisquestion each participant was asked two informationquestions about the clip, such as ‘How many eggs didNina [the actor] crack into the bowl?’ This was toencourage the participants to pay attention to the con-tent of each clip and to provide a post-hoc check ofthe degree to which each group attended to and followedthe story of the clips. Each participant was given ananswer booklet which provided space to write down anddescribe any continuity error associated with the clipand also to answer the information questions associ-ated with these clips. In addition to being read aloud bythe experimenter, each question also appeared on thescreen. The experimenter checked that each participanthad written an answer, or indicated ‘don’t know’ foreach question before continuing with the next clip.

Results

Figure 1 illustrates the percentage of errors detectedby each group broken down into change type (cen-tral/marginal and social/non-social). Figure 2illustrates the distribution of correct responses ineach group. Backward elimination log-linear modelswere applied to the 2 (group) · 2 (central or marginalerror) · 2 (error involving person or object) · 2 (errordetected or missed) data (Figure 1). This 4-way log-linear analysis produced a final model that retaineda 3-way interaction between group, central or mar-ginal error type, and error detected or missed, v2 (1) =6.45, p < .01. This interaction reflected the fact thatthe participants with ASD were more likely to detectthe error than the TD group, and, although bothgroups detected more errors in the central than themarginal condition, this effect was greater in the TDgroup. This interpretation was confirmed by two

chi-square analyses comparing the number of errorsdetected in each group which indicated that theparticipants with ASD detected more errors than theTD group in both the central and marginal condi-tions; v2 (1) = 15.2, p < .01, and v2 (1) = 53.1, p <.01 respectively. Odds ratios revealed that the ASDparticipants were 2.93 times more likely to detectcentral errors than the TD group and 8.47 timesmore likely to detect marginal errors than the TDgroup.

A further two chi-square analyses, comparing thenumber of errors detected in the central and mar-ginal conditions in each group, confirmed that therewas a significant association between error type(central and marginal) and whether or not the errorwas detected in both groups; TD group, v2 (1) = 34.4,p < .01, ASD group, v2 (1) = 5.6, p < .05. Oddsratios revealed that the TD participants were 5.63times more likely to detect central over marginalerrors whereas the ASD participants were only 1.95times more likely to detect central over marginalerrors. There was no significant association betweenthe number of errors detected and whether the errorinvolved the person or an object within the scene,and no interaction between this and group.

Only five of the participants with ASD answered aninformation question correctly for clip 18 (seeTable 2). The reason for this discrepancy is notknown, however the error detection data werere-anlaysed after removing clip 18 from the analysis.The pattern of results was similar to that of the fulldata: The model retained a significant 3-way inter-action between group, central or marginal, andnumber of errors detected, v2 (1) = 4.6, p < .05. Theparticipants with ASD detected more errors than theTD group, both in the central, v2 (1) = 16.6, p < .01,and in the marginal v2 (1) = 53.1, p < .01 conditions.However, in this analysis, there was no longer asignificant association between the number of errorsdetected in the central and marginal conditions bythe ASD group v2 (1) = 3.5, p = .063; however, thisassociation remained significant in the TD group v2

(1) = 23.9, p < .01.

0

20

40

60

80

100

Social Non-Social Social Non-SocialType of Error

% o

f Err

ors

Det

ecte

d

ASD

TD

Central Marginal

Figure 1 A graph showing the percentage of errorsdetected by each group across error type

0

1

2

3

4

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Number of Errors Detected

Num

ber

of P

artic

ipan

ts

ASDTD

Figure 2 Histograms illustrating the total number oferrors detected by participants in the ASD and TDgroups



No one in the ASD group described an error in aclip in which no error was present, although threeTD participants incorrectly identified one error in(different) clips when no error was present. Someparticipants erroneously described errors in clipswhich did contain errors but did not detect theactual error, i.e., they were making guesses as to thenature of continuity errors. The participants withASD made 4 (1.6%) false positives of this nature, theTD participants made 25 (10.4%). Chi square testindicated that there was a significant associationbetween group and the number of false positivesof this type, v2 (1) = 16.2, p < .01. Note that thenumber of errors detected was lower in clips two andthree, which may reflect decreased expectancy oferrors (clip one did not contain an error), or increasedlearning over trials in the ASD group. There was nosignificant difference between the number ofinformation questions answered correctly by eachgroup: ASD group = 79.8%, TD group = 79.3%, v2 (1)<1, p > .1. There was no relationship between autismsymptomatology as measured by the CARS and thetotal number of errors detected, r(15) = –.117, p > .1.

Discussion

The aim of this study was to measure change blind-ness in individuals with ASD by creating a stimuluswhich contained deliberate continuity errors. Sucherrors are not uncommon in films and televisionprogrammes, yet they typically remain undetected bythemajority of the audience. Explanations of this andother forms of change blindness imply that a limitednumber of features – usually the most contextuallysalient features (Rensink et al., 1997) – are availablefor conscious comparison and updating. The resultspresented here support existing findings from typicalobservers (e.g. Rensink et al., 1997; Simons & Levin,1998), as the overall error detection was low in the TDgroup (less than 50%), and both groups detectedmore errors in the central aspects of the scenecompared to the marginal aspects. Whether thechange involved the actor or the objects in the visualscene did not significantly affect error detection ineither group.

The participants with ASD showed superior abilityto detect the continuity errors compared with the TDgroup. These data demonstrate that when embeddedwithin a dynamic and ecologically valid display,individuals with ASD are more able than typicallydeveloping individuals to detect unexpected changeswhen they are instructed to do so. Such enhanceddetection of continuity errors suggests that morefeatures of the visual scene receive attentional focusand are therefore available for continuous compar-ison in individuals with ASD. This suggestion is inline with previous literature which has documentedan abnormally broad attentional spotlight (Burack,1994), and studies showing enhanced visual search

in individuals with autism (Plaisted et al., 1998).Although visual search and detection of continuityerrors may not reflect equivalent perceptual skills,both indicate enhanced feature detection in autism.

Despite beingpredictedbybehavioural andexistingexperimental evidence, the finding that participantswith autism detected significantly more errors thanthe typically developing participants, directly contra-dicts the results of previous investigation of changedetection in ASD (Fletcher-Watson et al., 2006).However there were several differences between thetasks used in the different studies which may explainthese contradictory results. Firstly, Fletcher-Watsonet al. used the flicker paradigm which involves pre-senting two very similar static images with a blankscreen in between. Here we investigated changedetection by measuring the number of continuity er-rors detected in a film. Asking participants to watch afilm is a more ecologically valid way of measuringattention in autism as it more closely resembles thereal world experience of perceiving the visual scene.Furthermore abnormalities in the top-down control ofattention to dynamic (butnot static) stimuli havebeenreported (Greenaway&Plaisted,2005). Inorder to testwhether the dynamic dimension of the video taskcompared to themore static flicker paradigmexplainsthe difference in findings between the two studies itwould be necessary to test change detection usingboth the flicker paradigm and a video paradigm in thesame participants.

Secondly, in the study presented here, in additionto detecting continuity errors, the participants wereasked questions about each clip such as ‘Whatingredients did Nina get out of the cupboard?’ Thatis, they were required to monitor the content of theclip, thus engaging in a deeper level of processing (cf.Craik & Lockheart, 1972), than merely looking at thesuperficial features of the scene. In the Fletcher-Watson et al. study the only task requirement was todetect the changes in the images. The attentionaldemands, and load on cognitive control, of the cur-rent study were therefore higher than in the Fletch-er-Watson et al. study. Increases in cognitive loadhave been shown to increase the influence ofdistractors on goal directed processing (Lavie, Hirst,de Fockert, & Viding, 2004). It is possible thatsuperior change detection in autism could arisebecause of the additional task demands of allocatingattention to the process of encoding and remember-ing semantically meaningful information about thescene. There is some evidence of deficits in cognitivecontrol in children with autism (Nyden, Gillberg,Hjelmquist, & Heimann, 1999), therefore, the dualnature of the task – answer questions about the clipand look for errors in the clip – may have increasedcognitive load to a greater extent (proportionally) inthe participants with ASD than the TD controls, thusincreasing awareness of the distracting items ofthe visual scene in this group more than in the TDgroup.



Finally, both the study presented here and thestudy presented by Fletcher-Watson et al. arepartially confounded by expectation effects asparticipants were told in advance to look for a changeor to look for something unusual. However, changeswere present in all trials of theFletcher-Watson study,whereas here, changes were present in only 80% oftrials in this study. Importantly, in the studypresented here participantswere told that not all clipscontained errors and were given no indication as tothe proportion of clips that did or did not contain anerror. Therefore the degree of expectationwas lower inthis study than that of Fletcher-Watson et al. Any ofthe above task differences, alone or in combination,could be crucial in influencing the different resultsbetween the current study and that of Fletcher-Watson et al. (2006). Furtherwork,measuring changedetection using both the flicker and film paradigms inthe same participants, will be useful to ascertainexactly which parameters are crucial.

Eye-tracking studies suggest that individuals withautism show different patterns of eye movementscompared with controls, especially when scanningsocial images (for a review see Boraston & Blake-more, 2007). It is therefore possible that the parti-cipants with ASD fixated different parts of the visualscene compared with the typically developingparticipants, and that this may contribute to theirenhanced ability to detect continuity errors. Futurestudies which measure the location of fixationsduring change detection experiments will be usefulin testing this suggestion. However, the distributionof fixation across the visual scene is driven by bothattentional and perceptual factors, so we wouldexpect eye-tracking data to corroborate our conclu-sion that the individuals with ASD spend more timelooking at marginal aspects of the scene than the TDgroup, rather than to explain it.

Both groups of participants identified more errorsin the central condition than in the marginal condi-tion, however this effect was much smaller in theparticipants with ASD (and became non-significantwhen the analysis was repeated after removing datafrom clip 18). It is possible that the smaller ratiobetween detection of central and marginal errors andthe overall superior error detection in the particip-ants with ASD, may reflect the fact that individualswith ASD have a different sense of what is ‘central’and what is ‘marginal’ in the visual scene comparedwith typically developing individuals. Joint atten-tion, i.e., the ability to establish a shared focus ofattention with another person, is critical for thedeveloping infant’s ability to learn to differentiatethose aspects of the environment which are worthyof focused attention and those which are not. Lack ofjoint attention is a key symptom of autism (Baird etal., 2000); it is therefore possible that the resultspresented here reflect atypical information process-ing originating from atypical joint attention. Alter-natively, accurate change detection is supported by

visual attention, which in turn is influenced by thesaliency of visual features. Given the many studieswhich demonstrate superior perceptual function inautism, it is possible that a more bottom-upexplanation, i.e., enhanced discrimination, or acombination of bottom-up (enhanced perception)and top-down (superior attentional control/abnormally broad attentional spotlight) factors, givesrise to the findings reported here. Future studieswhich investigate the influence of high and lowsaliency of visual features on change detection inautism and typical development will be important inteasing apart these hypotheses.

These data provide an insight into the experienceof perception in individuals with ASD. Changeblindness has been explained by the fact that only alimited number of items enter visual awareness atany one time. Thus, superior sensitivity to changesuggests that the number of items which enter visualawareness is greater in individuals with ASD than intypically developing individuals. While enhanceddiscrimination and awareness of environmentalstimuli can at times be valuable, such perceptualsensitivity may cause difficulty in situations where itis advantageous to devote attentional resources tothe more contextually relevant aspects of the envir-onment and to ignore extraneous and distractinginformation.

Individuals with autism often describe feelings ofsensory overload and over-arousal which have beeninterpreted as a result of inefficient attentional gatingand inefficient suppression of activity elicited byirrelevant distractors (Belmonte, 2000). Belmonteand Yurgelun-Todd (2003) have proposed a sequenceof factors which may lead to perceptual overload inautism. In their model, hyper-arousal, defined asabnormally intense primary sensory processingcoupled with impaired early selection of relevantstimuli, leads to overloading of higher level cognitiveprocesses (Belmonte&Yurgelun-Todd, 2003, p. 660).This model can be applied to the performance of theindividuals with ASD in this study, i.e., enhancedperception, coupled with reduced perceptual gating,gives rise to an increased ability to detect continuityerrors. However, whether the explanation for this isdue to superiority in certain cognitive skills, namelysuperior top-down control of visual search and/orenhanced perceptual function, or a deficit, i.e., re-duced attentional filtering, remains to be established.

In conclusion, the results of this study challengeprevious reports of typical change detection in indi-viduals with ASD. We have found increased ability todetect continuity errors, suggesting enhanced per-ception of local features and /or increased awarenessof the visual scene in individuals with ASD. In light ofexisting evidence that top-down attentional control isatypical in ASDwhenmeasuredwith dynamic but notstatic stimuli (Greenaway & Plaisted, 2005), we sug-gest that asking participants to detect changes in avideo presentation rather than in the flicker paradigm



provides a more realistic approximation to atten-tional control and is a more appropriate paradigm forfurther study of attentional capture in ASD.

Acknowledgements

We would like to thank the members of DONYWASPand Hall Cross School for participating in this study;Mrs S. Welch, Mr I. Day and Mr M. Cattral for theirhelp in recruiting participants, and Nina Hutchings.

Correspondence to

Elizabeth Milne, Department of Psychology, WesternBank, Sheffield, S10 2TN, UK; Tel: +44 (0) 1142226558; Fax: +44 (0) 114 2766515l; Email: [email protected]

References

Baird, G., Charman, T., Baron-Cohen, S., Cox, A.,Swettenham, J., Wheelright, S., & Drew, A. (2000). Ascreening instrument for autism at 18 months of age:A six year follow-up. Journal of the American Acad-emy of Child and Adolescent Psychiatry, 39, 694–702.

Belmonte, M.K. (2000). Abnormal attention in autismshown by steady-state visual evoked potentials.Autism, 4, 269–285.

Belmonte, M.K., & Yurgelun-Todd, D.A. (2003). Func-tional anatomy of impaired selective attention andcompensatory processing in autism. Cognitive BrainResearch, 17, 651–664.

Boraston, Z., & Blakemore, S.-J. (2007). The applica-tion of eye-tracking technology in the study of autism.Journal of Physiology, 581, 893–898.

Burack, J.A. (1994). Selective attention deficits inpersons with autism – preliminary evidence of aninefficient lens. Journal of Abnormal Psychology, 103,535–543.

Craik, F.I.M., & Lockheart, R.S. (1972). Levelsof processing: A framework for memory research.Journal of Verbal learning and verbal behaviour, 11,671–684.

Dawson, G., Meltzoff, A.N., Osterling, J., Rinaldi, J., &Brown, E. (1998). Children with autism fail to orientto naturally occurring social stimuli. Journal ofAutism and Developmental Disorders, 28, 479–485.

Downing, P.E., Bray, D., Rogers, J., & Childs, C. (2004).Bodies capture attention when nothing is expected.Cognition, B27–B38.

Downing, P.E., Jiang, Y., Shuman, M., & Kanwisher,N. (2001). A cortical area selective for visual pro-cessing of the human body. Science, 293, 2470–2473.

Duncan, J. (2006). Eps mid-career award, 2004. Brainmechanisms of attention. The Quarterly Journal ofExperimental Psychology, 59, 2–27.

Fletcher-Watson, S., Leekam, S.R., Turner, M.A., &Moxon, L. (2006). Do people with autistic spectrumdisorder show normal selection for attention? Evid-ence from change blindness. British Journal ofPsychology, 97, 537–554.

Frith, U. (1989). Autism: Explaining the enigma. Oxford:Blackwell Scientific Publications.

Greenaway, R., & Plaisted, K. (2005). Top-down atten-tional modulation in autistic spectrum disorders isstimulus-specific.Psychological Science,16, 987–994.

Kanwisher, N., McDermott, J., & Chun, M.M. (1997).The fusiform face area: A module in human extra-striate cortex specialised for face perception. TheJournal of Neuroscience, 17, 4302–4311.

Lavie, N., Hirst, A., de Fockert, J.W., & Viding, E. (2004).Load theory of selective attention and cognitivecontrol. Journal of Experimental Psychology: General,133, 339–354.

Levin, D.T., & Simons, D. J. (1997). Failure to detectchanges to attended objects in motion pictures.Psychonomic Bulletin & Review, 4, 501–506.

Levin, D.T., & Simons, D. J. (2000). Perceiving stabilityin a changing world: Combining shots and integratingviews in motion pictures and the real world. MediaPsychology, 2, 357–380.

Mack, A., & Rock, I. (1998). Inattentional blindness.London: MIT Press.

Mottron, L., Dawson, M., Soulieres, I., Hubert, B., &Burack, J.A. (2006). Enhanced perceptual function-ing in autism: An update and eight principles ofautistic perception. Journal of Autism and Develop-mental Disorders, 36, 27–43.

Nyden, A., Gillberg, C., Hjelmquist, E., & Heimann, M.(1999). Executive/attention deficits in boys withAsperger syndrome, attention disorder and reading/writing disorder. Autsim, 3, 213–228.

Plaisted, K., O’Riordan, M., & Baron-Cohen, S. (1998).Enhanced visual search for a conjunctive target inautism: A research note. Journal of Child Psychologyand Psychiatry, 39, 777–783.

Raven, J., Court, J., & Raven, J. (1988). Raven’sstandard progressive matrices. London: H.K. Lewis& Co. Limited.

Rensink, R.A., O’Regan, J.K., & Clark, J.J. (1997). Tosee or not to see: The need for attention to perceivechanges in scenes. Psychological Science, 8, 368–373.

Ro, T., Russell, C., & Lavie, N. (2001). Changing faces: Adetection advantage in the flicker paradigm. Psycho-logical Science, 12, 94–99.

Schopler, E., Reichler, R.J., & Renner, B.R. (1988). Thechildhood autism rating scale. Los Angeles: WesternPsychological Services.

Simons, D.J., & Levin, D.T. (1998). Failure to detectchanges to people in a real-world interaction. Psych-onomic Bulletin & Review, 5, 644–649.

Striano, T., & Reid, V.M. (2006). Social cognition in thefirst year. Trends in Cognitive Sciences, 10, 471–476.

Swettenham, J., Baron-Cohen, S., Charman, T., Cox,A., Baird, G., Drew, A., Rees, L., & Wheelright, S.(1998). The frequency and distribution of spontan-eous attention shifts between social and non-social stimuli in autistic, typically developing andnonautistic developmentally delayed infants. Journalof Child Psychology and Psychiatry, 39, 747–753.

Teunisse, J.-P., Cools, A., van Spaendonck, K., Aerts, F.,& Berger, H. (2001). Cognitive styles in high-function-ing adolescents with autistic disorder. Journal ofAutism and Developmental Disorders, 31(1), 55–66.

Manuscript accepted 7 May 2008



reduced change blindness suggests enhanced attention to detail in individuals with autism

Documents