B

SI

a

ARRAA

KMLGCC

1

‘eett(1m

1

esTuiatD

0d

Neuropsychologia 47 (2009) 1274–1281

Contents lists available at ScienceDirect

Neuropsychologia

journa l homepage: www.e lsev ier .com/ locate /neuropsychologia

ig heads, small details and autism

arah White ∗, Helen O’Reilly, Uta Frithnstitute of Cognitive Neuroscience, University College London, 17 Queen Square, London WC1N 3AR, UK

r t i c l e i n f o

rticle history:eceived 28 May 2008eceived in revised form 11 December 2008ccepted 7 January 2009vailable online 16 January 2009

eywords:acrocephaly

ocal

a b s t r a c t

Autism is thought to be associated with a bias towards detail-focussed processing. While the cognitivebasis remains controversial, one strong hypothesis is that there are high processing costs associatedwith changing from local into global processing. A possible neural mechanism underlying this processingstyle is abnormal neural connectivity; specifically reduced structural or functional connectivity betweenbrain regions might lead to good exemplar-based processing but poor generalisation. Abnormal neuralconnectivity has also been suggested to account for the increased incidence of macrocephaly in autism(increased head/brain size). The present study therefore investigated the effect of head size on the abilityto switch between global and local processing in autism. 49 high-functioning 7–12 year olds with autism

lobalonnectivityentral coherence

(12 with macrocephaly) were compared to 25 normally developing children in their performance on aLocal-Global Switching task. Those children with autism who also had macrocephaly showed a greaterprocessing cost when switching into global processing, or ‘zooming out’, than both the remaining childrenwith autism and the control children. A second experiment revealed that macrocephaly in the context ofnormal development is not associated with difficulty switching into global processing but rather occurs

callyal con

in children who are physimarker of abnormal neur

. Introduction

Individuals with autism are thought to be good at processinglocal’ details of information rather than applying the context andxtracting the ‘global’ meaning, an idea that has been stronglyndorsed by the autism community (e.g. Gerland, 1997). This abilityo perceive the elements rather than the whole was first attributedo a lack of drive for meaning, termed ‘weak central coherence’Frith, 1989) and more recently described as a local bias (Happé,999; Happé & Frith, 2006). There has, however, been little agree-ent on the mechanism that might support this processing style.

.1. What is the cognitive mechanism behind a local bias?

When reviewing the evidence to date, the majority of studiesmploy designs in which both local and global processing are pos-ible simultaneously and are in direct competition with each other.o pick a few notable examples, the Embedded Figures Task (firstsed in autism research by Shah & Frith, 1983) requires the partic-

pant to both ignore and inhibit Gestalt principles when viewingpicture that is designed to elicit global processing, and process

he local elements instead. This is also the case in traditional Blockesign tasks (first used in autism research by Shah & Frith, 1993),

∗ Corresponding author. Tel.: +44 20 7679 1168; fax: +44 20 7813 2835.E-mail address: s.white@ucl.ac.uk (S. White).

028-3932/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.oi:10.1016/j.neuropsychologia.2009.01.012

large. Macrocephaly in the context of autism may therefore be a biologicalnectivity, and of a local processing bias.

© 2009 Elsevier Ltd. All rights reserved.

where the global Gestalt image must be rejected in order to segmentthe design into component elements. Hierarchical figure tasks per-formed under divided attention (first used in autism research byMottron, Belleville, & Menard, 1999) similarly require participantsto attend to both local and global aspects of the stimuli.

A few studies of the local processing bias in autism haveattempted to separate the local and global constraints of tasks. Forexample, the segmented version of the Block Design task (Shah &Frith, 1993) removes the global image and therefore requires theparticipant to simply match the local elements, and hierarchicalfigures performed under selective attention (first used in autismresearch by Ozonoff, Strayer, McMahon, & Filloux, 1994) direct aparticipant’s attention towards either local or global aspects of thestimuli. Interestingly, although local and global task componentscan be isolated, it seems that these paradigms are unable to cap-ture the essence of a local processing bias; they indicate that globalprocessing is intact rather than deficient and local processing isnormal rather than enhanced in autism.

Instead, those tasks that appear to be most sensitive in detect-ing a local bias therefore either pit global and local processingagainst each other or require fast online responses that are capa-ble of picking up this bias (see Happé & Frith, 2006 for a review

of many recent studies). This body of research has led researcherswith quite different theoretical ideas to agree that, when both localand global processing are simultaneously possible, local process-ing often takes precedence over global processing, resulting in abias away from global and towards local stimuli (Happé & Frith,

cholog

2Sgttnwtfpiiatw

gbo2itcnlaiawamtwrcithf

oaTsuadahttucsstsaStsarbtl

S. White et al. / Neuropsy

006; Mottron, Dawson, Soulieres, Hubert, & Burack, 2006; Plaisted,aksida, Alcantara, & Weisblatt, 2003). While those studies findingroup differences have been useful in identifying and supportinghe idea of a bias towards an alternative processing style in autism,he presence of both local and global stimuli simultaneously doesot allow for an analysis of how this processing style arises andhat the mechanism behind it might be. Furthermore, in addition

o assessing processing capacity, the majority of tasks do not requireast online responses, allowing higher-order strategy use to alsolay a role in task performance; this confounds results and makes

t extremely hard to interpret which factors contribute to both pos-tive and negative findings. It is therefore still unknown how a biasway from global and towards local processing comes about, givenhat both global and local processing appear to be normal in autismhen processing in only one of these domains is required.

More recently, a handful of studies have together begun to sug-est a mechanism for this bias, in terms of difficulties broadening,ut not narrowing, the spread of attention or an inability to shiftut of local processing and into global processing (Mann & Walker,003; Rinehart, Bradshaw, Moss, Brereton, & Tonge, 2001). Once

ndividuals with autism are processing a small stimulus or element,hey are slower and have more difficulty in zooming out and pro-essing a large stimulus or element presented subsequently, butot vice versa. The suggestion being made is therefore that both

ocal and global processing can be performed normally; however,deficit in global processing may be seen when such process-

ng is required immediately after local processing, thus requiringfast switch from local to global. This deficit may also be presenthen a task involving local-global competition is open-ended, such

s the Embedded Figures Task, as an individual with a local biasay remain stuck in local processing once they have shifted into

his processing mode; the cost of switching into global processingill be high, making it easy to ignore global stimuli and possibly

esulting in an enhancement in local processing ability relative toontrols. While this shifting or switching deficit may at first be rem-niscent of the executive function deficits commonly attributed tohis disorder (see Hill, 2004 for a review), the crucial differenceere is that the shifting deficit is proposed to be unidirectional:

rom local to global but not vice versa.A particularly elegant novel paradigm, known as rapid serial

bject transformation (RSOT), has recently been introduced into theutism literature (López, Torres, & Valdés-Sosa, 2002; Valdés-Sosa,orres, Iglesias, & López, submitted for publication). This task avoidsome of the caveats of more traditional tasks: higher-order strategyse plays little role as fast online processing is necessary; and localnd global processing are temporally separated which allows for aetailed analysis of the information the participant is processingnd therefore a clearer interpretation of the results. The task usesierarchical figures in a novel way so that they can be transformedo uncover either global or local meaningful components, ratherhan both simultaneously, presented as consecutive pairs of stim-li in an attentional blink paradigm. Stimuli can be presented underonditions that do not require attentional switching (e.g. a globaltimulus followed by a global stimulus) or those that do (e.g. a localtimulus followed by a global stimulus). Participants often haverouble reporting a second stimulus that appears in quick succes-ion (300–400 ms) after a first, as few attentional resources remainvailable to direct towards it, hence the attentional blink (Raymond,hapiro, & Arnell, 1992). Moreover, if the participant is requiredo switch between two different modes of processing for the twotimuli, an additional attentional cost is required, lengthening the

ttentional blink and further reducing the probability of correctlyeporting the second stimulus (Ward, 1982). This paradigm haseen used to show that individuals with autism are less likelyhan controls to correctly identify a global stimulus following aocal stimulus but not vice versa (López, Torres, & Valdés-Sosa,

ia 47 (2009) 1274–1281 1275

2002; Valdés-Sosa et al., submitted for publication), indicating thatswitching from local into global processing has a higher processingcost for individuals with autism than typical observers.

1.2. Heterogeneity in the local bias

The literature regarding a local bias in autism is by no meansunanimous in its support, however; negative findings have beenreported alongside the positive ones (e.g. Brian & Bryson, 1996;Edgin & Pennington, 2005; Hoy, Hatton, & Hare, 2004; Kaland,Mortensen, & Smith, 2007; Ropar & Mitchell, 1999; Schlooz etal., 2006). Furthermore, one study found impaired rather thanthe expected enhanced performance on three visuospatial tasks(Burnette et al., 2005). While some of these findings appear to resultfrom methodological issues, such as group matching, task designand instructions, one explanation for such mixed results is the ideaof heterogeneity of this processing bias in the autistic population;perhaps one subset of individuals with autism may exhibit a localbias more than other subsets and the proportion of these individu-als may vary in different studies. It is also possible that a local biasmay not be a single construct but have a number of different pro-cesses contributing to it (e.g. a global deficit independent of a localenhancement), and that only subsets of individuals with autism dis-play an abnormality in any one aspect of it (Booth, Charlton, Hughes,& Happé, 2003; Lopez, Leekam, & Arts, 2008). While heterogene-ity has rarely been studied, two recent experiments have reportedthe presence of an abnormally local bias in almost all childrenwith autism tested (Jarrold, Gilchrist, & Bender, 2005; Pellicano,Maybery, Durkin, & Maley, 2006), while others have found only asmall proportion of individuals with autism to show this process-ing style (Edgin & Pennington, 2005; Jarrold & Russell, 1997; vanLang, Bouma, Sytema, Kraijer, & Minderaa, 2006). From these latterstudies, the studies which fail to find group differences and manyother studies finding group differences, it is possible to infer fromthe group means and standard deviations that the range of perfor-mances seen in autism and control groups tend to overlap greatly,indicating that heterogeneity in local/global processing may welloccur in autism.

1.3. What is the neural mechanism behind a local bias?

How is the subgroup of individuals with autism to be identi-fied that may be particularly liable to show a local processing bias?At the biological level, it has been suggested that a local bias mayresult from abnormal neuronal connectivity, due to either structuralor functional differences. These might involve a lack of synchroni-sation in activation between relevant brain areas (Brock, Brown,Boucher and Rippon, 2002) or reduced long-range and increasedshort-range physical connectivity (Just, Cherkassky, Keller andMinshew, 2004), both resulting in a lack of binding of parts intowholes; or numerous and inefficient feedback connections result-ing in a lack of top-down modulation of early sensory processingand a lack of integration of sensory processing with cognitive moni-toring (Frith, 2003). These ideas are supported by a growing numberof functional imaging studies showing reduced connectivity anda lack of top-down modulation particularly between frontal cor-tex and other brain regions (e.g. Bird, Catmur, Silani, Frith andFrith, 2006; Horwitz, Rumsey, Grady and Rapoport, 1988; Just,Cherkassky, Keller, Kana and Minshew, 2007; Koshino et al., 2008).All of these ideas predict that the abnormal connectivity would giverise to a preserved or enhanced ability for exemplar-based informa-

tion processing, in addition to a reduced ability to generalise acrossexamples or process information in context, reminiscent of a locallybiased style of processing. While these accounts were intendedto explain autism as a whole, it is possible that, like a local bias,there is heterogeneity in connectivity and that this may relate to

1276 S. White et al. / Neuropsychologia 47 (2009) 1274–1281

Table 1Means (and standard deviations) for group characteristics, experiment 1.

Control group (25) ASD without macrocephaly group (37) Macrocephalic ASD group (12)

Gender (M:F) 19:6 35:2 10:2Age (years) 9.9 (1.3) 9.7 (1.6) 9.2 (1.2)Verbal IQ 115 (16) 105 (21) 107 (16)Performance IQ* 104 (12) 96 (13) 94 (9)Head circumference (z-scores)*** +.72 (1.07) +.62 (.98) +2.66 (.81)Height (cm)a 143 (9) 137 (11) 141 (8)

Clinical diagnosis – 7 Autism 0 Autism19 AS 8 AS11 ASD 4 ASD

3Dib

Social*** 3.4 (2.1) 13.7 (4.2) 11.0 (3.9)Communication*** 3.5 (1.6) 14.8 (3.5) 14.9 (3.4)Repetitive behaviour*** .3 (.4) 5.4 (2.7) 4.2 (2.3)

* p < .05.

dw

wbatl2w9FSpfim(eMscB

aaahAfot(atli2nacosi

b

*** p < .001.a Data only available for 19, 21 and 8 children in each group, respectively.b Cut-offs for impairment are 10, 8 and 3 for the three domains, respectively.

ifferences in the degree of local bias shown by different individualsith autism.

One of the more consistent neurobiological findings in autism,hich was also noticed by Kanner (1943), is of increased head and

rain size. This is seen to different degrees in different individualsnd the distribution of head size in the autism population appearso be normal but broader than in the typically developing popu-ation, with a higher mean (Lainhart et al., 2006). Approximately0% of the autistic population are thought to have macrocephaly,hen defined as having a head circumference greater than the

7th percentile of the normal population (Bailey et al., 1995;ombonne, Roge, Claverie, Courty, & Fremolle, 1999; Lainhart, 2003;tevenson, Schroer, Skinner, Fender and Simensen, 1997) and twoostmortem studies of increased brain weight have supported thisnding (Bailey et al., 1993; Bauman & Kemper, 1985). This enlarge-ent appears to be general across the whole of the cerebral cortex

Hazlett et al., 2005) and be heritable, being present in the par-nts and siblings of individuals with autism (Lainhart et al., 2006;iles, Hadden, Takahashi, & Hillman, 2000). Very recently, a pos-

ible genetic mutation in the PTEN gene has been suggested as theause in some cases of macrocephaly in autism (Butler et al., 2005;uxbaum et al., 2007).

However, macrocephaly cannot normally be detected untilpproximately 2 years of age (Courchesne et al., 2001; Lainhart etl., 1997; Stevenson, Schroer, Skinner, Fender, & Simensen, 1997)lthough brain imaging studies indicate that the increased rate ofead growth starts around 12 months of age (Courchesne, Carper, &kshoomoff, 2003; Hazlett et al., 2005). There is evidence that feed

orward connections are established very early in normal devel-pment whereas feedback connections are continually refinedhrough neuronal elimination processes of pruning and apoptosisPrice et al., 2006), in order to eliminate faulty feedback connectionsnd optimise co-ordinated neural functioning. It is plausible thathere may be a decrease in these elimination processes in autism,eading to an excess number of synapses and coinciding with thencreased rate of head growth from about 12 months of age (Frith,003). This suggestion is therefore consistent with the abnormaleuronal connectivity hypotheses mentioned above and thereforelso with a local processing bias; this connection between abnormalonnectivity and macrocephaly is however speculative. In support

f this hypothesis, a computational model of autism has been con-tructed in which a lack of generalisation results from an increasen units (Cohen, 1994; Gustafsson, 1997).

Macrocephaly in autism has been linked to a number ofehavioural features with mixed results, including increased social

impairment and language delay (Lainhart et al., 2006) but alsoto reduced social impairment and improved adaptive functioning(Dementieva et al., 2005; Lainhart et al., 1997). At present, onlyone study has attempted to relate brain size with cognitive func-tion however; Deutsch and Joseph (2003) examined the IQ profilesof individuals with autism and found that relatively high perfor-mance IQ compared to verbal IQ was associated with larger headsizes.

1.4. Relating cognition and neural mechanisms

It is possible therefore that increased head size is related to thelocal bias seen in autism. Given that macrocephaly is not universalin autism, some individuals may display more of a local bias thanothers, which might help to account for the variable results seenin the literature on cognitive style in autism. As macrocephaly is soclear an index by which to identify individual cases, one approachis to focus entirely on such individuals and establish whether thereis any possible connection with a difficulty broadening attention or‘zooming out’. To this end, the effect of head size on the ability toshift from local into global processing was examined.

2. Experiment 1

2.1. Method

Ethical approval for the study was received from the UCLResearch Ethics Committee and consent was obtained from the par-ents of all participants prior to inclusion in the study. 49 childrenwith autism spectrum disorder (ASD) and 25 normally developingchildren took part in the study, aged 7–12 years (see Table 1). Thechildren with ASD had all received independent diagnoses froma qualified clinician and met criteria for an ASD on the Develop-mental, Dimensional and Diagnostic Interview (3Di: Skuse et al.,2004) at the time of this study, while none of the control chil-dren did (F(2,68) > 45.17, p < .001). The 3Di measure is similar tothe Autism Diagnostic Interview (ADI-R; Lord, Rutter, & Le Couteur,1994) with which it correlates highly (r for each area of the triadranges from .53 to .64; Skuse et al., 2004). The majority of childrenwith ASD also attended mainstream schools and had IQs within the

normal range so can be considered high-functioning. This age andability group was chosen in order to maximise the likelihood ofdetecting macrocephaly and of the child being able to complete thetask. Each child’s head circumference was measured with a flexi-ble tape measure and converted into z-scores adjusted for age and

S. White et al. / Neuropsychologia 47 (2009) 1274–1281 1277

al-loc

sJtvAsttawva(

hFhfgocswatii

((TsiS‘ilfmw(et(g

Fig. 1. Time line for a single trial of the glob

ex according to available norms (Farkas, 1994; used by Deutsch &oseph, 2003); in line with these norms, all children were selectedo be Caucasian. These norms are based on approximately 50 indi-iduals in each 1-year gender specific band, taken from a Northmerican sample. While local norms would have been preferable,uch data was not available. The children with ASD were split intohose with macrocephaly (head circumference equal to or greaterhan 1.88 S.D. above the normative mean; 24% of present sample)nd those without; the distribution of head size in the autism groupas unimodal however, with the median shifted towards higher

alues. Two children in the control group could also be describeds having macrocephaly but were not removed from the sampleF(2,71) = 20.89, p < .001).

All three groups were matched for age (F(2,71) = .92),eight (F(2,45) = 2.09) and verbal IQ (WISC-III: Wechsler, 1992;(2,71) = 2.19). However, there was a trend for both ASD groups toave slightly lower verbal IQs and the groups were not matched

or performance IQ (F(2,73) = 3.94, p = .024), again with both ASDroups scoring lower. The ASD groups were well matched to eachther for IQ though (vIQ t(47) = .42; pIQ t(47) = .58), enabling validomparisons between these groups; as the control group was noto closely matched and there was a wide range of ages and IQsithin the groups, both age and IQ were entered into all analyses

s covariates. The critical comparison however was between thewo well-matched ASD groups; the control group was interestingn as much as performance by these children was similar to thosen the ASD group without macrocephaly.

The previously mentioned task, designed by Valdés-Sosa et al.submitted for publication) and referred to here as the Local-GlobalLG) Switching task, was employed to assess local processing bias.he experiment was run on a laptop computer using E-primeoftware (Psychology Software Tools, Inc.) and the child sat approx-mately 40 cm from the screen. The stimuli were five letters (E, H, P,, U) presented either as a single large ‘global’ letter or many small

local’ letters. Each global stimulus was composed of many mean-ngless small elements (a rectangle divided in two by a horizontaline) arranged in a 3 × 5 grid (approximately 7◦ × 18◦ visual angle),or example the letter H was made up of two columns of five ele-

ents joined by a single element between them. Each local stimulusas composed of 15 small identical letters arranged in a 3 × 5 grid

each approximately 2◦ × 3◦ visual angle), for example five rowsach containing three letter Ss. Each trial consisted of two consecu-ive letters, presented in all four possible combinations of the globalG) and local (L) stimuli (GG, LL, GL and LG) with blanks (a 3 × 5rid of small meaningless elements) presented before, between and

al (GL) condition in the LG Switching task.

after the letters (see Fig. 1; a short animation of a trial can alsobe found at http://www.icn.ucl.ac.uk/dev group/LGSwitch.htm).Global stimuli had a shorter duration (50 ms) than local stimuli(200 ms) in order to produce similar levels of accuracy (80%) acrossthese two stimulus types (data from control adults; Valdés-Sosa etal., submitted for publication). The interstimulus interval, duringwhich a blank was presented, therefore also varied for the differ-ent stimulus types (350 ms after global and 200 ms after local) inorder to keep the stimulus onset asynchrony consistent (400 ms).The blanks presented at the beginning and end of each trial weredisplayed for 300 ms each.

The children were first introduced to the global and local stim-uli that they would see during the task and then completed fourblocked conditions in the following order; global-global (GG),local-local (LL), global-local (GL) then local-global (LG). Each blockconsisted of 5 slow practice trials followed by 25 randomly orderedtest trials with corrective feedback after each; these 25 trials pre-sented each possible combination of the 5 letters once each. Thechild was asked to identify both letters in each trial and select thecorresponding letters on the computer keyboard; this response wasnot under time constraints. The proportion of responses on whichthe second letter was correctly identified was calculated for onlythose trials on which the first letter was correctly identified. Itwas necessary to only include these trials in order to increase thelikelihood that the attentional blink had occurred as an incorrectresponse to the first letter indicates that the child has not attendedto that stimulus. An incorrect response to the second letter whenthe first letter has been correctly identified however indicates thatthe attentional blink was longer than the stimulus onset asynchronyand attentional resources could therefore not be directed towardsthe second letter.

2.2. Results

As an attentional blink could only be assumed to have takenplace on those trials in which a child responded correctly to the firstletter, only these trials were suitable for analysis; different numbersof trials were therefore entered into the analysis for different chil-dren. To ensure that any further analyses involved the same meannumber of trials in each group, the number of correct responses

to the first letter were therefore also analysed. While there was aslight trend for the children with ASD to respond correctly to fewerfirst letters than the controls, group differences were not signifi-cant after accounting for the effects of age and IQ (F(2,67) = 1.32;see Table 2).

1278 S. White et al. / Neuropsychologia 47 (2009) 1274–1281

Table 2Means (and standard deviations) for LG Switching task, experiment 1. Values given are raw data, not accounting for age and IQ.

Control group (25) ASD without macrocephalygroup (37)

MacrocephalicASD group (12)

No. of correct responses to first letter (25 max)

GG 16 (5) 13 (5) 11 (4)LL 20 (5) 18 (7) 16 (5)GL 16 (5) 14 (6) 12 (5)LG 18 (6) 17 (6) 15 (5)

P(.29)(.24)(.24)(.28)

otIoniwftv

rg(tt(s(awiswttc

rpsHgwg

2

wta

AbT

talt

2001; Valdés-Sosa et al., submitted for publication). Indeed, in allthree of these studies the group differences arise from a difficultyswitching into global processing or zooming out, but not from anenhanced ability to switch into local processing or zooming in, sup-porting the original conceptualisation of the local processing bias as

roportion of correct responses to second letter(given first letter was correct)

GG .64LL .79GL .55LG .56

No significant differences were found between the groups on anyf the four main conditions for the proportion of correct responseso the second letter (GG, LL, GL or LG; F(2,68) < 1.7; see Table 2).n order to calculate each individual’s ability to switch into a levelf processing (local or global), the difference between the relevanton-switching and switching condition was calculated; for switch-

ng into global, each participant’s score was calculated as GG-LG,hile for switching into local, LL-GL was used. This gave a measure

or each child of the cost of switching into a level in comparison tohat child’s own ability of processing within that same level, pro-iding a baseline to control for individual variation in performance.

A 2 × 3 repeated measures ANCOVA comparing all three groupsevealed no main effect of either condition (F(1,68) = 1.06) orroup (F(2,68) = 1.48) but a significant interaction between the twoF(2,68) = 3.05, p = .027; a one-tailed test was used given the direc-ion of this interaction was predicted). Post hoc tests indicatedhat this interaction was present between the two ASD groupsF(1,44) = 5.19, p = .014; one tailed), and the same pattern was alsoeen between the macrocephalic ASD group and the control groupF(1,32) = 3.05, p = .045; one tailed). Conversely, no sign of an inter-ction was present between the control group and the ASD childrenithout macrocephaly (F(1,57) = .13). For the two ASD groups, this

nteraction arose from a difference between the groups in the cost ofwitching into global processing (F(1,44) = 8.93, p = .003; one tailed),ith the macrocephalic group having a greater switching cost. For

he controls & macrocephalic ASD children, none of the post hocests within or between groups were significant due to a completerossover. This pattern of results is shown in Fig. 2.1

Additionally, within the ASD group only, head size was cor-elated to the cost of switching into global processing (r = .36,= .007; one tailed); children with larger heads had a greater cost of

witching into global processing than children with smaller heads.owever, head size was unrelated to the cost of switching intolobal processing in the control group (r = −.22). Furthermore, itas unrelated to the cost of switching into local processing in either

roup (ASD r = .02; control r = −.06).2

.3. Discussion

The relationship between head size and local bias in autismas investigated in a selected subgroup of individuals with ASD:

hose with macrocephaly. As predicted, the results from the inter-ction between head size and the LG Switching task confirm the

1 This analysis was repeated with a reduced sample of the non-macrocephalicSD group (n = 22) with head circumference z-scores of one or less; the interactionetween the three groups was still significant (F(2,53) = 2.98, p = .030; one tailed).his excludes any individuals who may have been on the borderline of macrocephaly.2 This task was part of a larger battery of tasks including tests of executive func-

ion and theory of mind ability. While no specific predictions were made regardingrelationship between these tests and macrocephaly, group differences and corre-

ations were analysed. Head circumference was found to be unrelated to either ofhese two cognitive domains.

.70 (.23) .64 (.17)

.83 (.18) .70 (.25)

.52 (.17) .52 (.19)

.55 (.30) .39 (.19)

hypothesis that a bias towards local processing may be charac-teristic of individuals with ASD and atypically large heads. Thecorrelation between head circumference and the cost of switch-ing into global processing also supports the idea that the sameprocess may underlie these two measures and be present to dif-ferent extents in different individuals. This raises the possibilitythat the increased head size, and therefore presumably brain size,seen in a proportion of individuals with ASD is a biological markerfor a locally oriented processing style. This was predicted from theidea that macrocephaly may result from abnormal neuronal con-nectivity early in development and lead to good exemplar-basedprocessing but poor generalisation and integration of information(Brock et al., 2002; Frith, 2003; Just et al., 2004).

The different conditions in the LG Switching task allow us toexamine where the difference in this processing style may lie. Indi-viduals with macrocephaly appear to have difficulties switchinginto a global level of processing or zooming out; they show a greatercost and presumably have to use greater resources to overcome thisattentional barrier. There was no evidence however that individ-uals with macrocephaly had an enhanced ability to process localstimuli or had difficulty processing global stimuli per se; this isconsistent with findings of normal global and local processing inautism under selective attention conditions (Plaisted, Swettenham,& Rees, 1999). Rather they had difficulty when having to switchout of local processing and into global processing, a finding whichsupports previous studies (Mann & Walker, 2003; Rinehart et al.,

Fig. 2. Interaction between the cost of switching into global versus local processingin the macrocephalic ASD group versus the ASD group without macrocephaly andthe controls. Error bars represent standard error of the mean. Cost of switching inthe global condition is calculated as the difference between staying in global andswitching into global processing mode (GG-LG) and the cost of switching in the localcondition is calculated in a similar manner. Low scores therefore indicate efficientswitching.

chologia 47 (2009) 1274–1281 1279

oeobpwuHttsumaA

twiicpAact(siptopt

mvaibsoc

3

3

mrwdaMNat(t

3

d

Table 3Means (and standard deviations) for group characteristics, experiment 2.

Group withoutmacrocephaly (13)

Macrocephalicgroup (12)

Age (years) 9.5 (1.5) 9.6 (1.2)Non-verbal IQ 111 (11) 108 (8)Head circumference

(z-scores)***+.27 (.49) +2.58 (.49)

Height (cm)* 140 (10) 148 (9)GG-LG +.09 (.25) −.04 (.21)

S. White et al. / Neuropsy

riginating from some form of global deficit (weak central coher-nce; Frith, 1989). This could still result in a preference for localver global processing and therefore in a local enhancement inehaviour, through an inability to zoom out or switch into globalrocessing, thus remaining in local processing. Furthermore, thisould also be compatible with the idea of impaired top-down mod-lation of incoming information in autism (e.g. Bird et al., 2006;orwitz et al., 1988; Just et al., 2007; Koshino et al., 2008), with

op-down modulation being possible but costly and therefore nothe default. The important difference to theories of local bias beinguggested here, however, is that this processing style may be partic-larly characteristic of those individuals with ASD who also showacrocephaly. Just as there is heterogeneity in macrocephaly, there

ppears to also be related heterogeneity in the local bias seen inSD.

These results also have implications for the executive functionheory of autism (see Hill, 2004 for a recent review). This theoryould have predicted impaired performance across both switch-

ng conditions, but the present results find little support for thisdea. Those individuals with macrocephaly had difficulties specifi-ally with switching in a particular direction, from local into globalrocessing, but not vice versa. Furthermore, those individuals withSD but normal head size were performing similarly to controlscross all conditions, if anything performing slightly more effi-iently on the switching conditions (see Fig. 2). This would indicatehat any problems seen in the ASD population with switching sete.g. Ozonoff et al., 2004) are likely to occur at a more conscious,trategic level, while the present task was assessing an individual’sntrinsic, automatic capacity to switch quickly between levels ofrocessing. It is possible that the impairments seen on executiveasks may be more to do with prompting oneself to switch levelsr choosing when to do so, while possessing an intact ability toerform the switch when prompted externally as in the currentask.

One question that remains is whether a local bias is a feature ofacrocephaly rather than autism and is therefore found in all indi-

iduals with macrocephaly regardless of whether they also have anutism diagnosis or not. Obviously, as the rate of macrocephaly isncreased in ASD, this would still lead to higher rates of a locallyiased processing style in ASD. In order to investigate this pos-ibility, experiment 1 was repeated but this time in two groupsf normally developing children: those with and without macro-ephaly.

. Experiment 2

.1. Method

In order to locate a group of normally developing children withacrocephaly, approximately 300 children aged 7–11 years were

ecruited through local primary schools. The parents of 12 childrenith macrocephaly agreed for them to take part and 13 further chil-ren without macrocephaly were also selected to be matched onge (t(23) = .026) and non-verbal IQ (Raven’s Standard Progressiveatrices: Raven, Court, & Raven, 1988; t(23) = 1.00) (see Table 3).one of these participants were reported or were known to haveny developmental disorders or family history of such difficul-ies. As expected, the groups differed on the basis of head z-scorest(23) = 11.79, p < .001). All 25 children performed the LG Switchingask and had their height measured.

.2. Results

A 2 × 2 repeated measures ANOVA revealed a main effect of con-ition in the LG Switching task (F(1,23) = 5.97, p = .023), with the

LL-GL +.28 (.33) +.26 (.25)

* p < .05.*** p < .001.

global condition requiring a lower switching cost. A trend towardsa main effect of group was also found (F(1,23) = 3.57, p = .072),with the children with macrocephaly performing slightly betteracross both conditions than the children without macrocephaly,although this effect was not significant for either condition indi-vidually (t(23) = .19 and .84). Importantly, no interaction betweenthe two was revealed (F(1,23) = .26). Group means were also notedto be similar to both the typically developing group and the ASDgroup without macrocephaly in the first experiment. However, adifference was found between the groups in height, with the chil-dren with macrocephaly being taller than the children without(t(23) = 2.17, p = .041; see Table 3).

3.3. Discussion

Macrocephaly appears to arise for different reasons in autismand in normal development. The results in this second experi-ment indicate that in normal development, macrocephaly does notresult in a locally oriented information processing style, as a rela-tive advantage in the local or disadvantage in the global switchingcondition was not observed. Furthermore, the difference in heightbetween the groups indicates that macrocephaly may occur as partof normal development in children who are physically bigger inoverall size. This cannot be said to be the case in autism however;the children with ASD and macrocephaly in the first experimentwere matched to the other two groups on the basis of height. Macro-cephaly in the presence of ASD therefore appears to be a sign ofabnormal neurodevelopment, resulting in a local bias. It would beof interest for future studies to address this relationship in otherclinical populations.

4. General conclusion

In a population of high-functioning children with ASD, a sub-group with macrocephaly was selected and used to test thehypothesis that large heads are associated with a local bias, pos-sibly due to abnormal neural connectivity. This was confirmedthrough the use of the LG Switching task. Those children with ASDand macrocephaly showed a bias towards local processing on theLG Switching task, portrayed as a greater processing cost whenswitching from local into global processing. Furthermore, this find-ing was restricted to ASD; macrocephaly in normal development,where abnormal neural connectivity is ruled out, did not resultin a local bias. If these results prove to be robust and replicablewith more direct measures of brain size, this would imply that alocal bias is restricted to a subgroup of individuals with ASD andshould not be considered universal to ASD. The tentative hypoth-

esis can therefore be proposed that head size may be a biologicalmarker of abnormal neural connectivity, resulting in a locally ori-ented processing style, and may provide a useful endophenotypefor investigating the genetic basis of a subgroup of individuals withASD.

1 cholog

A

uwNtgR0

R

B

B

B

B

B

B

B

B

B

B

C

C

C

D

D

E

FF

F

FGG

H

H

H

H

H

280 S. White et al. / Neuropsy

cknowledgements

We would like to thank Mitchell Valdés-Sosa for allowing us tose his LG Switching task, the many teachers, parents and childrenho made this research possible, and Elisabeth Hill, Sam Gilbert andilli Lavie for helpful comments and discussion at different stages of

he project. This research was funded by Medical Research Councilrants G78/8085 (S.W.) and G9617036 (U.F.) and a joint Medicalesearch Council/Economic and Social Research Council grant PTA-37-27-0107 (S.W.).

eferences

ailey, A., Le Couteur, A., Gottesman, I., Bolton, P., Simonoff, E., Yuzda, E., et al. (1995).Autism as a strongly genetic disorder: Evidence from a British twin study. Psy-chological Medicine, 25, 63–77.

ailey, A., Luthert, P., Bolton, P., Le Couteur, A., Rutter, M., & Harding, B. (1993). Autismand megalencephaly. Lancet, 341, 1225–1226.

auman, M., & Kemper, T. L. (1985). Histoanatomic observations of the brain in earlyinfantile autism. Neurology, 35, 866–874.

ird, G., Catmur, C., Silani, G., Frith, C., & Frith, U. (2006). Attention does not modulateneural responses to social stimuli in autism spectrum disorders. Neuroimage, 31,1614–1624.

ooth, R., Charlton, R., Hughes, C., & Happé, F. (2003). Disentangling weakcoherence and executive dysfunction: Planning drawing in autism and attention-deficit/hyperactivity disorder. Philosophical Transactions of the Royal Society ofLondon Series B-Biological Sciences, 358, 387–392.

rian, J. A., & Bryson, S. E. (1996). Disembedding performance and recognitionmemory in autism/PDD. Journal of Child Psychology and Psychiatry and AlliedDisciplines, 37, 865–872.

rock, J., Brown, C. C., Boucher, J., & Rippon, G. (2002). The temporal binding deficithypothesis of autism. Development and Psychopathology, 14, 209–224.

urnette, C. P., Mundy, P. C., Meyer, J. A., Sutton, S. K., Vaughan, A. E., & Charak, D.(2005). Weak central coherence and its relations to theory of mind and anxietyin autism. Journal of Autism and Developmental Disorders, 35, 63–73.

utler, M. G., Dasouki, M. J., Zhou, X. P., Talebizadeh, Z., Brown, M., Takahashi, T. N.,et al. (2005). Subset of individuals with autism spectrum disorders and extrememacrocephaly associated with germline PTEN tumour suppressor gene muta-tions. Journal of Medical Genetics, 42, 318–321.

uxbaum, J. D., Cai, G., Chaste, P., Nygren, G., Goldsmith, J., Reichert, J., et al. (2007).Mutation screening of the PTEN gene in patients with autism spectrum disordersand macrocephaly. American Journal of Medical Genetics. Part B, NeuropsychiatricGenetics, 144B, 484–491.

ohen, I. L. (1994). An artificial neural network analogue of learning in autism. Bio-logical Psychiatry, 36, 5–20.

ourchesne, E., Carper, R., & Akshoomoff, N. (2003). Evidence of brain overgrowthin the first year of life in autism. The Journal of the American Medical Association,290, 337–344.

ourchesne, E., Karns, C. M., Davis, H. R., Ziccardi, R., Carper, R. A., Tigue, Z. D., etal. (2001). Unusual brain growth patterns in early life in patients with autisticdisorder: An MRI study. Neurology, 57, 245–254.

ementieva, Y. A., Vance, D. D., Donnelly, S. L., Elston, L. A., Wolpert, C. M., Ravan,S. A., et al. (2005). Accelerated head growth in early development of individualswith autism. Pediatric Neurology, 32, 102–108.

eutsch, C. K., & Joseph, R. M. (2003). Brief report: Cognitive correlates of enlargedhead circumference in children with autism. Journal of Autism and DevelopmentalDisorders, 33, 209–215.

dgin, J. O., & Pennington, B. F. (2005). Spatial cognition in autism spectrum dis-orders: Superior, impaired or just intact? Journal of Autism and DevelopmentalDisorders, 35, 729–745.

arkas, L. (1994). Anthropometry of the head and face. New York, NY: Raven Press.ombonne, E., Roge, B., Claverie, J., Courty, S., & Fremolle, J. (1999). Microcephaly

and macrocephaly in autism. Journal of Autism and Developmental Disorders, 29,113–119.

rith, C. (2003). What do imaging studies tell us about the neural basis of autism?Novartis Foundation Symposium, 251, 149–166, discussion 166–176, 281–297

rith, U. (1989). Autism: Explaining the enigma. Oxford: Blackwell.erland, G. (1997). A real person: Life on the ourside. Souvenir Press Ltd.ustafsson, L. (1997). Inadequate cortical feature maps: A neural circuit theory of

autism. Biological Psychiatry, 42, 1138–1147.appé, F. (1999). Autism: Cognitive deficit or cognitive style? Trends in Cognitive

Sciences, 3, 216–222.appé, F., & Frith, U. (2006). The weak coherence account: Detail-focused cogni-

tive style in autism spectrum disorders. Journal of Autism and DevelopmentalDisorders, 36, 5–25.

azlett, H. C., Poe, M., Gerig, G., Smith, R. G., Provenzale, J., Ross, A., et al. (2005). Mag-netic resonance imaging and head circumference study of brain size in autism:

Birth through age 2 years. Archives of General Psychiatry, 62, 1366–1376.

ill, E. L. (2004). Executive dysfunction in autism. Trends in Cognitive Sciences, 8,26–32.

orwitz, B., Rumsey, J. M., Grady, C. L., & Rapoport, S. I. (1988). The cerebral metaboliclandscape in autism. Intercorrelations of regional glucose utilization. Archives ofNeurology, 45, 749–755.

ia 47 (2009) 1274–1281

Hoy, J. A., Hatton, C., & Hare, D. (2004). Weak central coherence: A cross-domainphenomenon specific to autism? Autism, 8, 267–281.

Jarrold, C., Gilchrist, I. D., & Bender, A. (2005). Embedded figures detection in autismand typical development: Preliminary evidence of a double dissociation in rela-tionships with visual search. Developmental Science, 8, 344–351.

Jarrold, C., & Russell, J. (1997). Counting abilities in autism: Possible implicationsfor central coherence theory. Journal of Autism and Developmental Disorders, 27,25–37.

Just, M. A., Cherkassky, V. L., Keller, T. A., Kana, R. K., & Minshew, N. J. (2007). Func-tional and anatomical cortical underconnectivity in autism: Evidence from anFMRI study of an executive function task and corpus callosum morphometry.Cerebral Cortex, 17, 951–961.

Just, M. A., Cherkassky, V. L., Keller, T. A., & Minshew, N. J. (2004). Cortical activa-tion and synchronization during sentence comprehension in high-functioningautism: Evidence of underconnectivity. Brain, 127, 1811–1821.

Kaland, N., Mortensen, E. L., & Smith, L. (2007). Disembedding performance inchildren and adolescents with Asperger syndrome or high-functioning autism.Autism, 11, 81–92.

Kanner, L. (1943). Autistic disturbances of affective contact. Nervous Child, 2, 217–250.

Koshino, H., Kana, R. K., Keller, T. A., Cherkassky, V. L., Minshew, N. J., & Just, M. A.(2008). fMRI investigation of working memory for faces in autism: Visual codingand underconnectivity with frontal areas. Cerebral Cortex, 18, 289–300.

Lainhart, J. E. (2003). Increased rate of head growth during infancy in autism. TheJournal of the American Medical Association, 290, 393–394.

Lainhart, J. E., Bigler, E. D., Bocian, M., Coon, H., Dinh, E., Dawson, G., et al. (2006).Head circumference and height in autism: A study by the Collaborative Pro-gram of Excellence in Autism. American Journal of Medical Genetics. Part A, 140,2257–2274.

Lainhart, J. E., Piven, J., Wzorek, M., Landa, R., Santangelo, S. L., Coon, H., et al.(1997). Macrocephaly in children and adults with autism. Journal of the AmericanAcademy of Child and Adolescent Psychiatry, 36, 282–290.

Lopez, B., Leekam, S. R., & Arts, G. R. (2008). How central is central coherence? Pre-liminary evidence on the link between conceptual and perceptual processing inchildren with autism. Autism, 12, 159–171.

López, K., Torres, R., & Valdés-Sosa, M. (2002). Medición directa del tiempo de tránsitoatencional entre distintos niveles de figuras jerárquicas: observadores normalesy autistas. Revista CENIC Ciencias Biológicas, 33(3), 111–117.

Lord, C., Rutter, M., & Le Couteur, A. (1994). Autism diagnostic interview-revised:A revised version of a diagnostic interview for caregivers of individuals withpossible pervasive developmental disorders. Journal of Autism and DevelopmentalDisorders, 24, 659–685.

Mann, T. A., & Walker, P. (2003). Autism and a deficit in broadening the spread ofvisual attention. Journal of Child Psychology and Psychiatry and Allied Disciplines,44, 274–284.

Miles, J. H., Hadden, L. L., Takahashi, T. N., & Hillman, R. E. (2000). Head circumferenceis an independent clinical finding associated with autism. American Journal ofMedical Genetics, 95, 339–350.

Mottron, L., Belleville, S., & Menard, E. (1999). Local bias in autistic subjects asevidenced by graphic tasks: Perceptual hierarchization or working memorydeficit? Journal of Child Psychology and Psychiatry and Allied Disciplines, 40, 743–755.

Mottron, L., Dawson, M., Soulieres, I., Hubert, B., & Burack, J. (2006). Enhanced percep-tual functioning in autism: An update, and eight principles of autistic perception.Journal of Autism and Developmental Disorders, 36, 27–43.

Ozonoff, S., Cook, I., Coon, H., Dawson, G., Joseph, R. M., Klin, A., et al. (2004). Per-formance on Cambridge Neuropsychological Test Automated Battery subtestssensitive to frontal lobe function in people with autistic disorder: Evidence fromthe Collaborative Programs of Excellence in Autism network. Journal of Autismand Developmental Disorders, 34, 139–150.

Ozonoff, S., Strayer, D. L., McMahon, W. M., & Filloux, F. (1994). Executive functionabilities in autism and Tourette syndrome: An information processing approach.Journal of Child Psychology and Psychiatry and Allied Disciplines, 35, 1015–1032.

Pellicano, E., Maybery, M., Durkin, K., & Maley, A. (2006). Multiple cognitive capa-bilities/deficits in children with an autism spectrum disorder: “Weak” centralcoherence and its relationship to theory of mind and executive control. Devel-opment and Psychopathology, 18, 77–98.

Plaisted, K., Saksida, L., Alcantara, J., & Weisblatt, E. (2003). Towards an understandingof the mechanisms of weak central coherence effects: Experiments in visual con-figural learning and auditory perception. Philosophical Transactions of the RoyalSociety of London Series B-Biological Sciences, 358, 375–386.

Plaisted, K., Swettenham, J., & Rees, L. (1999). Children with autism show local prece-dence in a divided attention task and global precedence in a selective attentiontask. Journal of Child Psychology and Psychiatry and Allied Disciplines, 40, 733–742.

Price, D. J., Kennedy, H., Dehay, C., Zhou, L., Mercier, M., Jossin, Y., et al. (2006). Thedevelopment of cortical connections. The European Journal of Neuroscience, 23,910–920.

Raven, J., Court, J., & Raven, J. (1988). Raven’s standard progressive matrices. London:J.C. Raven Ltd.

Raymond, J. E., Shapiro, K. L., & Arnell, K. M. (1992). Temporary suppression of visualprocessing in an RSVP task: An attentional blink? Journal of Experimental Psy-chology. Human Perception and Performance, 18, 849–860.

Rinehart, N. J., Bradshaw, J. L., Moss, S. A., Brereton, A. V., & Tonge, B. J. (2001). A deficitin shifting attention present in high-functioning autism but not Asperger’s dis-order. Autism, 5, 67–80.

cholog

R

S

S

S

S

S. White et al. / Neuropsy

opar, D., & Mitchell, P. (1999). Are individuals with autism and Asperger’s syndromesusceptible to visual illusions? Journal of Child Psychology and Psychiatry andAllied Disciplines, 40, 1283–1293.

chlooz, W. A., Hulstijn, W., van den Broek, P. J., van der Pijll, A. C., Gabreels, F., van derGaag, R. J., et al. (2006). Fragmented visuospatial processing in children with per-vasive developmental disorder. Journal of Autism and Developmental Disorders, 36,1025–1037.

hah, A., & Frith, U. (1983). An islet of ability in autistic children: A researchnote. Journal of Child Psychology and Psychiatry and Allied Disciplines, 24, 613–

620.

hah, A., & Frith, U. (1993). Why do autistic individuals show superior performanceon the Block Design task? Journal of Child Psychology and Psychiatry and AlliedDisciplines, 34, 1351–1364.

kuse, D., Warrington, R., Bishop, D., Chowdhury, U., Lau, J., Mandy, W., et al. (2004).The developmental, dimensional and diagnostic interview (3Di): A novel com-

ia 47 (2009) 1274–1281 1281

puterized assessment for autism spectrum disorders. Journal of the AmericanAcademy of Child and Adolescent Psychiatry, 43, 548–558.

Stevenson, R. E., Schroer, R. J., Skinner, C., Fender, D., & Simensen, R. J. (1997). Autismand macrocephaly. Lancet, 349, 1744–1745.

Valdés-Sosa, M., Torres, R., Iglesias, J., & López, K. (submitted for publication). Directtiming of attentional shifts between global and local vision in autism.

van Lang, N. D., Bouma, A., Sytema, S., Kraijer, D. W., & Minderaa, R. B. (2006). Acomparison of central coherence skills between adolescents with an intellec-tual disability with and without comorbid autism spectrum disorder. Research

in Developmental Disabilities, 27, 217–226.

Ward, L. M. (1982). Determinants of attention to local and global features of visualforms. Journal of Experimental Psychology. Human Perception and Performance, 8,562–581.

Wechsler, D. (1992). Wechsler Intelligence Scale for Children (WISC III-UK). New York,NY: Psychological Corporation.