Introduction
Several cognitive models of Autism Spectrum Disorder (ASD) suggest that attention to detail is related to the condition (Mottron et al.
2006; Plaisted
2001; Baron-Cohen
2002; Happé and Frith
2006). Detail focus is often referred to as local perception as opposite to global perception, i.e., focusing on the whole of e.g., a picture or integrating parts of a feature into a whole. Children with typical development (TD) can perceive both local and global features already at an early age (Stiles et al.
1991) and there are indications that individuals with TD develop a global visual default mode, perceiving global and local information as accurate when instructed to, but are faster at reporting global features in spontaneous answers (Campana et al.
2016). This global precedence has been found to remain in many individuals throughout life (Bruyer and Scailquin
2000) but results can vary depending on ethnicity (McKone et al.
2010) and method used (Dale and Arnell
2013). In individuals with ASD, on the other hand a visual
local processing default mode has been found in spontaneous responses (Wang et al.
2007; Happé and Frith
2006) although processing global information in ASD appear as accurate as in TD when explicitly and accurately demanded (Koldewyn et al.
2013). A local default mode has also been found in individuals with elevated autistic traits (Stevenson et al.
2016). A common paradigm to examine local–global default modes is to use stimuli where both global and local information levels are present, as they allow to investigate the trade-off effects in the perception and interference of one stimulus level in the presence of another. Findings of local-to-global and global-to-local interference in individuals with ASD or autistic traits have been mixed (Plaisted et al.
1999,
2006; Rinehart et al.
2000; Nayar et al.
2017; Deruelle et al.
2006; Stevenson et al.
2016; Van der Hallen et al.
2015,
2017).
Another approach is to examine local and global processing skills separately. Given evidence that local processing matures before global processing during typical development (Guy et al.
2016), and the indications that local and global processes are independent (Porporino et al.
2004), it appears reasonable to study these functions separately in ASD and TD, particularly in young children (Happé and Booth
2008). Local figure-ground tests (e.g., Shah and Frith
1983; Witkin et al.
1971) and visual search tasks have been quite common with results showing faster reaction times or higher accuracy in children with ASD in comparison to children with TD (Morgan et al.
2003; Kaldy et al.
2011; Pellicano
2006; Happé and Frith
2006; O’Riordan and Plaisted
2001) and positive correlations between autistic symptoms and local performance (Gliga et al.
2015; Cheung et al.
2016; Koldewyn et al.
2013; van Eylen et al.
2015). However, other studies find no group differences or inconsistent results (White and Saldana
2011; Muth et al.
2014; Horlin et al.
2016). Regarding research on global performance, tasks vary considerably, tapping into both visual and other cognitive abilities such as matching (Olu-Lafe
2015), drawing (Booth et al.
2003), puzzle-like tasks or deduction of the whole object from a part (Jolliffe and Baron-Cohen
2001; Nakano et al.
2010). A few studies have used visual global object/animate integration tasks in line with the definition of visual closure according to the Cattell-Horn-Carroll (CHC) factorial model of cognitive abilities (Schneider and McGrew
2012; Flanagan and Dixon
2013, p. 8). Here, tasks such as fragmented pictures (Snodgrass and Corwin
1988; Kessler et al.
1993) are used, i.e., animates or objects presented in a fragmented, shattered way that are to be recognised. Using fragmented stimuli is of particular interest when investigating children as possible confounds, such as of motor skills and deduction are controlled to a relatively high degree. Similar to results on local tasks, results differ from no difference between individuals with ASD and TD (Mottron et al.
2003) to global processing being slower or less accurate in individuals with ASD compared to TD (Booth and Happé
2016; Bölte et al.
2007; Scheurich et al.
2010) or mixed results (van Eylen et al.
2015). To the best of the authors knowledge, research including global performance, either as part of “trade-off” paradigms or real object/animate integration tasks in pre-school children with ASD has not been conducted.
In conclusion, the understanding of local and global processing in pre-school aged children with ASD is incomplete. Therefore, the objective of the current study was to examine local and global visual processing using separate measures for local and global perception, in 3-year-olds. An increasingly applied methodology to study early trajectories in ASD is high-risk (HR) for ASD sibling research (Bölte et al.
2013; Zwaigenbaum et al.
2015). HR-siblings are younger brothers or sisters to individuals diagnosed with ASD. Compared to about 1–2% in the general population (CDC
2016), the prevalence of ASD in siblings is about 14–20% (Messinger et al.
2013). In the current study we used cross sectional assessments of 3-year-olds that had been ascertained in a HR ASD longitudinal sibling design. We examined HR siblings with ASD (HR-ASD group), high-risk siblings without ASD (HR-noASD group) and low risk (LR) for ASD children, i.e., TD controls (LR group). We expected superior local performance in the HR-ASD group compared to the other groups. For global performance we tentatively expected superior performance in the HR-noASD and LR groups compared to the HR-ASD group.
Results
Mean,
SD and range for local and global measures and results are presented in Table
2. There was a between-group effect, after correcting for multiple comparisons across the one-way
ANOVAs, on HP accuracy (
F* (2, 39.46) = 7.34,
p = .018), with post hoc Games Howell tests showing the HR-ASD group to have higher scores than both HR-noASD (
p = .003,
s
p
= 2.69,
d = 1.23) and LR groups (
p = .001,
s
p
= 1.84,
d = 1.90). Applying correction for multiple comparisons also within the Games-Howell test for the HP- results, the results remained significant between the HR-ASD group on one hand, and the HR-noASD group (
p = .041) and the LR group (
p = .027), on the other. There were no group differences for accuracy and response latency for the other local measures (
F/
F* (2, 36–42) ≤ 2.18,
p ≥ .401), nor for the global measures (
F (2, 40–43) ≤ 1.80,
p ≥ .401).
As non-verbal IQ differed between the LR and HR-ASD groups (see Table
1), a between group ANCOVA was conducted with HP accuracy scores as dependent variable and MSEL’s nonverbal IQ as covariate. The effect remained between groups (
F (2, 39) = 3.99,
p = .027, partial
η
2
= 0.170), and pairwise comparisons with Bonferroni correction showed that the HR-ASD group had higher accuracy scores (
Ma = 13.68) than the LR (
Ma = 10.48,
p = .049) and HR-noASD (
Ma = 10.60,
p = .035) groups.
Discussion
In this study we examined visual processing in early ASD testing separate measures for local and global performance in siblings with ASD (HR-ASD), siblings with no ASD (HR-noASD group) and typically developing children (LR group) and found a group difference for the local measure
HP. Consistent with our prediction and in line with theories of enhanced visual local processing (e.g., Mottron et al.
2006), 3-year old children with ASD performed superior to the other groups. However, and against our expectations, there were no differences for global measures.
Unlike other studies (Morgan et al.
2003; Pellicano
2006), higher accuracy or shorter latencies in ASD were neither found on the CEFT nor the FG test. These tasks pose other executive demands than the HP, such as keeping attention when turning pages and processing different backgrounds, and, for the FG, also presenting different targets in each task. As executive functions are often found to be altered in individuals with ASD (Lai et al.
2017), executive control, flexibility and planning might thus have camouflaged potential superior local visual processing. It could well be that the simpler testing set-up in HP, such as looking for several targets in the same background, using only two pictures, leads to clearer differences between groups, particularly in young ages. However, talking against this, the result on the global GC did not differ between groups despite similar executive demands in the task. Unlike the other two local tasks, HP demands to find several of the same targets, different in size and angle. Superior HP performance in ASD could then either reflect that individuals with ASD show better visual form constancy (i.e., the ability to recognize similarity in form despite differences in sizes and angels) or enhanced visual search of finding several targets in a background.
Our results on global tasks are consistent with research reporting no global disadvantage related to ASD or autism symptoms (Mottron et al.
2003; Wang et al.
2007) but are inconsistent with specific fragmented picture research by Booth and Happé (
2016), finding inferior global performance in ASD. Compared to the latter, our sample differed in age, gender and IQ. It is also possible that divergent trajectories of global performance might not have emerged yet in 3-year olds.
Limitations
Our findings should be interpreted with adequate caution given that they are based on small sample size, limiting generalizability and power to detect small and medium differences on local and global tasks. Nevertheless, the result on HP remained sound, even after corrections for multiple comparisons both between and within
ANOVAs. There were also limitations concerning tasks used in this study, as some of them originally were constructed for use in older children and were here implemented in 3-year-olds for the first time. Moreover, latency measures were operationalized by only a few items, possibly limiting their accuracy, and the number of participants that completed the different tasks varied. In addition, unlike other studies, more girls than boys participated, possibly affecting results (Bölte et al.
2011; Lai et al.
2013; Kimchi et al.
2009).
Conclusions
This study shows that enhanced local performance is evident in children with ASD already at the age of 3, reflected by superior performance on the local measure Hidden Picture, independent of general developmental level and vocabulary. Our findings suggest that the testing of visual local performance, for example by a task as the HP, could add value to the clinical characterization of children with early suspicion of ASD. Further research on larger samples of young children with ASD is needed to investigate Hidden Pictures in relation to other local measures, visual search tasks as well as in relation to performance on form constancy.
Compliance with Ethical Standards