Assessing the Perceptual Origins of Cognitive Peaks in Autism

Background:  Individuals with autism recurrently demonstrate better performance on visuo-spatial cognitive tasks (referred to as cognitive peaks), as exemplified by superior performance on the Block Design Task (BDT) subtest of the Wechsler Intelligence Scale (Caron et al, 2006). Cognitive accounts suggest that peak BDT performance derives from a reduced “top-down” interference of perceptual cohesiveness of the global figure.  Whereas, perceptual accounts suggest that peaks may originate from superior local visual processing (bottom-up) of component blocks (Shah & Frith, 1993). The latter hypothesis is based off of differential sensitivity to low-level, elementary information characterized by luminance (increased sensitivity) or texture-defined (decreased sensitivity) visual attributes (Bertone et al, 2005; Vandenbrouke et al, 2008).  If manipulating visual attributes, of the BDT, differentially affects autistic performance, it can then be argued that superior visuo-cognitive abilities in autism may have a partially perceptual origin.

Objectives: To demonstrate the relative role of bottom-up and top down-processes in visuo-spatial peaks in autism by assessing whether performance on a computerized, reversed BDT is differentially affected by manipulating the visual attributes of its component blocks.

Methods: 10 participants with autism and 10 typically developing participants, matched for age and full-scale Wechsler IQ, were asked to complete a computerized reversed BDT (Caron et al, 2006). In addition to obtaining comparable performance relative to the traditional BDT, using a computerized reversed BDT allowed for precise manipulation of physical block characteristics defining the chosen visual attribute conditions. Trials consisted of a target block design with a matrix size of either 4 x 4, 9 x 9, or 16 x 16 presented at the center of a touch-sensitive screen, with four possible probes (one matching the target) presented simultaneously around it. Participants were asked to touch the matching probe as quickly and accurately as possible. Importantly, the visual attributes of the blocks were manipulated across these three matrix sizes, defined by either (a) traditional red/white surfaces, (b) luminance-defined (high-contrast) black/white surfaces, or (c) texture-defined surfaces. In addition, the perceptual coherence of block designs was also manipulated by varying the number of ‘adjacencies’ of opposite-colored edges.  The low-coherence (LC) designs necessitated increased local analysis to be resolved (block-by-block-processing), relative to high-coherence (HC) designs. Reaction times (RTs) for both groups were measured across all experimental conditions.

Results: On average, mean RTs for both autism and control groups were lower and comparable for HC conditions across visual attribute and matrix size conditions.  When collapsed across matrix size, mean RTs in the LC condition were significantly lower (cognitive peak) in the autism group for luminance-defined, black/white attribute condition (p=0.047), and lower for the traditional red/white condition (p=0.072). However, group differences were not evidenced for the texture-defined condition (p=0.19).    

Conclusions:  In the present study, a cognitive peak on a reversed BDT was not manifested in the autism group when constituent blocks were defined by texture-information, a visual attribute that is less efficiently processed in autism. Such results indicate that the differential processing of low-level perceptual information may have an effect on higher-level visuo-spatial performance in autism.