Abstract
<p>Face recognition in autism is well-researched and up to 1/3 of autistic individuals are reported to have severe face recognition difficulties. In comparison, there are few reports on object recognition performance in autism, and the findings are rather inconsistent. Here we report results from a series of object recognition tasks performed by autistic (N = 41) and matched non-autistic participants. Across tasks we varied the demand placed on perceptual differentiation, the structural similarity of the objects, and the salience and value of global shape information (the object’s bounding contour) relative to local shape information (details) in the recognition process. Our main hypothesis was that when local shape information was less salient than global shape information, and when both types of information needed to be integrated to support performance, autistic participants would perform less efficiently than non-autistic participants. This hypothesis was not fulfilled as the autistic group performed just as efficiently as the non-autistic group across all tasks. We discuss our findings in the context of previous reports of object recognition performance in autism and argue that former reports of impairments might reflect lexico-semantic rather than perceptual processes.</p>