Abstract
According to the two-visual-system hypothesis (Milner and Goodale, 1995), after V1 the visual system splits into “vision for action” and “vision for perception” modules. Perceptual modules may exhibit characteristic developmental pace and sequencing (Fodor, 1983). To test this hypothesis, we investigated two spatial tasks in typically developing school children, autistic children, and adults. The tasks consisted in blindwalking with no delay to a near target and in matching the frontal extent to the sagittal extent of an L-pattern formed by ropes on the ground. The blindwalking task was assumed to tap into the action module, whereas the L-pattern matching task was assumed to tap into the perception module. In the blindwalking task, typically developing children and adults were accurate whereas autistic children exhibited an underestimation bias. Conversely, in the L-pattern matching task adults showed underestimation, typically developing children showed even greater underestimation, and autistic children were accurate. Control experiments ruled out alternative interpretations based on differences in eye-height and testing rooms. Overall, these results are consistent with two visual modules that develop and function in different ways as a function of age or pathology. The surprising performance of autistic children suggest that autism may involve anomalies in the use of spatial reference frames in visual cognition.