Is there dissociation of perceptual and motor responses to figural illusions?

Open-loop reaching for locations within figural illusions was measured in three experiments. The experiments differed with respect to whether subjects were provided a visible target toward which to direct their reaching or were required to form a mental representation of the intended target. In the first experiment, subjects' reaching errors for vertices of a Müller-Lyer figure were similar to those for a nonillusory control stimulus. In experiment 2, subjects' errors while reaching to the imaginary bisector of the Judd illusion were consistent with the presence of an illusion of bisector location. However, when a bisector line was added to the Judd figure, reaching errors were similar to those obtained with a control figure. In experiment 3, subjects' open-loop reaching at the perceived midpoint of a triangle was biased toward its illusory perceptual midpoint. When a mark was placed at the midpoint between a vertex and the opposite side, reaching errors were similar to those obtained with a control figure. The results of the experiments indicate that hand-eye coordination is biased in the direction of illusions of bisector location only when no target is present at the intended goal of the reaching response and subjects are required instead to form a mental image of the target. Under these conditions, reaching responses appear to utilize the spatial map of the visual system, and are influenced by figural illusions of bisector location. The present data can be understood without invoking the notion of visual-motor dissociation.