Abstract
Some models of human vision propose a functional division of labor between vision-for-perception and vision-for-action (Milner & Goodale, 1995, The visual brain in action). This proposal is supported by neuropsychological, brain-imaging, and psychophysical studies. However, it has remained controversial in its prediction that actions are not affected by visual illusions. Here we re-analyze 16 studies on grasping the Müller-Lyer illusion (see Bruno, Bernardis & Gentilucci, in press, Neuroscience & Biobehavioral Reviews, for a similar meta-analysis on pointing). We find that median percent effects across studies are indeed slightly larger for perceptual than for grasping measures. However, all grasping effects are larger than zero and the two distributions show substantial overlap, with grasping effects showing a substantial variability. After detailed examination of methodological differences between and within the perceptual and grasping measures, we show that, as for pointing, critical roles in explaining such variability are played by the number of trials/condition (a learning-attentional effect), by the availability of visual feedback during movement, and by conditions at the programming phase of the action. We discuss to which degree these can explain differences between illusory effects on perception, grasping, and pointing, as well as their implications for the perception-action model.