Abstract
Some models of visual function propose that vision-for-action and vision-for-perception use different spatial representations, so that the visual guidance of action can be accurate even when there are systematic biases in conscious visual perception. Support for this counterintuitive prediction has come from behavioral studies suggesting that visually guided actions are largely immune from visual illusions. However, contrary to this prediction other studies have documented substantial illusory effects in motor tasks. Such large contradictions in the literature pose a challenge, and suggest that we might obtain useful insights about the visual guidance of pointing if we could understand the factors behind these differences of results. We have reanalyzed 33 independent studies of pointing on the Müller-Lyer and related illusions. Our reanalysis shows that estimates of the illusion effect average slightly above zero but vary widely across studies (from around zero to comparable to perceptual effects). In addition, our reanalysis shows that a substantial proportion of this variability can be explained by four factors: the initial position of the pointing finger, the quality of visual feedback during the pointing action, the number of trials per condition of the experiment, and the interaction of feedback and trials. These findings suggest that pointing is immune from the illusion under certain conditions, but it can be strongly affected by it under others. We propose, in particular, that immunity of pointing from visual illusions is not a matter of response mode but depends on a subtle interplay of a variety of factors, a crucial role beeing played by feedback-driven motor learning.
Supported by MIUR - PRIN 2005057573_004