Pointing and antipointing in Müller-Lyer figures: Why illusion effects need to be scaled

It has been suggested that goal-directed actions performed under full vision are immune to certain visual illusions, while movements relying on perception-based visual information are deceived by them (Milner & Goodale, 1995). Consequently, pointing movements should be deceived by visual illusions when a delay is introduced (memory demands) or when antipointing (spatial imagery) is required. In 2 experiments, participants performed either propointing or antipointing movements to different versions of the Müller-Lyer illusion in 2 vision conditions (open-loop vs. delay). Apart from open-loop propointing, all conditions should rely on perceptual processing and should therefore yield similarly illusion effects. While we observed illusion effects in all conditions, their magnitude varied in unexpected ways. Most surprisingly, introducing a delay seemed to reduce illusion effects in antipointing. We show that this decrease can be explained by the fact that pointing after delay is less responsive to physical size changes. After correcting for this, illusion effects in antipointing were similar in both vision conditions but still twice as large as in the delayed propointing task. Our findings highlight the necessity of employing a correction procedure when comparing illusion effects across tasks and do not conform well to the predictions derived from the perception-action model.