Error correction and interference in grasping illusions

摘要

There is a long-standing debate about whether pictorial illusions affect grasping, and to what degree. More recently, many studies have reported at least some effect of illusions on grasping, but also frequently a rather rapid decrease of the illusion effect on grasping over trials. Such a pattern is quite similar to what is typically found in studies on sensorimotor adaptation, and may be a consequence of participants learning to counteract illusory size distortions. If this were the case, one would expect adaptation to be slower or non-existent when opposite distortions are presented repeatedly in random order in the same location. To investigate this, we conducted a grasping experiment (N=40) using the M??ller-Lyer illusion with incremental and decremental M??ller-Lyer illusion displays presented either in (1) one block in pseudo-randomized order, or (2) two separate blocks of only one illusion type each. As predicted, we found an illusion effect on the maximum grip aperture, but a decrease of the illusion effect over trials as well as different decrease rates for single-illusion blocks and intermixed blocks. We applied a linear state-space error-correction model in which the illusion configuration was regarded as a constant visual size perturbation that determined the error signal. Our model was nicely able to qualitatively predict the data by assuming constant illusory perturbations, along with error-correction and some, but not full, error-generalization between illusion configurations. Consistent with the predictions of error-correction, we also found adaptation aftereffects and previous-trial effects of illusion configuration, but not of object size. This suggests that error-correction may be able to explain not only decreasing illusion effects, but also why some studies have found no such decreases.