Grasp Planning for Object Manipulation Without Simulation of the Object Manipulation Action

摘要

When an object is grasped, the grasp is usually adapted to upcoming object manipulations. We tested the hypothesis that grasp planning for object manipulation is based on simulations of body movements that could implement intended object manipulations. The simulation of body movements requires to map the desired object movement onto body movements at some stage of the planning process. Hence, manipulating this mapping should affect simulations and ultimately grasp selections. This hypothesis was tested in five experiments, in which participants grasped a circular knob and used it to rotate a pointer to various targets. In Experiments 1-3, we selectively manipulated the pointer-to-hand-rotation mapping with a "virtual rotation" procedure. During these virtual rotations, participants were exposed to an altered mapping between their hand movements and movements of the pointer. However, the exposure did not affect grasp selections in a subsequent test block. In Experiment 4, we verified that our manipulations of the mapping were sufficient to evoke substantial changes in grasp selections. In Experiment 5, we verified that adaptations in the virtual rotation procedure carried over to the test blocks, in which grasp selections were probed. In summary, participants adapted their grasps to different intended pointer rotations on a trial-to-trial bases, thus showing the end-state comfort effect. However, grasp selections were unaffected by the acquired mapping between pointer and hand movements. This suggests that anticipations of the body movements associated with specific object manipulation play no crucial role during grasp planning.