Abstract
When people reach out and lift an object, the variability of grip aperture, unlike the variability of perceptual estimates, is unaffected by variations in object size. According to the two-visual-systems hypothesis, Weber's law is violated because unlike for perception, visual coding for action is based on absolute metrics. A different hypothesis, which does not assume independent visual coding for perception and action, ascribes the violation mostly to biomechanical factors that affect finger aperture in grasping. Here, we contrast these hypotheses by introducing a task in which the role of biomechanical factors is eliminated, but object size is still a relevant feature. Participants (n = 31) had to either grasp or lift rods of different lengths (13, 24, 35 cm; 1x1 cm cross-section) with the requirement to achieve a balanced lift (grasping task) or simply indicate the rods' center (perceptual task). If the object's length for grasping is computed veridically, as predicted by the two-visual-systems hypothesis, we should find a dissociation between the grasping and the perceptual tasks: the variability in the grasping task should remain constant across objects, and the variability in the perceptual task should increase as a function of object length. Instead, we found that the variability increased with physical length in both the perceptual and the grasping tasks with identical slopes. Interestingly, grasping variability did not improve over successive repetitions even though both haptic and visual feedback about the object orientation was available during lifting. These results demonstrate that Weber's law is at play in both action and perception when length is essential to determine the center of the object and biomechanical factors are eliminated. Thus, these findings have implications for the fundamental assumption of the two-visual-systems hypothesis that action and perception rely on different computations of the visual input.