Online control of grasping for shapes defined by second-order contours

Abstract

When picking up smooth objects with complex shape, we tend to grasp at particular contact points that minimize slip and torsion. We have recently found that observers can quickly adjust movement and adopt ideal grasp points when an object is unexpectedly changed during an ongoing movement. This study tested whether shape processing for online control of grasping can utilize second-order contour information. The dorsal processing stream is thought to be less sensitive to second-order stimuli, so it might be difficult to make fast, online corrections to grasp points when shapes are specified by second-order contours. We recorded finger positions as observers reached to grasp virtual planar objects with smooth, random shapes. On perturbed trials, the initial object was replaced by a new object with different shape after the movement had begun, and observers adjusted their movement to grasp the new object. Three stimulus conditions were compared. In the baseline condition, stimuli were solid colored 2D shapes on a darker background. In the second-order condition, a visible contour was created by inverting a region of a low-pass noise pattern corresponding to the interior of a shape. A coherent shape can be perceived from the integrated boundary edges, which change luminance polarity along the contour. In a third condition, the base images were high-pass filtered, which removes low frequencies but preserves polarity of edges along the contour. In all conditions, we observed smooth corrective responses to perturbations, resulting in final grasp points that were near optimal and equivalent to the grasp points of unperturbed trials with the same object. We found no differences between the three stimulus conditions; performance was as accurate for the second-order contours as for shapes specified by luminance difference. The visual-motor system appears to be capable of robust shape processing for online control of grasping.