Visuomotor crowding: the resolution of grasping in cluttered scenes

Abstract

Reaching toward a cup of coffee while reading the newspaper becomes exceedingly difficult when other objects are nearby. Although much is known about the precision of visual perception in cluttered scenes, relatively little is understood about acting within these environments - the spatial resolution of visuomotor behavior. When the number and density of objects overwhelm visual processing, crowding results, which serves as a bottleneck for object recognition. Despite crowding, featural information of the ensemble persists, thereby supporting texture perception. While texture is beneficial for visual perception, it is relatively uninformative for guiding the metrics of grasping. Therefore, it would be adaptive if the visual and visuomotor systems utilized the clutter differently. Using an orientation task, we measured the effect of crowding on vision and visually guided grasping and found that the density of clutter similarly limited discrimination performance. However, while vision integrates the surround to compute a texture, action discounts this global information. We propose that this dissociation reflects an optimal use of information by each system.

Keywords: clutter; crowding; dual-visual systems; grasping; kinematics; perception-action dissociation.

Figure 1

Demonstration of stimulus and task. While fixating on the dot on the left (A), participants discriminated the central, target bar in each display by reaching and grasping the orientation of the target (B), or by a 3AFC perceptual response (C). The orientation of the target bar in each case above is vertical, yet it appears tilted in the direction of the ensemble, or average distractor orientation. Do we reach toward this ensemble orientation?

Figure 2

Group results for Experiment 1. The crowding effect (A) for perceptual (solid line) and visuomotor judgments (dotted line) of target orientation collapsed across the three eccentricities tested. There was a significant decrease in discrimination performance with increasing eccentricity and increased flanker density. Perceptual responses were more attracted to the mean ensemble orientation of the distractor bars (B), while visuomotor judgments revealed a relative repulsion to this ensemble orientation (negative Fisher z scores). Inset: data for perception (black) and visuomotor responses (gray) collapsed across densities. All error bars are between-observer SEM.

Figure 3

Visualization of the data plotted in Figure 2B across space. For each possible distractor position, this heatmap reveals how the flanker bars capture (red) or repel (blue) the perceived or grasped target bar orientation. Each panel shows the degree to which observers’ perceptual or visuomotor responses correlated with each of the six distractor positions at each eccentricity. While flankers tended to capture or attract the perceived target orientation, they tended to repel the grasped target orientation.

Figure 4

Group results for Experiment 2. Both perceptual (A) and visuomotor (B) discrimination performance was superior (less crowding) in the lower visual field. All error bars are between-observer SEM.