Abstract
We have previously shown that speed discrimination critically depends on how a stimulus is segmented (Verghese & Stone, 1996). When stimuli are seen as separate entities, independent information is available from each entity. But when separate stimuli are grouped into a single patch, only a single estimate is available. Here we determine whether a cue can overcome the effect of grouping and allow observers to select part of a grouped stimulus. Two stimulus configurations were presented at an eccentricity of 3 deg: the ungrouped stimulus was made up of 4 widely-separated Gabor patches, and the grouped stimulus was a single, large Gabor divided into 4 quadrants. All stimuli had a spatial frequency of 1.6 c/deg and drifted at 2.4 deg/s. The observer had to choose which of two intervals had a region (patch or quadrant) with a contrast increment. The potential test region was cued with a clearly visible cue in each interval. We used three different cues: a high contrast spatial cue, a speed cue where the target drifted at 4 times the speed of the other regions, and a direction cue where the target drifted in the opposite direction. The data across 4 observers showed a consistent pattern. When the patches were widely separated, cueing improved thresholds significantly over the uncued case, exactly as predicted from a perfect knowledge (no uncertainty) about the target. On the other hand, cueing one quadrant of the grouped stimulus barely improved contrast discrimination over the uncued case. The spatial cue had no effect at all, and the speed and direction cues improved contrast discrimination modestly. More importantly, thresholds with no cue in the grouped configuration were a factor of two higher than when the target was a separate, known entity. This increase is larger than predicted from uncertainty about which one of 4 quadrants is the target: it suggests that observers are unable to segment the grouped stimulus and therefore sum the information from all 4 quadrants.
This work was supported by NASA grant NAG9-1461 and by a Rachel C. Atkinson fellowship to AMW