Abstract
The visual system’s tendency to represent summary statistical information from crowds of objects has been well documented in recent research. However, this property of the visual system has the potential to provide insight into other visual phenomena as well. In the current set of experiments, we tested whether orientation information is automatically integrated across a region of space, and whether the spatial extent of integration depends on the allocation of attention. On each trial, observers saw either a single cue, which reliably directed observers to the visual field where a stimulus would appear, or saw two cues (one in each visual field), which revealed no stimulus location information. Observers then viewed a set of three oriented gabors (2 flankers, 1 target arranged horizontally) for 70 ms in either the left or right visual field, and were asked to adjust a test gabor to match the perceived orientation of the center target. Both flankers were 15 degrees clockwise or counterclockwise relative to the center target. In the single cue condition (reliable), observer adjustment to the center target was more precise than in the two-cue condition. Remarkably, in the two-cue condition, observer adjustment closely matched the average orientation of all three gabors. Modeling analyses revealed that this shift towards the mean reflected integration (i.e., averaging) of the three stimuli, and not substitution (i.e., incorrectly reporting a flanker). These results suggest that the size of the crowding region, or the integration field, is malleable and critically dependent on one’s attentional strategy. While attentional effects on the size of integration fields have been shown previously, the present results are the first to reveal attentional effects on the extent of integration, and also reinforces the link between crowding and summary statistical representations.
Meeting abstract presented at VSS 2012