Abstract
Introduction: Observers represent the average properties of sets even when they cannot identify individual elements (Ariely, 2001). Although it has been proposed that these representations complement limited capacity focused attention (e.g., Alvarez, 2011), there has yet been no empirical investigation of their functional role in visual perception. Indeed, the visual system could capitalize on statistical regularities inherent in the surrounding environment to create the illusion of stable and complete perception amidst constantly changing retinal imagery, despite its limited capacity to represent more than a handful of objects in detail. In support of this proposal, we present behavioral, eye tracking, Steady State Visual Evoked Potential (SSVEP), and patient data demonstrating that statistical stability facilitates visual search. Methods: We manipulated the statistical regularity of a visual scene over time while observers performed a search task. Specifically, we modulated the mean size of an array of Gabor patches while observers searched for a left or right tilted target among horizontal distractors. In 'stable' blocks, the mean size of the Gabor patches was constant over successive displays, whereas in 'unstable' blocks, it changed from trial-to-trial. We measured correct response times, eye movements (Experiment 1), and SSVEPs (Experiment 2) from healthy participants, and correct response times from visual neglect patients (Experiment 3). Results: When the mean size of the array remained stable over time, visual search was facilitated in both healthy participants (faster correct response times, fewer saccades, larger pupil sizes), and patients (faster correct response times), and the amplitude of both target- and background-related SSVEP amplitudes (indexing attentional allocation) were attenuated. Conclusions: Our results provide evidence that statistical representations can create a stable context, freeing limited capacity attentional resources to perform a perceptually demanding search task. Furthermore, results suggest statistical summaries of peripheral information may play a central role in visual perception.
Meeting abstract presented at VSS 2014