Abstract
The ability to identify an object in peripheral vision can be severely impaired by clutter. This phenomenon of crowding is ubiquitous and is thought to be a key limitation of form vision in the periphery. The interactions between target and flankers are complex, and no single model can account for the myriad of results. Recently we have argued that crowding is due to the improper encoding of image statistics in peripheral V1 (Nandy & Tjan, 2009 SfN). We hypothesize that image statistics are distorted due to a temporal overlap between spatial attention, which gates the acquisition of image statistics, and the subsequent saccadic eye movement it elicits. In terms of mutual information between edge orientations at neighboring positions, the distortion turns smooth continuation into repetition. By fixing all but one parameter in the model using well-known anatomical and eye-movement data unrelated to crowding, we found that the spatial extent of the distortion in orientation statistics precisely reproduces the spatial extent of crowding, with all its tell-tale characteristics: Bouma's Law, radial-tangential anisotropy, and inward-outward asymmetry. The reproduction is robust in the sole free parameter of the model (the hypothesized temporal overlap between attention and saccade). Here, we extended the model to quantify the effects of crowding on orientation discrimination with a Gabor target and different flanker configurations. We proceeded from first principles by using the saccade-distorted image statistics as priors in a Bayesian formulation of a simultaneous segmentation and estimation task within the computational framework of a random field. This model can account for the recent finding of Levi and Carney (2009, Curr. Biol.) that more flankers can cause less crowding. It can also match ordinally the varying levels of crowding induced by the different flanker configurations used in Livne and Sagi (in-press, JOV).
NIH R01EY016093, R01EY017707.