Abstract
PURPOSE: In the periphery, a letter target larger than the size of the observer's acuity can become unrecognizable when flanked by other letters. This “crowding” phenomenon is a vivid demonstration of the nonlinear nature of the deficits in form vision in the periphery. We used a combination of psychophysics and functional-imaging to determine the neural loci of crowding.
METHODS: Subjects performed a four-alternative-forced-choice letter identification task in the lower-right periphery at 5° eccentricity. The target-letter (Sloan Letters K, V, N, S) appeared for 100ms in four conditions: target-alone, or with letter flankers at 1.25x, 2x, and 3x letter height center-to-center separation.
RESULTS: Subject's psychophysical performance ranged from 50% in the crowded condition (1.25x separations) to near perfect in the non-crowded conditions (3x separation and ‘target-alone’). In early visual areas (V1, V2, V3) there was little or no difference in the peak amplitude of the BOLD response across conditions. However, in V4 there was a suppression in the crowded condition compared to the non-crowded conditions. This pattern of activity was reversed in anterior regions of LO and in the vicinity of IPS.
CONCLUSION: Our findings suggest a bottom-up origin of crowding: crowding impedes the formation of high-level features somewhere between V1 and V4, resulting in impoverished inputs to LO, which then recruits more spatial-attention to enhance and process the input. In essence, it does not appear to be a lack of attention that causes crowding; instead, crowding impedes bottom-up feature integration and increases attentional demands.
Supported by: NIH/NEI R03-EY016391 (BST) and R01-EY12810 (STLC)