Abstract
When multiple stimuli are present in the visual field, competitive interactions between stimuli could be biased by stimulus-driven and attention-driven processes (Desimone and Duncan, 1995; Beck and Kastner, 2009). To study the neural mechanisms underlying multiple-stimulus encoding, we used fMRI and forward encoding model to estimate the orientation-selective responses in human visual cortex to parafoveally presented Gabor arrays. In a heterogeneous array, a center Gabor (eccentricity = 2.5° right of fixation) was surrounded by eight flanking Gabors (0.75° from the center Gabor). The orientation of the flankers was horizontal, while the center Gabor was oriented horizontally, vertically, or ±45°. When attending to the center Gabor, observers could perfectly identify its orientation. In the main experiment, observers performed tasks unrelated to orientation. Using forward encoding model, we decomposed the multivoxel response in each visual area into responses of a hypothetical set of orientation-selective channels (Brouwer and Heeger, 2011; Saproo and Serences 2014). We found that when selective attention was on the entire array, the orientation response to the array was dominated by the orientation of the flankers. This bias increased from V1 to V4. However, when selective attention was directed to the center Gabor, the orientation response began to reflect the orientation of the center Gabor in extrastriate areas: the orientation selectivity for the center Gabor was strongest in V4 while weakest in V1. For a homogeneous array where the flanking and center Gabors had the same orientation, selective attention sharpened orientation tuning in extrastriate areas V2-V4. These findings support the idea that attention resolves the competition among multiple stimuli by counteracting the influences of nearby stimuli, resulting in enhanced information processing at the attended location. This process of biased competition is observable in as early as V2 and appears to be cumulative along the visual-processing hierarchy.
Meeting abstract presented at VSS 2016