Perceptual grouping weakens local feature representation

Abstract

Previous studies showed that shape perception could reduce activity in human primary visual cortex (Murray et al., 2002; Fang et al., 2008). But the functional role of the reduction is still unclear. Here we performed psychophysical adaptation experiments to address this issue. Subjects adapted to a thin diamond with its four corners hidden by three horizontal occluders. The diamond translated with a circular trajectory. Its speed was 3.38°/s and its direction (clockwise or counter-clockwise) reversed every 5 s. Due to the existence of the occluders, only four bars (part of the sides of the diamond) were visible, but subjects were able to group them into a translating diamond (the diamond stimulus). The width of the bars was 1.80° and their orientations were 25° tilted (left or right). Their luminance was modulated spatially by a sine function (2 cycles/°) and their Michelson contrast was 1. A second adapting stimulus (the non-diamond stimulus) was generated by slightly changing the positions of the occluders in the diamond stimulus, which then could not be perceived as a moving diamond. Both the diamond stimulus and the non-diamond stimulus were presented in the left visual field. We measured two local aftereffects induced by adaptation to one of the four bars—tilt aftereffect (TAE) and threshold elevation aftereffect (TEAE). We also measured high-level shape aftereffect (SAE) by presenting normal or close-to-normal diamonds (their aspect ratios were around 1) in the right visual field. Compared with adaptation to the non-diamond stimulus, adaptation to the diamond stimulus induced a stronger SAE, but weaker TAE and TEAE. These results suggest that perceptual grouping enhances shape representation in high-level shape selective areas (e.g. LOC), but meanwhile, weakens local feature representation in early visual areas (e.g. V1), which provides psychophysical evidence for the predictive coding theory (Mumford, 1992).