Neural representation of surface ordering in visual Areas V1, V2 and MT

Abstract

Visual cortical areas V1, V2 and MT may participate in the representation of surface ordering, the arrangement of one surface in front of another. This work investigates the role of neurons of V1 and V2 in figure-ground representation in static stimuli, as well as the role of MT in surface ordering in dynamic stimuli. Electrical recordings were made in V1 and V2 to determine whether neurons in these areas encode information about the identities of figure and ground, and also whether they respond to figure-ground cues. We recorded from 3 monkeys, one trained on a fixation task, and the other two on a match-to-sample task that ensured attention to the stimuli. The stimuli consisted of rectangles of differing contrast arranged in an unambiguous or ambiguous figure-ground configuration. The stimuli were positioned such that the cells' receptive fields were located either at the border between rectangles or in the interiors of rectangles. Cells demonstrating selectivity at borders or interiors of unambiguous figure-ground stimuli were considered selective for border ownership or figure vs. ground, respectively. Cells showing selectivity at borders or interiors of ambiguous figure-ground stimuli were considered selective for figure-ground cues. Preliminary experiments on the fixating monkey suggested that a small fraction of cells in V1 and V2 might play a role in figure-ground interpretation. The results from the awake behaving monkeys further support the hypothesis that V1 and V2 play a role in figure-ground perception. In both areas we found cells demonstrating selectivity for border ownership, and in V2 we found cells demonstrating selectivity for figure over ground. However, in V1 and V2 there was also evidence a separate population was responding to the presence of figure-ground cues in the stimulus. The experiments in MT were performed on two awake behaving monkeys. The stimuli were transparent rotating cylinders comprised of random dots moving along a sinusoidal gradient. The stimuli were bistable?perceived to rotate in one direction or its opposite. The monkeys indicated in which direction they perceived the cylinder?s front surface rotating. Cells were found whose firing correlated with the monkeys? bistable percept, even though the stimuli were identical.