Integral versus separable perceptual dimensional pairs are reflected in conjoint versus independent neural populations

Abstract

Some visual properties of objects are apprehended separately, whereas other dimensions are perceived as a composite; these have been termed separable and integral dimensions. We hypothesize that integral perceptual dimensions are represented by populations of neurons that represent the dimensions conjointly, while separable dimensions are represented by independent neural populations. To test this idea, we used a continuous carry-over fMRI design (GK Aguirre, 2007) to measure the recovery from neural adaptation associated with stimulus changes along a single perceptual dimension or combined across two dimensions. Stimulus changes along both perceptual dimensions should produce a recovery from adaptation that is the additive sum of the recovery for each axis in the case of separate representation, but which is sub-additive in the case of conjoint representation. In our first experiment, we presented a continuous stream of shapes that varied along two radial frequency component (RFC) dimensions which are characterized as behaviorally integral. Significant adaptation that was proportional to the perceptual similarity of the stimuli was observed in right, ventral LOC and in area V3A bilaterally (all p[[lt]]0.05). We then tested if shape changes along both perceptual dimensions resulted in a recovery from adaptation that was less than the sum of the recovery to changes along each axis in isolation. The significant sub-additivity observed (p[[lt]]0.05) indicates that neural populations within these regions represent the two, behaviorally integral, perceptual dimensions conjointly. In ongoing experiments, we are studying stimuli that vary along the dimensions of color and a single RFC component. These two dimensions are behaviorally separable. Preliminary data has identified recovery from adaptation for changes in shape and color in ventral area V2, with additive recovery from adaptation for stimulus changes that combine shape and color. If further replicated, these results indicate independent representation of shape and color information by neurons within V2v.