Abstract
One potential benefit of neural adaptation is the sharpening of neural representations to optimize performance in changing environments. This has been previously observed for low-level features such as color and orientation. Here we test whether adaptation sharpens complex object representations by characterizing the effect of adaptation on behavioral shape discrimination thresholds. We constructed a parametric 3D shape space whose axes correspond to different frequencies and orientations of sinusoidal modulations of a sphere. Two distinct reference shapes (A and B) were selected from this space, along with two “clouds” of fifty stimuli sampled near each reference by jittering the amplitude and orientation components of the references. Test stimuli were derived from a morph continuum between shapes A and B. In separate blocks, observers were first adapted to either cloud A or B with an initial 1-minute RSVP stream of adaptors, and then performed approximately 100 interleaved match-to-sample trials for the two references. Each trial began with a 5-second top-up adaptation, followed by a test stimulus from the morph continuum flashed for 200 ms at fixation, followed by either reference A or B and the same test item displayed on both sides of fixation until response. Observers indicated which item in the second interval matched in shape to the item in the first interval. Differences between the test and reference stimuli were controlled separately for the two references by interleaved staircases that converged on the 79% discrimination threshold. Results (N=3) indicate significant adaptation for both shape clouds: thresholds for reference A were lower after adaptation to cloud A compared to cloud B (median improvement 27%, range 14%-180%), and vice versa for reference B (median improvement 28%, range 20%-34%). These results offer novel evidence for the sharpening hypothesis in the perception of object shape.
Meeting abstract presented at VSS 2015