Abstract
PURPOSE: Perceptual grouping is highly sensitive to temporal synchrony between oscillating Gabor elements, yet the nature of mechanisms mediating synchrony perception remains unknown. Here we explore two competing hypotheses: Synchrony perception between two Gabor elements is either A) limited by the intersections-of-constraints (IOCs) that characterizes single MT cells, or B) defined by a velocity association field similar to the static association field inferred by edge grouping studies.
METHODS: Observers discriminated the synchrony of two spatially adjacent Gabors oscillating at 5 Hz in a temporal 2AFC task. Spatial phase varied sinusoidally with time as f(t) = A·cos(ωt + Δ). The A term controls the oscillation amplitude, ω the oscillation frequency, and Δ determines synchrony. We varied the Gabors' relative oscillation amplitudes, orientations, spatial frequencies, and spatial separation in a manner that either satisfied or violated a particular MT cell's IOC.
RESULTS: Synchrony discrimination thresholds for varying relative oscillation amplitude Gabor pairs follows a log-Gaussian tuning function, peaking at mid-range amplitudes. While orientation differences between Gabors reduced performance, satisfying or violating a single MT cell's IOC by varying oscillation amplitude had no effect. Spatial-frequency differences between Gabors reduced performance if an MT cell's IOC was satisfied, whereas performance remained high if the cell's IOC was violated but preserved spatial scale invariance. Spatial separation between Gabors had little effect.
CONCLUSIONS: Barlow suggested that, due to their finite temporal-integration window, neurons can act as coincidence (i.e. synchrony) detectors. MT cells therefore constitute potential candidates for synchronous motion oscillation processing. However, our data reveal that stimuli constructed to allow single MT cells to perform the relevant discrimination are ineffective or deleterious to performance. Our results suggest instead that synchrony perception is mediated by a spatial association field that groups synchronous signals as a function of their joint oscillation amplitudes, orientations, spatial frequencies, and separation.