Do We Need Another Neural Correlate of Contour Integration?

Abstract

Gilad and colleagues use an elegant combination of voltage-sensitive dyes and high temporal and spatial resolution optical imaging to visualize a differential response to collinear contour elements in monkey V1. This result adds to the literature on the neural correlates of contour integration, but does not yet tackle (or seek to tackle) the question as to whether contour integration is mediated by lateral connections within an area (e.g., V1), through pooling of feedfoward connections, or feedback mechanisms. Moreover, while Gilad et al. find that their differential response is correlated with the behavioral performance of each monkey, there are reasons to suspect that the correlation they observe is a consequence of processing in higher regions, and that the differential V1 response may not play a critical role in integrating contour elements, or in generating the monkey's response. Moreover, this differential V1 response was not observed in a monkey who was not trained on the task, a result that can only be reconciled, if one assumes that the monkey could not see the contour prior to training. If valid, this could raise doubts as to whether the study of contour integration really provides insights into the processes by which normal visual perception is achieved.