Contribution of receptive field center and surround to repetition suppression in macaque visual area V2

Abstract

Monkey inferotemporal cortex (ITC) neurons respond with declining strength to repeated presentations of relatively large and complex natural images. This phenomenon – repetition suppression – has often been assumed to arise at the level of ITC because ITC neurons possess the large receptive fields and sophisticated stimulus selectivity necessary for recognizing the image as a repetition. It has been recently discovered, however, that neurons in V2 exhibit repetition suppression under conditions identical to those in studies of ITC. This raises the question: How do V2 neurons, with classical receptive fields encompassing only a small fraction of the image, recognize it as a repetition? One possibility is that they are sensitive to repetition of image content not only in the classical receptive field but also in the receptive field surround. To assess this possibility, we monitored neuronal responses to sequential displays in which we controlled independently the repetition of elements in the classical receptive field and in the surround. Each stimulus consisted of a disk scaled to and centered on the classical receptive field and an adjoining annulus (8° outer diameter). The disks and annuli were taken from different natural scenes. The display on each trial consisted of a prime, a delay, and a probe, each 300 ms in duration. We found that repetition of the central disk was sufficient to produce repetition suppression but that suppression was enhanced by simultaneous repetition of the annulus. Suppression arising from repetition of the annulus occurred in a relatively late phase of the response, in accordance with the idea that it might have been mediated by horizontal connections in V2 and/or feedback from downstream areas. Overall, this work demonstrates that sensitivity to content outside the classical receptive field contributes to the image specificity of repetition suppression observed in V2 neurons to large natural images.