Published in Computational and Systems Neuroscience (CoSyNe), (T-41), Feb 2012.
Area V2 is the largest extrastriate visual area, but its functions remain enigmatic. We used experimental stimuli generated from a hierarchical model of natural image structure (Portilla and Simoncelli, 2000) to differentiate the responses of neurons in V2 from those in V1, and to explore their role in perception. The model has two stages: the first captures local orientation and spatial frequency content with rectified responses of oriented filters, and the second computes correlations between these responses at different orientations, frequencies, and positions, capturing naturalistic features found in visual textures. Starting from original texture images (leaves, fabric, bricks, etc) we computed model responses at both stages, and generated pairs of novel images that produce the same average responses at just the first stage, or both stages. Only the latter contain complex naturalistic features. We measured responses of V1 and V2 neurons in anesthetized macaque to these pairs of synthetic images. In V1, the images within each pair yielded similar single-unit responses, suggesting that they are approximately matched with respect to neural sensitivities in V1. In contrast, firing rates in V2 were reliably higher for the second (naturalistic) image in each pair (64% of single units in V2, compared to 10% in V1). fMRI responses to these image pairs also reliably differentiate V2 from V1 in humans. To relate these differential responses to perception, we asked human observers to discriminate between images drawn from the two ensembles. Strikingly, the texture categories for which perceptual discrimination is best are those that exhibit the largest differential responses, both in macaque V2 neurons and human fMRI. We conclude that neuronal responses in V2, but not V1, are sensitive to the joint statistics of local structure found in natural images, and may reflect an emerging specialization for these features in the extrastriate visual pathway.