Causal topography of visual cortex in perceptual learning

Abstract

Individuals are able to improve their visual skill with practice, a phenomenon called Visual Perceptual Learning (VPL). We previously observed that after training on a difficult shape identification task, the dorsal visual regions (i.e. right V2d/V3 and right lateral occipital, LO) corresponding to the trained visual quadrant, and their homologous in the opposite hemisphere, exhibited a selective activation at the end of the learning. By contrast, such modulation was not observed in the ventral visual regions, corresponding to the untrained quadrants. The causal role of the trained visual cortex was previously showed in a TMS study as its inactivation impaired behavioral performance to learned stimuli. Here, using the same experimental design, we employed TMS over the homologous of the trained area (i.e. left V2d/V3) as well as over the untrained region (i.e. right V4) to causally map the visual network during the perceptual learning. We report a decrease of accuracy after TMS over left V2d/V3 as compared to both right V4 and Sham (inactive stimulation) conditions. Importantly, TMS effect was correlated with the degree of learning, such that subjects with lower accuracy at the end of the training exhibited stronger TMS impairment. These results provide evidence that segregated regions within the visual network are causally involved in visual perceptual learning.