UC San Diego

Perceptual learning has classically been described as an improvement in sensitivity to a trained stimulus in a particular location in the visual field. However, the specificity of traditional perceptual learning makes it difficult to ascertain its relevance to real-world instances of perceptual expertise. Although more recent work has investigated the ability of perceptual learning to transfer across feature, task, and spatial location, the neural correlates this transfer still remain unclear. Furthermore, work regarding the nature of the relationship between perceptual learning and top-down attention suggests that there is substantial overlap between the mechanisms by which learning and attention modify perceptual sensitivity. Both perceptual learning and top- down attention have been shown to improve the integrity of sensory responses through enhanced gain and/or noise reduction (sensory modulation) and improve the read-out of sensory signals by downstream decision mechanisms (enhanced read-out), and reduce intrinsic noise correlations in sensory areas (noise reduction). Given this overlap, I propose that more generalized perceptual learning results from the refinement of top-down attentional control, specifically feature-based attention, which subsequently promotes the transfer of learned perceptual improvements across spatial location. Here, I present both psychophysical (Chapter 2) and neuroimaging results (Chapters 1 and 3) supporting the role of feature- based attention in the acquisition and transfer of learning. Together, these studies suggest that learning to attend leads to an enhancement in the selectivity of feature-based attention via sensory modulation, which in turn supports improved behavioral performance at trained and untrained spatial locations