Abstract
Perceptual learning is enabled by repeated practice. However, extensive practice is not the only route to make perfect: adapting memory reactivation-reconsolidation frameworks predominantly originating from fear-conditioning modulation studies in rodents, we showed that brief reactivations of the encoded visual skill memory are sufficient to improve human perceptual thresholds (Amar-Halpert et al., 2017).
Here, we aimed to reveal the underlying mechanisms of reactivation-induced perceptual learning. We reasoned that in contrast to use-dependent plasticity dominant in early visual areas, engagement of high-level regions mediates rapid perceptual learning, which in turn predicts facilitation of generalization patterns of learning.
To test this prediction, participants performed the texture discrimination task (Karni and Sagi, 1991), in which they decided whether an array of 3 diagonal bars embedded in an array of horizontal bars was horizontal or vertical. The stimulus was backward-masked, and target-to-mask asynchrony (SOA) was randomly changed within the session to obtain a psychometric curve, from which the SOA discrimination threshold was derived. Baseline performance was first measured in two target-array locations: upper left (location B) and lower right (location A) quadrants of the visual field. Participants returned for 3 daily location A reactivation sessions of only 5 trials each, at a near-threshold SOA. Final thresholds were measured in the "reactivation-trained" location A and the untrained location B.
Results indicate full learning in location A, replicating reactivation-induced learning. Furthermore, reactivation-induced learning transferred to the untrained location B, exhibiting reduced thresholds relative to baseline. To further evaluate the systems-level mechanisms of reactivation-induced learning and generalization, task-based and resting-state fMRI are measured before and after learning, analyzing engagement of higher-level regions and their communication with early visual areas. Together, the results suggest that reactivation-induced plasticity may unlock learning specificity and facilitate generalization patterns of learning.