Brain states predict individual differences in perceptual learning

Abstract

Brain states dynamically change with learning and these changes vary widely among individuals. Recent research proposes that electrophysiological measures of brain states can also predict individual variability in successful learning. This study was conducted to examine neural mechanisms of learning and neurophysiological indicators that predict success in a perceptual learning task. EEG was recorded over 20 blocks of trials while subjects learned to categorize a complex visual stimulus that required integration of multiple physical dimensions for successful categorization. For the analysis, final performance scores were used to median split subjects into high and low learners. By the 6th block, high learners began to diverge, eventually achieving 80% accuracy while low learners remained only nominally above chance. ERPs to the visual stimulus revealed a P3b that was significantly larger in high learners even before performance differences had emerged, but that did not vary with learning. Power spectral analyses showed that resting-state alpha was larger for high learners both before and during learning. Finally, alpha power increased for high but not for low learners as learning progressed. These results show that electrophysiological measures, especially alpha power, may not just reflect the learning process but also serve as predictors of eventual learning performance.