Abstract
Oscillatory activity in the alpha band (8–13 Hz) is among the most prominent aspect of scalp-recorded EEG signal. Changes in alpha power have been related to a spectrum of behavioral and cognitive processes. Recent research has converged to demonstrate that alpha power is selectively heightened when participants report states of mind-wandering, active imagery, and internal/anticipatory processing, as well as suppression of distracting stimuli. A major portion of this research has involved trial averaging and shown paradigm-specificity along with pronounced inter-individual differences. The present study (N=20) examined the trial-by-trial covariance between occipital alpha power, time-varying associative strength, and self-reported expectancy of aversive events in a conditioning paradigm. A Gabor patch (serving as the conditioned stimulus, CS) was randomly paired with an aversive sound (US) in 60 of 120 trials. Participants were asked to rate the likelihood of a US occurring in the subsequent trial. Occipital alpha levels in the inter-trial interval were analyzed separately as a function of the expectancy ratings and associative strength (previous CS-US pairings). Behavioral data followed the gambler’s fallacy effect, in which recently occurring outcomes are rated as less likely to happen in the future. In line with this finding, post-stimulus alpha power following two trials with no US was greater than alpha power following two trials with a CS-US pairing. By contrast, pre-stimulus alpha power following participant reports of high US likelihood in the upcoming trial was greater than alpha power following reports of low US likelihood. Fitting the trial-by-trial variability in alpha power with the Rescorla-Wagner model of associative learning converged with these findings, together suggesting that alpha power is sensitive to visuocortical changes during visual adaptive learning.