The brain is thought to optimise behaviour by generating predictions based on learned statistical regularities. Predictive processing seemingly explains expectation suppression (ES), the attenuation of neural activity in response to expected stimuli. However, the mechanisms behind ES are unclear, with conflicting evidence for alternative models. Sharpening models propose that expectations suppress neurons away from the expected stimulus, increasing the signal-to-noise ratio and boosting decoding for expected stimuli. In contrast, dampening models posit that expectations suppress neurons that are tuned to the expected stimuli, reducing overall response magnitude and decoding accuracy. The opposing process theory (OPT) suggests that both processes occur at different time points, namely that initial sharpening is followed by later dampening of the neural representations of the expected stimulus. Here we test this theory and shed light on the dynamics of expectation effects, both at single-trial level and over time. Thirty-one participants completed a statistical learning task in which a ‘leading’ image from one category predicted a ‘trailing’ image from a different category. Multivariate EEG analyses decoded stimulus information related to the trailing category. Within-trial, expectation increased decoding accuracy at early latencies and decreased it at later latencies, in line with OPT. However, across trials, stimulus expectation decreased decoding accuracy in initial trials and increased it in later trials. We theorise that these dissociable dynamics of expectation effects within and across trials support hierarchical learning mechanisms. While within-trial results support the OPT, across-trial results suggest that sharpening and dampening effects emerge at distinct stages of associative learning.
McDermott et al. (Tue,) studied this question.