Anticipatory reinstatement of expected perceptual events during visual sequence learning

Abstract

Being able to predict future events in learned sequences is a fundamental cognitive ability. Successful behavior requires the brain to not only anticipate an upcoming event, but to also continue to keep track of the sequence in case of eventual disruptions, (e.g., when a predicted event does not occur). However, the precise neural mechanisms supporting such processes remain unknown. Here, using multivariate pattern classification based on electroencephalography (EEG) activity and time-frequency amplitude, we show that the visual system represents upcoming expected stimuli during a sequence-learning task. Stimulus-evoked neural representations were reinstated prior to expected stimulus onset, and when an anticipated stimulus was unexpectedly withheld, suggesting proactive reinstatement of sensory templates. Importantly, stimulus representation of the absent stimulus co-occurred with an emerging representation of the following stimulus in the sequence, showing that the brain actively maintained sequence order even when the sequence was perturbed. Finally, selective activity was evident in the alpha-beta band (9-20 Hz) amplitude topographies, confirming the role of alpha-beta oscillations in carrying information about the nature of sensory expectations. These results show that the brain dynamically implements anticipatory mechanisms that reinstate sensory representations, and that allow us to make predictions about events further in the future.