ERPs differentiate the sensitivity to statistical probabilities and the learning of sequential structures during procedural learning

Procedural learning facilitates the efficient processing of complex environmental stimuli and contributes to the acquisition of automatic behaviors. Although earlier findings suggest different temporal trajectories of the multiple learning processes within procedural learning, this has not been clarified at the level of neurocognitive correlates. Therefore, we investigated whether two prominent learning processes - statistical learning and sequence learning - can be distinguished using event-related brain potentials (ERPs) within the same experimental setting. Healthy young adults (N = 40) performed the Alternating Serial Reaction Time task while RTs and ERPs were measured time-locked to the onset of the task's stimuli. Both RT and N2 effects reflected the rapid acquisition of statistical probabilities. At the same time, these effects reflected the gradual learning of sequential structures. The amplitude change of the P3 reflected only gradual sequence learning. The P1 component was sensitive to both learning processes, which did not change as the task progressed. Our results altogether indicate that statistical learning and sequence learning develop differently at the level of both ERPs and overt responses. These findings could provide insight to the dynamic change of multiple parallel learning processes that occur during procedural memory formation.