Date created: | Last Updated:

Category: Project

Description: We easily pick up sequential regularities that are present in the environment without being aware that we have learned them, a process typically referred to as implicit sequence learning. In the current study we target the question about whether or not predictive coding underlies the development of implicit sequence learning. Whereas it is increasingly clear that predictions follow from implicitly learned content, the question here is whether predictions also precede – and drive – the process of implicit learning. The latter question is notoriously difficult to address and seems constrained to indirect approaches. Here we reasoned that if predictions precede and drive implicit learning, that implicit learning would be dependent on the precision parameter – an estimate about the meaningfulness of prediction errors. We first introduced participants to random contexts (implicitly in Experiment 1, explicitly in Experiment 2) that indicated that there was nothing to learn. Second, in a subsequent learning phase (that was not saliently marked to differ from the preceding random phase), participants performed an implicit sequence learning task. From the above reasoning, we predicted that decreasing the precision of predictions (via random contexts) should slow down the development of implicit sequence learning. We found convincing evidence in favour of the null hypothesis (with Bayesian statistics): the random contexts did not influence the emergence of implicit knowledge in the subsequent learning phase. These findings, we argue, are in line with simple associative or Hebbian learning (i.e., not steered by predictions) being the mechanism underlying implicit sequence learning, and are consistent with a dissociation between implicit learning and (un)conscious expectations.