Abrahamse, E.
BCBL
The concepts of ´prediction´ and ´incidental learning´ share a rather fuzzy background. Against this background, it is especially challenging to figure out if and how predictions and prediction errors are a basic building block of the brain´s ability of incidental learning. Here I propose a (starting point to a) framework in which incidental learning is framed as goal-directed exploration - rather than the passive and collateral result of task processing. In brief, I postulate similar prediction-error based learning mechanisms underlying incidental and intentional learning, with the single difference being a more fine-tuned sampling from the environment in intentional learning as steered by selective attention. Hence, whereas intentional learning serves to confirm a model, incidental learning serves to build new promising models. This proposal fits the notion that few or no cognitive processes are truly insensitive to current goals, (temporarily) overcomes the methodological stalemate on implicit learning, and captures the recently suggested trade-off between incidental learning and cognitive control.