Abstract

When immersed in a new environment we are challenged to make sense of initially incomprehensible event streams. At first, it seems like a befuddling cacophony that leaves us completely unprepared for what will happen next. And yet, quite rapidly, the brain is able to find meaningful structures, helping us to predict and prepare for future actions. We combine behavioural and brain imaging measurements with computational modeling, to understand the dynamics of learning complex structures. We show that individuals adapt to changes in the environment’s statistics and extract predictive structures. Importantly, extracting complex structures relates to individual decision strategy: faster learning relates to selecting the most probable outcomes (i.e. maximising) and is implemented by interactions between executive and motor cortico-striatal mechanisms, while learning the exact stimulus statistics (i.e. matching) is implemented by visual cortico-striatal circuits. Our findings provide evidence for alternate routes to learning of behaviorally-relevant statistics that facilitate our ability to predict future events in variable environments.