Purpose: Experience is known to facilitate our ability to extract regularities from simple repetitive patterns to more complex probabilistic combinations (e.g. as in language, music, navigation). However, little is known about the neural mechanisms that mediate our ability to learn hierarchical structures.
Methods: Here we combine behavioral and functional MRI measurements to investigate the brain circuits involved in learning of hierarchically organized structures. In particular, we employed variable memory length Markov models to design temporal sequences of increasing complexity. We trained observers with sequences of four di?erent symbols that were determined first by frequency statistics (i.e. occurrence probability per symbol) and then by context-based statistics (i.e. the probability of a given symbol appearing relates to the context provided by the preceding symbol). Observers performed a prediction task during which they indicated which symbol they expected to appear following exposure to a sequence of symbols.
Results: Our results demonstrate that cortico-striatal mechanisms mediate learning of behaviorally-relevant statistics that are predictive of upcoming events. Importantly, we show that individual variability in learning relates to two di?erent learning strategies: fast learners adopt a maximization strategy (i.e. learning the most probable event per context) while slower learners focus on matching (i.e. memorize all presented combinations). Correlating fMRI activation with individual learning strategy demonstrates that learning by matching engages the visual cortico-striatal loop including hippocampal regions. By contrast, learning by maximization involves interactions between executive control and motor cortico-striatal loops.
Conclusion: Thus, our findings suggest dissociable cortico-striatal routes that promote structure- outperforms rote- learning and facilitate our ability to extract predictive statistics in variable environments.