Ruud Berkers | Modulatory influences on neural learning systems and long-term memory formation

The dominant view of long-term memory formation in the brain considers complementary learning systems: the hippocampus as a sparse, pattern-separated system for rapidly acquiring episodic memories, and the neocortex as a distributed system for gradually integrating information and overlap across episodes. However, other factors also influence which learning systems are engaged and how information is consolidated. I will present studies using fMRI and TMS that probe how these factors influence learning and memory. First, I'll present data from an artificial language learning task demonstrating how the brain acquires information about a latent semantic structure within a single learning session through striatal-neocortical interactions. Second, I will explore the mechanisms by which prior knowledge influences the manner in which knowledge is consolidated. Specifically, we found that encoding of novel information that is congruent with pre-established prior knowledge (or a schema) is rapidly assimilated into the long-term neocortical store, whereas novel incongruous information is preferentially processed by the hippocampus. Third, prior knowledge can bias memory encoding such that it can induce false memories. These schema-based memory inferences are presumably mediated by the medial prefrontal cortex, and we have explored the role of this region with TMS. Fourth, slow-wave sleep is important for memory consolidation through replay of associative memories. Using cued memory reactivation in sleeping participants we found functional network changes that explain overnight memory stabilization.