Johns Hopkins School of Medicine

Brain-wide computations for learning and memory 

Learning is a fundamental property of living organisms and is essential for survival in dynamic and uncertain environments. Even simple organisms can modify their behavior based on experience, but in more complex animals, learning supports a broader repertoire of sophisticated and flexible behaviors. These capacities likely arise from distributed brain networks in which interacting neuronal ensembles coordinate perception, memory, and action. In the Drieu Lab, we seek to uncover how neuronal ensembles distributed across the brain support learning and memory to enable adaptive behavior.

We study two core, yet traditionally distinct, components of adaptive behavior. On the one hand, we aim to understand how animals learn to use sensory cues from their environment to guide actions and predict outcomes, a process central to instrumental learning. On the other hand, we investigate how animals remember specific experiences – their spatial, temporal, and contextual features, through episodic memory. This form of memory depends critically on the hippocampus, which is also essential for spatial navigation. A unifying function of the hippocampus may be the construction of cognitive maps – internal models of the world that capture the spatial, temporal, and relational structure of experience. These maps support flexible decision-making by enabling inference, generalization, and prediction based on past experiences. From this perspective, the two ‘types’ of memories may be connected. Cue-guided instrumental learning may transiently rely on hippocampal cognitive maps during its initial stages, before behavior becomes stabilized through repeated experience in stable environments.

In the Drieu Lab, we combine innovative behavioral paradigms with state-of-the-art genetic, optical, and electrophysiological tools to uncover how distinct learning and memory systems interact during wakefulness and sleep to support adaptive behavior. We use multi-site, high-density electrophysiology in freely moving rats to investigate how brain rhythms coordinate distributed neuronal ensembles and how neural representations across brain regions reorganize with experience and sleep. Using dual-color two-photon calcium imaging in head-fixed mice, we examine cell-type-specific encoding and how long-range information shapes local network dynamics. We further leverage closed-loop optogenetic manipulations to determine how specific cell populations and neural dynamics causally contribute to behavior. To interpret the rich datasets generated by these experiments, we develop and apply advanced computational methods to reveal the structure and dynamics of brain-wide activity underlying flexible learning. Through collaborations with computational neuroscientists and theorists, we develop new models of distributed brain computation and generate testable predictions that guide experiments and link neural dynamics to behavior.

Our goal is to build a framework for understanding how interactions among distributed brain networks give rise to adaptive behavior. This research provides fundamental insights into cognition and has important implications for understanding and treating memory- and reward-related disorders.