A synaptic novelty signal to switch hippocampal attractor networks from generalization to discrimination

Episodic memory formation and recall are complementary processes that put conflicting requirements on neuronal computations, as the reliable retrieval of familiar representations, supported by robust attractor properties in hippocampal circuits, opposes the formation of new neuronal assemblies for the storage of novel episodic memories. To address this problem, we studied the activity of dentate gyrus granule cells in mice exploring familiar and novel environments. A consistent transient depolarization in these cells was found when mice are exposed to a novel environment for the first time. We show through a computational neural network model how this novelty signal can drive the downstream attractor networks from a familiar representation to a new state, not supported by strong synaptic inputs, thereby favoring the switch from memory retrieval to encoding. We then show that the introduction of Hebbian learning in the model is essential not only to consolidate the representation but also to reinstate network activity in the new neural state during subsequent explorations of the novel environment, winning the competition against the already consolidated attractors associated to familiar environments, even in the absence of additional inputs from the granule cells population.