Mechanisms underlying flexible, context-dependent timing in medial entorhinal cortex

The nervous system must perceive and learn durations to guide adaptive behavior. Prior work to investigate the neural substrates of interval timing has focused largely on relatively simple timing tasks whereby subjects learn to discriminate and reproduce the duration of events occurring at a single fixed interval. Although episodic memory requires accurate temporal coding and is mediated by structures in the medial temporal lobe (MTL), including the medial entorhinal cortex (MEC), a well-established role for MTL circuits in interval timing behavior is conspicuously lacking. MEC seems well-suited to perform timing via continuous attractor network dynamics. However, the role of MEC in timing remains unclear, perhaps because existing timing tasks do not require the type of flexible learning that MTL circuits have likely evolved to mediate. We therefore aim to establish the precise contribution of MEC to interval timing by testing the hypothesis that MEC is necessary for flexible interval timing behavior, which is driven by regular patterns of sequential neural activity. To test this hypothesis, we first developed a novel temporal delayed nonmatch to sample (tDNMS) task that requires mice to flexibly learn and compare the relative durations of pairs of stimuli. Chemogenetic inhibition of MEC completely abolished mice from learning to respond appropriately in each trial type, or temporal context, showing a clear necessity of MEC in flexible timing behavior. As hypothesized, cellular-resolution Ca2+ imaging revealed that populations of MEC neurons fire in regular, time-locked sequences across trials. Interestingly, many time cells display context-dependent timing activity that emerges with learning and supports task performance. Together, findings demonstrate that MEC is necessary to learn flexible, context-dependent timing, which is mediated by sequential activity of MEC time cells. We are now using these results to test circuit-level models, including that MEC time cell activity is driven by a continuous attractor network.