Medial entorhinal cortex plays a specialized role in learning of flexible, context-dependent interval timing behavior

In order to survive and adapt in a dynamic environment, animals must perceive and remember the temporal structure of events and actions across a wide range of timescales. Prior research to investigate the neurobiological basis underlying so-called interval timing (i.e. timing on the scale of second to minutes) has focused largely on simple timing tasks that require animals to discriminate and/or reproduce stimuli presented at single, fixed durations over thousands of trials. Perhaps not surprisingly, this work has uncovered circuits involved in procedural memory systems, such as the basal ganglia, that play a key role in learning and driving timing behavior. Episodic memory (i.e. the ability to remember specific, personal events that occur in spatial and temporal context) also requires accurate temporal processing and is known to require neural circuits in the medial temporal lobe (MTL), including entorhinal cortex. However, a clear role for medial temporal lobe structures in interval timing has been conspicuously lacking. In order to determine the specific role that MTL structures play in timing behavior, we developed a novel behavioral paradigm that requires learning complex temporal contingencies. Applying this novel interval timing task in mice, in concert with methods for manipulating neural activity and methods for large-scale cellular resolution neurophysiological recording, we have uncovered a specific role for medial entorhinal cortex (MEC) in flexible, context-dependent learning of interval timing behavior. We find neurons in MEC, termed time cells, fire in regular time-locked sequences during timing. Remarkably, over the course of learning, these neural dynamics become context-dependent with specific sub-populations of neurons becoming selectively active within particular trial contexts. We also find that this neural activity is disrupted when animals make errors in their timing behavior and that animals are completely unable to learn this timing behavior without MEC activity. Further, we find that MEC activity is not necessary for very rigid forms of timing behavior and that nearby structures, such as the hippocampus, are not necessary for learning this flexible timing task. Finally, the results identify the particular cognitive strategy employed by the animal to discriminate contexts that contain more complex temporal structure, that often exist in nature. Together these results establish a precise role of MEC in learning flexible timing behavior and demonstrate that learning is driven by context-dependent sequential neural dynamics of MEC time cells.