Candidate Neural Activity for the Encoding of Temporal Content in Memory

Abstract

We remember our lives in the context of time which allows us to predict upcoming events. However, there is no sensory organ or neural receptor specialized for measuring time. Instead, temporal content in memory arises from the integration of different time-varying neural activity profiles that can be found throughout the brain including, but not limited to, the medial prefrontal cortex and hippocampus memory system. In order to study temporal coding in memory, researchers have concentrated on extracellular recordings in freely behaving animals (including humans). This is because time is strictly an internal representation generated by the brain and has no single sensory or physical correlate. Thus, it necessary to use sophisticated approaches that can provide strong evidence for temporal content-based brain-behavior relationships. Without suitable behavioral measures, time-varying neural activity itself confounds the interpretation of any internal temporal representation. Additionally, time cannot be generated in the lab, thus diminishing experimental control, nor can it be easily separated from time-varying external covariates which themselves could be driving both behavioral and neural activity spuriously thought to be driven by the flow of time. In this chapter, we briefly review the literature on the neurobiology of time in the memory domain and summarize key criteria researchers have used to study temporal content. We first emphasize the behavioral aspects of studies that have focused on interval timing (scale = seconds), elapsed time memory (scale = minutes), and memory for sequences of events (temporal order). We then discuss three candidate temporally structured neural events that could underlie our ability to encode temporal content including ramping cells, time cells, and sequence cells. Ramping cells increase or decrease their activity leading up to an event or decision, time cells fire at circumscribed elapsed time intervals after an event, and sequence cells fire relative to the expected order of events. Ramping cells and time cells share key features with time perception, notably an increase in variance that is proportional to that magnitude of the elapsed time. Alternatively, sequence cells represent the flow of events as they had occurred in a match/mismatch fashion and thus are best conceptualized as contributing to memory retrievals. Lastly, we review some of basic systems architecture for time in memory.