Prefrontal representations of retrospective spatial working memory in a rodent radial maze task
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Working memory is the cognitive capacity for temporarily holding information in mind for processing or use. It has been theorized to depend upon executive and mnemonic subcomponents, although the contextual mapping of these subcomponents is not complete. Perturbations of prefrontal cortex (PFC) delay activity disrupt spatial working memory performance in rodent tasks. However, recordings of unperturbed PFC delay activity do not consistently contain mnemonic representations of spatial information in these tasks, calling into question the role that mnemonic PFC representations play in freely-moving spatial working memory. We hypothesized that increasing task complexity might increase the likelihood of mnemonic PFC representation emergence. We therefore used an automated eight-arm radial maze to implement a novel match-to-sample rodent spatial working memory task with seven options on each trial, and recorded calcium activity in PFC neurons during task performance. We found that the delay-phase activity of PFC neurons indeed contained mnemonic representations of spatial information at the population level. These representations were retrospective rather than prospective, and—surprisingly—were more evident on error trials. Together with previous results, these observations suggest that in freely-moving spatial working memory tasks, PFC mnemonic representations emerge to empower deviation from a routine behavioral strategy.
Significance Statement
Prefrontal cortex (PFC) activity is necessary for optimal performance of freely-moving spatial working memory tasks in rodents. Despite this, PFC representations of retrospective actions or stimuli—one quintessential working memory hallmark—are only variably observed during task delays, complicating our understanding of the PFC’s role in spatial working memory. Here, we examine cellular-resolution PFC activity in a high-optionality match-to-sample radial maze task and find retrospective delay representations. Strikingly, these delay representations are more evident in error trials. This suggests that in the freely-moving context, explicit PFC representations of retrospective information support deviations from an entrained behavioral strategy, rather than equally supporting all spatial working memory-based behavior.
