Investigating the role of highly reliable neurons in sequential activity during decision-making tasks

초록

Sequential neural activity has been widely observed across brain regions, mostly in rodents, performing various tasks. This temporally ordered population activity has also been reported to encode working-memory information shaped by sensory cues during decision-making tasks. Moreover, in a previous study, the sequence shown in averaged neural activities across correct trials was less obvious in those across error trials. However, whether sequences observed in averaged activity are maintained on a trial-by-trial basis has not been thoroughly examined, making it difficult to determine their importance for representing choice-relevant information. Here, we address this issue by analyzing neural population activity of mice acquired during decision-making tasks on a trial-by-trial basis. Using this method, we identified a small subset of 'Highly Reliable Neurons' (HRNs) that were not only highly predictive of the animal's choice but also exhibited significantly greater spike timing precision across trials compared to other neurons. In conclusion, the results of this study suggest that the sequential activity seen in trial averages is not a consistent chain-like propagation in single trials. Rather, it is the reliable firing order of HRNs across trials that gives rise to the appearance of sequential activity.