Reward-Predictive Value in Learning-Related Hippocampal Neural Activity

摘要

We use point process likelihood models to explain the spiking activity of individual hippocampal pyramidal neurons as a function of task-relevant covariates and their Temporal Difference (TD) reward-predictive value during reinforcement learning. Our results suggest that CA1 and CA3 neurons may signal information about the reward-predictive value of task-relevant signals pertinent to the acquisition of behavioral strategies that maximize reward, with more CA1 but not CA3 neurons doing so during novel versus familiar experiences.