Hot Off the Press – January 3, 2022
Zhang and Denman et al developed deep behavior mapping (DBM), a self-supervised learning method for fine-grained analysis of behavioral micro-states in video data. They combined DBM with miniScope in vivo calcium imaging in freely moving mice to show that neurons in the prelimbic cortex represent a wide variety of behaviors, especially the sequence of behaviors involved in pressing a lever for food reward and continuous sequences of exploratory behaviors. This demonstrates that prelimbic neurons are not only associated with the important events in a behavioral sequence, but with the entire sequence from start to finish in the operant learning. When animals learned a new behavior, neurons with weak and unstable tuning are recruited to encode newly learned behaviors.