Topic: demixed Principal Component Analysis

The ability to continually adapt to the environment is key for survival. How the brain learns continuously while retaining previous knowledge is not known. Using a combination of neural and behavioural data analysis together with deep learning modelling, we studied the role of the hippocampus in continual spatial reinforcement learning. First, we introduce a deep recurrent Q-learning agent consistent with the hippocampal architecture (hcDRQN) along with three control models: a non-recurrent model (hcDQN) and two standard machine continual learning algorithms. We trained all these reinforcement

Real-world decisions often require combining different cognitive computations. Recent findings suggest that, on the one hand, these computations are distributed across the cortex. On the other hand, cortical microcircuits seem to implement them via (near) orthogonal population-activity patterns. But how are these local geometries combined at the level of cortex-wide dynamics? To answer this, we developed a new task for mice navigating in virtual reality (VR) to temporally disentangle different cognitive computations. Mice first accumulate evidence to determine the predominant color of a random