Medial prefrontal cortex population activity is plastic irrespective of learning

Learning complex statistical relationships of the world, such as contingencies between actions and outcomes, is thought to depend on neural populations in the prefrontal cortex. The neural plasticity that underpins learning should alter the structure of population activity. But little is known about what changes to population activity in prefrontal cortex are specific to learning. To address this question, here we characterise the plasticity of population activity in medial prefrontal cortex of rats learning rules on a Y-maze. We show that the structure of joint population activity consistently changes between bouts of sleep before and after a session of training on the maze, irrespective of learning during training. Instead, only during learning are changes to joint activity during training carried forward to post-training sleep. Whereas non-learning population plasticity could be largely accounted for by changes to individual neuron excitability, plasticity in learning is driven by changes to both excitability and rate correlations between neurons. Our results suggest constant population-level plasticity in prefrontal cortex driven by changes to neuron excitability, and support the hypothesis that learning is underpinned by neural plasticity that creates long-lasting constraints on population dynamics.