TY - JOUR T1 - Task learning reveals neural signatures of internal models in rodent prefrontal cortex JF - bioRxiv DO - 10.1101/027102 SP - 027102 AU - Abhinav Singh AU - Adrien Peyrache AU - Mark D. Humphries Y1 - 2017/01/01 UR - http://biorxiv.org/content/early/2017/02/01/027102.abstract N2 - The inherent uncertainty of the world suggests that brains should internally represent its structure using probabilities. This idea has provided a powerful explanation for a range of behavioural phenomena. But describing behaviour in probabilistic terms is not strong evidence that the brain itself explicitly uses probabilistic models. We sought to test whether populations of neurons represent such models in higher cortical regions, learn them, and use them in behaviour. Combining theories of probabilistic learning and sampling, we predicted that trial-evoked and sleeping population activity respectively represent the inferred and expected probabilities generated from an internal model of a behavioural task; and that these distributions would become more similar as the task was learnt. To test these predictions, we analysed population activity from rodent prefrontal cortex before, during, and after sessions of learning rules on a Y-maze. We found that population activity patterns on millisecond time-scales occurred far in excess of chance in both waking and sleep activity. The distributions of these patterns changed between sleep episodes before and after successful learning. Changes were greatest for patterns expressed at the maze's choice point and predicting correct choice of maze arm to obtain reward, consistent with the population activity representing an internal model of the task. As predicted, these changes consistently increased the similarity between the distributions in trials and in post-learning sleep, compared to pre-learning sleep, implying that the underlying probability distribution had stabilised over successful learning. Our results provide evidence that prefrontal cortex contains a probabilistic model of behaviour, which is updated by learning. They thus suggest sample-based internal models are a general computational principle of cortex.Author Summary The cerebral cortex contains billions of neurons. The activity of one neuron is lost in this morass, so it is thought that the co-ordinated activity of groups of neurons – “neural ensembles” – are the basic element of cortical computation, underpinning sensation, cognition, and action. But what do these ensembles represent? Here we show that ensemble activity in rodent prefrontal cortex represents samples from an internal model of the world-a probability distribution that the world is in a specific state. We find that this internal model is updated during learning about changes to the world, and is sampled during sleep. These results suggest that probability-based computation is a generic principle of cortex. ER -