One promising theoretical framework to explain the function of cortex is predictive processing. It postulates that cortex functions by maintaining an internal model, or internal representation, of the world through a comparison of predictions based on this internal model with incoming sensory information. Implementing predictive processing in a cortical circuit would require a set of distinct functional cell types. These would include neurons that compute a difference between top-down predictions and bottom-up input, referred to as prediction error neurons, and a separate population of neurons that integrate the output of prediction error neurons to maintain an internal representation of the world. In this research project we test the framework of predictive processing and identify different putative circuit elements and cell types that are thought to form the circuit in mouse visual cortex. We use a combination of physiological recordings, optogenetic manipulations of neural activity, and gene expression measurements to determine the cell types that have functional responses consistent with different prediction errors, as well as those coding for the internal representation. Identifying the circuit elements underlying predictive processing in cortex may reveal a strategy to bias processing either towards top-down or bottom-up drive when the balance between the two is perturbed, as may be the case in neuropsychiatric disorders.
This research is funded by the European Research Commission under ERC-2019-COG - Grant Agreement-865617-CELPRED