Orbitofrontal cortex as a cognitive map of task space

Orbitofrontal cortex (OFC) has long been known to play an important role in decision making.  However, the exact nature of that role has remained elusive. The OFC does not seem necessary for almost anything---animals and humans can learn, unlearn and reverse previous learning even without an OFC, albeit more slowly than their healthy counterparts.  What role, then, can the OFC be playing such that its absence would cause subtle but broadly permeating deficits? We propose a new unifying theory of OFC function. Specifically, we hypothesize that OFC encodes a map of the states of the current task and their inter-relations, which provides a state space for reinforcement learning elsewhere in the brain. I will first use a simple perceptual judgement task to demonstrate that state spaces, a critical ingredient in any reinforcement learning algorithm, are learned from data. I will then use our hypothesis that the OFC encodes the learned state space to explain recent experimental findings in an odor-guided choice task (Takahashi et al, Nature Neuroscience 2012) as well as classic findings in reversal learning and extinction. Finally, I will lay out a number of testable experimental predictions that can distinguish our theory from other accounts of OFC function.