Title : Prefrontal cortex state representations shape human credit assignment 1

People learn adaptively from feedback, but the rate of such learning differs drastically 6 across individuals and contexts. Here we examine whether this variability reflects differences 7 in what is learned. Leveraging a neurocomputational approach that merges fMRI and an iterative 8 reward learning task, we link the specificity of credit assignment—how well people are able to 9 appropriately attribute outcomes to their causes—to the precision of neural codes in the 10 prefrontal cortex (PFC). Participants credit task-relevant cues more precisely in social compared 11 to nonsocial contexts, a process that is mediated by high-fidelity (i.e., distinct and consistent) 12 state representations in the PFC. Specifically, the medial PFC and orbitofrontal cortex work in 13 concert to match the neural codes from feedback to those at choice, and the strength of these 14 common neural codes predict credit assignment precision. Together this work provides a window 15 into how neural representations drive adaptive learning. 16 Significance Statement: Successful learning requires selectively attributing outcomes to their 17 cause—a process known as credit assignment. Little is known about how the brain performs this 18 credit assignment, or how the process might differ across contexts or individuals. Functional 19 neuroimaging analyses reveal that precise credit assignment is linked to high fidelity (i.e., 20 distinct and consistent) neural representations of causal cues in the prefrontal cortex (PFC), 21 which supports increased differentiation between stimuli during learning. Our results reveal why 22 individuals learn differently: differences are not driven by the magnitude of learning signals (i.e., 23 stronger prediction errors) as has been previously claimed, but by differences in the strength of 24 neural representations to which those learning signals are attributed. 25