Credit assignment is a critical component of learning in environments where the outcomes of particular choices may be delayed, and where the cause for any particular outcome may be but one among many possibilities. Properly assigning credit for speci?c outcomes underlies our ability to determine causality and make sense of our environment. However, the manner in which we connect causes to effects is not well understood. We believe the lack of knowledge about credit assignment and uncertainties in existing data arise from three main factors: 1) the absence of clear criteria neuronal activity must ful?ll to solve the credit assignment problem; 2) the paucity of single neuron neurophysiology, obtained during appropriate cognitive tasks, to reveal the low- level representations involved in credit assignment; and 3) the lack of direct comparisons across neuronal populations to determine the relative contributions of the lateral, orbital and medial PFC (?PFC, ?PFC and ?PFC) to this cognitive function. Based on previous work and our own data, we hypothesize the ?PFC is the PFC region whose activity is most consistent with credit assignment. Speci?cally, we expect neuronal activity ful?lling the requirements for credit assignment to be most prominent in the ?PFC, whereas activity in other PFC areas (?PFC and ?PFC) will fail to conform to those requirements, or do so in a delayed fashion relative to the ?PFC. Further, we predict credit assignment related activity in the ?PFC will be most predictive of learning behavior. We will record neurons in the PFC of nonhuman primates performing relevant cognitive tasks while applying a multi-level approach to elucidate the the circuit mechanisms of credit assignment. We will analyze individual neurons, population codes and dynamic inter-areal communication to understand these critical representations and their relationship to behavior. A better understanding of how the PFC contributes to this critical cognitive function will enable a more mechanistic assessment of frontal lobe impairments observed in frontal dementia, traumatic brain injury, stroke and other neurological diseases, as well as facilitate more principled therapies for these debilitating conditions.
Dysfunction of higher cognition, such as of learning and planning, are common in a wide variety of psychiatric and neurological diseases, and the prefrontal cortex is frequently implicated. By studying the precise, normal role of this structure in solving a particular cognitive problem ? namely, how outcomes are linked to their causes ? we aim to develop a more rational understanding of the manifestations of prefrontal dysfunction. Such an understanding could lead to more effective interventions for those suffering from these problems, ranging from behavioral therapies to neuromodulation.
