Behavioral Social Neuroscience Seminar

We have known since the case of Phineas Gage in 1848 that regions of the prefrontal cortex (PFC) are essential for learning and decision-making. Converging evidence from lesion, neuroimaging and electrophysiology studies emphasize that a set of PFC areas – the anterior cingulate cortex (ACC), dorsolateral PFC (DLPFC), orbitofrontal cortex (OFC) and ventromedial PFC (VMPFC) – are modulated when subjects make value-based decisions. Yet what remains relatively elusive is a neural signature of "how" each PFC area contributes to learning and decision-making. Here I will discuss recent electrophysiology experiments in our lab to examine this issue. In a first experiment, I will discuss the activity patterns of ACC and OFC neurons while subjects learned new stimulus-outcome associations, and then made choices either between newly learned stimuli or between overlearned stimuli. Whereas ACC chosen value neurons responded similarly irrespective of cue novelty as if value was encoded abstractly, OFC chosen value neurons differentiated between novel and overlearned cues. We also found evidence of a dynamic "stimulus code" present in OFC during both learning and choices between novel cues, suggesting a mechanism by which value might be ascribed to novel stimuli in OFC during learning for use in decision making. In a second experiment, we explored how value signals evolved as subjects gathered information in a sequential multi-attribute decision making task. A significant proportion of neurons throughout ACC, OFC, DLPFC and VMPFC exhibited value-related selectivity during the task. However, neurons in ACC and DLPFC differentially encoded cue value based on an action reference frame (Left/Right response), while OFC neurons differentially encoded cue value based on the decision attribute (reward probability vs reward magnitude). As more information was obtained toward making a choice, value signals evolved from encoding the value of the cue that was currently attended to encoding the value of the option that would be eventually chosen. By utilizing the natural dynamics of the information gathering and choice strategies, we provide evidence of "how" different PFC subpopulations make simultaneous (but specialized) contributions to value computation and comparison.