Neuronal Mechanisms of Decision Making in the Prefrontal Cortex

This thesis examines several aspects of decision computations which are critical for understanding the processes by which decisions are made. It will show that subjects engaged covert attention to bias both saccadic and choice processes during simple decision making tasks even when these stimuli were novel. This saccadic behaviour was overridden when one presented stimulus is relatively more novel than the other implying the existence of separate value comparison circuits in the brain which deal with making value based decisions about attention and choice respectively. Even when the task was made more complex by introducing multiple decision variables this phenomenon of covert attention was maintained. This thesis will demonstrate that subjects controlled both the amount and manner of information gathering during decisions. This behaviour showed features of a confirmation bias. Single cell neuronal recordings were performed while subjects executed a multiattribute decision making task. ACC neurons represented action values and different populations of OFC neurons encoded attribute and attentional values. These neurons did not just reflect value (i.e. an input into a decision process) but instead evolved their coding to represent final choice thereby implying the existence of a parallel decision making circuit which compares value in different frames of reference. Information gathering strategy was also computed in the same frames of reference implying the existence of a common value comparison system which simultaneously drives both choice and information gathering. At the outcome of the decision ACC neurons encoded both categorical reward outcome and positive prediction errors. vmPFC neurons encoded prediction errors while OFC and ACC neurons encode fictive value when rewards were withheld. Finally frame of reference specific computations were observed in LPFC and OFC. The results in this thesis therefore provide novel insight into the role of valuation circuitry during value based decision making and outcome monitoring