Genetic Dissection of Neural Circuits Underlying Value Based Decision Making

Decision making is the fundamental process that we utilize to accomplish objectives in everyday lives. To understand the neural substrates of this process, we developed a behavioral task for mice that required repetition of the processes of action initiation, action selection, and learning. The task is a two-option choice task with stochastic reward delivery and reversals. To map brain areas involved in this type of value-based learning, we inactivated neuronal activity in the prefrontal cortex (PFC) and the nucleus accumbens (NAc) while mice performed the task. Inactivation of the NAc resulted in altered action initiation and learning but had subtle effects on choice. To dissect underlying neural circuitry, specific cell types and inputs of the NAc were inactivated. Inactivation of two dominant cell types in the NAc, direct and indirect pathway medium spiny neurons (MSNs), showed partially overlapping but distinct behavioral effects. Inactivation of direct pathway MSNs showed the stronger effect on learning while inactivation of indirect pathway MSNs also showed the effect. In contrast, only inactivation of indirect pathway MSNs affected behavioral measures of action initiation. The contribution of specific inputs to the NAc, dopaminergic and glutamatergic inputs, were also studied. While both experiments affected behavioral measures of initiation of action, only the inactivation of dopaminergic inputs affected learning. The effect on learning was specific to trials after reward omissions, and the effect was more prominent in trials which animals spent less time to initiate. These results provided new insights into the function of the NAc in processing information about reward values. In contrast, inactivation of two subregions in the PFC, the anterior cingulate cortex (ACC) and the orbitofrontal cortex OFC, affected action initiation and action selection. The action initiation was affected by inactivation of both areas, but OFC inactivation affected more behavioral measures. In contrast, action selection was affected more prominently in ACC inactivation. These differential effects on action initiation and action selection suggested the functional distinction between these two areas. In this study, we have developed a behavioral assay that allowed us to dissect different aspects of cognitive functions for decisions in mice and revealed roles of distinct circuit elements in the NAc and the PFC. Utilizing temporally precise inactivation, we found that the same circuit element was used for different cognitive processes depending on the timing. Although this type of behavioral task has been used extensively in rats and primates to understand decision making, identification of cell types and circuits required for these behaviors has been difficult in these species due to the lack of the powerful genetic methodologies. The approach we have demonstrated here is important because it enables genetic dissection of complex behaviors in mice, allowing studies of circuit properties that are executed by specific cell types in the cerebral cortex and basal ganglia. Since the approach taken in this study can be expanded to other neural circuits and behavioral paradigms, this and future studies will reveal the neural basis of decision making and, perhaps, lead to new approaches to treatments for maladaptive behaviors