Content area
Abstract
Dopamine has become widely known for its role in reward-based reinforcement learning. This thesis will add to the growing literature arguing that midbrain dopamine neurons are functionally diverse. My work began with the hypothesis that functionally distinct populations should receive inputs from different neurons. To look for this type of organization, I used a virus-based tracing method to label the inputs to dopamine neurons with specified projection targets and developed an automated pipeline for whole-brain imaging and analysis to quantify the data. This study (Chapter 1), demonstrated that many populations of dopamine neurons receive similar patterns of inputs, but that those projecting to the posterior tail of the striatum (TS) were unique and did not receive significant innervation from the ventral striatum, a region known best for its role in reward and addiction. To look for clues regarding the function of each population, I assayed the activity of dopamine neuron axons in different projection targets using a GCaMP calcium indicator (Chapter 2). This study demonstrated that dopamine axons in the ventral striatum (VS) encode a pure value-based prediction error that is not contaminated by novelty or stimulus intensity. Surprisingly, I found that dopamine axons in TS are exclusively modulated by novelty and stimulus intensity and are not related to reward. To look for the function of this unusual population, I performed optogenetic stimulation and drug-based lesion experiments (Chapter 3). This study demonstrated that TS dopamine stimulation caused avoidance, and that TS dopamine is required for retreat from novel stimuli or high intensity stimuli. Together, these data suggest that dopamine neurons are diverse and do not all reinforce reward. This thesis will argue that dopamine performs reinforcement of different types of neural activity depending on the target region.
