Neural Representation of Economic Parameters by Dopamine Release: a Quantitative Analysis

Abstract

The goal of the field of neuroeconomics is to understand the neural representations of variables relevant to theories of decision-making. These variables include the expected value of a prospect and the uncertainty associated with a choice. Recordings from dopamine neurons in awake, behaving monkeys strongly indicate that dopamine plays a role in representing these parameters. A complementary role of dopamine is the encoding of a reward prediction error signal, which can be used to iteratively update the expected value associated with a cue that predicts reward. While these roles of dopamine neurons have been qualitatively described, there are very few quantitative analyses that relate dopamine to these variables, and none have been conducted at the level of dopamine release. In the current thesis, I combined simple Pavlovian and operant behavioral tasks with recordings of dopamine release in the nucleus accumbens core of rats, using the electrochemical method fast-scan cyclic voltammetry. To address the relationship between dopamine release and expected value and uncertainty, I applied regression analyses to conditioned stimulus-evoked dopamine signals during learning. To address the relationship between dopamine release and reward prediction error signaling, I applied an axiomatic reward prediction error model to reward-evoked dopamine release from rats performing an operant task with probabilistic rewards. I report that dopamine release correlates with expected value, uncertainty, and reward prediction error signals. Expected value and uncertainty signals may be blended at the level of dopamine release, and reward prediction error signals can be represented in a balanced or an imbalanced manner depending on the range of errors studied and the behavioral task.