Dopamine signals are thought to be critical for learning. They consist of a baseline tonic firing rate on which a phasic pattern of firing is superimposed. The phasic component of the signal encodes a prediction error, which is the difference between what you expected to happen and what actually happened. Prediction errors can be used to reinforce future behavior. If an outcome was better than expected, you should be more likely to repeat that behavior in future. In contrast, if an outcome was worse than anticipated, you should be less likely to repeat that behavior. Delineating the role of dopamine in reinforcement learning could have important implications for addiction. A common feature of drugs of abuse is their ability to activate the dopamine system and theories of addiction suggest that the inappropriate activation of dopamine is what causes the strengthening of addictive behaviors. Recently, we have discovered that neurons in one part of the frontal lobe, the anterior cingulate cortex (ACC), also encode prediction errors. ACC is one of the heaviest recipients of dopaminergic input in the frontal lobe and is thought to play an important role in enabling rewards to guide the selection of actions. Thus, a critical question is to what extent these frontal lobe prediction error signals depend on dopamine. One problem with answering this question is that ACC is relatively inaccessible. It is located several millimeters from the surface of the brain, on the medial wall of the cerebral hemisphere. However, recent advances in electrode technology have led to the development of an injectrode that can infuse pharmacological agents while simultaneously recording the electrical activity of single neurons. This project will use this device to determine the effect of dopamine on ACC encoding. Specifically, we will train two animals to make choices between pictures that vary in terms of their probability of delivering a reward. During the performance of this task, we will simultaneously record the activity of ACC neurons while infusing either vehicle or the D1 antagonist SCH23390. We predict that infusion of the D1 antagonist will attenuate prediction error signals in ACC, thereby demonstrating that ACC prediction error signals depend on dopamine.
This project focuses on understanding how dopamine modulates encoding in anterior cingulate cortex. Specifically, we will examine whether learning signals in anterior cingulate cortex depend on dopaminergic input. Dopamine and the anterior cingulate cortex are heavily implicated in reward-based decision-making and neuropsychiatric disorders that involve impaired decision-making such as addiction, and so understanding the interaction of these brain systems could allow us to develop novel therapeutic strategies for these disorders.
| Cavanagh, Sean E; Wallis, Joni D; Kennerley, Steven W et al. (2016) Autocorrelation structure at rest predicts value correlates of single neurons during reward-guided choice. Elife 5: |
| Hunt, Laurence T; Behrens, Timothy E J; Hosokawa, Takayuki et al. (2015) Capturing the temporal evolution of choice across prefrontal cortex. Elife 4: |
