Postsynaptic mechanisms underlying negative prediction error

点击翻译按钮获取中文摘要

ABSTRACT The role of the dopamine system in reward has been well established. Whether dopaminergic mechanisms may underlie reinforcement learning or incentive motivation has been a central issue in the field. Evidence that supports the reinforcement learning hypothesis mostly come from recordings of dopamine neuron firing or dopamine release that are often temporally correlated with prediction error. In contrast, causal evidence that supports the incentive motivation hypothesis mostly come from pharmacological manipulations of dopaminergic signaling that often show altered motivation in reward seeking. One study that stands out is Roy Wise’s 1978 study that used dopamine antagonist and unambiguously demonstrated that decreased dopaminergic signaling is sufficient in causing extinction learning. It is often cited as one of the best causal evidence that supports reinforcement learning but not incentive motivation hypothesis. However, a number of questions remain unanswered. Is it dose dependent? Is it receptor specific? Is it post-synaptic signaling pathway specific? Does such manipulation represent negative prediction error? To address these questions, the present application will employ operant conditioning paradigms similar to what’s used in the Wise paper and test the following hypothesis: lack of dopaminergic signaling--> negative prediction error--> lack of dopamine D2 receptor activation-->lack of inhibition of adenylyl cyclase 5 (AC5) --> elevated cAMP in D2 striatal neurons-->extinction learning. In Aim 1, we will use D2 antagonist and use Gs DREADD expressed in D2 striatal neurons to cause extinction (to test sufficiency). We will also use AC5 knockout mice, AC5 inhibitor and Gi DREADD expressed in D2 striatal neurons to prevent extinction (to test necessity). The objective of Aim 1 is to establish that “D2-AC5-cAMP elevation” is indeed necessary and sufficient in causing extinction learning using natural reward. However, those studies will not demonstrate if this pathway is in fact processing the negative prediction error signal in causing extinction learning. Aim 2 studies will use experimentally generated phasic dip in dopamine release to test this hypothesis. In addition, we will use light induced phasic dopamine neuron firing during no reward condition. We will test if such a manipulation will be able to prevent extinction caused by lack of reward.