Computational Substrates of Addiction and Reward

描述（由申请人提供）：滥用药物的部分运作方式是破坏通常用于选择行动和做出决定的适应机制（算法）。否则，为什么一个正常的人会放弃食物、性、住所和一般的福祉，把白色粉末放进鼻子里，或者把海洛因放进血管里？很明显，成瘾者的决策算法出现了故障，从这个意义上说，吸毒代表了一套病态的算法，用于奖励学习和奖励依赖决策。众所周知，许多（如果不是大多数）滥用药物在多巴胺受体水平或直接在将多巴胺递送到靶结构的神经元上影响中脑多巴胺系统。因此，重要的是要了解这些系统执行的算法，算法正常工作的方式，以及它们与潜在的细胞和分子事件的联系方式。这项提议的广泛目标是通过提供对中脑多巴胺系统构建和传播的信息种类以及这些信号对下游神经靶点的影响的计算理解来建立这样一种联系。使用功能性磁共振成像（fMRI）在人类受试者中，这项工作将进行详细的测试中脑多巴胺能系统的计算模型：预测误差模型。该模型已被证明可以解释非人类灵长类动物在奖励依赖学习和决策任务中神经活动的快速变化。这个计算模型的一个中心思想是，腹侧被盖区和黑质多巴胺神经元的一个大子集的活动变化代表了对未来奖励事件预测的错误。这个模型的细节做出了许多重要的预测，这些预测从未在人类受试者中进行过测试。这项工作将确定参与人类奖励条件反射的神经通路，并将用于测试和扩展多巴胺能功能的预测误差模型。此外，通过研究人类对奖励期望的基本神经生物学过程，这一提议对理解学习和其他认知过程的各个方面具有广泛的意义。

DESCRIPTION (provided by applicant): Drugs of abuse operate in part by subverting the adaptive mechanisms (algorithms) that are normally used to choose actions and make decisions. Otherwise, why would a normal human give up food, sex, shelter, and general well being to put white powder in their nose or heroin in their veins? Clearly an, addict's decision making algorithms are malfunctioning, and in this sense, drug taking represents a set of diseased algorithms for reward-learning and reward-dependent decision-making. It is well known that many if not most drugs of abuse affect midbrain dopamine systems either at the level of dopamine receptors or directly on the neurons that deliver dopamine to target structures. It is therefore important to understand the algorithms that these systems carry out, the way that the algorithms normally function, and the way they are linked to underlying cellular and molecular events. The broad objective of this proposal is to establish one such link by providing computational understanding of the kinds of information constructed and broadcast by midbrain dopamine systems and the influence of these signals on downstream neural targets. Using functional magnetic resonance imaging (fMRI) in human subjects, this work will carry out detailed tests of a computational model of midbrain dopaminergic systems: the prediction error model. This model has been shown to account for rapid changes in neural activity in non-human primates during reward-dependent learning and decision-making tasks. One central idea of this computational model is that activity changes in a large subset of dopamine neurons of the ventral tegmental area and substantia nigra represent errors in the predictions of future rewarding events. The details of this model make numerous important predictions that have never been tested in human subjects. This work will identify those neural pathways involved in reward conditioning in humans and will serve to test and extend the prediction error model of dopaminergic function. Moreover; by studying a fundamental neurobiological process of reward expectancy in humans, this proposal has broad implications for understanding aspects of learning and other cognitive processes.