A translational determination of the mechanisms of maladaptive choice in cocaine use disorder

摘要 可卡因使用障碍（CUD）的特征是决定使用可卡因而牺牲其他活动。 因此，以实验室为基础的解决这一问题的努力包括可卡因选择自我管理程序 结合非药物替代品来模拟这一定义特征。使用这些程序的研究 通常调度竞争的调度器以使得概率是确定的。然而，这种确定性 结果不代表现实世界的情况，其中与药物有关的决定的后果 往往是不可预测的。重要的是，在动态的、不确定的背景下做出决策， 价值的选择选项，并要求不断更新的选项价值，这涉及学习过程 和相关的皮质纹状体网络，可能在CUD中功能异常。动态决策 环境已经成功地使用概率学习选择（PRLC）任务建模。 这些任务与重复学习（RL）建模的集成已被用于捕获 时刻变化的机制，动态选择，和应用神经科学技术， 开始识别潜在的神经生物学。这种方法揭示了基于生物学的决策 在多种脑部疾病的异常，但尚未被系统地应用于实验研究， CUD，将RL和神经科学方法结合起来翻译成CUD是合乎逻辑的， 适应不良的选择行为，典型的障碍，在可卡因使用者的不同强化概率 自然环境，以及学习障碍和神经适应，已经记录在 个人CUD因此，在我们对动力学机制的理解方面存在着严重的差距。 可卡因使用决策，以及应用RL框架来填补这些空白的强有力的科学前提。这 项目提出严格的PRLC任务、RL建模、神经调节/fMRI神经成像技术， 在大鼠和人类中进行互补的转化研究设计，以研究CUD中的动态选择。第一组 跨物种实验将证明可卡因使用问题对动态决策的影响 并揭示这种异常任务表现背后的神经行为和神经生物学过程。的 第二组实验将使用PRLC任务，其中静脉注射可卡因可作为替代品， 非药物评估者确定与使用决定相关的行为和神经“概况” 可卡因和减少可卡因的选择在治疗期间。苯丙胺维持和非药物替代 更好的治疗减少了可卡因的选择，这将在这里被用来揭示行为和神经 这些机制可以作为未来治疗开发的目标。这个项目将产生重大影响 通过建立实验应用的协同学习理论的研究领域， CUD中的不适应动态决策。

ABSTRACT Cocaine use disorder (CUD) is characterized by the decision to use cocaine at the expense of other activities. Lab-based efforts to address this problem have therefore included cocaine choice self-administration procedures that incorporate a non-drug alternative to model this defining feature. Studies using these procedures have typically scheduled competing reinforcers so that the probabilities are certain. However, such deterministic outcomes are not representative of real-world scenarios in which the consequences from drug-related decisions are often unpredictable. Importantly, decision-making in a dynamic, uncertain context significantly alters the value of choice options and requires continuous updating of option values, which engages learning processes and related corticostriatal networks that might be functioning abnormally in CUD. Decision-making in dynamic environments has been successfully modeled using probabilistic reinforcement-learning choice (PRLC) tasks. The integration of these tasks with reinforcement-learning (RL) modeling has been used to capture moment-to- moment changes in the mechanisms of dynamic choice, and the application of neuroscience techniques has begun to identify the underlying neurobiology. This approach has uncovered biologically-based decision-making abnormalities in multiple brain disorders, but has yet to be systematically applied to the experimental study of CUD, The translation of combined RL and neuroscience approaches to CUD is logical considering the maladaptive choice behavior that typifies the disorder, the varying reinforcement probabilities in cocaine users’ natural environments, and the learning impairments and neuroadaptations that have been documented in individuals with CUD. Thus, there are critical gaps in our understanding of the mechanisms underlying dynamic cocaine use decisions, and a strong scientific premise for applying an RL framework to fill these gaps. This project proposes rigorous PRLC tasks, RL modeling, neuromodulation/fMRI neuroimaging techniques and complementary, translational study designs in rats and humans to study dynamic choice in CUD. The first set of cross-species experiments will demonstrate the impact of problematic cocaine use on dynamic decision-making and reveal the neurobehavioral and neurobiological processes underlying this abnormal task performance. The second set of experiments will use a PRLC task in which intravenous cocaine is available as an alternative to the non-drug reinforcer to determine the behavioral and neural “profile” associated with the decision to use cocaine and reduced cocaine choice during treatment. Amphetamine maintenance and non-drug alternative reinforcer treatments reduce cocaine choice, which will be leveraged here to uncover behavioral and neural mechanisms that can be targeted for future treatment development. This project will have a significant impact on the field by establishing the experimental application of reinforcement-learning theory to the study of maladaptive dynamic decision-making in CUD.