Influence of Reward and Punishment on Goal-directed and Habit Learning in Adolescent Anorexia Nervosa

项目总结/摘要 神经性厌食症（AN）是一种使人衰弱的、通常是慢性的、有时是致命的进食障碍（艾德），其典型地 在青春期发育。AN的病因可能是复杂的，但仍然未知，因此，目前 治疗缺乏足够的疗效。存在显著的症状异质性，但目前的研究不能很好地捕捉到这些症状。 诊断标准因此，识别潜在的艾德症状及其神经元的维度结构， 相关性对于提高我们对艾德的机制理解和识别新的治疗靶点至关重要， 通知精准医疗AN的行为症状，如食物限制（AN-R）和暴饮暴食/腹泻 事件（AN-BP），来自适应不良的决策。决策由两个系统管理：1） 目标导向的学习反映了基于预期结果的灵活和有目的的行动，以及2）习惯 学习反映了由先前强化建立的自动、有效和不灵活的选择。这些 学习系统是由分离但重叠的皮质纹状体回路介导的， 使用两步顺序决策任务建模。很难在这些系统之间进行仲裁， 导致对一种策略的过度依赖，可能导致不同的AN症状。这 一项提案测试了一个新的假设，即在奖励和损失的条件下，目标导向和习惯学习 在AN中不同，并且与不同的症状和皮质纹状体连接唯一相关。青少年 患有AN-R、AN-BP或健康对照的14-17岁女孩将接受神经影像学检查， 状态皮质纹状体网络的功能连接和微结构完整性，然后进行 在奖励和损失条件下，扫描仪外的两步学习任务。对于每个参与者， 因子（ω）代表目标导向学习和习惯学习之间的相对平衡，将计算增益 的损失条件。扫描后，AN参与者将完成为期14天的生态瞬间评估 评估自然的、瞬间体验的艾德症状的方案。目标1将决定行为 AN患者在奖励和惩罚条件下目标导向和习惯学习的差异- R、AN-BP和健康对照。目标2将检验目标导向学习和奖励习惯学习之间的关系 和惩罚与症状和现实世界的经验，以确定是否学习差异 导致不同的症状目标3将确定皮质纹状体回路是否与目标相关- 奖励和惩罚的定向和习惯学习因组和条件而异，以告知神经机制 在AN中改变的强化学习。数据将支持未来的纵向R 01应用程序，以测试 基于强化学习的行为表型的维度方法， 限制和放纵型ED，以告知神经发育模型。在艾德中定义有意义的生物标志物 这将是该领域的一个重大进展，因为它将允许开发新的治疗方法， 可怜的结果。

PROJECT SUMMARY/ABSTRACT Anorexia nervosa (AN) is a debilitating, often chronic, and sometimes fatal eating disorder (ED) that typically develops in adolescence. The etiology of AN is likely complex, yet remains unknown, and as a result, current treatments lack sufficient efficacy. Significant symptom heterogeneity exists, yet is not well captured by current diagnostic criteria. Thus, the identification of dimensional constructs underlying ED symptoms and their neural correlates is critical to improve our mechanistic understanding of ED and identify novel therapeutic targets to inform precision medicine. Behavioral symptoms of AN, such as food restriction (AN-R) and binge/purge episodes (AN-BP), arise from maladaptive decision-making. Decision-making is governed by two systems: 1) goal-directed learning reflects flexible and purposeful actions based on anticipated outcome, and 2) habit learning reflects automatic, efficient, and inflexible choices established by previous reinforcement. These learning systems are mediated by separate, yet overlapping, corticostriatal circuits and can be computationally modeled using a two-step sequential decision-making task. Difficulty arbitrating between these systems, resulting in an over-reliance on one strategy over the other, may contribute to divergent AN symptoms. This proposal tests the novel hypothesis that goal-directed and habit learning under conditions of reward and loss differs in AN and is uniquely associated with divergent symptoms and corticostriatal connectivity. Adolescent girls ages 14-17 with AN-R, AN-BP or who are healthy controls will undergo neuroimaging to assess resting state functional connectivity and microstructural integrity of corticostriatal networks, followed by performing the two-step learning task outside the scanner under conditions of reward and loss. For each participant, a weighting factor (ω), representing the relative balance between goal-directed and habit learning, will be calculated for gain and loss conditions. After scanning, AN participants will complete a 14-day ecological momentary assessment protocol to evaluate naturalistic, momentary experiences of ED symptoms. Aim 1 will determine the behavioral differences in goal-directed and habit learning under conditions of reward and punishment in individuals with AN- R, AN-BP and healthy controls. Aim 2 will examine the association of goal-directed and habit learning for reward and punishment with symptoms and real-world experiences in AN to determine whether learning differences contribute to divergent symptoms. Aim 3 will determine whether corticostriatal circuits associated with goal- directed and habit learning for reward and punishment differ by group and condition to inform neural mechanisms of altered reinforcement learning in AN. Data will support a future longitudinal R01 application to test a dimensional approach of behavioral phenotyping based on reinforcement learning in a wider spectrum of restricting and binge-type EDs to inform a neurodevelopmental model. Defining meaningful biomarkers in ED would be a major advance in the field, as it would allow the development of new treatment approaches to reduce poor outcome.