Computational dissociation of the causes of cognitive rigidity in depression

抑郁症认知僵化原因的计算分离

• 批准号: 10684272
• 负责人: Alexander Herman
• 金额: $ 18.83万
• 依托单位: UNIVERSITY OF MINNESOTA
• 依托单位国家: 美国
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-08-15 至 2024-07-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10684272
• 关键词: Address; Algorithms; Anhedonia; Animal Model; Behavior; Behavioral; Clinical; Cognitive; Computer Models; Data; Data Set; Decision Making; Depressed mood; Dissociation; Early Diagnosis; Early Intervention; Environment; Equipment and supply inventories; Face; Failure; Feeling hopeless; Future; Goals; Human; Impairment; Individual Differences; Lead; Learning; Link; Major Depressive Disorder; Mental Depression; Mind; Modeling; Moods; Mus; Outcome; Patients; Pattern; Perception; Persons; Physiological; Population; Process; Psychological reinforcement; Research; Rewards; Sampling; Severities; Shapes; Suicide; Suicide attempt; Symptoms; Testing; Translating; Treatment outcome; United States; Variant; Work; anxious; base; cognitive control; cognitive process; cognitive rigidity; computer framework; depressive symptoms; educational atmosphere; environmental change; expectation; experience; flexibility; innovation; model design; models and simulation; novel; pre-clinical; premature; prognostic; response; success; suicidal risk; theories

Depression distorts perceptions of the past and future, manifesting in symptoms such as hopelessness and anhedonia, along with increasingly rigid patterns of decision-making. However, the mechanistic link between important prognostic symptoms in depression and depressive rigidity remains poorly understood. In large part, this is because we lack validated computational models that explain how appraisals of the past and views of the future can mechanistically contribute to cognitive rigidity or flexibility. Here, we construct and test a mechanistic model that will allow us to quantify the impact of depressive symptoms on cognitive flexibility and open new avenues for early diagnosis and intervention. Previous work modeling decision-making in depression with reinforcement learning (RL) has shed light on how depressive symptoms like anhedonia alter reward-based judgements. However, standard RL lacks the validity needed to explain depressive rigidity. Here, we develop a novel model from another powerful computational framework: foraging theory. We designed this model with depressive symptoms in mind to explicitly link learning from the past and estimating the future to cognitive flexibility. The central hypothesis is that learning from past rewards and estimating future rewards are dissociable mechanisms that control cognitive flexibility. The specific aims of this proposal are to (1) Determine how appraisals of the past and future shape cognitive flexibility and (2) Examine how variations in the environment constrain cognitive flexibility. To accomplish these aims, we will characterize how judgements about the past and future influence decision-making rigidity and respond to environmental changes through model simulation and analysis. We will determine how individual differences in these cognitive processes are learned from the environment and if they predict rigidity by administering a flexible decision-making task to clinical depression and large online samples. Collecting depressive symptom inventories along with task data will allow us to interrogate the mechanisms by which depressive symptoms like anhedonia lead to rigidity. To test cross-species validity for preclinical work, we apply this model to a previously collected mouse behavioral dataset. Innovation: Our novel model will determine how views of the past and future, and their responses to the environment, contribute to cognitive rigidity, and how they are impacted by depressive symptoms in an online sample. The results will guide future hypothesis-driven research into the algorithmic basis of depressive rigidity in patients. Specifically, a future R01 application will test the model’s utility for predicting depression subtypes and treatment outcomes in a clinical population. This model will also enable us to study the physiological bases of these computational processes in animal models and humans undergoing invasive and non-invasive neurophysiolgoical studies.

抑郁症扭曲了对过去和未来的看法，表现为绝望和 快感缺失，以及日益僵化的决策模式。然而，两者之间的机械联系 抑郁症和抑郁强直的重要预后症状仍然知之甚少。在很大程度上， 这是因为我们缺乏经过验证的计算模型来解释如何评估过去和现在的情况。 对未来的看法可以机械地促进认知僵化或灵活性。在这里，我们构造 并测试一个机制模型，该模型将使我们能够量化抑郁症状对认知的影响 灵活性并为早期诊断和干预开辟新途径。 之前通过强化学习（RL）对抑郁症决策进行建模的工作已经揭示了这一点 关于快感缺失等抑郁症状如何改变基于奖励的判断。然而，标准强化学习缺乏 解释抑郁僵化所需的有效性。在这里，我们从另一个强大的模型中开发了一个新颖的模型 计算框架：觅食理论。我们设计这个模型时考虑到了抑郁症状 将过去的学习和对未来的估计与认知灵活性明确联系起来。 中心假设是从过去的奖励中学习和估计未来的奖励是分离的 控制认知灵活性的机制。该提案的具体目标是 (1) 确定如何 对过去和未来的评估塑造了认知灵活性，并且 (2) 检查 环境限制了认知的灵活性。为了实现这些目标，我们将描述判断如何 关于过去和未来的影响决策刚性并通过以下方式应对环境变化 模型模拟与分析。我们将确定这些认知过程中的个体差异如何 从环境中学习，如果他们通过管理灵活的决策任务来预测僵化 临床抑郁症和大量在线样本。收集抑郁症状清单以及任务数据 将使我们能够探究快感缺失等抑郁症状导致僵化的机制。到 为了测试临床前工作的跨物种有效性，我们将该模型应用于之前收集的小鼠行为 数据集。 创新：我们的新颖模型将决定如何看待过去和未来，以及他们对未来的反应 环境，导致认知僵化，以及他们如何受到在线抑郁症状的影响 样本。研究结果将指导未来以假设为驱动的抑郁僵硬算法基础研究 在患者中。具体来说，未来的 R01 应用程序将测试该模型预测抑郁症亚型的实用性 以及临床人群的治疗结果。该模型还将使我们能够研究生理基础 动物模型和人类经历侵入性和非侵入性的这些计算过程 神经生理学研究。