Karthik Swathi K, Pereira Katherine M K, Culbreth Adam J, Barch Deanna M, Moran Erin K
Department of Psychological and Brain Sciences, Washington University in St. Louis, St. Louis, MO, 63130, USA.
Department of Psychology, University of Wisconsin-Madison, Madison, WI, 53706, USA.
Cogn Affect Behav Neurosci. 2025 May 22. doi: 10.3758/s13415-025-01302-3.
Decision making is driven by factors such as motivation, pleasure, and cognitive skill. The current study evaluates how these factors are related to decision making in a community population. In recent years, work in the field of reinforcement learning has identified two main pathways that drive decision making: model-based and model-free learning. Model-free learning updates action values retrospectively, after a reward is received. In contrast, model-based learning updates action values prospectively, by weighing contextual factors, the overall structure of the situation, and reward received. The current study utilizes a two-stage decision-making task to assess the relative contribution of model-free versus model-based learning in relation to measures that assess motivation, pleasure, and cognition in a community sample (n = 127). Generalized linear mixed-effect models showed that individuals high in motivation and pleasure had significantly greater reliance on model-free decision making (p = 0.0267). In contrast, individuals with better working memory, as measured by a running span task, had significantly greater reliance on model-based learning (p = 0.0003). These findings provide evidence that individual differences in motivation and cognition are associated with reliance on particular learning pathways. It has been suggested that lower levels of motivation, pleasure, and cognition in various forms of psychopathology (e.g., depression) can impair decision making. Our results show these relationships transcend clinical contexts. Specifically, these findings suggest that individuals who experience low motivation and pleasure may be less sensitive to immediate rewards, and that working memory capacity is highly relevant to model-based learning.
决策受动机、愉悦感和认知技能等因素驱动。本研究评估了这些因素与社区人群决策的关系。近年来，强化学习领域的研究确定了驱动决策的两条主要途径：基于模型的学习和无模型学习。无模型学习在获得奖励后回顾性地更新行动价值。相比之下，基于模型的学习通过权衡情境因素、整体情境结构和获得的奖励来前瞻性地更新行动价值。本研究采用两阶段决策任务，以评估无模型学习与基于模型学习在社区样本（n = 127）中相对于评估动机、愉悦感和认知的指标的相对贡献。广义线性混合效应模型显示，动机和愉悦感较高的个体对无模型决策的依赖程度显著更高（p = 0.0267）。相反，通过连续广度任务测量，工作记忆较好的个体对基于模型学习的依赖程度显著更高（p = 0.0003）。这些发现提供了证据，表明动机和认知方面的个体差异与对特定学习途径的依赖有关。有人提出，各种形式的精神病理学（如抑郁症）中较低水平的动机、愉悦感和认知会损害决策。我们的结果表明这些关系超越了临床背景。具体而言，这些发现表明，经历低动机和低愉悦感的个体可能对即时奖励不太敏感，并且工作记忆容量与基于模型的学习高度相关。
