关键词: 抑郁症, 快感缺失, 动机损害, 身体努力, 认知努力, 计算模型

Abstract: Motivation deficits are a common symptom of depression, often leading to abnormal effort-related reward processing in individuals with depression. Understanding the cognitive neural mechanisms of the willingness to expend cognitive or physical effort to obtain rewards is essential for helping patients recover their social functioning. However, research in this area is currently hindered due to a lack of appropriate methods for determining the roles of cognitive and physical effort expenditure in motivation deficits. In recent years, computational modeling, such as reinforcement learning models, drift diffusion models, cost-benefit optimization models and models with utility discounting functions, has been applied to this field and proved to be a highly promising method of exploring the potential mechanisms underlying effort-related behaviors in depression. Unlike traditional methods, computational modeling allows for a trial-by-trial analysis of behavioral data, providing a more precise and objective assessment of motivational variables. Studies using computational modeling approaches found that, compared with healthy controls, individuals with depression exhibited lower willingness to expend physical and cognitive effort, and abnormal effort-related behaviors in depressed patients were associated with altered neural activity of the prefrontal cortex, striatum, cingulate gyrus and insula. Combining models of utility discounting functions and functional Magnetic Resonance Imaging (fMRI) method, studies in healthy populations showed that model results of encoding of subjective value for both physical and cognitive effort costs were linked to neural activity in brain regions such as the ventromedial prefrontal cortex, anterior cingulate gyrus, ventral striatum and dorsolateral prefrontal cortex. Studies utilizing reinforcement learning models and fMRI method in healthy populations found that specific brain regions, such as the ventromedial prefrontal cortex and anterior insula, were involved in encoding reward and effort prediction errors during physical effort tasks, and the fronto-parietal network was involved in encoding effort prediction errors during cognitive effort tasks. Studies using computational models (i.e., models of utility discounting functions and drift diffusion models) and non-invasive brain stimulation techniques in healthy populations found that neural activity in the dorsolateral prefrontal cortex was related to increased cognitive effort sensitivity, and neural activity in the dorsomedial prefrontal cortex was related to increased physical effort sensitivity. And a recent study combining models of utility discounting functions, transcranial magnetic stimulation (TMS) and electroencephalography (EEG) techniques to explore the role of the left dorsolateral prefrontal cortex in depressed patients showed that, depressed patients with enhanced left dorsolateral prefrontal cortex activation exhibited reduced sensitivity to effort, along with increased amplitudes of the P300 wave to effort-related information and increased amplitudes of the contingent negative variation and stimulus-preceding negativity to reward outcomes. These findings suggest that motivation deficits in patients with depression occur in both physical and cognitive effort domains and highlight the application potential of combining computational modeling approaches with cognitive neuroscience techniques to uncover the shared and divergent mechanisms underlying abnormal physical and cognitive effort-based motivation in depression. Future research combining computational models and cognitive neuroscience techniques could help further dissect the complex cognitive processes and neural foundations of effort-related behaviors, hereby not only providing a more nuanced understanding of the underlying causes of motivation deficits but also helping personalize treatment in individuals with depression.

Key words: depression, anhedonia, motivation deficits, physical effort, cognitive effort, computational modeling