【作者】 于权浩； 高峰； 曹妍媛； 王向； 樊洁； 朱熊兆；

【Author】 Quanhao Yu;Feng Gao;Yanyuan Cao;Xiang Wang;Jie Fan;Xiongzhao Zhu;Medical Psychological Center, the Second Xiangya Hospital, Central South University;Medical Psychological Institute of Central South University;National Clinical Research Center for Mental Disorders;

【机构】 中南大学湘雅二医院医学心理研究中心； 中南大学医学心理研究所； 国家精神心理疾病临床医学研究中心；

【摘要】 背景:基于行为学习双系统理论的框架下,强迫行为被认为是一种因刺激诱发的且对目标变化不敏感的过度习惯性行为。强迫行为的产生源自行为学习双系统的失衡。过去关于强迫症行为学习双系统失衡的研究较多关注行为反应阶段,在建立刺激—反应—结果联结的行为学习中强迫症患者表现出的特征尚不清楚。此外,过去关于强迫症行为学习双系统失衡的研究大多数为行为学研究,其失衡的神经机制尚未明确。方法:本研究纳入49名强迫症患者（Obsessive-Compulsive Disorder, OCD）和38名健康对照（Healthy Controls,HC）。被试首先完成强迫症状严重程度评估,然后完成SOAT的行为学习以及行为反应部分。行为反应阶段同时进行磁共振扫描。通过强化学习模型对行为学习阶段进行计算认知建模,采用分层贝叶斯的建模方法比较两组被试在学习阶段的学习率、强化敏感性以及坚持参数等指标,分析上述指标对行为反应阶段的影响。比较两组被试在行为反应阶段中负责习惯性行为和目标导向行为的脑区激活差异。结果:（1）在SOAT任务的行为学习阶段,OCD组的正确率和反应时与HC组无显著差异(正确率:F_（1,82）=0.104, p> 0.05;反应时:F_（1,82）=0.185, p> 0.05);基于强化学习模型结果发现,OCD组的平均学习率、坚持参数,以及3种刺激—反应—结果的联结学习率以及显著低于HC组,强化敏感性参数显著高于HC组。（2）强迫症患者组平均学习率和坚持参数与Y-BOCS强迫行为得分表现出显著负相关（r=-0.311,p <0.05; r=-0.315,p <0.05）,强化敏感性参数与强迫症状相关不显著。（3）在行为反应阶段,OCD组的贬值敏感性指数以及对有价值结果反应率显著低于HC组（p <0.001;p <0.05）,对贬值结果反应率显著高于HC组（p <0.001）。（4）OCD组平均学习率与对有价值结果反应率呈显著相关关系（r=0.373,p <0.05）,坚持参数与对贬值结果反应率呈显著负相关关系（r=-0.342,p <0.05）;OCD组与HC组中均发现强化敏感性参数与对贬值结果反应率（OCD:r=-0.382, p <0.05; HC:r=-0.403, p <0.05）、贬值敏感性指数（OCD:r=0.432, p <0.01;HC:r=0.397, p <0.01）呈显著相关关系。（5）任务态磁共振结果发现,OCD组在目标导向性行为与习惯性行为相竞争时以及执行习惯性行为时,均表现出眶额叶以及脑岛的激活显著低于HC组（团块水平p <0.05,FWE校正）,并且在执行习惯性行为时的条件下右侧眶额下回的激活与强迫行为严重程度显著负相关（r=-0.394,p <0.05）;在对有价值刺激进行反应时,OCD组在中央前后回的区域激活显著高于HC组（团块水平p <0.05,FWE校正）。结论:（1）强迫症患者目标导向性行为和习惯性行为的学习能力受损,表现出无法根据预测误差及时更新行为联结强度,更容易怀疑已经建立的正确联结。强迫症患者双系统行为学习能力受损主要与强迫行为而非强迫思维的症状严重程度有关。（2）强迫症患者在行为反应阶段,目标导向性行为表现不佳,过度依赖习惯性行为,表现出行为学习双系统失衡,强迫症患者在行为学习阶段的特征可以预测其行为反应阶段的行为模式。（3）强迫症患者过度依赖习惯性行为可能是因其眶额叶、脑岛区域的激活减弱从而导致对结果价值变化不敏感,提示强迫症患者可能由于编码预期结果价值的能力受损从而导致过度依赖刺激—反应联结。更多还原

【Abstract】 Background: Based on the dual-system theory, compulsive behavior is viewed as overly habitual behavior that is stimulus-driven and insensitive to changes in goals. The production of compulsive behavior arises from an imbalance between the two behavior systems. Previous research on the imbalance of the dual-systems in obsessive-compulsive disorder(OCD) has mainly focused on the behavioral response stage, and the characteristics exhibited by OCD patients in establishing the stimulus-response-reinforcement association are not yet clear.Additionally, most previous studies on the imbalance of the dual systems of behavior learning in OCD have been behavioral studies, and the neural mechanisms underlying the dual-system learning imbalance in OCD patients have not been elucidated. Methods: This study included 49 patients with obsessive-compulsive disorder and 38healthy controls. Participants completed assessments of the severity of their compulsive symptoms, anxiety and depression, and demographic data collection. All participants completed the behavior learning phase of the Slip of Action Test(SOAT) in the laboratory, ensuring that each participant memorized the relationships between stimuli,behaviors, and outcomes. After that, they completed the outcome devaluation phase of SOAT in functional magnetic resonance imaging(fMRI). Reinforcement learning models were used for cognitive computational modeling of the learning phase, and hierarchical Bayesian modeling was used to compare the two groups of participants’ learning rates, reinforcement sensitivity, persistence parameters, and their effects on behavioral response in the outcome devaluation phase. The impact of these parameters on behavioral responses during the reward devaluation stage were analyzed. Compare the differences in brain region activation between two groups of subjects responsible for habitual behavior and goal-directed behavior during the behavioral response phase.Results:(1) During the learning phase of SOAT, OCD patients showed no significant differences in accuracy and reaction time compared to healthy controls(ACC: F(1,82) = 0.104, p > 0.05; RT: F(1,82) = 0.185, p > 0.05). Based on the results of the computational cognitive model, OCD patients’ three types of associative learning rates, average learning rate, and persistence parameter were significantly lower than those of healthy controls, while their reinforcement sensitivity parameter was significantly higher than that of healthy controls.(2) Average learning rate and persistence parameters were significantly negatively correlated with the Y-BOCS compulsive behavior score(r=-0.311, p < 0.05; r =-0.315, p < 0.05), while the reinforcement sensitivity parameter was not significantly correlated with OCD symptoms.(3) During the behavioral response phase, the OCD group had significantly lower devaluation sensitivity and reaction rate to valuable results than the HC group(p < 0.001; p < 0.05), and a significantly higher reaction rate to devalued results than the HC group(p < 0.001).(4) The average learning rate in the OCD group was significantly positively correlated with the reaction rate to valuable results(r = 0.373, p <0.05), and the persistence parameter was significantly negatively correlated with the reaction rate to devalued results(r =-0.342, p < 0.05). Both the OCD and HC groups showed a significant correlation between the reinforcement sensitivity parameter and the reaction rate to devalued results(OCD: r =-0.382, p < 0.05; HC: r =-0.403, p < 0.05), and the devaluation sensitivity index(OCD: r = 0.432, p < 0.01; HC: r = 0.397, p < 0.01).(5)The results of the task-state functional magnetic resonance imaging(fMRI) showed that in the competition between goal-directed behavior and habitual behavior, and during the execution of habitual behavior, patients with obsessive-compulsive disorder(OCD) exhibited significantly reduced activation in the orbitofrontal cortex and the insula(cluster level p < 0.05, family-wise error correction). Moreover, during the execution of habitual behavior,the activation of the right orbitofrontal cortex was significantly negatively correlated with the severity of compulsive behavior(r =-0.394, p < 0.05). When responding to valuable stimuli, patients with OCD showed significantly enhanced activation in the anterior and posterior cingulate regions(cluster level p < 0.05, family-wise error correction).Conclusion:(1) The learning abilities of both goal-directed and habitual behaviors are impaired in patients with OCD, as they exhibit an inability to update the strength of behavior associations according to prediction errors and are more likely to doubt the correctness of established associations. The impairment of the dual-system learning abilities in patients with OCD is mainly related to the severity of compulsion rather than obsession.(2) In the behavior response stage, patients with OCD perform poorly in goal-directed behavior and overly rely on habitual behavior, leading to an imbalance in dual-system behavior. The characteristics of patients’ behavior during the learning stage can predict their behavior patterns during the response stage.(3) The excessive reliance of patients with OCD on habitual behavior may be due to the reduced activation in the orbitofrontal cortex and insula, which leads to insensitivity to the change of the value of results. This suggests that patients with OCD may have impaired ability to encode the expected value of results, resulting in excessive dependence on stimulus-response associations.更多还原