Graduate School of Medical Science and Engineering, Korea Advanced Institute of Science and Technology, Daejeon, Republic of Korea.
Department of Psychiatry, Uijeongbu St. Mary's Hospital, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea.
Depress Anxiety. 2020 Jul;37(7):609-619. doi: 10.1002/da.23007. Epub 2020 Mar 18.
Despite the well-known association between anxiety and risk-avoidant decision making, it is unclear how pathological anxiety biases risk learning. We propose a Bayesian inference model with bias parameters of prior, learning, and perception during risk learning in individuals with pathological anxiety.
Patients with panic disorder (PD, n = 40) and healthy control subjects (n = 84) completed the balloon analog risk task (BART). By fitting our computational model of three bias parameters (prior belief, learning rate, and perceptual bias) to the participants' behavior, we estimated the degree of bias in risk learning and its relationship with anxiety symptoms.
Relative to the healthy control subjects, the pathologically anxious participants exhibited a biased underestimation of perceptual evidence rather than differences in priors and learning rates. The degree of perceptual bias was correlated with the anxiety and depression symptom severity in the patients with PD. Furthermore, our proposed model was the winning model for BART data in an external data set from different patient groups.
Our results showed that individuals with pathological anxiety demonstrate perceptual bias in evidence accumulation, which may explain why patients with anxiety overestimate risk in their daily lives. This clarification highlights the importance of interventions focusing on perceptual bias, such as enhancing the clarity of favorable outcome probabilities.
尽管焦虑与风险规避决策之间存在众所周知的关联，但病理性焦虑如何影响风险学习尚不清楚。我们提出了一种贝叶斯推断模型，该模型具有病理性焦虑个体在风险学习期间的先验、学习和感知偏置参数。
惊恐障碍患者（PD，n=40）和健康对照受试者（n=84）完成了气球模拟风险任务（BART）。通过将我们的三个偏置参数（先验信念、学习率和感知偏置）的计算模型拟合到参与者的行为中，我们估计了风险学习中的偏置程度及其与焦虑症状的关系。
与健康对照组相比，病理性焦虑组表现出对感知证据的低估偏见，而不是先验和学习率的差异。感知偏差的程度与 PD 患者的焦虑和抑郁症状严重程度相关。此外，我们提出的模型在来自不同患者群体的外部数据集的 BART 数据中是获胜模型。
我们的研究结果表明，病理性焦虑个体在证据积累中表现出感知偏见，这可能解释了为什么焦虑症患者在日常生活中高估风险。这种澄清强调了关注感知偏见的干预措施的重要性，例如提高有利结果概率的清晰度。
