风险决策的预测者-行动者差异：认知的视角

Predictor-actor difference in risky decision-making have gradually become one of the hot topics in the field of decision-making. It refers to the discrepancy between the predictor’s prediction of risky decision-making of the actor and the actual risky decision-making of the actor. Such difference often lead to errors in negotiations, investment, competition and even policy making. However, there are many contradictions or inconsistencies in the prior studies on decision-making tasks, findings, and theoretical explanations, and there is a lack of exploration of the deep rules and cognitive mechanisms behind such difference. Based on construal level theory, this project uses behavioral experiments, Mouselab technology, and model fitting, to exploring the difference characteristics of risky decision-making between predictor and actor (difference of value function and difference of probability weighting function), cognitive mechanism underlying such differences (difference of information focus and difference of information processing), and strategies to reduce such differences to improve forecast accuracy (changing information focus and changing information processing). It is hoped theoretically to construct predictor-actor value function model and weight function model, provide a cognitive mechanism, and practically to help countries, organizations and individuals to understand the predictor-actor differences in negotiation, investment, competition and policy making, and grasp the ways to improve the forecast accuracy.

风险决策的预测者-行动者差异逐渐成为决策领域研究的热点之一，它是指预测者对行动者风险决策的预测与行动者实际风险决策之间存在的差异。这种差异常常造成谈判、投资、竞争甚至政策制定的失误。然而，目前对该差异的研究在采用的决策任务、得到的研究发现及理论解释上均存在诸多矛盾或不一致，并且缺乏对差异背后深层规律及认知机制的探索。本项目基于建构水平理论，采用行为实验、Mouselab技术、模型参数拟合等手段，探究预测者和行动者风险决策的差异特征（价值函数差异和概率权重函数差异）、差异背后的认知机制（信息关注差异和信息加工差异）以及消减差异优化预测力的策略（改变信息关注和改变信息加工）。希望从理论上构建预测者-行动者差异的价值函数模型和概率权重函数模型，并提供认知视角的解释；从实践上帮助国家、组织和个体在谈判、投资、竞争及政策制定中了解预测者-行动者差异规律、掌握消减差异提高预测力的策略。

风险决策的预测者-行为者差异反映了预测者对行动者风险决策的预测与行动者实际风险决策之间的偏差。它常常造成谈判、投资、竞争、合作甚至政策制定的失误。本项目基于认知的视角，运用行为实验、Mouselab技术、模型参数拟合等手段，探究了预测者-行动者风险决策的差异特征、认知机制和优化策略。.本项目：1）比较了预测者和行动者在价值感知（价值函数）和风险感知（权重函数）的差异及其决策偏差模式，发现预测者比行动者对价值变化更敏感、损失厌恶感更小，这导致预测偏差呈现三重模式：获益时预测者比行动者更冒险；损失时行动者比预测者更冒险；混合损益时，预测者比行动者更冒险。行动者比预测者更高估小概率事件、更低估大概率事件，这导致预测偏差呈现四重模式：小概率获益时和大概率损失时行动者比预测者更冒险；大概率获益时和小概率损失时预测者比行动者更冒险。2）结合Mouselab技术，测量了预测者和行动者在决策过程中信息关注和信息加工策略的差异。发现预测者关注“渴望性”信息（如可能的“大收益”信息），行动者关注“可行性”选项（如“确定”的获益信息）；预测者更倾向于采用“果率同肥”的补偿性策略（即关注金额比较又关注概率比较），行动者更倾向于采用“果肥率瘦”的非补偿性策略（关注结果比较、忽视概率比较），这些认知差异导致两者的价值和风险感知之差。3）凸显“可行性”信息比凸显“渴望性”信息提升了预测准确性；使用“果肥率瘦”的非补偿性策略比“果率同肥”补偿性策略改善了预测力。4）除认知差异外，预测偏差还与预测者-行动者的情绪体验差异有关。5）预测偏差可拓展至社会互动领域，如为他人决策、合作行为，发现预测偏差会泛化到为他人决策和合作行为之中，进而影响个体的社会互动行为。.本项目构建了预测者-行动者差异的价值、概率权重函数模型，揭示了两者差异的三重和四重模式，并提供认知机制的解释，为国家、组织和个体在社会交互中优化预测力提供了理论依据和优化策略。