[论文解读] Manipulating decision making of typical agents

V. I. Yukalov, Didier Sornette|arXiv (Cornell University)|Sep 2, 2014

Decision-Making and Behavioral Economics参考文献 27被引用 4

一句话总结

本文将量子决策理论（QDT）扩展至社会决策情境，通过操控‘吸引力因子’——一种主观且潜意识的偏见，来解释在风险与不确定性下如何影响集体决策。通过调整收益权重或提供战略性信息，可改变吸引力因子的符号，从而在效用不变的情况下反转偏好，为预测和引导群体决策提供定量框架。

ABSTRACT

We investigate how the choice of decision makers can be varied under the presence of risk and uncertainty. Our analysis is based on the approach we have previously applied to individual decision makers, which we now generalize to the case of decision makers that are members of a society. The approach employs the mathematical techniques that are common in quantum theory, justifying our naming as Quantum Decision Theory. However, we do not assume that decision makers are quantum objects. The techniques of quantum theory are needed only for defining the prospect probabilities taking into account such hidden variables as behavioral biases and other subconscious feelings. The approach describes an agent's choice as a probabilistic event occurring with a probability that is the sum of a utility factor and of an attraction factor. The attraction factor embodies subjective and unconscious dimensions in the mind of the decision maker. We show that the typical aggregate amplitude of the attraction factor is $1/4$, and it can be either positive or negative depending on the relative attraction of the competing choices. The most efficient way of varying the decision makers choice is realized by influencing the attraction factor. This can be done in two ways. One method is to arrange in a special manner the payoff weights, which induces the required changes of the values of attraction factors. We show that a slight variation of the payoff weights can invert the sign of the attraction factors and reverse the decision preferences, even when the prospect utilities remain unchanged. The second method of influencing the decision makers choice is by providing information to decision makers. The methods of influencing decision making are illustrated by several experiments, whose outcomes are compared quantitatively with the predictions of our approach.

研究动机与目标

• 将量子决策理论（QDT）从个体决策推广至社会决策情境。
• 识别在效用值稳定的情况下，决策者偏好可被系统性影响的机制。
• 量化信息与收益结构对群体决策中行为偏见的影响。
• 提供一个数学模型，通过操控潜意识因素预测并实现偏好反转。
• 通过实证实验验证模型，显示高预测精度。

提出的方法

• 采用类量子数学技术，将决策概率建模为效用与吸引力因子之和。
• 引入‘吸引力因子’以表征决策中的潜意识偏见、情绪及认知扭曲。
• 通过‘典型个体’模型群体行为——即具有特征吸引力振幅1/4的社会平均决策者。
• 应用收益权重操控以诱导吸引力因子值的变化，可能反转决策偏好。
• 分析信息影响：正确信息降低吸引力因子，错误或误导性信息则增加之。
• 通过实证数据检验预测结果，显示与实验误差范围内的定量一致性。

实验结果

研究问题

• RQ1 在不改变前景客观效用的情况下，如何实现决策偏好的反转？
• RQ2 吸引力因子在不确定性下的群体决策中发挥何种作用？
• RQ3 信息质量与结构在多大程度上可操控集体决策结果？
• RQ4 收益权重的安排如何影响吸引力因子的符号与大小？
• RQ5 QDT能否以高精度定量预测社会情境下的现实决策行为？

主要发现

• 决策者间吸引力因子的典型总体振幅被量化为1/4，其符号取决于相对偏好强度。
• 对收益权重的微小调整可反转吸引力因子的符号，从而在效用不变的情况下实现偏好反转。
• 提供错误或误导性信息会增大吸引力因子的幅度，放大非理性偏见，引导决策朝向期望结果。
• 正确信息可降低吸引力因子，但无法完全消除，证明由于信息有限，决策者无法实现完全理性。
• QDT的理论预测与实证实验结果高度吻合，展现出强大的定量有效性。
• 该模型成功解释了经典决策理论中的悖论，并为个体与群体决策行为提供了统一框架。