EAGER: A Mathematical Model of Privacy Decisions: A Behavioral Economic Perspective

在做出有关信息隐私的决策时，人们并不总是按照自己的最大利益理性行事。因此，重要的是要理解为什么人们表达了对隐私的担忧，但往往采取了与他们声明的意图相反的行为。本研究从行为经济学的角度研究个人的隐私决策：1)调查人类大脑中的两个思维系统--情感系统(系统1)和认知系统(系统2)在隐私决策过程中是如何运作的；2)开发情感认知算法，以数学方式描述系统1和系统2在隐私决策中的操作；以及3)测试和评估这些算法的准确性。这是形成一种新的隐私数学理论的第一步，该理论可以描述人们实际如何做出隐私决定，以及他们被期望如何做出这样的决定。利用离散数学、理论计算机科学和行为经济学的技术，并适应有限理性的概念，私人投资机构将开发情感认知算法，以模拟人类在隐私决策中的经验效用(即，他们在披露隐私时感知的风险和收益)。这与现有的隐私决策数学模型根本不同，现有的隐私决策数学模型假设人类具有稳定的偏好，总是选择预期效用最高的选项(即具有最大隐私的选项)。PI使用行为博弈论中的技术来严格测试和评估他们算法的准确性。PI应用行为经济学、数学心理学和之前关于信息隐私的研究成果，将系统1和2的运行转换为数学模型。这项研究旨在弥合三个研究流派，即信息隐私、理论计算机科学以及行为和实验经济学。这项研究的影响具有转化为应用于真实在线环境和在决策支持工具中实施的潜力，例如推荐系统。

When making decisions about information privacy, people do not always act rationally according to their best interests. It is thus important to understand why people express concerns about privacy, but often act contrary to their stated intentions. This research investigates individuals' privacy decisions, from a behavioral economic perspective, by: 1) investigating how two systems of thinking in human minds, affective (system 1) and cognitive (system 2), operate during privacy decisions; 2) developing affective-cognitive algorithms to mathematically describe the operation of systems 1 and 2 in privacy decisions; and 3) testing and evaluating the accuracy of these algorithms. This is the first step in forming a new mathematical theory of privacy that can describe how people actually make privacy decisions versus how they are expected to make such decisions. Using techniques from discrete mathematics, theoretical computer science and behavioral economics, and adapting the concept of bounded rationality, the PIs will develop affective-cognitive algorithms to model human experienced-utility in privacy decisions (i.e., risks and benefits they perceive in disclosing privacy). This is fundamentally different from the existing mathematical models of privacy decisions that assume humans have stable preferences and always choose the option with the highest expected utility (i.e., the option with the maximum privacy). The PIs use techniques from behavioral game theory to rigorously test and evaluate the accuracy of their algorithms. The PIs apply findings of behavioral economics, mathematical psychology, and previous research on information privacy to translate operation of systems 1 and 2 into mathematical models. This research is intended to bridge three streams of research, namely information privacy, theoretical computer science, and behavioral and experimental economics. The impacts of this research have translational potential for application to real online environments and implementation in decision support tools, such as recommender systems.