基于认知过程视角的跨期决策实验、模型与应用研究

Intertemporal decision making is ubiquitous in our everyday life. Such decisions not only affect one’s health, wealth, and happiness, but also determine the economic prosperity of nations. However, most of the existing studies on intertemporal decision making are built on the outcome-based black-box models, which only care about the outcome of the choice, but not the cognitive process which leads to the decisions. Recently, economists have realized that the cognitive process underlying decisions helps us to better understand human behavior. This project will study intertemporal decision making systematically from the perspective of cognitive processes. First, using behavioral and eye-tracking experiments, we will show that the cognitive process of intertemporal decision making complies with both evidence accumulation models and dual-process theories. That is, when making intertemporal decisions, the evidence accumulation process, the intuitive process, and the deliberative process act on different levels and therefore coexist in the brain. Second, based on the experimental results, we will develop process-based models for intertemporal decision making, one is the dual-process evidence accumulation model, which is based on the accurate description of the cognitive process; the other is the intertemporal choice model based on process data and machine learning. Third, we will investigate the potential applications of the study above using a nudge experiment and business data analysis. This project will further reveal the black box of the brain, and further enrich and develop intertemporal decision theories, which has great theoretical and practical significance.

跨期决策在现实中广泛存在，它不仅影响个人的健康、财富与幸福，也决定着一个国家经济繁荣的程度。然而，现有的跨期决策研究大多是建立在基于决策结果的黑箱模型基础之上的，只关注决策者的最终选择，而忽略了决策背后的认知过程。近年来，学者们逐渐意识到认知过程可以帮助我们更好的理解人类行为。本项目将从认知过程的视角对跨期决策进行系统深入的研究。首先，通过行为与眼动实验，验证跨期决策的认知过程既遵循证据累积模型，也遵循双系统加工理论，揭示跨期决策背后的证据累积与双系统加工共驱的认知机制；在此基础上，构建基于认知过程的跨期决策模型，一是建立在对认知过程描述基础上的双加工-证据累积决策模型，二是基于认知过程数据与机器学习的决策模型；最后，通过助推实验和商业数据分析，研究上述认知过程与决策模型研究的潜在政策应用和商业价值。本项目将进一步揭示大脑决策黑箱，丰富和发展跨期决策理论体系，具有重要的理论和现实意义。

与建立在效用理论与基于决策结果的显示性偏好理论基础上的跨期决策理论（例如指数贴现模型、双曲贴现模型）不同，本项目聚焦跨期决策背后的认知加工过程，从认知加工过程视角来研究跨期选择行为，推进经济学与自然科学的相互渗透与融合，拓展经济学研究的范畴和边界。首先，通过认知过程追踪实验，采集行为数据、决策时间数据、鼠标轨迹数据和眼动数据，解构出跨期决策背后认知加工过程中的关键成分，倾向偏差和属性潜伏。在此基础上，识别出跨期决策背后“证据累积加工与双系统认知加工共驱”的认知加工机制，揭示了时间偏好的形成过程，给出了跨期选择的微观认知基础。其次，构建了基于上述认知加工机制的跨期决策认知计算模型，即“带有倾向偏差的漂移扩散模型”和“起始时间漂移扩散模型”，通过模型拟合与交叉验证，证明了与传统模型相比，基于认知加工机制的跨期决策模型具有更好的解释和预测能力，丰富和发展了跨期选择模型与理论。然后，在揭示认知加工机制的基础上，设计了基于认知加工机制的助推政策，通过干预个体决策背后的认知加工过程来影响其跨期决策行为，并通过实验验证了助推政策的有效性。此外，将上述研究由跨期决策扩展至社会决策，从认知加工视角研究了社会决策，解决了社会决策背后“单系统”与“双系统”之间的重大争议。最后，针对数字经济背景下的现实问题，从认知加工视角开展了应用研究。相关研究成果发表在《Nature Communications》、《Experimental Economics》、《Journal of Business Research》、《管理工程学报》等期刊，并得到国内外主流媒体的广泛关注和报道，还有多篇论文正在《Management Science》、《Manufacturing & Service Operations Management》等期刊修改中，并获得包括浙江省哲学社会科学优秀成果二等奖在内的多个奖项。