RI: SMALL: Prospective and retrospective mechanisms in complex planning by humans

RI：小：人类复杂规划中的前瞻性和回顾性机制

• 批准号: 2008331
• 负责人: Wei Ji Ma
• 金额: $ 49.5万
• 依托单位: New York University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-10-01 至 2023-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2008331&HistoricalAwards=false
• 关键词: RI; SMALL; Prospective; retrospective; mechanisms

Planning is an important part of everyday life: we plan how to get from A to B, how to develop our careers, how to write a text, and how to approach a sports match. In all these cases, a sequence of decisions has to be made, with many possibilities at each step. For these reasons, planning taxes people’s cognitive abilities. In recent years, the field of artificial intelligence has made impressive progress on hard planning problems, such as playing chess and go. By contrast, the field of cognitive science has lagged behind in understanding how people solve difficult planning problems. To plan well, a person has to imagine situations that might happen in the future, even though they might never have experienced those situations before. At the same time, people also use their memory of similar situations to decide what to do. The goal of this project is to understand how these two processes -- imagining the future and recalling memories -- contribute to people’s planning decisions. This project will also support associated educational activities for high school and general audiences.This project will break down the mechanisms of human decision-making in complex planning tasks into two components: mental simulation of possible futures (prospective decision-making) and learning directly from past experiences (retrospective decision-making). Through a mobile app, the investigators collected data from people performing an engaging, competitive task that requires planning. The data set consists of hundreds of millions of decisions, allowing for statistically highly robust conclusions. The investigators will analyze these data using statistical methods and computational modeling. The prospective component of the model will be based on heuristic tree search algorithms. The retrospective component will be based on repeating successful actions in previously experienced states. The researchers will compare to alternative models and aim to account for human decisions and reaction times. It is expected that this project will lead to a better fundamental understanding of the cognitive processes involved in complex planning.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

计划是日常生活的重要组成部分：我们计划如何从A到B，如何发展我们的职业生涯，如何写一篇文章，以及如何接近一场体育比赛。在所有这些情况下，必须做出一系列决定，每一步都有许多可能性。由于这些原因，计划税人的认知能力。近年来，人工智能领域在硬规划问题上取得了令人印象深刻的进展，例如下棋和围棋。相比之下，认知科学领域在理解人们如何解决困难的规划问题方面落后了。为了做好计划，一个人必须想象未来可能发生的情况，即使他们以前可能从未经历过这些情况。与此同时，人们也会利用他们对类似情况的记忆来决定该怎么做。这个项目的目标是了解这两个过程-想象未来和回忆记忆-如何有助于人们的规划决策。该项目还将支持面向高中和普通受众的相关教育活动，将人类在复杂规划任务中的决策机制分解为两个部分：对可能未来的心理模拟（前瞻性决策）和直接从过去经验中学习（回顾性决策）。通过一个移动的应用程序，研究人员收集了人们执行一项需要计划的吸引人的竞争性任务的数据。该数据集由数亿个决策组成，可以得出统计上高度可靠的结论。研究人员将使用统计方法和计算建模来分析这些数据。该模型的预期组成部分将基于启发式树搜索算法。回顾部分将基于在以前经历的状态中重复成功的行动。研究人员将与其他模型进行比较，旨在解释人类的决策和反应时间。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Improving a model of human planning via large-scale data and deep neural networks

通过大规模数据和深度神经网络改进人类规划模型

• 发表时间: 2022
• 期刊: Proceedings of the 44th Annual Conference of the Cognitive Science Society
• 作者: Kuperwajs, I.;Schuett, H.H.;Ma, W.J.
• 通讯作者: Ma, W.J.

A joint analysis of dropout and learning functions in human decision-making with massive online data

海量在线数据联合分析人类决策中的dropout和学习函数

• 发表时间: 2022
• 期刊: Proceedings of the 44th Annual Conference of the Cognitive Science Society
• 作者: Kuperwajs, I.;Ma, W.J.
• 通讯作者: Ma, W.J.

Planning to plan: a Bayesian model for optimizing the depth of decision tree search

规划到规划：优化决策树搜索深度的贝叶斯模型

• 发表时间: 2021
• 期刊: Proceedings of the Annual Meeting of the Cognitive Science Society
• 作者: Kuperwajs, I.;Ma, W.J.
• 通讯作者: Ma, W.J.