Classifying Categorization Using State Trace Analysis and Hierarchical Bayesian Modeling

使用状态跟踪分析和分层贝叶斯建模进行分类

• 批准号: 1256959
• 负责人: Michael Kalish
• 金额: $ 52.19万
• 依托单位: University of Louisiana at Lafayette
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2013
• 资助国家: 美国
• 起止时间: 2013-03-15 至 2014-10-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1256959&HistoricalAwards=false
• 关键词: Classifying; Categorization; Using; State; Trace

Humans must categorize the world in order to understand it. Categories like healthy versus poisonous, valuable versus overpriced, or even apples versus oranges coordinate our thinking and guide our actions in the world. In line with their centrality in cognition, current theories of categorization and category learning are numerous and diverse. The goal of this research project is to gather together these diverse theories and subject them to direct comparison using new and innovative statistical theories, computer implementations, and experimental designs. This project aims to serve Socrate's goal of carving nature at its joints by finding out why different people, at different times, learn different kinds of categories in different ways. This project has many potential scientific and practical ramifications. In particular, the techniques to be developed are directly relevant to research in the psychology of memory, judgment, and decision-making. More broadly, knowing if different category types are reliably different in the way they are learned, or if people are reliably different in the way they learn categories, will allow development of programs to help people learn difficult categories (like cancerous vs benign radiographs) more quickly and accurately.

人类必须对世界进行分类才能理解它。健康与有毒、贵重与高价，甚至苹果与橙子等类别协调了我们的思维，并指导着我们在这个世界上的行动。与认知的中心性相一致，当前的范畴化理论和范畴学习理论也是多种多样的。这个研究项目的目标是收集这些不同的理论，并利用新的和创新的统计理论、计算机实现和实验设计对它们进行直接比较。该项目旨在通过找出为什么不同的人在不同的时间以不同的方式学习不同的类别来服务于Socrate在关节处雕刻自然的目标。这个项目有许多潜在的科学和实际影响。特别是，将要开发的技术与记忆、判断和决策的心理学研究直接相关。更广泛地说，了解不同类别的学习方式是否可靠地不同，或者人们学习类别的方式是否可靠地不同，将有助于开发程序，帮助人们更快、更准确地学习困难的类别(如癌变与良性放射照片)。