Cognition, Learning and Memory: A Parallel Distributed Processing Approach

认知、学习和记忆：并行分布式处理方法

• 批准号: 8812048
• 负责人: James McClelland
• 金额: $ 13.31万
• 依托单位: Carnegie-Mellon University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1988
• 资助国家: 美国
• 起止时间: 1988-08-01 至 1992-01-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=8812048&HistoricalAwards=false
• 关键词: Cognition; Learning; Memory; Parallel; Distributed

This project will consider two topics that arise in the development of a theoretical framework, known as the parallel distributed processing (PDP) framework, for understanding learning, memory, and cognitive processes. In this framework, cognitive processes emerge from the simultaneous activity of a large number of simple processing units; the knowledge that governs processing is represented by the numerical strengths of connections among the units. The approach will be applied to recall of information from memory and to language acquisition. From a PDP perspective, recall becomes a process of constructing the representation of a to-be-remembered item from the contributions of a number of prior experiences. This view predicts that information acquired at different times may be combined in the same act of recall, producing "blend" errors in memory. The first part of the project will verify that such errors occur and will examine the circumstances under which they occur. The experiments will examine the effects of similarity of memory traces, will assess the tendency for blend errors to occur in recognition as well as recall, and will determine whether blend errors occur more frequently when the different experiences to be combined complement, rather than conflict, with each other. These experiments will help clarify when and how the mechanisms of memory can give rise to errors that could invalidate eyewitness testimony. From a PDP perspective, learning language is not a matter of learning a system of linguistic rules per se; rather, it is a matter of adjusting connection strengths among units. The second part of the project will examine whether this view is consistent with children's acquisition of one aspect of language, namely the use of inflected forms of verbs (e.g., the past tense). This part of the project will attempt to develop a revision of an earlier model of acquisition by connection strength adjustment and to use the model to account for errors that children make during acquisition. This part of the project should help clarify the nature of the mechanisms that underlie the acquisition of language.

这个项目将考虑两个主题，出现在 发展的理论框架，称为平行 分布式处理（PDP）框架，用于理解 学习记忆和认知过程 在这一框架内， 认知过程来自于一个人的同时活动， 大量的简单处理单元;知识， 支配处理由以下数值强度表示： 单位之间的联系。 该方法将应用于 从记忆到语言习得的信息回忆。 从PDP的角度来看，回忆成为一个构建过程， 要记住的项目的表示 一些先前经验的贡献。 这种观点 预测在不同时间获得的信息可能 在同一个回忆行为中结合起来，产生“混合”错误， 记忆 该项目的第一部分将核实， 错误发生，并将审查的情况下，他们 发生. 这些实验将研究相似性的影响， 记忆痕迹，将评估混合错误发生的趋势 在承认以及召回，并将确定是否 当不同的体验 是相辅相成的，而不是相互冲突的。 这些实验将有助于阐明何时以及如何 可能会导致错误， 目击者的证词 从PDP的角度来看，学习语言不是一个问题， 学习语言规则本身的系统;相反，它是一个 调整单元之间的连接强度的问题。 第二 该项目的一部分将检查这种观点是否一致 与儿童习得语言的一个方面，即 动词的屈折形式的使用（例如，过去式）。 这 该项目的一部分将试图开发一个修订版的 通过连接强度调整的早期获取模型 并利用这个模型来解释孩子们犯的错误 在收购过程中。 项目的这一部分应该有助于澄清 获得的机制的性质 语言