Computational and Behavioral Approaches to Cognition

认知的计算和行为方法

• 批准号: 7101050
• 负责人: LYNNE Marie REDER
• 金额: $ 19.06万
• 依托单位: CARNEGIE-MELLON UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 1998
• 资助国家: 美国
• 起止时间: 1998-07-01 至 2008-06-30
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7101050
• 关键词: Computational; Behavioral; Approaches; Cognition

DESCRIPTION (provided by applicant): The purpose of the proposed predoctoral and postdoctoral programs is to train the next generation of cognitive psychologists both to develop formal computational models and to test and refine these models, by rigorously comparing the simulation data to carefully collected empirical data. The field is ready to benefit from formal, computational models of cognitive processes. The tools are being developed that enable this formalism, and the end product will not only deepen the empirical and conceptual basis of cognitive psychology, but will also provide stronger links between psychology, neuroscience, and the treatment of problems in mental health. Carnegie Mellon is especially suited to provide this next generation of cognitive scientists with these tools. There is a long tradition at CMU to strive for complete cognitive models to account for a wide range of phenomena using a small common set of theoretical assumptions. The proposed program would be our first focused on training modeling skills. One of the distinctive features of psychological research at CMU is the dual concern for experimental methodology and theoretical models, not just each in isolation. We have promoted the development of both production system (symbolic), connectionist (sub-symbolic) and hybrid models of the human information processing architecture as well as many specific models of performance in particular tasks. In all cases, the researchers have tested and refined their models based on behavioral and physiological data collected here at CMU and elsewhere. Methodologies that have been developed and refined within by our department include: the automatic coding of verbal protocols, the analysis of eye fixations while thinking and problem solving, and functional MRI measurements of higher cognitive processes. Some of these models address the data at the grain size of individual responses, with few subject-specific parameters. The program's goal is to develop skilled researchers who are both competent and comfortable combining the approaches of behavioral research with development of computationally implemented models of cognitive performance. Participation in research, both empirical and modeling, is a fundamental component of helping students achieve this goal. Formal courses and seminars play an important role as well. We will formally instruct and demonstrate the skills of comparing the data derived from a simulation to the human data collected from behavioral research, providing students with the skills to evaluate the quality of the fit and the sensitivity to know when and how to revise one's model based on these comparisons. In addition we will ensure that trainees are conversant in multiple computational approaches and recognize the strengths and weaknesses of each.

描述(由申请人提供)：拟议的博士后和博士后课程的目的是培训下一代认知心理学家，通过严格比较模拟数据和仔细收集的经验数据，开发正式的计算模型，并测试和改进这些模型。该领域已经准备好从认知过程的正式计算模型中受益。实现这种形式主义的工具正在开发中，最终产品不仅将深化认知心理学的经验和概念基础，还将在心理学、神经科学和心理健康问题的治疗之间提供更强有力的联系。卡内基梅隆大学特别适合为下一代认知科学家提供这些工具。在CMU有一个悠久的传统，即努力建立完整的认知模型，以使用一小部分共同的理论假设来解释广泛的现象。拟议的项目将是我们第一个专注于培训建模技能的项目。CMU心理学研究的一个显著特点是对实验方法和理论模型的双重关注，而不是各自孤立地关注。我们促进了人类信息处理架构的生产系统(符号)、连接主义(次符号)和混合模式的发展，以及在特定任务中的许多具体绩效模式。在所有情况下，研究人员都基于在CMU和其他地方收集的行为和生理数据来测试和改进他们的模型。我们部门开发和完善的方法包括：言语协议的自动编码，思考和解决问题时眼睛凝视的分析，以及高级认知过程的功能磁共振测量。其中一些模型以个体回答的粒度处理数据，几乎没有特定于受试者的参数。该计划的目标是培养既称职又熟练的熟练研究人员，将行为研究的方法与开发通过计算实现的认知表现模型相结合。参与研究，无论是实证的还是建模的，都是帮助学生实现这一目标的基本组成部分。正规课程和研讨会也发挥着重要作用。我们将正式指导和演示将来自模拟的数据与从行为研究收集的人类数据进行比较的技能，为学生提供评估匹配质量的技能，以及了解何时以及如何基于这些比较修改自己的模型的敏感度。此外，我们将确保受训人员熟悉多种计算方法，并认识到每种方法的优点和缺点。