Methodological Issues in Cognitive Neuropsychology

SES 99-10931 - Clark Gavanour（匹兹堡大学）“认知神经心理学的方法论问题”与博士后助理合作，首席研究员将把科学哲学和计算机科学的结果和框架应用于认知神经心理学的一组方法论问题。这项工作将导致（i）基于脑损伤受试者的行为证据，哪些心理加工模型是不可区分的算法表征;（ii）从个体和聚合的心理加工模型推断的可靠性表征。（以各种方式）来自脑损伤受试者的行为的数据;（iii）数学上可能的数据集和解释条件的表征，在这些条件下，损伤的神经网络不能解释想象的正常和脑损伤的行为;以及（iv）研究使用成像和诱发反应电位数据来推断认知处理结构特征的策略。

SES 99-10931 - Clark Glymour (University of Pittsburgh) "Methodological Issues in Cognitive Neuropsychology"In collaboration with a post-doctoral assistant, the principal investigator will apply results and frameworks from philosophy of science and computer science to a group of methodological problems in cognitive neuropsychology. This work will result in (i) algorithmic characterizations of which mental processing models are indistinguishable on the basis of behavioral evidence of brain damaged subjects; (ii) characterizations of the reliability of inference to mental processing models from individual and aggregated (in various ways) data from the behavior of brain damaged subjects; (iii) characterizations of mathematically possible data sets and conditions on explanation under which lesioned neural nets cannot explain imagined normal and brain damaged behavior; and (iv) studies of strategies for using imaging and evoked response potential data for inferences to features of cognitive processing architecture.