The highest level goal of the proposed research is an understanding of the processes and structures of human memory and the processes involved in simple decision making. Investigation is planned in five theoretically related domains. In the first domain, three models of choice reaction time (the diffusion model, the OU model, and the accumulator model) will be tested against data from simple perceptual decision tasks and from higher level cognitive tasks. Second, the models'abilities to elucidate the processes by which information is retrieved from long term memory will be examined.
The aim i s to see whether the reaction time models can be usefully combined with the new memory models that have been developed recently, and if so, provide simultaneous empirical tests of both kinds of model. Third, relatively little work has been done on testing models for situations in which there are more than two response choices, in part because multichoice models of reaction time are often mathematically intractable and they have many parameters. But with fast computers and large experiments, we will be able to constrain and test the models. Fourth, our work with reaction time models has naturally led to tests of the currently leading models for categorization, exemplar models and distance from criterion models. We plan to examine ways to distinguish between these classes of models and also add to them mechanisms for predicting decision response times. Fifth, we continue work on implicit memory models. Research on implicit memory has usually progressed without any specific models of the processes involved in the tasks used to assess implicit memory. By proposing such models, we hope to raise the debate about implicit memory to consideration of the mechanisms by which priming effects come about. We plan to develop models for the frequently used picture naming and stern completion tasks. Overall, the proposed research represents the interaction of two methodologies: the development of explicit models of processing and representation and the development of empirical tests and data bases for those models. An important theme is the use of new models to serve as competitors for well established views, with the aim of driving research in new directions. The proposed research is relevant to themes described in the NIMH """"""""Decade of the Brain"""""""" report to Congress. The broad class of models to be examined can be seen as neurally inspired, and these models have been applied to various pathologies (e.g., amnesia and dyslexia). Future applications of well-validated models could help discriminate such issues as whether memory deficits are due to encoding or retrieval problems or whether rapid automatic processes or slower more conscious ones are faulty. Reaction time models especially might lead to diagnostic techniques that are non-invasive and relatively inexpensive. In general, the more we know about the human cognitive system, the more we can improve and augment traditional diagnostic methods.
Showing the most recent 10 out of 68 publications
