Modeling the Mind: Computational Models of Cognition
Seminar presentations: Wednesdays 1:30-3:00 pm
Classroom: FedEx Institute of Technology #225
Phone: 678-3142
EMAIL: franklin@memphis.edu
Web site: www.cs.memphis.edu/~franklin/CognitiveScienceSeminar-Fall04.htm
Computational models have long been a major, and perhaps indispensable, tool in cognitive science. Many of these model some psychological theory of a particular aspect of cognition, attempting to account for experimental data. Others, such as the construction-integration model [[130 (Kintsch 1998)]], SOAR [[54 (Laird et al 1987)]], Clarion [[409 (Sun 1997)]] and ACT-R [[149 (Anderson 1990)]], aspire to be a general computational model of cognition. The IDA model, a recent entry into this field of general computational models of cognitio, is conceptual as well as computational, since portions of the model have been designed but not yet implemented computationally. The IDA model differs from the other general models in that it is conceived of as an autonomous software agent [[53 (Franklin 1997); 58 (Franklin & Graesser 1997)]]. This agent is designed to implement a theory of cognition and to automate a practical task typically performed by humans. IDA automates a rather complex personnel assignment task for the US Navy [[155 (Franklin 2001); 113 (Franklin et al 1998)]]. The Neural Schema model [[248 (McCauley 2002)]], like the IDA model, is qualitative, rather than quantitative, but is not a practical software agent.
This seminar will be devoted to the description, discussion and comparison of these broad model of cognition, their architectures and mechanisms, the cognitive theories they implements or support, and the hypotheses they generate. We'll also illustrate their use in analyzing cognitive tasks.
I would hope that participants in the seminar would, by the end of the semester, be comfortable enough with the various models to use them as an aid to their own thinking about cognitive processes and as a tool for analyzing cognitive tasks. I would further hope that some might be motivated to test hypotheses arising from the various models.
Each enrolled student will participate in a two (on rare occasions perhaps three) person team project. Each student will make one presentation during the semester of some portion of a model chosen in consultation with the instructor. Computer science students will get one of the models, chosen in consultation with the instructor, up and running on a local machine, and learn to use it. Psychology students will choose one of the model in consultation with the instructor, and use the model to emulate the data from an existing experiment. Deliverables from these projects will include from each team an essay describing the project and a brief oral presentation to be made during the final exam period on Wednesday, December 15 th , 10-12. Grades will be based on the student's attendance and participation during the classes, on the presentation, on the project essay, and on the project presentation.
The schedule on the web site lists the presentations to be made from 1:30 to 3:00 pm each Wednesday and will be updated as needed. A list of references and suggested reading materials also can be found on the web site. During the 12:40-1:30 and or 3:00-3:40 period enrolled students will discuss the previous week's topic and the various student team projects. The first such period will be a time for getting acquainted, the forming of project teams, and consultations about project topics.