Spring 2006:  Tues/Thurs, 11:15-12:30



Instructor: Dr. Michael Jones

Email: jonesmn@indiana.edu  ; Office: Psyc 357





Overview

Q550 is intended to be a “capstone” class of cognitive science that
integrates skills from Q530 (programming methods) and Q560
(empirical methods) with cognitive modeling. The course is also
intended to be a “hands on” experience for students to actually do
some modeling, not just read about it.



The course will be a mixture of instructor-directed lectures, and
student-led presentations.  The lectures will cover the major topics
in the art of computational modeling (e.g., curve fitting, parameter
estimation, learning algorithms, etc.), and the student
presentations will focus on specific areas of application of this
art (e.g., models of perceptual learning, etc.).



Pre-requisites

Students should have already taken Q530 (programming methods) and
Q560 (empirical methods), or have equivalent experience and
knowledge. Most programming languages will be acceptable (if you
have a favorite), however, demonstrations will work with MATLAB,
which you can learn quickly if you have previous computer
programming experience. You should also have some experience with
collecting and analyzing data from human subjects (rats will do) or
Q560 (empirical methods).



Goals

Along with gaining an understanding of concepts of cognitive
modeling, students will get experience:

Designing and running a simple experiment relevant to research
interests
Analyzing the data using standard inferential statistics
Programming a simple cognitive process model(s) for the experiment
task
Fitting the model(s) to the data


Reading Materials

There is no textbook for this course. Selected articles will be
listed online.



Grading

There will be no exams or tests during this course. Evaluation will
be based on a final poster, paper, in-class presentations, and
discussions. The final exam period will be dedicated to a poster
presentation session.









Topics

Foundations and goals of cognitive modeling
Evolution of math models
Model fitting and parameter estimation
Model comparison
AI and machine learning techniques applied to cognitive modeling
Approaches:
Connectionist modeling
Bayesian modeling
Vector space modeling
Localist and production system modeling
Genetic algorithms/evolutionary modeling
Dynamic systems modeling
Models of:
Memory, attention, learning, categorization
Vision and perception
Pattern recognition
Decision, judgment, reasoning, and problem solving
Language learning, representation, and comprehension
Anything else of interest that comes up




Individual Project

Each student will select a domain that is of personal interest.
There are three approaches to the individual project: 1) identify a
model and the phenomena it explains, collect data, and simplify the
model to the basic mechanisms necessary to address the core
phenomenon; 2) identify two competing models of a phenomenon, code
the models, and create an experiment which will allow you to
constrain between the models; or 3) identify a robust phenomenon for
which there is currently no explanatory model, and create/fit one.



Students will program a model and collect/analyze experimental data
(using the other students and instructor as subjects). Each student
will present two short talks: one proposing the model and
experiment, and one reporting the results after the model and
experiment are completed. At the end of the course, each student
will present a poster on his/her work, and will submit a final
paper.