Cognitive Science

Research by Kathie Galotti

Dyadic and individual decision making:
Parents making school choice decisions (funded by the National Science Foundation)

The project is a longitudinal study of parents making school and program placement decisions for their children. It is a descriptive study of an important, everyday decision, made by non-experts. The specific decision chosen for study is one made by parents of public school kindergartners as they select from among options in which to place their child for first grade. In Northfield, Minnesota, public school parents have a choice of three different educational programs for their children within the regular public schools. All three options are available without cost. In addition, there are some other options (private or religious schools) potentially available that charge tuition, as well as three other non-traditional, no-fee choices available (2 charter schools and home schooling).

This study has 4 specific aims:

1. To compare the decision making of couples to that of individuals.
2. To compare the information gathering and decision structuring of people who report themselves as having different decision-making styles.
3. To examine the relationship between the performance variables (amount of information gathered,complexity of the decision structure) and affective response variables (ratings of stress, satisfaction, comfort, enjoyment with the decision-making process itself) and to eventual, retrospective evaluations of the decision making.
4. To compare the decision-making of novice and experienced parents, and of less educated and more educated parent

A longitudinal study of college students' major decision making
(in collaboration with Elizabeth Ciner)

This project follows first-year Carleton students and tracks their thinking about which major to declare over a two- year period. Questions to be addressed include:

1. From what sources do students gather information?
2. How complex (in terms of number of alternatives considered and criteria used) is the thinking of different students?
3. How well do measures of cognitive performance predict affective responses to the decision (e.g., feelings of stress, pressure, uncertainty), either during the process or retrospectively
4. How much of an influence is there of different decision-making styles, epistemological styles, or planfulness in either cognitive performance measures or affective response measures?
5. In what ways do students' performance and affective reactions to this decision change over the course of a year

Presentation of Paper by Roy Elveton

"Prolog and the Language of Thought:
Logic as an Introduction to Cognitive Science"

I have been invited to present a paper at the first European meeting of the Computing and Philosophy (CAP) organization. This group was originally sponsored by the American Philosophy Association and is usually hosted in the U.S. by Carnegie-Mellon University. My paper is entitled "Prolog and the Language of Thought: Logic as an Introduction to Cognitive Science."

The conference will be hosted by the Philosophy Department of the University of Glasgow, Scotland and will take place the end of March, 2003. Conference details can be found at www.gla.ac.uk/departments/philosophy/ECAP.html

Final Projects in Cognitive Processes Lab, PSYC/CGST 233

Memory on the Wall

With the help of the Recreation Center's Climbing Wall Staff, Carolyn Speidel '05 and Christine Linnerud '05 tested the effects of contextual stimuli on memorization and recall.

In-class Poster Presentations

Sarah Glass '02 and Grant Anderson '02 present their study on the correlation between decision-making styles and implicit learning.

Cassie McMillan '03 and Mary Harvey '03 discuss their research on music perception.

Rosalyn Claret '04 presents her research on harmonic perception.

Projects by Dave Musicant

"Extracting Questions from Discussion Groups via Text Mining" (joint research with faculty member Jeff Ondich and Carleton students Sarah Allen, Ester Gubbrud, Janet Campbell, Rachel Kirby, and Lillian Kittredge)

The goal of this ongoing project is to automatically generate Frequently Asked Question lists for online discussion groups. The first stage in doing so, automatically identifying which message group postings contain questions, was presented in a poster by Carleton students at the Grace Hopper Celebration of Women in Computing 2002. In order to identify such messages, we tagged a number of messages to generate a training set and techniques such as decision trees and support vector machines to identify which messages contained questions.

"Weka-Parallel: Machine in Learning in Parallel" (joint research with Carleton student Sebastian Celis)

Weka-Parallel is an enhancement to Weka, a popular machine learning software package. The original Weka suite provides capabilities for doing machine learning with a variety of well-known algorithms, such as decision trees and neural networks. Weka-Parallel expands upon the original software by allowing one to do n-fold cross validation (an important experimental technique) in parallel across multiple computers. This work will be presented by Sebastian in a poster at the 2003 SIGCSE Technical Symposium on Computer Science Education.

"Optimizing F-Measure with Support Vector Machines" (joint research with Vipin Kumar and Aysel Ozgur at University of Minnesota)

A popular machine learning algorithm, known as the support vector machine (SVM), is often used for classification of data where one class is present with much higher frequency than the other. "F-Measure" is a common metric used in determining how well a classifier performs on a given set of data. This theoretical work provides evidence that common heuristic techniques in using SVMs to optimize F-measure, are in fact, "the right thing to do."