Cognitive Modeling

Welcome to the cognitive modeling section of the AI department of the University of Groningen!

Here are our research focuses in a nutshell:

Learning

The main focus of our research is learning in a broad sense. How do people learn new skills? How can someone, after only brief and incomplete instructions, do a complicated task? There are no simple answers to this question, as a whole range of learning mechanisms, strategies and phenomena are involved. Some of our learning seems automatic, outside our control, as our brains adapt themselves to new situations. But other learning is definitely under control, as students working for an exam know very well.

The goal of our research effort is to better understand all the learning processes involved, not as separate entities, but as multi-level system in which the components work in concert.

ACT-R

In order to study learning as a complex interaction of mechanisms and strategies, computer simulation is necessary. The ACT-R architecture, developed by John Anderson at CMU, is an ideal platform for simulating and understanding cognitive processes.

Individual differences

Although it is generally acknowledged that there are great differences between people with respect to problem solving, learning and other cognitive abilities, many theorist see individual differences as an unwanted source of noise.

In cognitive modeling, however, individual differences are an additional source of information and constraint. Where traditional experimental psychology tries to reduce the data to the extend that a small set of hypotheses can be tested, cognitive modeling tries to capture as much as possible available information in the data, including differences between individuals.

Applied research

Once we decently understand human learning, the knowledge can be used in applied research. There are several possibilities to do this:

• use a learning model to test user interfaces
• program adaptive agents
• maintain a user model in the application in order adapt the application to the state of knowledge of the user

The latter of these options is explored in the OPTIMA project.