Abstract / Summary
The aim of this HEiKA collaboration was the development of a cognitive user model of a complex learning task as basis for assistive computer technology. The user model brings together expertise from both cognitive psychology and human-centered computer science by combining two modeling approaches:
A psychologically well-founded computational model generates a-priori predictions of user states based on the simulation of cognitive processes. This prediction is combined with real-time statistical EEG-based analysis of the user’s brain activity to detect neural markers for certain internal states to yield a more accurate overall prediction.
During the project, we developed a prototype system that models learning performance in a complex reinforcement learning task using a Q-learning algorithm augmented by spectral analysis of concurrently recorded EEG.