Seminar Artificial Intelligence - Dr. M. Ragni

Title: Can we reconstruct or even predict thought processes of individuals?

Abstract:

The way humans process information, make decisions, and reason is highly relevant in our everyday life. Cognitive psychology and AI have started to develop cognitive and computational models to explain the most frequent answer given in psychological experiments. Existing cognitive models of reasoning are often limited by their inability to actively adapt to an individual user’s cognitive system, e.g., they cannot predict how a specific user’s mind will (mis-)process information. Hence, the time is ripe for a cutting-edge challenge: To develop novel cognitive AI models that are able to smoothly adapt themselves to the reasoning processes of the human mind and are able to predict an individual's future inferences based on few observations on the reasoning on some past examples. To be successful a new research paradigm is necessary: Data from psychological experiments needs to be reanalyzed with the focus on the individual information processes. Existing models need to be modularized and parametrized and new models need to be developed that comprise the best of AI and Cognition. Based on this, major steps towards a predictive cognitive modeling to predict a single reasoner’s inference can be made.