Abstract Cognitive Models

• Applicant: Professor Dr. Andreas Podelski
• Subject Area: Software Engineering and Programming Languages
• Term: from 2016 to 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 316490981

The project will provide the foundation for the formal, semantics-based analysis of models in Cognitive Science. Specifically we are interested in automated methods based on model checking and SMT solving for the analysis of psychological theories and of hybrid cognitive models. A psychological theory can be viewed as the informal description of a high-level model for human behaviour in problem solving tasks. A hybrid cognitive model can be viewed as a low-level implementation of a psychological theory. In Cognitive Science, one uses the simulation of a hybrid cognitive model (in, say, ACT-R) in order to validate a psychological theory against empirical data from behavioural experiments. The analysis approach envisioned by the project can possibly enhance or complement the existing, simulation-based approach. We will develop the notion of an abstract cognitive model which allows us to describe a psycho- logical theory by a formal model (a model on which we can apply formal, semantics-based analysis methods). We will investigate an approach for the analysis of abstract cognitive models (an analysis based on hybrid system model checking and SMT solving). The analysis can be used for a first check of a psychological theory before the implementation and the simulation of hybrid cognitive models. We will also investigate how the approach, by the synthesis of numerical parameters in an ab- stract cognitive model, can automatically synthesise a `correct-by-construction' hybrid cognitive model for a given psychological theory.

• DFG Programme: Research Grants