Optimisation of ACT-R for simulating human computer interaction in healthcare

RDI PhD Studentship

Research student: Jean-Claude Golovine

Principal supervisor: Professor Patrik O’Brian Holt

Second supervisors: Dr. John McCall, Dr. Stuart Watt

External advisors: Dr. Gus Ferguson (Heriot-Watt University) and Dr. Albert Burger (MRC)

The Adaptive Control of Thought-Rational (ACT-R) architecture has been developed by Anderson and colleagues at Carnegie Mellon University in the USA. ACT-R is without a doubt the most sophisticated model currently available to simulate human cognitive processes and behavioural responses. There is a very active research community in this area that is focused both on the development of ACT but also on its application, e.g. to simulate human responses when interacting with computers or other artefacts such as remote controlled vehicles.

This project aims to adapt ACT-R to support a User Interface Design Toolkit to aid interface development in healthcare, in particular, clinical genetics. The results will be transferable to other application areas.

An important novel aspect of this project is use of optimisation to improve the performance of ACT-R to simulate user interactions. The optimisation process is delivered either by an Ant Colony or an Evolutionary algorithm - using ACT-R as a human simulator, thus gaining human cognitive predictions for particular tasks which are used to assess the quality of potential optimised solutions.