UNIVERSITY OF HERTFORDSHIRE
		 COMPUTER SCIENCE RESEARCH COLLOQUIUM


		     "Evolving Computational Models
			  of Human Behaviour"

			      Peter Lane
		       School of Computer Science
		       University of Hertfordshire

			19 October 2005 (Wednesday)

			  Lecture Theatre E350
		       Hatfield, College Lane Campus
			       3 - 4 pm

	       Coffee/tea and biscuits will be available.
			 [Catering Permitting]

		    Everyone is Welcome to Attend
			  [Space Permitting]


Abstract:

One of the aims of cognitive science is to create a unified explanation
of different human behaviours.  Computational models are used to embody
these explanations, and can generate empirical predictions of behaviour 
in future tasks.  However, current research has two problems: first, to 
ensure that optimal models are produced for each behaviour, and second, 
to compare the performance of each model across all observed behaviours.  
I shall present a novel methodology for describing observed behaviours, 
and for developing cognitive models.  This novel methodology allows the 
problem of finding optimal cognitive models to be framed as a multi-
objective optimisation problem, which may be tackled using evolutionary 
techniques (specifically, a non-dominated sorting genetic algorithm).  
Experimental results demonstrate how optimal models from three different
cognitive theories can be evolved for a set of behaviours in a classic 
psychological problem of categorisation.  The evolved models not only 
outperform those previously published in the literature, but also enable 
informed comparisons to be made between the competing theories.
--
Hertfordshire Computer Science Research Colloquium
http://homepages.feis.herts.ac.uk/~nehaniv/colloq