UNIVERSITY OF HERTFORDSHIRE
		 COMPUTER SCIENCE RESEARCH COLLOQUIUM

	 "Synthesizing Behaviour in a Complex Evolving Ecosystem"

			 Peter-Paul Pichler
		     (School of Computer Science,
		     University of Hertfordshire)


			   6 December 2006

			  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:


The last decades have brought about a change on how we look at
cognition in animals as well as in artifacts. Substantial progress in
various fields (notably in neuroscience, cognitive science and
evolutionary biology) has led to a process of consolidating what was
traditionally considered to be separate and even opposed. The dualisms
of body and mind, and affect and reason are being reconciled to form a
more holistic picture of the 'mechanisms and machinery' that constitute
living organisms.

This growing consensus about the importance that organisms have evolved
as whole systems by the means of natural selection, spans across
multiple disciplines. The neurosciences, evolutionary biology, and
psychology usually take life on earth as the starting point and try to
reverse engineer the mechanisms and situations that led to the
adaptations which shaped life as it actually occurred. The findings of
these fields have also had effects on, and are complemented by,
different research on animat design. In robotics the prominent rank of
the concept of embodiment is widely accepted and also the modeling of
the emotions as regulatory components plays an increasingly eminent
role.

We adopt an artificial life perspective and utilize evolution as the
means to evolve artificial creatures. The emphasis hereby lies on
modeling a coherent framework of an evolving ecosystem to evolve
complex agent behaviour based on the principle of natural selection. We
view behaviour as an emergent phenomenon which depends on the circular
relationship between body, mind and environment. We are interested in
behaviour in general rather than any specific behaviour. In order not
to needlessly bias or constrain the evolutionary process the agents
start with very limited functional design and morphology and neural
controllers are evolved as functional wholes. The agents are `grounded'
in an increasingly complex environment by a model metabolism and rich
interaction dynamics. Furthermore, we introduce a novel criterion for
evaluating differential reproductive success aimed at maximizing
evolutionary freedom.
In this talk we will present first experimental results suggesting that
this approach may be conducive to widening the scope of artificial
evolution for the generation of agents exhibiting non-trivial 
behaviours in a complex ecosystem.




--------------------------------------------------
Hertfordshire Computer Science Research Colloquium
http://homepages.feis.herts.ac.uk/~nehaniv/colloq