Dr Jason Noble believes that the secret of making progress in the world of computing lies in understanding the animal kingdom.
'If we are going to make real advances in artificial intelligence, we had better make sure that we understand rats before moving on to more ambitious systems,' he said.
Jason has just joined the Science and Engineering of Natural Systems Group (SENSe) in the School of Electronics and Computer Science (ECS), where he is currently working on a paper on natural and artificial agents for the journal, Biology and Philosophy.
He believes that the work that he has done in observing the social behaviour of animals will be useful when it comes to designing autonomous computerised agents.
'Animals and people don't face the challenges of their environment all alone; they generally work in groups,' he said. 'Therefore, if we are going to have, for example, computerised agents bidding on eBay, we need to look at how they will interact with one another and issues such as trust and reciprocity become paramount.'
Jason has always had an interest in animal social behaviour and in 1998 completed his PhD on the evolution of animal communication at the University of Sussex under the supervision of Professor Dave Cliff, who now leads the SENSe group.
During his PhD, he discovered that the 'selfish gene' does not necessarily imply the 'selfish animal', as all kinds of co-operative behaviour can be found in the animal world. He went on to find that this approach could also be applied to computerised agents.
'If autonomous agents take off, they will face many of the issues that animals face as they will not operate in an empty world,' he said.
Jason also realised that he could use ideas from biology to solve optimisation problems and that he could use evolutionary theory to 'grow' good solutions to practical problems.
He went on to do a post-doctoral research fellowship at the Max Planck Institute in Berlin, where he looked at social learning in animals. He looked in particular at how animals learn not just from their experience of the world but from each other.
'Rats learn what to eat by smelling the breath of other rats, and animals send alarm signals to one another when there is danger looming,' he said.
His research made him more convinced that animal behaviour could provide useful models for agents.
'The "you scratch my back and I'll scratch yours" philosophy which exists in the animal kingdom will certainly be relevant in the world of computer agents.'
After undertaking a research fellowship in the School of Computing at the University of Leeds, where he developed his animal behavior work further, Jason joined ECS in January 2007.
‘I joined the SENSe group because it stands out in the UK as being the place for biologically-inspired computing,' he said. 'ECS has been successful enough and brave enough to promote this exciting area of work with all of its associated risks. It takes a University of the calibre of Southampton to do this.’
Jason plans to continue to develop models of social behaviour and evolutionary computing. He will focus on the dual approach of using computing to look at social behaviour, and using social behaviour as the basis for novel computer systems.
'I’m confident that over the next 100 years, computing and biology will become increasingly intertwined,' he said.
