skip to content

Computational models you can cuddle

Presented by: 
S Grand [Cyberlife Research]
Thursday 11th October 2001 - 13:30 to 14:30
INI Seminar Room 1
Despite a lingering 'physics envy', most biologists now recognise that cells, organs, organisms and ecosystems are simply too non-linear to be reducible to their parts alone and need to be understood as the totality of the relationships between those parts. As a consequence, analysis is increasingly being supplemented by synthesis in the form of computer models. Nevertheless, a requirement for biological veracity, and the complexity and detail that this implies, can sometime make it hard to see the wood for all the trees.
One way to separate the general from the specific, and hence hope to discover some of the abstract organisational and self-organisational principles that underlie biological systems, is to start instead with a blank sheet and a pile of biologically plausible building blocks, and attempt to build a complete working organism (albeit a relatively simple one) from scratch.
Ten years ago I set myself such a goal, under the cunning disguise of writing a computer game. From a heap of 'genes', 'neurons', 'chemicals' and 'receptors' I constructed a species of artifical creatures that could learn, behave, interact, reproduce and evolve. Today I'm building a robot creature with the goal of understanding some of the general principles underlying the mammalian brain. Whether these toy creatures tell us much about real biology is questionable, but they are certainly some of the most complex and complete computational models of organisms yet attempted, and I hope you may be interested in hearing about the things I learned from building them.
University of Cambridge Research Councils UK
    Clay Mathematics Institute London Mathematical Society NM Rothschild and Sons