The agent population is being given three types of ability to learn; individual learning, evolutionary learning and social learning. And one of the many innovative parts of the project is the ability agents will be given to pass knowledge from one to another, for example through 'parents' teaching their 'children'.
The possibility of social learning could also provide evidence of an emergent "culture" developed jointly by multiple agents.
"Until now a project like this was impractical and the stuff of science fiction, no one thought that it would be possible to do. But now that the massive computing resources needed to grow an artificial society are much more accessible, and work on the evolution of language has progressed, we are now in a position to attempt to grow an artificial society on this scale," said team member Nigel Gilbert of the University of Surrey, in a statement.
"One of the greatest scientific challenges of this project is not only to evolve a language but to understand it," said the project's coordinator, Gusz Eiben of the Free University Amsterdam, in the same statement.
Eiben concluded: "If language evolution indeed takes off as we hope, how can we understand what the individuals discuss with each other? We intend to put up a number of monitors to look at how language is developing and try to look not only at the end result, but also at the process. Thereby, we hope to be able to keep up with the language's evolution."
The website for the NEW-TIES project could be found at http://www.new-ties.org/mambo when this story was first posted.