Title: Active control of complexity growth in Language Games
Links to manuscript, slides and related articles can be found on: [ Ссылка ]
Abstract:
How do we agree on social conventions? Negotiating a consensus can sometimes lead to a complex situation with many competing options, and it is complicated for individuals to remember all of them. In this thesis, we use computational models of language emergence to show that an active choice of what individuals talk about can have a significant impact on the speed of the agreement process and the memory needs. The presented policies are close to an optimal behavior. We finally show that people do behave in a similar manner, through an online user experiment presented as a game. (The game is still accessible: [ Ссылка ])
