The grounded naming game

This chapter shows a concrete example of a language game experiment for studying the cultural evolution of one of the most basic functions of language, namely to draw attention to an object in the context by naming a characteristic feature of the object. If the object is a specific recognizable individual, then the name is called a proper name, and this is the case that is studied in this chapter. We investigate a concrete operational language strategy, with a conceptual as well a linguistic component, and show that a population of agents endowed with this strategy is able to self-organize a vocabulary of grounded proper names from scratch. The example provides a clear example of the role of alignment in stimulating self-organization and how expressive adequacy, cognitive effort, learnability, and social conformity act as selectionist forces, driving the population towards an effective language system.This research was carried out at the AI Lab of the University of Brussels (VUB) and the Sony Computer Science Laboratory in Paris.Peer Reviewe