A simulation of language evolution, wherein agents develop a common language through local interactions called language games.
The basic kind of simulation involves a population of autonomous agents. Each agent has a list of labels for talking about objects in its environment. Initially agents have a randomised interpretaion, meaning their labels refer to different kinds of objects. During the simulation, the agents engage in language games with each other. During these language games, a pair of agents observe an object; one agent describes it with a label, and the other updates its meaning of that label accordingly. As the simulation progresses, the agents gradually converge on a shared set of labels.