PATH DEPENDENCE AND LEARNING FROM NEIGHBORS

We study the long-run properties of a class of locally interactive lea rning systems. A finite set of players at fixed locations play a two-b y-two symmetric normal form game with strategic complementarities, wit h one of their ''neighbors'' selected at random. Because of the endoge nous nature of experimentation, or ''noise,'' the systems we study exh ibit a high degree of path dependence. Different actions of a pure coo rdination game may survive in the long-run at different locations of t he system. A reinterpretation of our results shows that the local natu re of search may be a robust reason for price dispersion in a search m odel. (C) 1996 Academic Press, Inc.