Autonomous Tags: Language as Generative of Culture

Abstract

The first intelligent agent social model, created by the author in 1991, used tags with emergent meaning to simulate the emergence of institutions based on the principles of interpretive social science. This symbolic interactionist simulation program existed before Holland’s Echo, however, Echo and subsequent programs with tags failed to preserve the autonomy of perception of the agents that displayed and read tags, as first program did. These subsequent tag programs include Epstein and Axtell; Axelrod; Hales; Hales and Edmonds; Riolo, Cohen and Axelrod, as well as the works on contagion originating in Carley, etc. The only exceptions are the author’s 1995 SISTER program, and Axtell, Epstein, and Young’s 2001 program on the emergence of social classes, which was influenced by the symbolic interactionist simulation program at George Mason University, and Steels’ 1996 work. Axtell Epstein and Young’s program has since been credited for strong emergence (Desalles et al.). This paper explains that autonomy of perception is the essential difference in the symbolic interactionist implementation of tags that enables a strong emergence to occur, and that is why strong emergence has occurred in the works of Duong and of Axtell, Epstein and Young. This paper explains the important differences in existing tag models, pointing out the qualities that enable symbolic interactionist models to become social engines with strong emergence, and also introduces new work that puts the SISTER program in a spatial grid, and explores what happens to prices across the grid. In half of the runs, a standard of trade, or money, emerges.