Chapter 53 Equilibrium Convergence in Normal Form Games

For all three types of one-dimensional games and their two-dimensional analogues, the states reliably achieve a loose behavioral equilibrium (BE) even within the first half-run of 5 periods. Most of the loose BE are also tight BE, the main exceptions occurring in two-dimensional games with unique Nash equilibria (NE). Most BE coincide with NE, and most of the observed NE are indeed evolutionary equilibria (EE). In general, the evolutionary treatments of mean-matching (MM) and feedback (Hist) appear to improve convergence to EE. Thus the main tendencies of the data are consistent with evolutionary game theory. All treatments are held constant within a run to test for convergence. Runs are separated by obvious changes in the player population and/or the payoff matrix, and the history box is erased at the beginning of a new run.