Does knowing the opponent's strategy guarantee optimal play? (Vol. 46 No. 4)

Methods of statistical physics are proving indispensable for the study of evolutionary games in structured populations. The evolution of cooperation and the phase transitions leading to favorable evolutionary outcomes depend sensitively on the structure of the interaction network and the type of interactions, as well as on the number and type of competing strategies. Now, physicists have solved the puzzle of the availability of information in evolutionary games. In a new theoretical model, the authors answer whether knowing the strategy of an opponent is indeed the holy grail of optimal play in social dilemmas, or whether the situation is in fact more complex. It is indeed the latter, as final evolutionary outcomes depend sensitively not just on individual relations between the competitors as determined by payoff elements, but equally strongly on the spatiotemporal dynamics of defensive alliances that emerge spontaneously as a result of strategic complexity. Reentrant phase transitions highlight the fact that the viability of an alliance depends sharply on the invasion speeds between group members who cyclically dominate each other.

A. Szolnoki and M. Perc, Reentrant phase transitions and defensive alliances in social dilemmas with informed strategies, EPL, 110, 38003 (2015)
[Abstract]