Shaped Prisoner's Dilemma Automata
Submitted to CIBCB 2005
Wendy Ashlock,
with Daniel Ashlock

Abstract PDF eprint

Finite state automata are one of the common representations used to encode agents to play the iterated prisoner’s dilemma. A shape for a finite state automaton is a restriction on what transitions are permitted out of each state in the automata. Eight shapes, including a baseline shape with all possible transitions, are tested by enforcing the shape during the training of agents that play iterated prisoner’s dilemma with an evolutionary algorithm. All eight shapes yield distinct distributions of behaviors in the evolved agents.