The study examines two approaches to the development of behavioral strategies: i) the evolutionary approach manifested in a Genetic Algorithm, which accounts for gradual development and simultaneous refinement of an entire population; and ii) the behavioral learning approach, which focuses on reinforcements at the individual's level. The current work is part from an ongoing project dealing with the development of strategic behavior. The reported study evaluates the potential of differential reinforcements to provide probabilistic noisy Tit-For-Tat strategies with the motivation to adopt a pure Tit-For-Tat strategy. Results show that provocability and forgiveness, the traits that account for Tit-For-Tat's successes, also prevent it from gaining relative fitness and become an attractor for noisy (non-perfect) Tit-For-Tat strategies.
|State||Published - Jan 2003|
ASJC Scopus subject areas
- Computer Science (miscellaneous)
- Social Sciences (all)