Apology and forgiveness evolve to resolve failures in cooperative agreements

When interactions are repeated mistakes, whether intentionally or not, tend to occur. Researchers have argued that revenge, apology and forgiveness are mechanisms that humans have acquired to ensure that intentional mistakes are avoided and that mutually beneficial relationships can continue. We have shown in the context of the iterated prisoners dilemma wherein agents can decide to make cooperative agreements that these three behaviours emerge spontaneously. Concretely our work reveals that apology and forgiveness are very efficient even in a very noisy environment and ensure long lasting relationships. Yet in order for apology to work, it needs to be sufficiently costly otherwise taking revenge by defecting is the most profitable behaviour. This research has direct implications for online socio-technological systems who’s success depends on the trust users (and agents) have in the other users (or agents) participating in the system.