Modeling collective morality via evolutionary game theory

We have been addressing problems in machine ethics dealt with by using computational techniques. In the preceding chapters, our research has focused on Computational Logic, particularly Logic Programming, and its appropriateness to model morality, namely moral permissibility, its justification, and the dual-process ofmoral judgments regarding the realm of the individual. Now, in the sections of this chapter, we address the collective realm computationally, using Evolutionary Game Theory in populations of individuals, to report on norms and morality emergence. These populations, to start with, are not equipped with much cognitive capability, and simply act from a predetermined set of actions. Our research has shown that the introduction of cognitive capabilities, such as intention recognition, commitment, apology, forgiveness, and revenge, separately and jointly, reinforce the emergence of cooperation in populations, comparatively to their absence. We then prospect future work concerned with adding guilt. In particular, we show: • how learning to recognize intentions and committing resolve cooperation dilemmas; • the evaluation of two strategies in the emergence of cooperation in groups, viz., avoidance versus restriction; • the role of apology in committed versus commitment-free repeated interactions; • how apology and forgiveness evolve to resolve failures in cooperative agreements; • the role that guilt may play to prime apology and forgiveness.