Social computations and the origin of moral norms

Abstract

The author proposes a conception in which society and a person are understood as a nested computing system of a multi-agent architecture. The ‘Computational’ as a con­cept extends beyond the Turing machine and is considered as the ability to compute and predict statistical distributions that best match the characteristics of the external en­vironment to achieve the optimal value of an assumed control variable. Society induces the rules of behavior in agents similarly to the way by which a magnetic field induces a current in a moving conductor. Induction is carried out by selecting the patterns of be­havior empirically derived by individuals and certifying the patterns that are most use­ful for achieving the target indicators of the system. Certified rules are perceived by in­dividuals as, on the one hand, immanent to them, and on the other, as generally allow­ing for their violation, since both individual agents and the whole society are statistical machines. The objective probability of non-compliance with a rule is perceived as moral freedom, and hormonal reinforcement of adherence to the general rule is per­ceived as a sense of moral duty. The computational approach to morality overcomes the inductivism of naturalistic and utilitarian theories, and at the same time scientifi­cally explains the ‘transcendentality’ of moral law, without resorting to idealistic meta­physics, as does Kantian deontism.