Biologically Plausible Modelling of Morality

Neural computation has an extraordinarily influential role in the study of several human capacities and behavior. It has been the dominant approach in the vision science of the last half century, and it is currently one of the fundamental methods of investigation for several higher cognitive functions. Yet, no neurocomputational models have been proposed for morality. Computational modeling in general has been scarcely pursued in morality, and existent non-neural attempts have failed to account for the mental processes involved during moral judgments. In this paper we argue that in the past decade the situation has evolved in a way that subverted the insufficient knowledge on the basic organization of moral cognition in brain circuits, making the project of modeling morality in neurocomputational terms feasible. We will sketch an original architecture that combines reinforcement learning and Hebbian learning, aimed at simulating forms of moral behavior in a simple artificial context.