The Role of Cognitive Architectures in the Modelling of Human Virtue

Abstract

There has been increasing interest in virtue, both from the perspective of theology, and from human sciences such as psychology. This paper focuses on the possibility of modelling human virtue computationally, in terms of a particular cognitive architecture, Philip Barnard’s Interacting Cognitive Subsystems. It is unusual in being a macro cognitive architecture; it has been formulated with a computational level of precision; and its evolutionary development has been described. Though it is possible that computers could acquire virtue in a distinctive way, the focus here is on modelling human-like virtue. That needs to be grounded in empirical research on human moral functioning, and research is discussed on two main topics: whether or not virtue functions as a cross-situational, characterological trait; and the tendency of virtue to fragment into components such as values and behaviour which are only weakly connected. An important feature of Interacting Cognitive Subsystems is the distinction between two different modes of central cognition, one intuitive-embodied and the other conceptual. The interplay between these two different modes of cognition in the acquisition of virtue is discussed.