An Integrated Reasoning Approach to Moral Decision Making

Abstract

We present a computational model, MoralDM, which integrates several AI techniques in order to model recent psychological findings on moral decision-making. Current theories of moral decision-making extend beyond pure utilitarian models by relying on contextual factors that vary with culture. MoralDM uses a natural language system to produce formal representations from psychological stimuli, to reduce tailorability. The impacts of secular versus sacred values are modeled via qualitative reasoning, using an order of magnitude representation. MoralDM uses a combination of first-principles reasoning and analogical reasoning to determine consequences and utilities when making moral judgments. We describe how MoralDM works and show that it can model psychological results and improve its performance via accumulating examples.

Explainable, Normative, and Justified Agency

“…The AI literature reports some work on such normative reasoning, with the earliest focused on legal inference (e.g., Branting, 2000). Equally relevant has been research on machine ethics and moral reasoning (e.g., Anderson et al, 2006;Bringsjord et al, 2006;Dehghani et al, 2008;Guarini, 2005;McLaren, 2005). Some researchers have developed new representations and mechanisms to support the ability to make normative judgements and decisions, but others (Iba and Langley, 2011;Liu et al, 2013) have instead treated moral reasoning as a form of everyday cognition.…”

Extending SME to Handle Large‐Scale Cognitive Modeling

“…Automatically generated representations can be taken from other AI systems that are carrying out some task. In systems that have SME as a component, SME's representations can be produced by another part of the system itself, through interpreting natural language text (e.g., Dehghani, Tomai, Forbus, & Klenk, ) or sketches (e.g., Lovett et al., ). This has the advantage that in cognitive simulation, the materials given to the system are the same stimuli given to human subjects (Lovett et al., ; Sagi et al., ).…”

“…MoralDM makes choices that are compatible with the majority choices made in multiple experiments (Dehghani et al., ). Moreover, as the size of the case library grows, the proportion of correct responses improves (Dehghani et al., ), demonstrating that the use of analogy matters in the model's performance.…”

A moderate role for cognitive models in agent-based modeling of cultural change

“…Current models of culture typically represent cultural variation with a few simple variables, in aggregate or simple-agent models (e.g. (Acerbi A et al 2012;Afshar M and Asadpour M 2010;Alexander JM 2007;Richerson PJ 1985, 2005;Cavalli-Sforza LL and Feldman MW 1981;Enquist M et al 2010;Grim P et al 2004;Mueller ST et al 2010;Nakahashi W 2010)), or represent interactions between cultural components without reference to the role of interacting individuals in a society (Dehghani M et al 2008;Klüver J and Klüver C 2010;Thagard P 2012).…”