Learning a commonsense moral theory.

Area of Science:

• Cognitive Science
• Moral Psychology
• Computational Neuroscience

Background:

• Understanding moral learning is crucial for explaining social cohesion and ethical development.
• Existing models often struggle to capture the complexity and adaptability of human moral reasoning.
• The societal impact of moral learning dynamics necessitates a robust theoretical framework.

Purpose of the Study:

• To introduce a novel computational framework for modeling moral learning.
• To elucidate the cognitive capacities essential for navigating complex welfare trade-offs.
• To explain the development of broad moral competence from limited observations and the dynamics of societal moral change.

Main Methods:

• Development of a computational model integrating utility calculus, hierarchical Bayesian inference, and meta-values.
• Formalization of cognitive capacities for quantitative welfare representation and value alignment.
• Simulation of moral learning trajectories from childhood to societal levels.

Main Results:

• The framework successfully models the acquisition of moral competence from sparse data.
• It explains generalization of moral judgments to novel individuals and situations.
• The model provides insights into both rapid and gradual societal moral change.

Conclusions:

• The proposed computational framework offers a parsimonious explanation for moral learning and its societal implications.
• Key cognitive capacities identified are fundamental to human ethical development and adaptation.
• This work advances our understanding of the mechanisms driving individual and collective moral evolution.