Fairness Is Not Innate: A Three-Variable Framework for Human and Artificial Moral Behavior

This paper proposes a three-variable framework for fairness capacity in humans and artificial systems: F(t) = I(t) × D(t) × S(t) where I represents intelligence and prosocial reasoning capacity, D represents informational accuracy and reliability, and S represents perceived survival safety. The framework formalizes fairness as a dynamically constrained process rather than a fixed innate trait. A time-evolution formulation is introduced to model recursive feedback effects, including scarcity-driven fairness cascades. The paper synthesizes findings from behavioral economics, social psychology, neuroscience, primate fairness research, and AI alignment literature into a unified formal structure. Several testable predictions are derived and compared with prior empirical findings. Exploratory analyses using Global Preferences Survey (GPS), European Social Survey (ESS), and EVS/WVS longitudinal panel data provide partial support for the framework and for a proposed social-bond extension (B_social). Cross-validation and train-test analyses indicate stable out-of-sample performance, while Random-Intercept Cross-Lagged Panel Models (RI-CLPMs) provide more conservative longitudinal estimates. The framework is presented as a formal synthesis and exploratory modeling approach rather than a claim of definitive causal proof. Implications are discussed for human moral behavior, social trust, institutional stability, poverty dynamics, and AI alignment under conditions of informational distortion and perceived threat.