The evolutionary advantage of guilt: co-evolution of social and non-social guilt in structured populations

TL;DR

The paper tackles how guilt—implemented as social and non-social emotional states—can evolve to promote cooperation in multi-agent systems. It uses an IPD-based model with guilt thresholds $G \in \{+\\infty,0\\}$, guilt level $g$, and associated costs $\\gamma$ and $\\gamma_s$ to define six strategies that differ in sociability and adaptiveness. Analytical results via risk-dominance and stochastic fixation, together with extensive agent-based simulations on well-mixed, lattice, and scale-free networks, show that social guilt can thrive in well-mixed populations only under low social costs and suitable guilt levels, while structured populations foster both social and non-social guilt through clustering, improving cooperation. In particular, lattice and scale-free networks enable non-social guilt to persist and cluster with emotionally inclined strategies, providing protection against exploiters and elevating overall cooperation. These findings illuminate how population structure shapes the evolution of guilt-driven cooperation, informing the design of ethically aligned, distributed AI.

Abstract

Building ethical machines may involve bestowing upon them the emotional capacity to self-evaluate and repent on their actions. While apologies represent potential strategic interactions, the explicit evolution of guilt as a behavioural trait remains poorly understood. Our study delves into the co-evolution of two forms of emotional guilt: social guilt entails a cost, requiring agents to exert efforts to understand others' internal states and behaviours; and non-social guilt, which only involves awareness of one's own state, incurs no social cost. Resorting to methods from evolutionary game theory, we study analytically, and through extensive numerical and agent-based simulations, whether and how guilt can evolve and deploy, depending on the underlying structure of the systems of agents. Our findings reveal that in lattice and scale-free networks, strategies favouring emotional guilt dominate a broader range of guilt and social costs compared to non-structured well-mixed populations, so leading to higher levels of cooperation. In structured populations, both social and non-social guilt can thrive through clustering with emotionally inclined strategies, thereby providing protection against exploiters, particularly for less costly non-social strategies. These insights shed light on the complex interplay of guilt and cooperation, enhancing our understanding of ethical artificial intelligence.