Norm Enforcement with a Soft Touch: Faster Emergence, Happier Agents

TL;DR

The paper tackles how norms emerge in multiagent systems by leveraging a richer spectrum of social signals beyond sanctions. It introduces Nest, a framework that integrates sanctions, tell messages, and hints with reward shaping to accelerate normative learning. Through a pandemic-scenario simulation, Nest achieves faster norm emergence, reduces infections and deaths, and yields higher agent satisfaction compared to baselines, even when information is balanced across types. The results suggest that softer, inferential communications (hints) substantially enhance cooperation and robustness, offering a scalable approach to modeling normative dynamics in complex environments.

Abstract

A multiagent system is a society of autonomous agents whose interactions can be regulated via social norms. In general, the norms of a society are not hardcoded but emerge from the agents' interactions. Specifically, how the agents in a society react to each other's behavior and respond to the reactions of others determines which norms emerge in the society. We think of these reactions by an agent to the satisfactory or unsatisfactory behaviors of another agent as communications from the first agent to the second agent. Understanding these communications is a kind of social intelligence: these communications provide natural drivers for norm emergence by pushing agents toward certain behaviors, which can become established as norms. Whereas it is well-known that sanctioning can lead to the emergence of norms, we posit that a broader kind of social intelligence can prove more effective in promoting cooperation in a multiagent system. Accordingly, we develop Nest, a framework that models social intelligence via a wider variety of communications and understanding of them than in previous work. To evaluate Nest, we develop a simulated pandemic environment and conduct simulation experiments to compare Nest with baselines considering a combination of three kinds of social communication: sanction, tell, and hint. We find that societies formed of Nest agents achieve norms faster. Moreover, Nest agents effectively avoid undesirable consequences, which are negative sanctions and deviation from goals, and yield higher satisfaction for themselves than baseline agents despite requiring only an equivalent amount of information.

Figures (11)

• Figure 1: Disease model with state transition probabilities. The transition for healthy agents applies when coming in contact with those who are infected. Here, $\alpha$ and $\beta$ parameterize the transition probabilities for vaccination and home rest, respectively. In our study, $\alpha = 0.5$ means vaccinated, and $\alpha = 1.0$ means not vaccinated. And, $\beta = 2.0$ means at home and $\beta = 1.0$ means not at home. For example, the edge weight from Mild to Critical can be read as 0.01 for vaccinated agents and 0.005 for unvaccinated agents. The probability of remaining in a state is $1 -$ the probability of evolving to the next state.
• Figure 2: Nest results in the least infected and deceased agents with the highest vaccination rate among all societies. However, despite a lower fraction of vaccinated agents, Emote has fewer infected and deceased agents, and more healthy agents than other baselines. The effect is large for the comparisons of MDeceased and MInfections. For MInfected, the effect is negligible for Emote and small for Penalty and Tell and large for Primitive. Appendix \ref{['sec:additional-results']} includes plots for the first 500 steps where the differences are noticeable.
• Figure 3: Isolation is higher in Emote and Nest (effect is small) than in societies that lack hints. Nest puts fewer agents in quarantine to achieve stable cooperation than Penalty and Tell. The effect is negligible for Emote and small for Penalty and Tell.
• Figure 4: Nest yields more goal satisfaction than Primitive, Penalty, Emote, and Tell. The effect is small for Emote and large for the other societies.
• Figure 5: Comparing the average number of infections (MInfections) in various societies. Nest yields fewer infections on average than other societies. The effect is large.

Theorems & Definitions (3)

• Example 1
• Example 2
• Example 3