Social rules for agent systems

When creating (open) agent systems it has become common practice to use social concepts such as social practices, norms and conventions to model the way the interactions between the agents are regulated. However, in the literature most papers concentrate on only one of these aspects at the time. Therefore there is hardly any research on how these social concepts relate and when each of them emerges or evolves from another concept. In this paper we will investigate some of the relations between these concepts and also whether they are fundamentally stemming from a single social object or should be seen as different types of objects altogether.

💡 Research Summary

The paper “Social Rules for Agent Systems” investigates the relationships among several social concepts—social practices, conventions, norms, rituals, and habits—that are increasingly used to regulate interactions in open multi‑agent environments. The authors observe that most prior work focuses on a single type of social rule, leaving a gap in understanding how these concepts interrelate, whether one can evolve from another, and how they should be chosen for implementation.

In the introductory section the authors argue that agents are only partially autonomous; therefore, hard constraints alone are insufficient to guide behavior. Social rules provide both a constraining and a motivational component, influencing planning (e.g., traffic‑rule norms prompting early departure). Because different applications exploit different aspects of social regulation, a uniform implementation is difficult; the internal mechanisms required by agents and the surrounding environment vary widely.

Section 2 provides a taxonomy of five major social rule types.

• Social Practices are described as comprising three elements—resources (time, place, objects, actors), activities (the set and possible ordering of actions), and meaning (the social interpretation and purpose). Practices are routine, often unconscious patterns that structure everyday life (e.g., morning routines, shopping trips). They require direct interaction among participants, though the authors note that some literature treats solitary activities as practices.
• Conventions are defined as coordination solutions that emerge from common expectations. The classic example is walking on the right side of a pavement. Conventions are externally beneficial: the more agents follow them, the higher the payoff for each. The authors liken conventions to Nash equilibria in game theory, where mutual expectation and mutual compliance are essential.
• Norms are split into four sub‑categories: social norms, moral norms, legal/regulative norms, and constitutive norms. Social norms describe regular, observable behavior patterns and their societal consequences. Moral norms add a value‑based justification, often becoming internalized judgments that can differ across groups. Legal norms formalize expectations with explicit sanctions and are usually more abstract, covering a broader set of situations. Constitutive norms define the very terms of other norms (e.g., defining an electric bike as a bike). The paper discusses how norms can evolve into moral norms, become institutionalized as law, and how “regimented” norms can be enforced physically.
• Rituals are characterized by prescribed sequences of gestures, words, and objects performed in a designated space, emphasizing formalism, tradition, invariance, rule‑governance, symbolism, and performance. Unlike practices, rituals are primarily symbolic, reinforcing group identity and demarcating insiders from outsiders. Their functional efficiency is secondary to their social meaning.
• Habits are presented as individually learned, relatively fixed patterns of thought, willingness, or feeling that arise from repeated mental experience. While habits are personal, they can influence social expectations because others may anticipate an agent’s habitual behavior. The authors note that from an external viewpoint, synchronized habits may appear as a coordinated practice, but participants are aware of the broader social context, which habits themselves do not encode.

Section 3 explores the hypothesized evolutionary pathways among these rule types. The authors propose that a common “expectation of behavior” serves as a root from which more specific rule forms emerge. For instance, repeated practices can crystallize into conventions; conventions, when reinforced by mutual benefit and expectation, may become social norms; norms that acquire moral justification become moral norms; and institutionalization transforms norms into legal rules. Conversely, rituals can be seen as symbolic enactments of underlying practices or norms, while habits may provide the micro‑level cognitive substrate that supports the stability of higher‑level social rules.

Section 4 discusses implementation implications. Because each rule type demands distinct internal mechanisms—such as monitoring and sanctioning for norms, coordination protocols for conventions, meaning‑interpretation modules for practices, symbolic execution engines for rituals, and habit‑learning components for habits—system designers must select or combine modules according to the application’s needs. The authors deliberately avoid formal modeling to prevent bias toward any particular formalism, but they acknowledge that future work should develop formal representations tailored to each rule type, enabling simulation, verification, and comparative analysis.

The conclusion reiterates that social rules, while distinct in their characteristics, share a common foundation of expected behavior and can evolve from one another. Designers of open agent systems should therefore consider the specific functional and symbolic requirements of their domain when choosing which rule(s) to implement, possibly employing hybrid approaches that capture both coordination efficiency and normative motivation. The paper sets the stage for subsequent research on formalizing these relationships and testing them in realistic multi‑agent scenarios.