From the Selfish Gene to the “We”: Cooperation Often Needs Suppressors

ABSTRACT

Human cooperation is often explained through motivations and values, but it is also a strategic and evolutionary problem: locally advantageous behavior can erode the collective arrangements on which everyone depends. Dawkins’ gene‐centered view remains useful here, not as a moral theory of selfish people, but as a way to track how lower‐level selection can destabilize higher‐level organization. I argue, however, that cooperation is stabilized by a broader repertoire than external punishment alone. Repeated interaction, partner choice, aligned incentives, and architectural constraints can all make cheating unattractive; in many settings, additional suppressors then make defection costly or unprofitable. On this view, game theory and a multilevel evolutionary lens are complementary. The Folk Theorem shows how cooperative outcomes can be sustained endogenously under patience and monitoring, while biology shows how the informational and structural conditions for such equilibria are themselves produced or maintained. The point is diagnostic rather than fatalistic: cooperation is typically conditional, costly to sustain, and shaped by trade‐offs, but humans can redesign incentives and institutions, especially in digital environments where misinformation and attention now propagate with unusual speed and fidelity.

Why does cooperation break down, and what holds it together? From the selfish gene to cancer to misinformation, the same multilevel logic recurs: cooperation lasts only when cheating is detected, deterred, or made unprofitable. Suppressors and stabilizers, from immune surveillance to institutions, turn defection from profitable to costly.

1. Introduction: A Metaphor and a Category Error

The phrase “the selfish gene” was never a thesis about human character. It was a disciplined shift in explanatory perspective. Follow the entities that are copied with sufficient fidelity across generations, and many patterns of conflict and cooperation become intelligible without invoking intention or moral psychology [1].

Yet the metaphor predictably invites a category error. The technical sense of “selfish” (differential copying) is smuggled back into ordinary language and mistaken for a claim about motives. That confusion fuels symmetrical errors such as genetic determinism (“people are selfish, so cooperation is naive”) and moralism (“biology excuses bad behavior”). Both miss Dawkins’ point: evolutionary self‐interest is a bookkeeping term for what spreads, not a theory of what people want. A replicator lens does not deny agency, and it does not imply adaptationist overreach; humans act for reasons, and those reasons are shaped by environments, institutions, drift, and historical contingency [1, 2, 3, 4].

Game theory already supplies much of the formal analysis that this Perspective needs. In repeated games, cooperative outcomes need not depend on an external punisher; with sufficiently patient players and adequate monitoring, they can be sustained by endogenous strategies and credible threats [5, 6, 7]. My claim is therefore narrower than “cooperation needs an outside policeman.” Cooperation lasts when the strategic environment makes defection unattractive, detectable, or self‐limiting.

For that reason, I use “suppressor” in a broad but not unlimited sense: a mechanism that reduces the payoff to defection by sanctioning it, detecting it, or blocking it. Some suppressors are exogenous institutions; others are endogenous equilibrium strategies; still others are architectural features that narrow the feasible space for cheating. This broader usage is intended to complement, not replace, game‐theoretic analysis [5, 6, 8, 9].

Biology remains useful because it shows how often such conditions must be built rather than assumed. Major transitions repeatedly rely on bottlenecks, developmental constraints, policing, and reductions in within‐group competition [10, 11, 12, 13]. At the same time, evolution does not optimize in a vacuum: it works through compromises, inherited constraints, and path‐dependent solutions of convenience [8, 12] (Figure 1).

FIGURE 1.

From genes to the “we”: cooperation is a contingent equilibrium, not a default.

The framework contrasts local drivers of conflict with a heterogeneous set of stabilizers. At genomic and organismal scales these include silencing, checkpoints, tissue architecture, and immune control; at kin and social scales they include relatedness, partner choice, reputation, norms, and formal institutions. The point is not that all cooperation is policed or that societies are literally organisms. It is that higher‐level cooperation persists when lower‐level exploitation is deterred, rendered unprofitable, or made difficult to realize [9, 10, 11, 14].

2. From Replicators to Vehicles: Why Cooperation Is an Achievement

The replicator‐vehicle distinction clarifies why cooperation is precarious. Replicators are what gets copied; vehicles are the transient structures that carry them forward. Selection can therefore favor locally successful strategies even when they erode the larger organization that makes copying possible [1, 3, 4]. Price's equation makes the bookkeeping explicit by separating within‐group from between‐group selection [3, 4].

Intragenomic conflict makes the logic concrete. Selfish genetic elements can bias transmission even when they reduce organismal fitness, and genomes evolve silencing, imprinting, and meiotic control to limit that damage [15, 16]. The general point is not that policing is universal, but that whenever lower‐level units can gain by exploiting a larger system, higher‐level order must be stabilized somehow.

Multicellularity is a major transition precisely because cells had to give up unrestricted replication [10, 13]. This can be achieved through differentiation, resource sharing, developmental bottlenecks, programmed cell death, and limits on cell division. In some cases, those mechanisms look like suppressors; in others they simply make profitable cheating harder to realize.

3. Stabilizers: How Cooperation Becomes Durable

In organisms, stabilizers include cellular checkpoints, tissue architecture, and immune surveillance; in societies, they include rule enforcement, accountability, reputation, partner choice, and market design. The analogy is functional, not literal [14, 17].

This broader language matters because cooperation is not stabilized only by external punishment. Classical evolutionary and game‐theoretic accounts identify several routes: kin selection, reciprocity, repeated interaction, by‐product benefits, and partner choice [9, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27]. In repeated games, the Folk Theorem shows that many cooperative outcomes can be sustained endogenously when players are sufficiently patient and can monitor one another [6, 7]. The question, then, is not simply whether there is punishment, but what informational and institutional conditions make cooperative equilibria credible.

Some forms of cooperation are stable because cheating is unprofitable or difficult. By‐product benefits can align private and collective payoffs; partner choice can allow cooperators to exit exploitative relationships; and the architecture of a task can leave little room for profitable free‐riding [9, 26, 27]. Calling all these mechanisms “suppressors” would be misleading. They are better seen as a family of stabilizers, of which punishment is only one subtype.

Costs also matter. Punishment consumes time, energy, and legitimacy, and it can create second‐order collective‐action problems because enforcers bear burdens that others can avoid [24, 28]. Strong cooperation, therefore, does not always come from harsher sanctions. In many settings, it is better to redesign incentives ex ante, through matching rules, verification systems, pricing, defaults, or choice architecture, than to punish defection ex post [14, 17, 29, 30, 31, 32, 33].

This is why I retain a control‐language frame only in a limited sense. Cooperation persists when a system contains feedback that keep opportunism from dominating. Those feedback may be endogenous, decentralized, and incentive‐compatible rather than coercive.

4. Cancer as a Case Study in Multicellular Governance

Cancer is a useful case study because it shows both the need for control and the cost of it. Tumors emerge when cell lineages evade restraints on proliferation, survival, differentiation, and microenvironmental maintenance [34, 35, 36, 37].

But the lesson is not that “bad cells” are metaphysically exceptional. Variation, competition, and local gain are ordinary products of somatic evolution [35, 36, 37, 38, 39, 40]. What matters is whether surrounding architectures keep those dynamics compatible with organism‐level persistence.

Trade‐offs are central. Tumor‐suppressive mechanisms can limit regeneration, promote senescence, or generate chronic inflammatory costs; immune control can improve surveillance while also risking collateral damage and loss of tolerance [37, 41, 42]. The relevant question is therefore comparative: which combination of repair, surveillance, and tolerance best manages risk in a given tissue and life‐history context?

This trade‐off language also helps at the social level. Verification, moderation, auditing, and bureaucracy can protect the commons, but they can also add friction, exclude legitimate participation, or become tools of domination. The issue is not maximal suppression. It is calibrated governance.

5. From Genes to Culture: Institutions as the Human “We”

Human cooperation is distinctive because stabilizers can be represented symbolically. Rules can be taught, justified, contested, and redesigned [43, 44, 45]. That flexibility expands the design space for cooperation, but it also means norms can be hijacked, weaponized, or made brittle by rapid environmental change [29, 30, 45, 46].

Institutions are therefore not just punishers. They also store information, coordinate expectations, reduce transaction costs, and align incentives [14, 17, 29, 30]. Ostrom's design principles are valuable precisely because they combine monitoring and sanctions with boundary definition, conflict‐resolution procedures, and legitimate participation [14].

This framing helps reconcile two empirical realities. People often display cooperative dispositions in experiments and everyday life, yet societies can still drift toward extractive equilibria when monitoring fails, when exit options are unequal, or when concentrated interests repeatedly externalize costs onto diffuse publics [47, 48].

6. Digital Environments and the Information Commons

Digital infrastructures sharpen these issues because they alter population structure, observability, and propagation costs [49, 50, 51, 52, 53, 54, 55, 56, 57]. High‐fidelity, low‐cost copying can reward attention‐capturing variants even when they degrade shared digital goods.

Still, effects are heterogeneous and claims of uniform collapse should be resisted. Evidence linking digital media to polarization, trust, and democratic quality is real but context‐dependent, varying across institutions, media systems, and countervailing safeguards [57].

Here again, the practical question is not simply whether to punish. Provenance cues, rate limits, sharing prompts, audit trails, and accuracy nudges can sometimes outperform blunt takedowns by changing payoffs and reducing accidental amplification [32, 55, 57, 58].

In the language of contagion, many prosocial norms require repeated reinforcement from trusted peers, whereas sensational or incendiary signals often diffuse more easily after sparse exposures [53, 54, 55, 56, 59]. Cooperation‐oriented communication may therefore need deliberate reinforcement rather than moral exhortation alone.

A Vocabulary for a Multilevel Cooperation Lens

Replicator: Any entity that is copied with enough fidelity for selection‐like processes to operate (genes, but also some cultural variants).

Vehicle: The transient structure that carries replicators forward (organisms for genes; organizations and media infrastructures for many cultural variants).

Cheating (free riding): Gaining locally while contributing less to the maintenance of the system that generates the gain.

Stabilizer: Any mechanism that makes cooperation robust by aligning incentives, constraining action, or supporting reciprocal expectations.

Suppressor: A subset of stabilizers that actively reduces the payoff to defection through detection, friction, sanction, or exclusion.

Commons: A shared resource whose benefits are diffuse and whose degradation is individually tempting (institutions, public trust, or information integrity).

Equilibrium: A stable pattern of behavior under given incentives; changing incentives can move the system to a different equilibrium.

7. Implications and Testable Agenda

As a Perspective, this article does not report new experimental results; instead, it synthesizes prior empirical findings across biological, social, and digital systems to define testable predictions. If this Perspective is useful, it should generate hypotheses rather than slogans. One set concerns measurement: which features of institutions or platforms, observability, repeated interaction, switching costs, sanction legitimacy, and auditability, predict whether cooperative equilibria are robust across settings [5, 6, 32, 57, 58, 60]?

A second set concerns intervention: when does ex ante alignment outperform ex post punishment? In digital environments, for example, do rate limits, provenance cues, and partner‐choice mechanisms preserve information quality more effectively than removal alone [32, 40, 55, 57, 58, 61, 62, 63, 64]?

Scope note: I use “selection” and “fitness” metaphorically when discussing institutions and information. Cultural change is also shaped by strategic choice, law, power, and historical contingency. The value of the analogy is not to reduce politics to biology, but to ask when lower‐level incentives overwhelm higher‐level cooperation and which stabilizers restore it [4, 43, 44, 45, 57].

8. Conclusion: Keeping the Future Within the Bounds of the “We”

Taken together, these strands suggest a more modest unifying frame: cooperation can be studied as a control problem without displacing game theory. Repeated‐game analysis explains how cooperation can be self‐enforcing; multilevel evolutionary analysis explains why the conditions for self‐enforcement are often fragile, and why systems repeatedly invest in devices that create observability, patience, alignment, and sanction [5, 6, 7].

The practical agenda is not to fix human nature or to maximize suppression. It is to build institutions in which cooperative behavior is easier, cheating less profitable, and enforcement costs tolerable [14, 17, 29, 30, 32, 57].

That agenda requires attention to trade‐offs. Suppressors can be captured by elites, moderation can be weaponized, bureaucracy can slow renewal, and friction can exclude as well as protect [29, 30, 57]. The goal is therefore calibrated, revisable governance rather than blanket control. The “we” is neither a natural default nor an impossible ideal. It is a conditional achievement.

Author Contributions

Michel Salzet: conceived the argument and wrote the article.

Funding

This research was supported by grants from the ministère de l'Enseignement Supérieur et de la Recherche (MESR), Institut national de la recherche médicale et de la santé, Région Hauts‐de‐France, University of Lille, Institut Universitaire de France, and La Ligue Contre le Cancer (équipe labellisée 2024).

Conflicts of Interest

The authors declare no conflicts of interest.

Use of Generative AI and AI‐Assisted Technologies in the Writing Process

Generative AI tools were used for language polishing and structural editing; the author reviewed and takes responsibility for the final content.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.