Contribution
A synthesis with a staged model (Fig. 2), and no new data. It puts two contrasts side by side. One is dyad against larger group (Table 1). The other is cooperation within groups against cooperation between them. It separates mistrust, distrust, hostility, aggression and hate as stages, and keeps in-group favouritism apart from costly competition.
Key results
- The thesis. “cooperation is often strong in dyads but more fragile between groups” (p. 547).
- Dyads beat groups, with counter-evidence acknowledged. “dyads are more cooperative than groups in single-trial social dilemmas” (p. 550). The reasons are “reciprocity, empathy and perspective-taking, and psychological safety” (p. 550). Exclusion costs both partners: “one cannot exclude the other person in a dyad without bearing the similar cost of cut ties” (p. 551). Large groups “rely on institutions to promote and maintain cooperation” (p. 551).
- Mistrust from categorization alone. It “can arise just from categorization and unfamiliarity with outgroup members” (p. 552).
- Bounded reciprocity. “Ingroup favouritism primarily stems from the heuristic expectation that cooperation is reciprocated and rewarded exclusively by ingroup members” (p. 553).
- The interindividual–intergroup discontinuity effect. Individuals nested in groups are less cooperative towards outsiders than they would be by themselves. The in-group supports their non-cooperation and spreads responsibility for it, and reputational checks are weak (p. 553). One mechanism among about seven, not the centrepiece.
- Favouritism is not harm. “such ingroup favouritism need not imply that people show any harmful behaviour towards outgroups” (p. 553).
- Conflicting interests drive escalation, as perceived. Costly competition “becomes more common in situations where the interests of both groups conflict” (p. 554). There, “ingroup cooperation can require disadvantaging outgroup members” (p. 554). What counts is “people’s perceptions of their interdependence” (p. 554).
- In-group cooperation fuels intergroup conflict. “the same processes that can establish ingroup cooperation can also fuel intergroup conflict” (p. 554). Peers punish members who will not invest in the conflict. Peer punishment “fails to inhibit self-serving costly competition towards the outgroup by a small subset of the ingroup” (p. 554), and retaliation then targets the whole group.
- Institutions as a precondition for peace. The review reports, from ref. 206, that “a prerequisite for the emergence of peaceful intergroup relations is the establishment of more centralized institutions at the level of the group that could constrain such behaviours by individual group members” (p. 554).
- Not inevitable. Groups “compete, cooperate or coexist independently” depending on the situation, expectations, and “the ability of groups to prevent proactive aggression against other groups” (p. 555).
- Group-based morality. It “is primarily defined in terms of the quality of outcomes for one’s own group, with little or no regard for the quality of outcomes for outgroups” (p. 555).
- Remedy, and how it could end. Cooperative cross-group dyads, with reputations spreading, “as long as individuals are not punished for friendly relations with outgroup members by their peers” (p. 555). Frequent cooperation “might even lead the groups to merge into larger collectives that transcend former group boundaries” (p. 555).
Assumptions
- Cooperation in the economic-game sense: “providing good outcomes for others at some cost to oneself” (p. 547). Laboratory games and minimal groups are taken to bear on real intergroup relations.
- “Among individuals” is mostly dyads, so the model joins two comparisons: group size and the group boundary.
- In-group favouritism is close to a default, and expected reciprocity matters more than identity or esteem (p. 553).
Concepts
- schema-based mistrust against evidence-based distrust (p. 552);
- bounded reciprocity (p. 553);
- interindividual–intergroup discontinuity (p. 553);
- negative interdependence (p. 554);
- psychological safety, in dyads: no exclusion, no subgroups (p. 551);
- group-based against individual-based morality (p. 555).
Bearing on the record
Source of THEORY-1352.
- The synthesis’s §7 is right in substance, with one refinement. Groups cooperate less with each other even when their interests do not conflict: categorization and bounded reciprocity are enough. Conflicting interests, and above all perceived ones, are what turn lower cooperation into costly competition and hostility (pp. 553–554).
- CLAIM-547. Supports one link: “the same processes that can establish ingroup cooperation can also fuel intergroup conflict” (p. 554). The review is about cooperation and conflict between groups, not their persistence. It says nothing on extinction, Nowak’s rule or impartial concern, beyond the contrast of group-based and individual-based morality. That contrast backs the claim’s remark that THEORY-468’s group-loyalty domain can work against the coexistence of groups. The review’s own success case, groups merging into larger collectives, is a second route to fewer groups: by merger, not extinction. The claim’s note of 2026-11-04 records this.
- QUESTION-371. No answer. There is no norm that bears harder on a dominant group. The nearest it comes is institutions within a group that restrain its own aggressive minority (p. 554, ref. 206), and intergroup tolerance as risk buffering (refs. 227–229). Power asymmetry appears only as one group “more deprived of resources or status” (p. 554), a source of negative interdependence.
- THEORY-1114 (self-categorization). A loose fit. The review uses categorization and fit, but not metacontrast or a varied frame of reference, so it does not bear on that theory’s promote_when. Its account of favouritism rests on reciprocity rather than identity.
- THEORY-1112 (entitativity). Indirect. Minimal-group favouritism from a shared preference is compatible with a mental-state cue, but the review does not separate similarity from shared intention.
- CLAIM-493. A different route by which a minority sets a group’s outward stance: unchecked aggression by a few, conformity, and retaliation against the whole group (p. 554). That is not prototypicality, so it is a rival or complement, not evidence.
Limitations
- A Perspective. The literature is the authors’ selection, and no effect sizes are pooled. The stages of Fig. 2 are proposed and not tested together.
- The dyad advantage has counter-evidence, which the authors cite (p. 550). Extortion and exploitation are also easier in dyads (p. 551).
- The evidence is mostly laboratory games. The applications to war and politics come from cited work.
- The cross-group-dyad remedy is untested, and the authors name its limit: the partner may be taken as “an exception to the rule” (p. 555). CASE-146 is that case.
- The copy read may be a late proof.