Contribution
The chapter (LIT-503) turns a commonplace into a taxonomy. Its Table 1 collects fifteen entries, from Aristotle to Greene, who said that morality has something to do with cooperation. It then uses the game theory of cooperation to say which problems there are, and so how many moral domains to expect: seven, one per game. Each domain is laid out the same way: the game, the animal evidence, the human adaptations, the cultural inventions, and quotations showing that moralists have praised the solution. Table 2 is the “periodic table of ethics”. Tables 3 and 4 argue that earlier reviews of cooperation and earlier moral taxonomies (Fiske’s relational models, Shweder’s CAD triad, Haidt’s moral foundations) each omit or conflate some of the seven.
Key insight
If morality is the set of solutions to cooperation problems, then the list of moral domains is not an empirical inventory to be gathered but a consequence of the list of games. That gives the theory its reach and its risk. A moral domain that corresponds to no game is a counterexample, and the chapter names its candidates in advance (p. 40).
Assumptions
- Identity, not cause. The solutions to cooperation problems “constitute human morality” (p. 29). “Morality” is the label philosophers have attached to them. This is a stipulation about what the word picks out. It leaves Bloom’s question in NOTE-034, why cooperation is moralized rather than merely liked, outside the theory’s scope.
- Completeness of the game list. The seven are the problems game theory had identified by 2016. The chapter expects new games to add domains (“extending the foundations”, p. 40).
- Two levels at once. “Biological and cultural solutions” are treated together. A domain counts whether it is served by an evolved mechanism or by an invented institution, such as land registries for possession.
- Positive evidence is illustrative. Every “there is evidence to suggest that they are” (pp. 30–37) is followed by quotations from moralists and ethnographers, chosen to show that a solution has been praised. No search for contrary cases is described.
Key results
The chapter reports no study. Its results are its taxonomy and its stated predictions.
- Seven problems (pp. 30–37): kin allocation (Hamilton), coordination to mutual advantage (Lewis, Schelling, Skyrms), exchange (Trivers, Axelrod), and conflict resolution by hawk–dove contests (Maynard Smith and Price), division (Nash, Skyrms) and possession (Gintis, Maynard Smith).
- Seven moral domains (Table 2): obligations to kin and the incest prohibition; friendship, loyalty and conformity; reciprocity with trust, gratitude, guilt, apology, forgiveness and patience; hawkish virtues (bravery, fortitude, generosity); dovish virtues (humility, respect, obedience); fairness and compromise; property rights and the ban on theft.
- Predictions (pp. 39–40):
- the seven are morally good and their failures morally bad;
- “behaviour that has nothing to do with cooperation … will be regarded as morally neutral”;
- the seven are good “in every human culture, at all times and in all places”, with no culture where morality “is about something other than cooperation — say, aesthetics or nutrition”;
- societies weight the domains by the problems they face (family size, warfare, inequality);
- sexual, political, international and religious morality will turn out to be cooperative. Religious morality here includes “ancestor worship, food taboos, karma, reverence”.
- Against rival taxonomies (pp. 42–43): Fiske, Shweder and Haidt conflate kinship with mutualism and exchange with division, and omit hawkish traits and possession. MFT’s purity domain has “no apparent connection to cooperation” (p. 42).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Morality is a collection of biological and cultural solutions to recurrent problems of cooperation and conflict | assertion (a definition plus a research programme) | argued from the history of cooperative accounts (Table 1) and from game theory; no test |
| C2 | Game theory identifies seven distinct cooperation problems, so there are seven moral domains | moderate as game theory, weak as a count of domains | the games are standard; why these seven and only these is not argued, and the chapter expects more |
| C3 | Each of the seven is regarded as morally good | weak–moderate | quotations selected to agree; no search for contrary cases is described |
| C4 | Non-cooperative behaviour is regarded as morally neutral, in every culture | untested prediction | stated (p. 40); the later tests (LIT-047, LIT-496) do not code non-cooperative content |
| C5 | Sexual, religious and dietary morality will turn out to be cooperative | conjecture | stated as a prediction (p. 40); this also absorbs the cases C4 would count against the theory |
| C6 | MAC is better than relational models, the CAD triad and MFT, which conflate, omit and misconstrue types of cooperation | informal argument | Table 4; a comparison of category lists, not of predictions |
Concepts
- Problem-centred approach: deriving moral domains from a list of cooperation problems rather than from a survey of moral content.
- Periodic table of ethics: Table 2, problem × theory × animal examples × human examples × morals.
- Hawkish and dovish virtues: the two halves of the display-and-defer strategy, after Hume’s “heroic” and “monkish” virtues (p. 35).
- Tools and rules: cultural inventions that improve on natural selection’s solutions (p. 29).
Connections
- The test. Curry, Mullins and Whitehouse (LIT-047) test C3 in sixty societies. NOTE-034 judged that test weakly diagnostic because nothing non-cooperative was coded. This chapter shows that the omission matters on the theory’s own terms: C4 is the theory’s stated prediction, and the test left it out.
- The questionnaire. The MAC-Q (LIT-496) operationalizes the seven domains and, in its MFQ studies, finds that purity items are rated morally relevant, which its authors concede “appear[s] to contradict” the prediction that purity will not be.
- Purity. Fitouchi, André and Baumard (LIT-101) argue that puritanical norms are cooperative after all, because indulgence is believed to erode the self-control cooperation needs. That is a C5-style absorption, and the first one in the record that is worked out and tested. This chapter already lists patience among the components of reciprocity, citing Curry, Price and Price (2008) on cooperative people’s lower discount rates (p. 33). That gives LIT-101’s mechanism a foothold inside the exchange domain. LIT-101 itself reduces puritanism to harm and fairness concerns, and the MAC-Q argues that harm is not a domain. The two accounts therefore agree that purity is cooperative and disagree about where it sits.
- Coordination. The coordination domain rests on Lewis’s Convention (LIT-473), which the record holds unread.
- Obligation. Tomasello appears in Table 1 only through Tomasello and Vaish (2013). His “The moral psychology of obligation” (LIT-125) gives an account of why obligations feel binding, which is the proximate story this chapter does not give.
- Why cooperation pays. Peters and Adamou (LIT-035) offer a different, variance-reduction account of why cooperation pays. This chapter’s games are the standard expected-payoff ones, and it does not engage that alternative.
- Not held here. Haidt and Graham’s moral foundations, Fiske’s relational models, and the evolutionary sources (Hamilton, Trivers, Axelrod, Nowak’s five rules in Table 3) are being filed in parallel by other contributions.
Bearing on the record
There is nothing here for ML practice, and the anthology does not hold the chapter. For the morality-as-cooperation thread it is the source to cite for what the theory predicts. Two of those predictions go beyond what LIT-047 and LIT-496 test: that non-cooperative content is morally neutral, and that nothing outside cooperation is moralized in any culture. A THEORY document built on MAC should quote p. 40, and should say that no study in the record has tested it yet.
Note of 2026-10-16: THEORY-468 and THEORY-632 now cite LIT-503 as a source.
Limitations
- No data. Every “there is evidence to suggest” is a set of selected quotations.
- The falsifier is pre-absorbed. The same page that names “nutrition” as a domain no culture’s morality will be about predicts that food taboos will turn out cooperative (p. 40). Until the cooperative account of a given taboo is written and tested, the prediction cannot fail.
- The count of seven is open-ended. New games add domains, so finding a further cooperative domain does not count against the theory. Only a non-cooperative domain would.
- The definition carries the theory. Because morality is identified with the cooperative solutions, the chapter has no account of what makes a solution moral rather than merely preferred or useful. That is Bloom’s question in NOTE-034, and the chapter does not address it.
Open questions
- Coded blind and in the same sample, are non-cooperative domains (purity, sexual conduct, treatment of the dead, harm to strangers) moralized as often as the seven? This is NOTE-034’s open question. This chapter makes it the theory’s own test.
- For which food taboos, if any, has a cooperative account been written and tested?
- Does domain weighting track the ecological variables named on p. 40?
Corrections
- none to a seeded skim (there was no seed)
- The citation. The owner’s citation (in Shackelford and Hansen, eds., Springer, pp. 27–51) is correct. The volume is in Springer’s Evolutionary Psychology series. The copy on Curry’s site is not a preprint. It is the publisher’s typeset chapter.
- “Cooperation and conflict”. The chapter’s definition is solutions “to the problems of cooperation and conflict” (pp. 29, 44). Later papers in the programme drop “and conflict” (LIT-496, abstract), though conflict resolution is still three of the seven domains.
- The hawk–dove attribution. The chapter credits hawk–dove contests to Maynard Smith and Price 1973. That paper (LIT-505, NOTE-404) has no Hawk–Dove game: its strategies are Mouse, Hawk, Bully, Retaliator and Prober-Retaliator, between identical contestants. It also gives Hamilton 1964’s pages as “1–16, 17–52” with only Part II’s DOI.