6 ideas
A social convention is fragile while people see it as man-made and arbitrary. It becomes a stable institution when it is grounded in an analogy to nature, as when a division of labour is mapped onto left and right hands or male and female bodies. Once that analogy holds, the arrangement feels like the way things are, and challenging it feels like challenging reality rather than a rule.
Sameness is not given by objects themselves. Any two things are alike in some respects and different in others, so someone must decide which resemblances matter. Institutions settle this by fixing the categories that members then treat as natural kinds, and people come to fit and inhabit those labels, so the classifications partly create the things they claim to describe.
Institutions store the memories that support their current structure and systematically drop what does not fit. Genealogies get telescoped to match present alliances, and scientific fields forget precursors whose work lacked a place in the reigning theory. Public memory is therefore a filtered product of the institution's needs, not a neutral record.
Rational-choice accounts predict free-riding and so cannot explain why people sacrifice for groups. Douglas argues that cooperation is sustained by the shared classifications and thought style an institution instills, which define what individuals count as their interest in the first place. Small voluntary groups without coercive power hold together by making exit the main threat, which pushes them toward strong boundaries and suspicion of outsiders.
Treat an individual's perceptions, judgments, and sense of what is obvious as largely supplied by the thought style of the collective they belong to. The mind thinks with institutional categories it did not choose and mostly cannot see. It then experiences the conclusions those categories produce as its own independent reasoning.
In a hypothetical case, trapped cave explorers kill and eat one of their party to survive. Judges reach opposite verdicts on it, and each verdict follows from a different institutional framework: positive law, natural law, or the explorers' own compact. The case shows that there is no institution-free standard of justice, because life-and-death judgments of fairness are constituted by the institutions within which people reason.