Multi-level Selection

Multi-level selection (MLS) is a framework in evolutionary biology that treats natural selection as operating simultaneously at multiple hierarchical levels — genes, cells, organisms, groups, and species — rather than exclusively at the level of the individual organism or the gene. The framework holds that the unit of analysis in evolution is not fixed: selection pressure can be partitioned across levels using the Price equation, and the empirical question is which levels contribute meaningfully to the direction and rate of evolutionary change in a given case.

The key distinction, formalized by Samir Okasha, is between MLS1 (selection among individuals within groups, where group membership affects individual fitness) and MLS2 (selection among groups as collective units, where groups reproduce differentially and their offspring groups are recognizably descended from parent groups). Altruism, cooperation, and division of labour are most naturally explained by MLS2 — the case where the group as a whole succeeds or fails as an entity, not merely as an environment for individual competition.

The relationship between MLS and inclusive fitness theory is the most contested question in modern evolutionary biology. The gene-centric view (associated with W.D. Hamilton, John Maynard Smith, and Richard Dawkins) holds that the two frameworks are mathematically equivalent for additive fitness effects — they are different bookkeeping systems for the same underlying causal process, and the gene-level account is more parsimonious. D.S. Wilson and E.O. Wilson argue that the frameworks are not equivalent under non-additive fitness functions, and that MLS provides the more natural account of major evolutionary transitions (from prokaryotes to eukaryotes, from single cells to multicellular organisms, from solitary animals to supercolonies) where the transition itself is a shift in the relevant unit of selection.

The most important application of MLS to human evolution is the cultural group selection hypothesis: that cultural variants (norms, practices, beliefs, institutions) are transmitted within groups more readily than between groups, creating the conditions for selection to act on groups as units. If true, this explains human prosociality, large-scale cooperation among non-kin, and the co-evolution of genetic and cultural dispositions toward group-level behaviour — without requiring implausibly high genetic relatedness among cooperators.

The skeptic's note: the debate between MLS and inclusive fitness has produced more heat than light partly because both sides have conflated the mathematical question (are they equivalent?) with the explanatory question (which framing better guides research?). These are separate questions, and the answer to the second does not follow from the answer to the first.