Once light has been shed on their epistemological differences, three interpretations of methodological individualism (MI) can be distinguished: the “constitutive” version, the “minimalist” version and the “reductionist” version. The first, originally theorized by Max Weber, requires that social phenomena should be explained based on a (generally rational) interpretation of the subjective meaning of individual actions. The second, founded on a broad conception of individual action, is essentially opposed to the misuse of collective concepts. It tends to only find an ontological justification with no methodological legitimacy, except for that found in the empiricist bias of the agent-based modelling promoted by analytical sociology. The reductionist version, which dominates contemporary Anglo-Saxon philosophy, is for its part a pure philosophical construct resulting from the layered interpretation of science and the world inherited from positivism. It is shown that the constitutive, or Weberian, version shares its premises with active epistemologies, bringing into play capacities or causal powers with transsituational properties. Émile Meyerson’s anthropology of knowledge allows us to offer a ultimately non-metaphysical justification of this version.