Ontological and methodological constraints on a theory of cognition that would generalize across species are identified. Within these constraints, ecological arguments for animal-environment mutuality and reciprocity and the necessary specificity of structured energy distributions to environmental facts are developed as counterpoints to the classical doctrines of animal-environment dualism and intractable nonspecificity. Implications of and for a cognitive theory consistent with Gibson's programme of ecological psychology are identified and contrasted with contemporary cognitivism.