I think part of the problem here is the vagueness of the term "cognition". Levin has chosen to adopt terms for higher-level competencies like "decision-making" , "memory", and "problem solving" toward explaining very simple adaptable behaviors. Are the invariant features across unicellular organisms and complex multicellular organisms all the way up to humans? I think this is a legitimate empirical question that need not invoke "magic". Personally, I think a cell seems to structurally instantiate certain causal dynamics that form a precondition for more complex forms of cognition, enabling the emergence of nervous systems and stratified structures within these that enable stratified information processing with much greater computational power. I'm not sure though that Levin has quite identified the invariants he seeks in morphogenesis. As a biologist, I'd be curious to know what you think of the concept of autopoiesis and whether you find it useful as a formal stepping stone in filling in some of the explanatory gaps in biology?