Science Undermines Itself When It Denies Higher Levels of Reality

A performative contradiction sits at the heart of eliminative reductionism: the claim that only fundamental physics is ontologically real delegitimizes the very epistemic processes — measurement, inference, communication — through which that claim is produced. Scientists operate at macroscopic levels of organization. If those levels are epiphenomenal, then the knowledge generated there about the microphysical has no epistemic warrant. The argument is self-defeating in a precise logical sense: it uses the causal efficacy of higher-level processes to deny that higher-level processes are causally efficacious.

Erik Hoel's work in causal Emergence, grounded in Shannon Information theory, provides formal teeth for this intuition. Reliable information transfer requires redundancy and error correction — operations that abstract away from the full microstate. A purely bottom-level communication channel is overwhelmed by noise; only by compressing into higher-level macrovariables can a system achieve effective causal coordination. This is not an observer-dependent choice of description. The higher-level variables possess causal powers — measured by effective information — that the lower-level variables provably lack. Emergence, on this account, is mathematically demonstrable, not merely philosophical assertion.

Ned Block's regress argument sharpens the point: reductionism offers no principled stopping place. Neurons reduce to molecules, molecules to atoms, atoms to subatomic particles, particles to quantum fields, and at the putative absolute bottom — the pre-Big Bang singularity — physical law itself becomes undefined. A coherent naturalism must therefore be a layered naturalism, one that grants genuine Ontological standing to multiple levels of causal organization. Extended naturalism, understood this way, is not an expansion beyond science but a correction demanded by science's own operational presuppositions.