For more than a century, a quiet assumption has governed much of scientific and philosophical thought: that the fundamental entities described by physics — quarks, leptons, fields, forces — are all that really exist. Everything else, from the wetness of water to the redness of a sunset to the pangs of conscience, is just those micro-level entities doing their thing in complicated arrangements. It’s a tidy picture. It’s also, according to a growing body of philosophical argument, almost certainly wrong — or at least radically incomplete.
The thesis is called microphysicalism, and its most rigorous contemporary critique comes from philosopher Edward Feser, whose detailed paper on the subject mounts a case that the position is not merely unproven but incoherent in ways that working scientists and technologists rarely confront. Feser’s argument isn’t anti-science. It’s anti-reductionism of a particular and pervasive kind — the kind that says the micro-level is the only real level, and that macro-level phenomena are either identical to or wholly determined by microphysical goings-on.
This matters far beyond philosophy seminars. It matters for artificial intelligence research, for neuroscience, for anyone building systems premised on the idea that if you model the small parts precisely enough, the big picture takes care of itself.
The Structure of the Problem
Microphysicalism, as Feser frames it, comes in several flavors. The strongest version — eliminativism — says that only microphysical entities exist and that talk of tables, minds, and organisms is just a convenient shorthand. A softer version holds that macro-level things do exist, but they’re nothing “over and above” their microphysical constituents; they supervene on them, meaning no macro-level difference without a micro-level difference. Still another version says macro-level properties are real but causally inert — all the causal work happens at the bottom.
Each version faces trouble.
Start with eliminativism. If only microphysical entities exist, then microphysicalism itself — a thesis held by human minds, expressed in language, defended by arguments — doesn’t exist. There are no arguments at the microphysical level. There are no truths or falsehoods among quarks. The position is self-undermining in the most direct way possible: it eliminates the very cognitive and semantic apparatus required to assert it. Feser draws on a long tradition here, from Aristotle through the twentieth-century philosopher Elizabeth Anscombe, noting that rational inference requires the existence of genuine logical relations between propositions. Propositions aren’t microphysical entities. Neither are logical relations. If eliminativism is true, then no one has ever believed anything, including eliminativism.
This isn’t a parlor trick. It’s a structural problem that has dogged reductive materialism since at least the logical positivists. And it has never been satisfactorily resolved.
The supervenience version looks more promising but runs into what philosophers call the “exclusion problem” in reverse. If macro-level properties are real and supervene on micro-level properties, what explanatory work are the macro-level properties doing? If they’re causally redundant, then the sciences that traffic in them — biology, psychology, chemistry — are describing epiphenomena. But no working biologist thinks natural selection is epiphenomenal. No chemist thinks valence bonds are merely along for the ride. The actual practice of science presupposes that higher-level properties have genuine causal relevance. Microphysicalism, taken seriously, tells those scientists they’re wrong.
Feser’s paper pushes harder. He argues that the mathematical and structural character of microphysics is itself a reason to doubt that it gives a complete account of physical reality. Modern physics describes entities in terms of their relational and dispositional properties — an electron is characterized by its mass, charge, spin, and how it interacts with other entities. But what is the electron, intrinsically? Physics doesn’t say. It gives you the equations governing behavior. It doesn’t give you the nature of the thing behaving.
This observation has been made by figures as varied as Bertrand Russell, Arthur Eddington, and more recently Galen Strawson. Russell pointed out in The Analysis of Matter (1927) that physics tells us about the causal structure of the external world but is silent on its intrinsic character. Eddington, the physicist who confirmed general relativity, wrote that physics gives us “pointer readings” — quantitative relationships — but not the qualitative stuff those readings are about.
Feser connects this structural gap to the classical philosophical tradition, particularly to Aristotelian-Thomistic metaphysics. On that view, the microphysical level is an abstraction from a richer reality that includes formal and final causes — the organizational principles and goal-directedness that make a living organism something more than a collection of particles. This isn’t vitalism. It’s the claim that the organizational structure of a thing is as real and causally relevant as its material composition.
The implications for AI and cognitive science are direct. If minds aren’t reducible to microphysical processes, then simulating microphysical processes — or even higher-level neural processes — may not be sufficient to produce minds. The so-called “hard problem of consciousness,” articulated by philosopher David Chalmers in the 1990s, is a symptom of the same underlying issue. You can describe every physical process in the brain and still face the question: why is there subjective experience at all?
Recent debates in AI have circled this question without quite landing on it. Large language models produce outputs that mimic understanding. But mimicry and understanding are different things — unless you’re already committed to the view that the micro-level (in this case, matrix multiplications and weight adjustments) is all there is. If Feser and the anti-reductionists are right, that commitment is a philosophical choice, not a scientific finding.
And it’s a choice with consequences. Engineers building autonomous systems need to know whether the systems they’re building can, even in principle, possess the kind of understanding that would make them genuinely reliable agents rather than sophisticated pattern-matchers. The answer depends, in part, on whether microphysicalism is true.
Feser also addresses a common dodge: the claim that future physics will eventually explain everything that current physics can’t. This, he argues, is not a scientific prediction but a promissory note — and one that has been outstanding for a very long time. Every generation of physicists has discovered that the fundamental level is stranger and more incomplete than the last generation thought. Quantum mechanics introduced indeterminacy. Quantum field theory dissolved particles into field excitations. String theory, if it ever becomes testable, would dissolve those into vibrating strings. At no point has the progression converged on a theory that explains, say, why water is wet or why pain hurts. The promissory note keeps getting extended.
None of this means physics is wrong. Physics is extraordinarily successful at what it does. But what it does, Feser contends, is describe one aspect of reality — the quantitative, structural, mathematically expressible aspect. Mistaking that aspect for the whole of reality is like mistaking a blueprint for a building.
The philosophical literature on this topic is vast and growing. Recent work by philosophers like Philip Goff at Durham University has explored “panpsychism” — the view that consciousness is a fundamental feature of matter — as one possible response to the incompleteness of microphysicalism. Others, like Tim Maudlin at NYU, have pushed back, defending a version of physicalism that takes the structure of physical law more seriously as an explanation of higher-level phenomena. The debate is active, contentious, and nowhere near settled.
For the technology industry, the stakes are practical. Billions of dollars are flowing into AI systems built on the implicit assumption that intelligence is substrate-independent — that if you replicate the right computational patterns, you get the real thing. That assumption may be correct. But it rests on a philosophical foundation that serious thinkers have been contesting for decades, and the contest isn’t going well for the reductionists.
So the next time someone tells you that everything is “just physics,” ask them a simple question: which physics? The physics we have, which explicitly disclaims knowledge of intrinsic natures? Or the physics we hope to have someday, which no one can describe? Either way, the claim that microphysics explains everything is not a conclusion drawn from evidence. It’s a faith commitment dressed in a lab coat.
And faith commitments, however widespread, deserve scrutiny — especially when they underwrite trillion-dollar industries.
WebProNews is an iEntry Publication