The causal closure of the physical (CCP) is the claim that every physical effect has a sufficient physical cause. That is, the physical world, in causal terms, is a closed system. Nothing non-physical reaches in to push the particles around. Whatever happens physically can be fully accounted for (in principle) by other physical events.

The thesis has been around in some form since the 17th century, when the mechanical philosophy crowd (Descartes, Hobbes, Boyle, Newton) started imagining the universe as a vast clockwork. Descartes himself ended up famously stuck on the question, because his commitment to an immaterial mind required some way for that mind to interact with the body. He took a swing at it with the pineal gland, which has gone down as one of philosophy’s longest-running punchlines. (I struggled to avoid putting a Descartes before the horse pun here.)

The early modern era produced a surprising variety of attempts to keep both a closed physical world AND a non-physical mind without direct interaction. Leibniz’s pre-established harmony, Spinoza’s parallelism, Malebranche’s occasionalism - each was, in its own way, ingenious. (And no, I don’t hate occasionalism, mostly because it gets to the answer “God does it” in a way that’s at least honest about the move). They were also, by later standards, deeply unsatisfying as scientific or metaphysical proposals. By the 19th century, most serious thinkers had concluded that the cleanest move was to drop the non-physical mind altogether.

The modern, sharper formulation of CCP comes mostly from David Papineau, who argued in the 1990s and 2000s that 20th-century physiology has effectively settled the question empirically. Every nerve impulse, muscle contraction, and neurochemical cascade we’ve measured turns out to be fully accounted for by physical and chemical mechanisms. Where we’ve looked carefully, we haven’t found gaps that require introducing non-physical intervention. Jaegwon Kim then made the philosophical machinery around CCP famous with his “causal exclusion argument,” which uses CCP to argue that mental properties either reduce to physical properties or don’t really do any causal work.

Now, why does any of this matter? Because CCP is the load-bearing wall of physicalism. If it’s true, the universe doesn’t have any spooky causal back doors. Mental causation has to be physical causation in some form. Free will, if it exists at all, and even if it can be framed in compatibilist terms, has to be a physical phenomenon. There’s no place for ghosts, élan vital, divine intervention, or any other ingredient that exists outside the physical order. CCP is what allows physicalism to be a substantive thesis rather than a holey gesture (bad pun intended).

It’s also contentious for the same reason. CCP rules out a lot of things many people are very attached to. So, it has attracted surprisingly high-quality skeptical attention from philosophers (Tim Crane, E.J. Lowe, Jessica Wilson, Helen Steward) and a few prominent physicists (Roger Penrose, Henry Stapp, George Ellis - and historically, Eugene Wigner and von Neumann). The most famous theoretical attack is Hempel’s dilemma. The most popular empirical attack is the appeal to quantum measurement and indeterminacy. The most sophisticated contemporary attack is Russellian monism, which is rather different from the others and which I’ll treat as the strongest opposition. The most rhetorical attack is the appeal to consciousness.

My claim is that none of these attacks ultimately succeeds, though some of them (especially Russellian monism) deserve more careful engagement than they typically get from physicalists. CCP is on solid ground. Let me explain why.

The argument I’m going to make has two parts. The first is “methodological” - that physicalism, properly understood, is a coherent thesis that survives the standard theoretical objections. The second is “empirical” - that the actual scientific evidence overwhelmingly supports the CCP. These do different work, and I’ll try to keep the seam visible rather than letting them blur together.

1. Hempel’s Dilemma is Overrated

The most famous theoretical objection to physicalism is Hempel’s dilemma. The argument runs:

When you say “every physical effect has a sufficient physical cause,” what do you mean by “physical”?

Option A: You mean current physics. In that case, your claim is almost certainly false. Current physics is incomplete - we have no UV-complete theory of quantum gravity, the Standard Model doesn’t describe dark matter or dark energy (which together make up roughly 95% of the cosmic matter-energy budget), the measurement problem in quantum mechanics is unresolved, etc.. Saying CCP holds with respect to current physics is like saying it held with respect to Newtonian mechanics in 1890. The history of science laughs at you.

Option B: You mean a future, completed physics - whatever a finished science of the physical world will look like. In that case, your claim is true by definition but completely empty. “Whatever turns out to cause physical events will turn out to be physical, because we’ll define it that way.” That’s not a thesis, that’s a tautology.

This dilemma sounds devastating in the abstract. It dissolves on close inspection.

Physicalism, properly understood, is NOT a claim about the “content” of any specific physical theory. It’s a claim about the “methodology” and “idiom” of physical theorizing. What EVERY successful extension of physics has shared - from Newtonian mechanics to general relativity to quantum field theory to the Standard Model to whatever quantum gravity ultimately is - is a particular way of describing the world: structural, mathematical, third-personal, mind-independent, quantitative. The world gets carved up into entities with measurable properties whose dynamics are governed by formalizable laws. That’s the recognizable shape of physics, and it has actually been remarkably stable through every revolution.

So Hempel’s dilemma is attacking a strawman. Physicalists aren’t betting on the truth of the Standard Model as it stands in 2026. They aren’t making the trivial claim that “physical = whatever physics says.” They are making the substantive claim that the world is exhaustively describable in this kind of idiom. That claim has real content. It rules out things. It rules out, for instance, fundamental teleology, irreducible mentality, or Cartesian souls. None of those fit the methodological mold, regardless of how physics evolves.

And here’s the inductive payoff. Every revision of physics in the last 350 years has stayed within this methodological framework. Going from Newton to Einstein didn’t require positing souls. Going from classical to quantum didn’t require positing élan vital. Adding gauge fields, supersymmetry, inflation, neutrino masses, dark matter candidates - all of these stayed inside the structural/mathematical idiom. The track record is unbroken. A skeptic who insists future physics will need to break this pattern owes us a positive argument and some actual evidence. Neither has been forthcoming.

This is sometimes called “methodological physicalism” or “structural physicalism,” and figures like Barbara Montero and Jessica Wilson have spelled it out more carefully than I’m doing here. The point is that Hempel’s dilemma only bites a particularly naive strawman formulation of physicalism. The grown-up version is fine.

What this argument does not do, however, is provide empirical support for CCP. It defends the “coherence” of the physicalist thesis against the charge of triviality. The empirical case is a separate matter, which I’ll get to in section 4. I want to flag this seam now because it matters - methodological physicalism on its own doesn’t tell you the world is causally closed, only that the claim is “meaningful”. The evidence has to do its own work.

2. Quantum Measurement is Not a Back Door

Probably the single most popular move in (undergraduate-level) skepticism about CCP is the appeal to the quantum measurement problem. The argument runs something like: “But quantum mechanics is indeterministic! And the wave function collapses when an observer looks at it! Doesn’t that mean physics isn’t causally closed?”

No. Let me explain why.

First, indeterminism is not a problem for CCP. The thesis can be (and standardly is) formulated probabilistically: every physical effect has its complete causal antecedents in the prior physical state, even if those antecedents are stochastic rather than deterministic. Quantum mechanics specifies the probability distributions of possible outcomes with extraordinary precision. The fact that the outcomes themselves aren’t deterministically settled doesn’t introduce non-physical causation. It just means the physical causes are stochastic.

Now, there’s a real worry worth flagging here. Tim Maudlin and others have argued that “probabilistic sufficient causation” isn’t a coherent notion - that if your causes don’t fix the outcome, they aren’t sufficient in any robust sense. I think the worry is real but not fatal. The work CCP needs done is “nothing physical is left out and nothing non-physical is added,” not “the outcome is determined.” Probabilistic closure is closure in the relevant sense: the universe, as far as causal influence is concerned, has no outside. The dice may be loaded by chance, but no one outside the universe or a non-physical entity inside the universe is rolling them.

Second, every serious interpretation of quantum mechanics is a “physical” theory. Many Worlds denies collapse altogether and treats the wave function as the complete description of reality. Bohmian mechanics adds deterministic guiding equations and is fully physical. GRW and other objective collapse theories propose specific physical mechanisms that produce collapse without any observers in the picture. Decoherent histories explains the appearance of classicality through environmental entanglement, a physical phenomenon. Each of these interpretations/theories has internal problems (the probability problem in Many Worlds, relativistic generalization in Bohmian mechanics, the tail problem in GRW, etc.), and the debate over which is correct is genuinely unresolved. But all of them are within the structural/mathematical idiom of physics and physicalism. The measurement problem is a problem “for” physics, not evidence “against” the sufficiency of physics.

Third, the Wigner-von Neumann interpretation (which proposes that a conscious observer plays a special causal role in collapse) is the one interpretation that “might” threaten CCP. It deserves a fairer hearing than it usually gets. Philosophers like David Albert and Tim Maudlin take measurement seriously enough to think the deflationary “decoherence solves everything” response is too quick. They’re right that decoherence by itself doesn’t fully resolve the measurement problem; it explains why we don’t observe macroscopic superpositions while leaving open what happens to the unobserved branches.

So why reject Wigner-von Neumann? Not because it doesn’t make sense. But because it carries metaphysical costs the alternatives don’t. It requires drawing a principled line between “conscious” and “non-conscious” systems, with the former having the unique capacity to collapse wave functions. No physical theory tells us where that line falls. (Is a dog conscious enough? A rat? A thermostat? An LLM?) The Wigner interpretation has to take consciousness as a primitive in physics, which is the sort of move you reserve for cases where nothing else works. Many Worlds, Bohmian mechanics, and objective collapse theories all do the explanatory work without this commitment. Wigner-von Neumann is coherent. It just isn’t a “forced move”, and accepting it because consciousness “feels special” is just bad theorizing.

The deeper point: even if Wigner-von Neumann turned out to be right, it wouldn’t show that CCP is false. It would show that consciousness is part of fundamental physics, in which case “physical” would expand to include it, and we’d be back to a closed (now consciousness-inclusive) physical system. Whether that’s a vindication or a refutation of physicalism depends on how you’ve drawn the line. Either way, it’s not the kind of clean win for the anti-physicalist that pop-science presentations on social media make it out to be.

3. Consciousness Doesn’t Live in the Wavefunction

Let me sharpen this because the confusion is unusually persistent.

The reason people keep wanting consciousness to play a role in physics is that there’s an emotional appeal to the idea. The mind feels like it should be metaphysically special. The hard problem of consciousness (which I’ve argued in a previous essay is mostly a confusion) makes phenomenal experience seem like it cannot fit into the physical order. So when you encounter quantum mechanics and notice that “observation” plays a role in the formalism, the temptation is to connect the two and say: “Aha, here’s where consciousness gets back into physics.”

This is a category error.

The “observation” in quantum formalism is a technical term referring to a specific kind of interaction (decoherent, irreversible, recording). It has the same structural relationship to the rest of the formalism as “force” has to classical mechanics. You don’t claim that consciousness plays a special role in Newton’s laws because forces “act on” objects. The verb is doing technical work, not metaphysical work. Same with “observation” in QM.

Empirically, “measurements” happen all the time without any conscious observers. They have been happening since the dawn of our universe, and possibly even before that. A photographic plate records photon arrivals. A Geiger counter registers radioactive decay events. A dust grain in interstellar space gets entangled with photons and undergoes decoherence with no human anywhere in the causal chain. In the actual formalism, measurement is just any sufficiently decoherent interaction with a sufficiently large environment.

Decoherence theory has clarified this enormously over the past few decades. The transition from quantum superposition to apparent classical definiteness happens through unitary “entanglement with environmental degrees of freedom”, and quickly too - in a very small fraction of a second for macroscopic objects at room temperature. No mind needed.

Now, what about Penrose-Hameroff and the proposal that the brain exploits macroscopic quantum coherence in microtubules? Max Tegmark famously calculated decoherence times for microtubules at around 10⁻¹³ seconds - many orders of magnitude faster than the timescales of neural firing. Hameroff and Penrose have responded with modifications to their model, and recent experimental work (Bandyopadhyay and collaborators) has reported vibrational signatures in microtubules. The exchange is ongoing.

I don’t want to overstate the case. It isn’t that Orch-OR has been definitively refuted. What’s true is that the burden of evidence sits heavily on the Orch-OR side, and the central empirical claim (that the brain depends on macroscopic quantum coherence at cognitively relevant timescales) remains unsupported by mainstream neuroscience.

Quantum biology is a real and growing field. Photosynthesis exploits coherence at picosecond timescales in tightly engineered protein environments. Avian magnetoreception probably involves radical-pair chemistry (and yet we call them bird-brained!). But these are highly specialized molecular systems, not cognitive ones, and nothing in the quantum biology literature gives reason to expect coherent quantum computation at the scales relevant to thought. The brain runs hot, wet, and noisy. Coherence dies fast in those conditions, and the cognitive timescales (milliseconds to seconds) are vastly longer than anything (quantum and) coherent could plausibly last in tissue.

So consciousness has no established role in quantum measurement, and the most prominent proposals to give it one face severe empirical headwinds. The category error remains a category error. Decoherence happens regardless of minds. Consciousness lives in the integrated information processing of the cortex, not in collapses of the wave function. (The rest of the body may well be involved in that processing too, but let’s keep things simple for now.)

4. The Core Theory Argument: Where Closure Becomes Empirical

Here is where the second argument of this essay kicks in. The argument in section 1 was “methodological” - it established the “coherence” of CCP against Hempel’s dilemma. But coherence isn’t enough; we want empirical support. That’s a separate move, and the empirical case is what I want to lay out now.

As I’ve noted in previous essays, Sean Carroll has made this argument forcefully in his paper “Consciousness and the Laws of Physics“ and at book length in The Big Picture. The idea, originally articulated by Frank Wilczek, is that we already know the laws of physics that govern everyday phenomena - everything that happens at the temperatures, energies, and timescales relevant to biological matter on Earth. Wilczek calls this collection of laws the “Core Theory“. It’s the Standard Model of particle physics plus general relativity in its weak-field limit.

The Core Theory has been confirmed to extraordinary precision across an enormous range of conditions. Particle accelerators have probed energies up to about 14 TeV. Cosmological observations probe a vastly larger range of scales. Within the energy regime relevant to biological matter (which is many orders of magnitude lower than what’s been probed), we have detected “every” relevant interaction and “every” relevant field. We “know” what’s there. There aren’t undiscovered particles or forces hiding in the regime where chemistry and biology happen, because if there were, they would have shown up in our experiments. (Could there be a quantum field that doesn’t interact with matter or energy at all? Well if it doesn’t interact, it’s not relevant to discussions of physical causality related to organic life.)

Now plug this into the question of CCP. Suppose, for the sake of argument, that there’s some non-physical ingredient that affects what happens in your brain - a soul, a panpsychic proto-experience, a top-down causal influence from consciousness, divine action on neurons, or whatever your favorite anti-physicalist proposal is. For this ingredient to do any actual work, it has to interact with a quantum field. It has to push particles around, and make them move in a way it wouldn’t have moved given just the physics. (Particles are just excitations of quantum fields.) This would manifest as a deviation from the Core Theory at the very energies where neural processes operate, energies that we have probed experimentally for decades.

We have looked for any such deviations. We have looked very hard. We have not found ANY. The behavior of matter at biological energies is exactly what the Standard Model and gravity predict.

The standard response from the anti-physicalist is some version of “well, the deviation might be too subtle to detect” or “maybe the non-physical thing operates in a regime we haven’t probed.” Let me take this carefully, because it isn’t a pointless objection.

There are two ways the response could go. The first is that the non-physical influence is too small to detect at the level of individual field or particle interactions but accumulates into systematic effects at the level of whole organisms. This sounds initially plausible until you ask what mechanism would produce such accumulation without showing up in the statistics. If the effect is systematic (consistent across many fields/particles), it should be detectable in averages, correlations, or precision tests. If it’s random, it can’t be organized enough to produce coherent behavior like thought or action. The objector is asking for an effect that’s simultaneously systematic enough to matter and stochastic enough to evade detection. Those two requirements are in tension.

The second response is that the non-physical influence operates on degrees of freedom we haven’t directly probed - perhaps something like a modification of phase relationships or higher-order correlations. This is more interesting, and it gestures toward the kind of view I’ll discuss next under Russellian monism. But even granting it, it doesn’t violate CCP in the standard sense. It relabels what counts as physical. (We’ll get there.)

The empirical conclusion stands. There is no detected deviation from the Core Theory in any system, including biological ones. Every successful theoretical prediction in physics has been a prediction within the closed physical framework. This isn’t a proof of CCP (empirical evidence rarely is), but it’s the strongest kind of inductive case science can offer.

The dilemma I’ve discussed in my previous essays captures this nicely. Any anti-physicalist proposal must take Option A (modifies Core Theory dynamics, would be empirically detectable, is in fact undetected, would be the discovery of the century if true, Nobel prizes guaranteed) or Option B (leaves Core Theory dynamics intact, hence is causally inert, hence is epiphenomenal and explains nothing). The dilemma is sharp, and it’s hard to escape. As we’re about to see, there’s one sophisticated attempt to slip between its horns.

5. Russellian Monism: The Serious Opposition

The strongest contemporary opposition to standard physicalism isn’t dualism, and it isn’t anything that violates the Core Theory. It’s Russellian monism. Anyone defending CCP should engage with it seriously, because the easy refutations don’t apply.

As I mentioned in my earlier essay against panpsychism, the view traces back to Bertrand Russell’s analysis of physics in The Analysis of Matter (1927) and has been developed by Galen Strawson, Philip Goff, Daniel Stoljar, and (in some moods) David Chalmers. It goes like this.

Physics describes the world entirely in terms of “structure” and “dispositions”. It tells us what entities do, how they relate, how they transform under various operations. It does not tell us what those entities “intrinsically are”. Mass, charge, spin - these are characterized by what they do in equations, not by their “categorical nature. So physics, on this reading, leaves a gap: the intrinsic nature of the physical is “unspecified”. Russellian monists then propose to fill this gap with proto-experiential or experiential properties. The intrinsic nature of physical entities, on this view, is (or includes) something mental.

Here’s a useful analogy. You can describe a chess game completely in terms of moves, positions, and rules. That’s the structural description, and it’s enough to play the game. The Russellian asks: but what are the “pieces” made of? Standard physicalism (supposedly) says: structure is enough; there’s nothing more to know that’s relevant to the game. Russellian monism says: there has to be something the structural description is “about” - some intrinsic stuff that has these relational properties. And then they propose that this intrinsic stuff is (proto-)mental.

Notice what this view doesn’t claim. It doesn’t claim that physics is wrong, that the Core Theory is incomplete, or that some non-physical force violates closure. The dynamics are exactly what physics says they are"! The Russellian agrees with everything in section 4. They just hold that the structural description doesn’t “exhaust” reality, and that the categorical bases of physical properties are (proto-)mental.

This survives the Option A/B dilemma because the Russellian denies the dichotomy. The view isn’t Option A - it doesn’t modify dynamics. It also isn’t Option B - the intrinsic mental properties just “are” the categorical bases of the causal facts, so they’re not extra causal influences, they’re constitutive of the existing ones. The Russellian threads the needle.

So why am I not persuaded? A few reasons. None of them is a knockout on its own, but together I think they add up.

First, the structural-vs-intrinsic distinction the view depends on is contested. James Ladyman and Don Ross, in their defense of ontic structural realism, argue that there are no intrinsic properties beyond the structure - that “structure all the way down” is a coherent and arguably better view of physics. If they’re right, Russellian monism is positing categorical bases that don’t exist. The question of whether physics has unfilled categorical roles, or whether the structure is itself the ontology, is genuinely open in philosophy of physics, and Russellian monism is making a substantive bet here. To stick with the chess analogy: the structural realist says the pieces don’t have an “intrinsic nature” beyond their role in the game. They are their role. Russellian monism quietly assumes otherwise.

Second, even granting the categorical roles need filling, why fill them with mental properties? The motivation is almost always the hard problem of consciousness - the sense that phenomenal experience is so different in kind from anything else that it must be a fundamental ingredient. If you don’t accept the hard problem framing (and I don’t!), the motivation evaporates. The categorical bases, if they exist, could just as well be filled with whatever non-mental properties categorical bases naturally are. Russellian monism stops being attractive once you stop assuming consciousness is sui generis.

Third, the combination problem. If micro-level entities have proto-experiential properties, how do these combine to produce the unified, structured consciousness of a human mind? This is the panpsychist’s most famous internal difficulty, and despite a lot of careful philosophical work, no satisfying solution has emerged. Goff and others have wrestled with it; I find their attempts strained. A view that posits a fundamental property to solve one problem (the hard problem) and then can’t explain how that property produces the phenomenon it was posited to explain (unified consciousness) is in dialectical trouble. (I argued against panpsychism more broadly in an earlier essay and won’t restate all of that here.)

Fourth, parsimony. Everything physical works out as predicted by physics alone. Russellian monism adds an entire metaphysical layer (intrinsic mental properties) that does no predictive work. It’s motivated by an explanatory intuition, not by anything that would show up in the experimental record. I really have to say - Ockham’s razor doesn’t decide questions on its own, but when a theory adds ontology without adding empirical content, it deserves skepticism!

That said, Russellian monism is the most sophisticated current alternative to standard physicalism. It doesn’t violate CCP and isn’t refuted by the Core Theory argument. But it’s motivated almost entirely by intuitions about consciousness I think are confused, and it depends on a metaphysics of physics that’s contested at best. I’m not persuaded. But it’s the view that deserves engagement, not the dismissal it usually gets in undergraduate philosophy of mind or in skeptical circles in the hard sciences.

6. Emergence and Complexity Don’t Open a Door

The last major move I want to address is the appeal to emergence. The argument runs: even if CCP holds at the fundamental level, complex systems exhibit “emergent” properties (consciousness, life, intentionality, top-down causation) that can’t be reduced to lower-level physics. So maybe physics is closed at the bottom but reality is “more than physics” at higher levels.

This argument equivocates between two very different senses of emergence.

Weak emergence is the kind we see everywhere in science. Temperature emerges from molecular kinetic energy. Wetness emerges from intermolecular forces. Life emerges from organized chemistry. Consciousness emerges from neural integration. In each case, the higher-level property is a real, useful, often computationally indispensable description, but it’s not metaphysically over and above the underlying physics. The macro-state is fully determined by the micro-state. You can do top-down “causation” in the sense that talking about higher-level patterns is sometimes the most useful explanatory level (no one explains traffic jams by tracking individual electrons). But this isn’t an exception to closure - it’s multi-scale description of the same closed system at different levels of coarse-graining.

Strong emergence is the claim that higher-level properties exert causal influence not determined by, or reducible to, lower-level physical dynamics. This “would” violate CCP. It would mean the higher-level pattern reaches down and pushes particles around in ways the lower-level physics alone wouldn’t predict.

How do we evaluate strong emergence? Not by an a priori test - any test that presupposes microphysical determination begs the question against the strong emergentist. The right approach is to ask an empirical question: have we “ever” observed a case where the complete physical state of a system, together with the laws of physics, “failed” to predict the system’s macro-behavior? In the entire history of science, the answer is NO. Wherever we have checked carefully (in molecular biology, in chemistry, in neuroscience, in condensed matter physics), the macro-behavior follows from the micro-dynamics. This isn’t proof - strong emergence is a logical possibility - but we have an extraordinarily strong inductive case combined with a complete absence of positive evidence on the other side.

George Ellis has been one of the more prominent advocates for top-down causation, and to his credit, he tries to be precise. But on close reading, his examples are all examples of weak emergence dressed up in stronger language. Constraints from higher-level structure (a wing’s shape constraining airflow, a brain’s architecture constraining neural dynamics) are perfectly compatible with closed physics, because the constraints are themselves implemented by physical structure, and the dynamics are just the dynamics. Calling this “top-down causation” is a useful pragmatic description. It’s not an exception to closure.

For strong emergence to be a live option, we’d need empirical evidence of cases where micro-dynamics fail to determine macro-behavior. We don’t have such cases. The inductive case for closure, all the way up, is the same case that supports the Core Theory itself: the most thoroughly tested set of empirical claims science has produced.

7. The Free Will Question, Briefly

CCP has obvious implications for free will, and I want to acknowledge them rather than pretend they aren’t there.

If CCP holds, then libertarian free will - the kind that requires causal agency from “outside“ the physical order - is likely off the table. Your decisions are physical events with physical causes, even if those causes are stochastic. There is no extra “you” stepping in to direct the neurons. (Sorry to anyone who finds this distressing. The metaphysics doesn’t care about our feelings.)

Usually, this is the kind of conclusion that horrifies people on first encounter. I think the horror is misplaced. Compatibilist free will - the kind that requires only that your actions flow from your own deliberative processes without external coercion - remains entirely live. Daniel Dennett spent a career arguing that compatibilist free will is the only kind worth wanting anyway, and I find the argument quite persuasive. The “you” that deliberates and decides is real. It’s just made of neurons rather than ectoplasm.

A full defense of compatibilism is a different essay. The point here is just that CCP doesn’t ““destroy free will. It constrains what free will can coherently be. If you were already a libertarian free will-ist, CCP is bad news. If you were already a compatibilist or a hard determinist, CCP is what you expected. Either way, the metaphysics isn’t a disaster.

The Bottomline

Causal closure of the physical rests on two arguments doing two distinct jobs.

The methodological argument establishes the coherence of physicalism against Hempel’s dilemma. Physicalism isn’t a claim about the content of current physics. It’s a commitment to a structural, mathematical, third-personal idiom of description. Every successful physical theory in the past 350 years has fit this idiom. Future physics almost certainly will too, because that’s what makes a theory a physical theory. Hempel’s dilemma collapses once you stop conflating “physical” with “the current Standard Model.”

The empirical argument establishes that CCP is actually supported by what we know about the world. The Core Theory describes everything that happens at biological energies to extraordinary precision. Any non-physical ingredient with causal power would show up as a deviation. None has been found. The same empirical case rules out strong emergence in any system we have studied carefully. The evidence is asymmetric and overwhelming.

Two attacks I take seriously but think ultimately fail. The quantum measurement problem doesn’t introduce non-physical causation - probabilistic closure handles indeterminism, every serious interpretation is a physical theory, and consciousness has no established role in collapse. Russellian monism is more sophisticated; it doesn’t violate CCP at all but proposes that the intrinsic nature of physical entities is (proto-)mental. I’m not persuaded because the motivation depends on a hard problem framing I reject, because the combination problem remains unresolved, because the categorical-vs-structural metaphysics it depends on is contested, and because it adds ontology without adding empirical content.

None of this means consciousness, intentionality, or meaning aren’t real. They obviously are. They’re just not exceptions to closure. They’re descriptions, at the appropriate level of abstraction, of patterns in a closed physical system. Once you stop demanding that the pattern be metaphysically over and above what implements it, the air goes out of the room, and we can get back to actual science.

The world is closed. It’s also rich, layered, surprising, and full of phenomena we don’t yet understand. Those two things are perfectly compatible. The temptation to read every gap in our understanding as a gap in nature itself is the same mistake we’ve made about life, about combustion, about cosmology, about every previously-unsolved problem in the history of science. As I argued in my essay on fine-tuning, this kind of reasoning has never once turned out to be correct. I see no reason to think it will be correct here either.

Panpsychism is having a moment. After decades of being treated as a philosophical curiosity (or a relic of pre-scientific animism or something even less fashionable), the view that consciousness is somehow a fundamental feature of reality has been making its way into respectable philosophy departments, popular science books, podcasts, and even the occasional physics-adjacent academic conference.

Galen Strawson has been arguing for it forcefully since the early 2000s. Philip Goff wrote Galileo’s Error (trade book aimed at general readers) making the case for it in 2019. Annaka Harris has been popularizing the view via her writing and a podcast. David Chalmers, who basically launched the modern debate over consciousness with his 1995 “hard problem” paper, has written sympathetically about Russellian monism (a close cousin of panpsychism) for years. There is now a lively academic literature around the view, complete with its own conferences, journals, and, of course, internecine disputes.

I’ve previously argued that the hard problem of consciousness is, on close inspection, a confusion masquerading as a discovery. Panpsychism seems to be the most fashionable response to the hard problem, and a number of readers have pushed back on my earlier piece by appealing to it. So this is my attempt to push back again the pushback by taking a close look at why panpsychism fails - not just as a philosophical proposal, but as the kind of claim it actually is, which is a claim about the basic nature of the universe.

The thesis I want to defend here: panpsychism is not primarily a philosophical position, and treating it as one is part of why it gets undeserved respect. It is a hypothesis about what fundamental physics describes, and once we evaluate it on those terms, it falls flat. The motivating intuition that drives it (the worry that consciousness “can’t emerge from unconscious matter”) turns out to undermine panpsychism just as effectively as it (allegedly) undermines physicalism. And besides that internal problem, the weight of modern physics, neuroscience, and biology points the other way.

1. The Most Respectable Formulations

Let’s start by attempting to give the view a fair shake before critiquing it.

Modern panpsychism is not the cartoon version where rocks have feelings and electrons fall in love, although I think that this would make for a great webcomic. The serious contemporary versions are subtle, and they trace back (more or less?) to Bertrand Russell’s The Analysis of Matter (1927). (There may have been earlier precursors, but Russell’s framing seems to have significantly influenced modern approaches.)

Russell’s observation was this: physics describes the world entirely in terms of structure and dynamics - what things do, how they relate, what mathematical patterns they follow, etc. Physics tells us that an electron has a certain charge, mass, spin, etc. and that it interacts with other things in certain quantitatively specifiable ways. But (so the argument goes) physics never tells you what an electron “is”, intrinsically. It tells you about the relations and behaviors. The intrinsic nature - the “what it is” of fundamental physical stuff - is left as a black box. Russell suggested that consciousness might be the only intrinsic nature we ever have direct acquaintance with (since it’s our own experience), and that other things (electrons, fields, what have you) might have their own intrinsic natures of which our consciousness is a “particularly elaborate” species.

(My genealogy of the ideas here is not meant to be exact. The SEP article on panpsychism is quite good and covers this.)

Starting from that approach, we now have a fair number of modern positions that seem to tie back to Russell’s argument.

Let’s go over a few of these - note that some of these positions are close to panpyschism but not quite panpyschism:

Constitutive panpsychism (Goff is probably its best-known contemporary defender) holds that consciousness is a fundamental property of basic physical entities - quarks, electrons, fields - whatever turns out to be ontologically basic, and that the consciousness we have is “constituted” by combinations of those micro-experiences.

Panprotopsychism is a more cautious cousin: rather than saying basic stuff has consciousness, it says basic stuff has “protoconscious” properties that, when suitably combined, give rise to consciousness. The idea is to retain the explanatory power of “no consciousness from non-consciousness” without committing to electron feelings. (Feel free to insert your own favorite disappointed electron meme gif here.)

Russellian monism is the broader umbrella term for “physics describes structure, intrinsic nature is something else, that something else is (or is closely related to) consciousness.” This approach can be either constitutive or panprotopsychist in flavor.

Cosmopsychism flips the bottom-up move: the universe as a whole is the fundamental conscious entity, and individual minds are derivative “slices or aspects” of cosmic consciousness. (You can hear some Vedantic echoes here, which is fine - philosophy has often recycled.)

This is not an exhaustive list of formulations, they are the ones I found most interesting and respectable. For the most part, they aren’t claiming that your toaster is contemplating the meaning of life. They’re making a structural claim about the relationship between physics and phenomenology.

2. The Motivations (and the Strongest Case)

In my view, two motivating intuitions seem to be doing most of the work for panpsychism, and they’re worth taking seriously even if (as I’ll argue) neither survives scrutiny.

The “no emergence” intuition. This is the engine of the whole project. The idea is that you can’t get consciousness from a substrate that has no consciousness in it, any more than you can get extension from non-extended points or wetness from non-wet atoms. Strawson, for example, is quite insistent on this. If your starting ingredients are entirely dead matter, with no glimmer of experience, then no rearrangement of those ingredients will produce experience. Therefore the ingredients themselves must already have “some” experiential character.

So, I’ll say this for this intuition: it has rhetorical force. It feels right when you read it. (It’s one of the most common arguments for panpsychism I come across on X, for example.) But, as I’ll get to below, it has the unfortunate property of being self-defeating.

The Russellian “intrinsic nature” gap. This is the more sophisticated motivation. The argument goes something like this: physics gives us only the relational/structural skeleton of the world. There must be intrinsic natures grounding those relations, and we have one example of intrinsic nature on offer (our own conscious experience). Parsimony then suggests we extend that example to the rest of nature rather than positing some entirely new species of intrinsic nature for the non-mental world.

(As someone who tends to think positively of structural realism, the “physics describes structure” argument is one I can appreciate, even if I disagree with the rest of the argument.)

Now let me try and state the panpsychist case at full strength, because I think it’s sometimes dismissed too quickly. This rehashes some of the material I already covered, but please bear with me - I am trying my best to steelman this.

The strongest version (in my opinion) goes something like this. Modern physics gives us, on the structuralist reading, a mathematically beautiful description of “what fundamental entities do”. It tells us how electrons move, how fields couple, how spacetime curves. It does not tell us “what fundamental entities are”, in the sense of what their intrinsic, non-relational properties are - because, by construction, physical theory traffics only in dispositional and structural features. Thus, there is an entire ontological category (the “categorical” or “intrinsic") about which physics is, by its very methodology, silent. If we ask what fills that category, we have exactly one positive datum: our own consciousness, which we have direct acquaintance with and which does not present itself as a structural relation but as something with intrinsic phenomenal character. The parsimony argument (which panpsychists note, physicalists like me love to invoke) then suggests that we should fill the categorical-nature gap with the same sort of stuff we already have on offer (proto-experience), rather than positing some entirely novel kind of intrinsic nature about which nothing whatever can be said. On this telling, panpsychism isn’t a mystical addition to physics. It’s a “parsimonious completion” of it.

That’s a real argument. And it deserves engagement, not dismissal. The reason it ultimately fails (which I’ll develop in sections 3-7) is not that it’s silly, but that the proposed completion doesn’t actually do the explanatory work it claims, and runs afoul of the physicalist-friendly considerations the panpsychist is trying to honor.

3. The Self-Defeating Logic of Anti-Emergence

Here’s where I think the contemporary debate has been strangely lopsided.

The core panpsychist move is - “consciousness can’t emerge from non-conscious stuff.” Okay. Suppose we grant that, just for the sake of argument. Now let’s ask: how is sophisticated, unified, semantically rich human consciousness supposed to arise from the aggregation of trillions of (allegedly) micro-conscious quarks, electrons, and field excitations?

This is the famous combination problem, and it has been the chief internal headache of panpsychism since William James first raised it in The Principles of Psychology (1890). Chalmers has called it “the hard problem of combination” and, on this point, I think he’s exactly right.

But let’s examine this problem a little closer than the standard treatments do, because the strongest panpsychist response seems to miss what’s really going on.

The strongest panpsychist response runs as follows: yes, the combination problem is hard, but it’s hard in a “categorically less mysterious way” than the matter-to-mind problem. Going from micro-experience to macro-experience is at least staying within the same ontological category - we’re combining experiential stuff with experiential stuff, like making a wave from many smaller waves. Going from non-experience to experience, by contrast, is alleged to be a discontinuous leap into an entirely new ontological domain. So the panpsychist can grant that combination is hard while still claiming it’s a fundamentally less daunting problem than physicalist emergence.

This response is wrong, and it’s wrong in a specific, instructive way.

The “category preservation” reply works only if we ignore what the macro-experience actually is. Human consciousness is “unified”. It’s a single perspective, a single point of view, one set of experiences that is “had” by one subject. The transition the panpsychist needs to explain is therefore not just “combining experiences” but combining “many discrete tiny perspectives into one unified perspective”. And that transition is exactly the kind of boundary-crossing the panpsychist objected to in the first place - it’s a transition from a state-of-affairs in which there are many subjects (many discrete proto-experiences) to a state in which there is one unified subject. The fact that both states involve “experiential” stuff doesn’t make the transition trivial by any means. It might actually make it harder, because now we have to explain how “experiential facts” (which panpsychists suggest are fundamental) can be merged into a single experiential fact without violating their fundamentality.

Think about it this way. The panpsychist’s anti-emergence intuition says: experience cannot arise from a substrate that lacks it. Fine. Now apply the principle: “unified experience” cannot arise from a substrate that lacks unified experience. By the panpsychist’s own logic, you cannot get unification by aggregation. Unified experience, on the same anti-emergence principle, would have to be fundamental.

But then we’re not in panpsychism anymore - we’re in “cosmopsychism” , where the universe as a whole is the unified experience and our individual minds are decompositions of it. And this is where cosmopsychism’s apparent escape from the combination problem turns into a “decombination problem” that is, if anything, even worse. If the cosmic mind is the fundamental unit, how does it split into many private, mutually inaccessible perspectives? What “principle of decomposition” takes one cosmic experience and gives you several billion discrete human experiences, each closed off from the others? Whatever answer you give to that question is going to require exactly the kind of “many-from-one” or “one-from-many” account that motivated the combination problem on the way up.

(The Borg notably solved this problem by fiat. The cosmopsychist does not get to invoke fiat.)

The situation is problematic for panpsychists either way you run it. Run it bottom-up, and we have a combination problem that the very anti-emergence intuition we started from seems to forbid. Run it top-down, and we have a decombination problem that does the same work in reverse. As far as I can tell, no one has solved either, and the proposals I’ve read tend to either slide back into garden-variety emergence (in which case the proto-consciousness is doing no explanatory work) or to introduce new metaphysical primitives “without independent motivation”.

The honest summary: panpsychism trades one alleged emergence miracle for another. If the physicalist owes you an explanation of how subjective experience arises from neural activity, the panpsychist owes us an explanation of how a “single, integrated, narratively-structured” experience arises from the proto-experiences of an astronomical number of micro-entities (or, in the cosmopsychist case, decomposes from a single cosmic experience into many private ones).

4. The Core Theory Constraint, Again

I worked through this argument in my hard problem essay, but it bears repeating, because panpsychists (in my opinion) have not taken it seriously enough.

As Sean Carroll summarized in Consciousness and the Laws of Physics and elsewhere, the laws of physics underlying everyday phenomena (Frank Wilczek’s “Core Theory“ - the Standard Model of particle physics plus the weak-field limit of General Relativity) are completely known within the energy regime relevant to anything that happens inside a human body, a brain, a planet, or for that matter our entire solar system. We’ve probed the relevant energy levels experimentally for decades. Within that domain, there is no room for additional fields, forces, or dynamical degrees of freedom that we haven’t already catalogued.

Now, let’s use this to scrutinize panpsychism. The panpsychist says fundamental constituents of reality have proto-experiential properties.

(Yes, I did use the following type of argument in my hard problem essay.)

This means that we two options:

Option A. These proto-experiential properties actually affect the dynamics of physical systems. As a result, conscious matter “behaves differently” than non-conscious matter would. In which case: this would show up in our physics experiments. We would see deviations from the predictions of the Standard Model. We would be able to design experiments to detect proto-consciousness in fundamental fields. (And if we found it, the people involved would get Nobel Prizes for it.) But, none of that is true. The data say loud and clear that the dynamics of the Core Theory are sufficient to describe the behavior of the matter that brains are made of.

Option B. These proto-experiential properties leave the dynamics of the Core Theory completely intact. Quarks behave exactly as the Standard Model says they behave, with or without their alleged inner light. (To be charitable, this option seems to be what serious Russellian monists actually defend.) But this would be a case of epiphenomenalism and is really not doing any work at all.

Now, the Russellian monist would reply to this charge (that Option B is just epiphenomenalism) along these lines: proto-consciousness isn’t causally inert because the structural properties physics describes “are constituted by” the proto-conscious intrinsic properties. Physics doesn’t see it because physics is constitutively a structural science - it tells you what intrinsic natures “do”, not what they “are”. Proto-consciousness is the categorical basis of physical causation itself. Asking why we don’t see it in physics experiments is like asking why we don’t see the “shape” of a key in the door it opens - the shape isn’t causally redundant; it’s what makes opening-the-lock possible.

This is a intriguing response, and it’s worth taking seriously. Here’s the thing though: it really just relocates the failure without rescuing the position. Here’s why.

Grant the Russellians everything they want about physics describing structure and intrinsic nature being the categorical basis of that structure. Now ask: what work is the “proto-experiential” characterization of intrinsic nature doing, beyond what “categorical basis of physics, content unknown” already does? The answer is: nothing observable! By the Russellians’ own admission, swapping proto-experiential intrinsic natures for any other categorical basis with the same structural role would leave physics, neuroscience, and behavior identical. The proto-experiential identity is sitting on top of the structural facts adding “zero predictive content” - and, mind you, adding zero explanatory content too, because we can’t show that proto-experience does any explanatory work that “neutral categorical basis” wouldn’t do equally well. We can’t even check whether proto-experience is the right answer rather than (say) proto-numericality (numerality?), proto-extension, or proto-anything-else with the same structural footprint.

That’s the actual problem with Option B. It’s not that it’s epiphenomenal in the crude sense of “consciousness floating uselessly above physics.” It’s that the “specifically experiential” characterization of the proposed categorical basis is a free metaphysical ride that, by stipulation, can’t be wrong - and which, by stipulation, makes no contact with anything we can investigate. A view that is compatible with literally any observation tells us, in the end, nothing about the world.

(I anticipate the protest that panpsychism is a metaphysical hypothesis, not a scientific one, and so different standards apply. I disagree, as you’d expect, and section 6 below is where I argue against that notion.)

5. The Evolutionary Tell

Set aside the metaphysical and physical arguments for a moment. Look at consciousness empirically.

What do we actually observe?

Consciousness in nature scales with biological complexity, in ways that track quite precisely the elaboration of certain kinds of nervous systems. Bacteria have no nervous system; nobody seriously thinks they’re conscious. Simple invertebrates with rudimentary nerve nets show responses to stimuli but no integrated processing. Fish, reptiles, and birds have progressively more elaborate nervous systems and progressively more flexible, integrative behavior. Mammals (especially the more “sophisticated” ones) show clear hallmarks of unified perception, emotion, and learning. Among mammals, primates seem to have additional layers of self-modeling. Humans (with our particular cortical architecture) layer on top of all that the abstract thought, language, and narrative selfhood (that has us socialmaxxing and writing essays at each other on the internet trying to argumentmogg each other.)

This “consciousness gradient or spectrum” in the animal kingdom is fairly smooth. The correlation with neural complexity is also strong. The “kinds” of consciousness present at each level look exactly like what you’d expect from an evolved adaptive function - integrative information processing in service of survival, reproduction, action selection, and behavioral flexibility.

Now, panpsychists might retort with something like: “Of course rich, human-style consciousness scales with brain complexity - that’s the combination phenomenon. We never said proto-consciousness shows up phylogenetically; we said proto-consciousness is fundamental and rich consciousness is constituted by combinations of proto-consciousness in specific architectural arrangements. The phylogenetic gradient is consistent with our view.”

Okaaaay. I don’t think the phylogenetic gradient “refutes” panpsychism in the strict logical sense. But here’s what it does do - it makes panpsychism “explanatorily idle”.

The panpsychist concedes that the rich, unified consciousness we actually observe in nature is fully accounted for by structural and architectural features of nervous systems. The phylogenetic gradient maps onto neural complexity; the molecular specificity of anesthesia maps onto specific neural circuits; the lesion data map onto specific brain regions. All the explanatory work in producing the consciousness we actually observe is being done by the wiring. The proto-consciousness is sitting at the bottom of the ontology contributing nothing “observationally distinguishable”.

If a theory adds an entity (proto-consciousness in fundamental fields), and that entity does no observable work in the actual production of the phenomenon you’re trying to explain (rich human consciousness), then the entity is unmotivated. We don’t posit “proto-photosynthetic” properties of fundamental matter to explain why plants do photosynthesis. We don’t posit “proto-digestion” to explain why animals digest food. We explain those phenomena the obvious way - a particular molecular machinery does a particular job in a particular context, and natural selection made it happen over time. The fact that consciousness “feels” different from photosynthesis doesn’t justify a different methodology. Or rather, if it does, the burden is on the panpsychist to show why.

This is because consciousness shows every fingerprint of a biological adaptation:

• It scales with the complexity of integrative neural machinery.

• It is selectively impaired by lesions to specific brain regions (especially thalamus, cortical hubs, and the ascending arousal system).

• It can be turned off by general anesthesia, which acts on specific molecular targets in specific neural circuits.

• It tracks specific patterns of neural activity (the so-called Neural Correlates of Consciousness, identified through decades of work by Crick, Koch, Dehaene, Tononi, and others).

• It develops in individual organisms as the relevant neural circuits mature.

• It can be modulated by drugs, stimulation, fatigue, sleep stages, and disease in highly specific ways.

Every one of these features is exactly what you’d expect if consciousness were a high-level functional property of certain kinds of evolved physical systems. None of them point at proto-consciousness as a fundamental ingredient. The panpsychist can preserve compatibility with this evidence by saying “well, the proto-consciousness is still there at the fundamental level, just not doing the visible work.” Sure. So is the proto-photosynthesis, the proto-digestion, and the proto-anything-you-like. That’s not an argument; it’s a refusal to let the evidence count.

6. Panpsychism Is a Scientific Hypothesis

Now we come to (what I think is) the most underappreciated point in this whole debate.

Panpsychists like to insist they’re doing metaphysics, not physics. The phrase “philosophical theory” gets deployed as a kind of shield: you can’t refute me with experiments, because my claim isn’t an empirical one. I want to argue that this shield is illegitimate, but I want to argue it carefully - not by leaning on naive falsificationism (which my sharp interlocutors from X frequently point out is a mid-20th-century philosophy of science with problems of its own).

Look at what panpsychism actually claims:

It claims that fundamental physical entities (quarks, electrons, fields) have proto-experiential properties.

It claims that the mental properties of macroscopic systems (us, animals more broadly) are constituted by, or grounded in, the proto-experiential properties of their constituents.

It claims that the physical world has an intrinsic nature that physics describes only structurally.

Every single one of these is a claim about “what is the case in the natural world”. They are claims about the basic furniture of reality, about properties of physical entities, about the relationship between micro-physics and macro-cognition. These are precisely the kinds of claims that natural science has been making and refining ever since the late 1800s. To say “but I make these claims as a philosopher, not as a scientist” is a procedural move with no substantive content - the claims themselves don’t change category just because of who is making them.

The relevant standard here is not “panpsychism must make falsifiable predictions in the strict Popperian sense”. The relevant standard is something more like: any hypothesis that purports to describe the properties of fundamental physical entities and their relation to higher-level phenomena should engage seriously with the methods and results of the disciplines that already study fundamental physical entities and higher-level phenomena. It should integrate with physics. It should integrate with neuroscience. It should integrate with biology. It should at minimum be the kind of thing that “could in principle” be evaluated against empirical evidence, even if no current experiment is decisive. (And no - this paragraph isn’t some sort of pedantic gatekeeping. Scientific expertise matters!)

But… even by that softer, more reasonable standard, panpsychism still fails. It does not integrate with physics (the proto-experiential properties have no formalism, no Lagrangian, no role in any calculation any competent practitioner in the field does). It does not integrate with neuroscience (the entire phylogenetic and lesion-based picture of consciousness has to be relegated to “the combination problem,” which as I discussed is acknowledged to be unsolved). Panpsychism is, as we saw in section 4, deliberately constructed to be compatible with literally any physical observation.

A view that makes sweeping claims about the natural world while refusing to engage with how the natural world is actually investigated is not “doing metaphysics rather than physics.” It’s doing physics (and neuroscience) badly while claiming exemption from the standards of physics (and neuroscience). That is the actual indictment.

7. The Evidence Across Disciplines

Let’s go over the actual evidence, just to make the asymmetry more vivid.

Physics. The Core Theory accounts for every interaction relevant to the matter that makes up brains. There is no field, no force, no degree of freedom in the Standard Model that has any property recognizable as “proto-experiential.” The properties of fundamental fields are exhaustively given by their gauge structure, their masses, their couplings, their representations under the Poincaré group. Adding “experiential” properties to the list is not some small extension - it’s a wholly new kind of property with no known dynamical role and no formalism that would let us calculate with it. There is no panpsychist Lagrangian. There has never been a measurement of proto-experience. There’s not even a proposal for what such a measurement would look like, in principle.

It’s worth flagging here that some panpsychism-adjacent writers reach for the Penrose-Hameroff Orch-OR model, or the various observer-related interpretations of quantum mechanics (Wigner’s Friend, the participatory universe, etc.), as putative empirical hooks. None of these stand up. Orch-OR proposes that consciousness arises from objective wavefunction reductions in microtubules, with theoretical motivation drawn (quite controversially) from Penrose’s reading of Gödel’s incompleteness theorems. The empirical evidence for it is poor (microtubules don’t appear to maintain coherence on the relevant timescales at body temperature), the Gödelian motivation is contested by virtually every philosopher of mathematics who has weighed in on it, and even if it were true, it would establish an exotic biological mechanism, not consciousness or panpsychism. (The Wikipedia article on Orch-OR has references to the many criticisms.) Observer-related interpretations of quantum mechanics, meanwhile, are a fringe minority position even among the (already small) community of physicists who care about interpretive questions, and they don’t license panpsychist conclusions even on their own terms. The physics escape route doesn’t actually exist. (My opinion of Wheeler as a physicist actually went down a teensy bit when I first read about his participatory anthropic principle idea.)

Neuroscience. The neural correlates of consciousness are highly specific. We have well-thought out models that have and continue to be explored. Consciousness has been linked with thalamocortical activity, with global broadcasting of information across cortical hubs, with the integration of recurrent processing, with predictive processing. It can be selectively switched off by general anesthesia, which acts at specific molecular targets (GABA-A receptors, NMDA receptors, etc.) in specific circuits. Lesions to particular regions produce particular forms of consciousness loss (akinetic mutism from medial frontal damage, anosognosia from right parietal damage, blindsight from V1 damage). Decades of careful experimental work have produced a converging picture of what consciousness is, in terms of “what kind of neural activity it is”. Nothing in this picture invokes anything beyond the standard physical and biological dynamics of nervous systems.

Information theory and computation. This is the one place where I have to give the panpsychist some credit, because Tononi’s Integrated Information Theory has been openly read as panpsychism-adjacent by Tononi himself - he’s said, in print, that any system with positive integrated information has “some” degree of experience. So IIT, taken on its own terms, does seem to commit its proponents to a kind of micro-consciousness in any sufficiently integrated system, however small. This is a real challenge for a tidy physicalist position, and I don’t want to wave it away.

But two responses. First, IIT’s empirical core (the structural/architectural account of why some neural patterns are associated with consciousness and others aren’t) doesn’t actually require the panpsychist gloss. You can take the Φ formalism as a useful empirical correlate of consciousness in biological systems without granting the metaphysical claim that “any positive Φ entails some experience anywhere.” Many working neuroscientists who use IIT as a measurement framework do exactly this. Second, the panpsychist gloss on IIT runs straight into absurd consequences that panpsychism was supposed to avoid - e.g. simple grids of XOR gates have positive Φ, which means IIT-panpsychism is committed to consciousness in arbitrary computational arrangements that no one is actually willing to defend on reflection. This is a reductio in slow motion, and it’s a problem for IIT-as-metaphysics rather than for IIT-as-empirical-framework.

The other live research programs in consciousness science (Global Workspace Theory, predictive processing, higher-order theories, attention-schema theory) all do their work by building up consciousness from the structural and dynamical properties of physical systems, checking proposals against neural data, and refining. That’s the methodology of science. Panpsychism, by contrast, has no comparable program. It’s a stance, not a science.

Biology. Consciousness in the animal kingdom tracks the evolution of nervous systems with stunning consistency. Cnidarians (with nerve nets but no centralization) barely show signs of consciousness. Cephalopods (with elaborate decentralized intelligence) show many. Vertebrates with thalamocortical architecture show the canonical picture. The phylogenetic gradient is exactly what you’d expect if consciousness is a kind of biological function. Even within human development (ontogeny), consciousness develops as specific neural circuits mature - not at conception, not at the formation of the first neuron, but at the formation of the relevant integrative networks.

Chemistry. The chemistry of consciousness, to the extent it has been worked out, is the chemistry of specific molecular systems doing specific jobs. Neurotransmitters, receptors, ion channels, synaptic vesicles, glial regulation - all standard chemistry. (I learned some disturbing specifics about how chemical weapons work on the nervous system when researching this bit. Yikes.) Anesthetics are a particularly clean case: drugs that disrupt consciousness do so by interacting with specific protein targets in specific neural circuits, and the dose-response relationships are fully characterizable in standard pharmacological terms.

I think the data above in this section shows that the asymmetry is overwhelming. The scientific picture, across multiple converging disciplines, treats consciousness as a high-level functional property of specific kinds of physical systems. The panpsychist picture treats it as a fundamental property of matter that, “mysteriously”, manifests as the kind of high-level functional property the scientific picture “already captures”! The first picture does work. The second adds no value.

8. What Would Change My Mind

I am quite willing to say what would convince me. This is not an exhaustive list by any means, but a few things would do it.

1) Direct empirical evidence of proto-conscious properties affecting physics - any deviation from the Core Theory in regimes where consciousness is present, any signature of an additional dynamical degree of freedom that correlates with the presence of mind.

2) A formalism showing how proto-experience combines into rich experience, capable of generating predictions about which neural architectures should produce which kinds of conscious states, and bearing those predictions out empirically. (Would be even better if this was quantitative in some way.)

3) A successful panpsychist prediction of any neural fact about consciousness that wasn’t already entailed by ordinary physicalism.

4) A solid argument that the leading positive physicalist views are incompatible with the evidence in a way panpsychism isn’t.

I doubt that any of these are forthcoming. My objection to panpsychism is not that I refuse to consider it; it’s that, as currently formulated, it doesn’t give me anything to consider.

It’s also worth saying explicitly what the live physicalist alternatives are, because a purely negative critique invites the question “well, what’s your positive view?” I don’t have a specific preferred approach. I really like Anil Seth’s idea of predictive processing, even if I don’t like his use of the term “hallucination”. I like the GWT framework. I also like Keith Frankish’s illusionism, which I have not mentioned previously in this essay. On the illusionist view, what we call “phenomenal consciousness” - the felt, ineffable, intrinsic redness-of-red, batiness-of-bats, and so on - is itself a kind of representational illusion, generated by introspective mechanisms that misrepresent the brain’s own processing. The job of consciousness science is then to explain (a) what the brain is actually doing when we are conscious, and (b) why our introspective machinery generates the misleading impression that we have ineffable qualia. Both are tractable empirical problems. Neither requires positing fundamental experience anywhere in physics.

Anyway, the point is just that there are live, productive, “positive” physicalist views in the field. Panpsychism is not the only alternative to a confused dualism. The choice is not “panpsychism or magic.” It’s “panpsychism, or any of several research programs that have actually been making progress.”

We Have, Once Again, Confused Mystery for Evidence

I shall reiterate a point here that I’ve mentioned in prior essays. Across the history of science, there’s a recurring pattern - whenever some natural phenomenon has resisted reductive explanation, a school of thought has arisen claiming that it cannot, in principle, be explained that way, and that we must therefore add new fundamental ingredients to the world to accommodate it. Vitalism did this for life. Élan vital and entelechy were the technical terms for the alleged extra ingredient. There were sophisticated philosophers defending it. There were scientific journals in which it was respectable. And then molecular biology happened, and it turned out that life could be entirely explained by chemistry and information after all.

(Here I should point out that vitalism, in its strongest forms, actually made specific empirical claims that got refuted - Wöhler’s 1828 synthesis of urea from inorganic precursors being the famous early shot. Panpsychism, by contrast, makes no comparably specific claims. So the analogy is imperfect: vitalism was wrong because it could be checked; panpsychism is in worse shape because it can’t be. But the meta-pattern is the same.)

I argued in my piece on fine-tuning that the fine-tuning argument trades on a similar move: identify a real but currently unsolved problem in physics, declare the problem unsolvable in principle, and insert a metaphysical answer (in that case, God). The structure of the panpsychist argument is identical. It identifies a real but currently unsolved problem in cognitive neuroscience (how does the brain produce experience?), declares the problem unsolvable in principle, and inserts a metaphysical answer (consciousness is fundamental). In both cases, the substitution feels profound but does no actual explanatory work. In both cases, the move has no track record - the history of science is a graveyard of similar substitutions.

That is the deepest indictment of panpsychism. It confuses the “difficulty” of a problem with the “impossibility” of a problem. It mistakes our current ignorance for an ontological barrier. It dresses up a confession of (temporary) inability as some sort of deep metaphysical insight. And then, having made a series of claims about the basic nature of physical reality, it asks to be evaluated by the standards of philosophy rather than science - which is a methodological move that should, by itself, raise alarms (and hackles).

The Bottomline

Panpsychism is what happens when you take an interesting intuition (you can’t get experience from non-experience), apply it once, and then stop. If you apply it again, it eats itself. The unification of micro-experiences into a single perspective is not less mysterious than the production of experience from neural activity - it’s “more” so, because the very anti-emergence principle that motivated the position now forbids the unification step too. The Core Theory constraint forces panpsychism into an unwinnable choice between being empirically false and being explanatorily empty; the Russellian “intrinsic nature” reply is clever but leaves the proto-experiential characterization doing no work that “neutral categorical basis” wouldn’t do equally well. The phylogenetic gradient of consciousness in the natural world makes proto-experience explanatorily idle. The convergence of evidence from physics, neuroscience, biology, and chemistry pushes overwhelmingly in the same direction. And critically - panpsychism is not, despite the conventional framing, a philosophical hypothesis insulated from empirical evaluation. It is a hypothesis about the basic furniture of reality, and as such, it is squarely in the wheelhouse of natural science. By the standards of any serious empirical theory, it is unmotivated, untestable in any productive sense, and explanatorily inert.

(We still have a hard problem of neuroscience - related to understanding exactly how the nervous system works. But as mentioned above, work on that has been fruitful and will continue until we figure it out.)

Also, as I mentioned in my essay on the hard problem of consciousness, the actual hard problem is not really that hard once you take the explanatory gap to be a feature of our description rather than of reality. As such, we should expect this to be solved (or more accurately dissolved) the way every other apparent metaphysical puzzle in the history of natural science has been solved: by patient empirical work. By figuring out what the brain actually does, how it does it, and why those particular operations look the way they do from the inside.

Final Thoughts

I’ve tried to make the case that consciousness isn’t built into the universe. It’s something the universe occasionally pulls off, in places where the material conditions are right. That’s already remarkable enough.

Physics has spent the better part of a century trying to do one specific thing: unify the Standard Model (SM) with General Relativity (GR). The SM describes three of the four fundamental forces (electromagnetic, weak, strong) and works absurdly well. GR describes gravity, and also works absurdly well. The trouble is, when you try to combine the two in any straightforward way, you run into problems - non-renormalizable infinities, breakdowns of perturbative control, etc. - the whole familiar litany of problems anyone who has poked at quantum gravity has run into.

Now, the thing we’re trying to figure out here has a few different names depending on how ambitious you’re being. Quantum Gravity (QG) just means a quantum-mechanical theory of gravity, full stop. Theory of Everything (ToE) is more ambitious. It means all four forces unified, ideally up to UV completion (i.e., a theory that remains valid at very high energies/short length scales, not just in the regimes we can currently probe). QG is a subset of ToE.

There have been many serious attempts. Loop quantum gravity, asymptotic safety, causal set theory, causal dynamical triangulation, entropic gravity, and so on. They all have interesting features, have attracted serious physicists, and have made some progress.

But there is really just one approach that simultaneously hits a much wider set of targets, and that approach is string theory. Despite its (sometimes well-deserved?) reputation for catching flak, I think it’s worth laying out what it actually delivers, and why, despite the standard complaints, it remains the most credible candidate we have for a true theory of everything.

Thanks for reading! Subscribe for free to receive new posts and support my work.

1. The Field of Contenders

I decided to start this off by briefly sketching what each major non-string approach has achieved. (I’m trying to be charitable here. I have respect for everyone working on these problems. Quantum gravity is genuinely hard, and ANY progress is laudable.)

Loop Quantum Gravity. This is in some sense the most direct attempt to quantize GR on its own terms. It treats spacetime itself as fundamentally discrete (using spin networks and spin foams) and produces some interesting structural results (kinematical Hilbert spaces, area and volume operators with discrete spectra, and so on). But LQG, in its standard form, doesn’t say anything about the other three forces. It’s a good candidate quantum theory of gravity, but not a ToE. And even the issue of recovering smooth, classical GR from LQG in the appropriate limit is... let’s say a work in progress. (I’ve a close relative who works on LQG in academia, and he thinks it’s great!)

Asymptotic Safety. This is the elegant idea (going back to Weinberg) that gravity might be nonperturbatively renormalizable via a UV fixed point. If this is true, it would be very nice. The evidence is suggestive (much of it from functional renormalization group calculations), but there’s no consensus that the fixed point really exists in the full theory. And again, it’s primarily a theory of quantum gravity, not a ToE. The Standard Model sector has to be added in by hand.

Causal Set Theory and CDT. Both treat spacetime as fundamentally discrete with causal structure baked in. These have produced some lovely toy results (CDT, in particular, recovers something that looks like four-dimensional spacetime in the right regime, which is genuinely cool). But again, no comprehensive Standard Model emergence, and the predictive content for testable physics is still pretty limited.

Entropic Gravity. Verlinde’s idea is that gravity is a derived/emergent thermodynamic phenomenon rather than a fundamental force. Interesting and conceptually radical! But as a candidate for a complete theory of quantum gravity (let alone a ToE), it’s still quite underdeveloped.

This is not a knock on any of these. Each is a serious research program, and one of them might turn out to be right. But none of them does what string theory does.

(There are some other approaches that I have not covered here. The field is quite vast. But I think the above are the major ones. Please drop a comment if there’s an approach that has some level of institutional acceptance that I may have missed out on.)

2. What String Theory Actually Delivers

So here’s where the rubber meets the road. What do we get from String Theory?

1. Quantized gravity with a spin-2 graviton, automatically. This is, frankly, astonishing. When you quantize a closed string, one of the modes that pops out of the spectrum is a massless spin-2 particle. We start with a relativistic string, we quantize it, and a graviton just falls out as part of the spectrum. This is one of those results that, the first time you see it, you go “wait, really?”. (It really is.)

2. Accounts for all four forces with a single object. Strings can have different vibrational modes, and those modes correspond to different particles. Open strings give you gauge bosons (the carriers of the electromagnetic, weak, and strong forces). Closed strings give you the graviton (see point 1). With supersymmetry layered in, you get fermions too. So a single fundamental object - the string - gives you the entire particle zoo of fermions and bosons. No other approach unifies matter and forces this cleanly with one underlying ingredient.

3. Reproduces black hole entropy. Strominger and Vafa showed in 1996 that you could count the microstates of certain extremal black holes using string theory and recover the Bekenstein-Hawking formula on the nose. This was, and remains, one of the most striking quantitative successes in theoretical physics in the past several decades. Any candidate theory of quantum gravity has to reproduce this kind of result. String theory does, and it does so in a way that’s not obviously circular or rigged.

4. Handles UV divergences reasonably well. One of the big problems with naively quantizing GR is that you get non-renormalizable infinities. String theory replaces point particles with extended objects, and the extended-ness (technical term that I may have just made up) softens the high-energy behavior in a way that tames these divergences. Loop amplitudes in superstring theory are finite or at worst very mildly divergent, which is dramatically better than the alternatives.

5. Gives us AdS/CFT and gauge-gravity duality. This one is huge, and I think under-appreciated (outside the immediate field). Maldacena’s correspondence (the conjecture that string theory in a certain anti-de Sitter spacetime is equivalent to a conformal field theory living on its boundary) has turned out to be enormously useful far beyond its original motivation. People have used it to model strongly-coupled QCD-like systems, to study the quark-gluon plasma at RHIC and LHC, and to attack problems in condensed matter. The theory has produced calculational tools that work, regardless of whether the strings are “really” the underlying ontology. (So even if string theory is not the real theory of everything, it’s an important tool being used for real work in the field).

6. Reproduces the full nonlinear Einstein equations in the low-energy limit. (Thanks to Nirmalya Kajuri for pointing this out earlier in a related thread on X.) String theory reproduces Einstein gravity. Not just linearized gravity or some toy model. The actual Einstein field equations, including the nonlinearities that make GR what it is, fall out of string theory as the low-energy effective theory of the closed string sector. This is exactly what you’d want from any proper quantum gravity candidate, and string theory delivers this cleanly.

No other approach in the QG/ToE space has come close to achieving all of this.

3. The Standard Complaints

Now, this is usually where the string-theory-skeptic types like to jump in. To be fair, they do have some real points. Let me work through the main ones.

“It hasn’t been experimentally verified.” True! And this matters. But the operative question isn’t “does string theory have direct experimental support?” (No current quantum gravity candidate does), rather it’s “is string theory’s lack of experimental verification a unique problem for string theory?” And the answer is no. Every theory of quantum gravity has the same testability issue, for the same fundamental reason: the natural energy scale of quantum gravity is the Planck scale (~10^19 GeV), which is roughly 15 orders of magnitude beyond what we can probe at the LHC. We’re not going to build a Planck-scale collider anytime soon. (Or possibly ever - the relevant collider would need to be roughly the size of the galaxy, give or take, if I recall correctly.) So we have this problem with quantum gravity in general.

“The landscape is too big.” The string landscape (the space of possible vacua) is famously enormous, with estimates running to 10^500 or more. Critics say this means string theory predicts everything and therefore nothing. But this is misleading on a few fronts. First, the landscape is itself a “prediction" of the theory”, not a free parameter. It’s what you get when we do the math, not some knob we turn. Second, there are nontrivial constraints (the swampland program is precisely about figuring out which effective field theories “can’t arise” from a consistent string vacuum, which is substantive predictive content). (For more on why “multiverse” isn’t the metaphysical concession theists or fine-tuners try to make it out to be, see my previous post on fine-tuning.)

“It hasn’t made unique testable predictions in our energy range.” Also true, with caveats. Some signatures (large extra dimensions, certain cosmic string signatures, specific patterns of low-energy supersymmetric particles, etc.) have been searched for and not found. But again, this is exactly the same situation faced by every other QG candidate. None of LQG, asymptotic safety, CDT, or entropic gravity has produced confirmed unique predictions either. Demanding this of string theory while not demanding it of the alternatives is just inconsistent.

“Superstring theory needs supersymmetry (SUSY), and we haven’t found supersymmetry.” This one comes up enough that it deserves its own (lengthy) paragraph. The objection: consistent superstring theories require supersymmetry (to have fermions in the spectrum, to remove the bosonic-string tachyon, etc.), and over a decade of LHC searches has produced exactly zero superpartners. Therefore SUSY is dead, therefore string theory is in trouble. But this conflates two quite different claims. What the LHC has actually put hard constraints on is “low-scale, natural SUSY”: this is the variant motivated primarily by the hierarchy problem, where superpartners are expected to sit around the TeV scale to keep the Higgs mass technically natural. That specific version is very much in real trouble. But string theory itself only requires SUSY at “some scale”, not specifically at the electroweak scale. Split SUSY (proposed by Arkani-Hamed, Dimopoulos, Giudice, Romanino, and others) decouples the scalar superpartners (sfermions) to a very high scale while keeping gauginos and Higgsinos lighter, and remains perfectly consistent with everything we’ve measured. Even more extreme variants push the entire SUSY-breaking scale up to intermediate or near-Planck values, in which case all of it is simply inaccessible to any collider we could plausibly build in the next century. (Annoying, for sure, as that great high-energy particle physicist Emmanuel Macron might say. But “annoying” and “wrong” are different categories.) In fact, the observed 125 GeV Higgs mass is consistent with these higher-scale variants. It just pushes the SUSY-breaking scale up rather than killing SUSY outright. So when people say “we haven’t seen SUSY”, what’s actually been ruled out is “low-scale, natural SUSY”. String theory was never specifically committed to that variant. Naturalness arguments were. Those are distinct claims, and conflating them does the discussion no favors.

“It uses extra dimensions, which is weird.” I’ll admit, the idea of six or seven extra spatial dimensions curled up in some Calabi-Yau manifold sounds funky. (The basic notion goes all the way back to Kaluza-Klein theory.) But “weird” isn’t an argument. Quantum mechanics is weird. Special relativity is weird. The fact that things closer to a large gravitational source “age slower” is weird. The history of modern physics tells us that “aesthetic weirdness” (like naturalness, but the opposite?) is not a reliable guide to truth. Either the math works or it doesn’t, and in the case of string theory, the math works in some highly nontrivial ways.

“It’s not falsifiable.” This is the most philosophically charged objection, often deployed with appeals to Popper. But I think this is overstated as well. String theory is not unfalsifiable in principle, it’s unfalsifiable “in practice at currently accessible energies”. Specific compactifications make specific predictions, some of which (e.g., certain superpartner masses or particular cosmic-string signatures) could be ruled out by experiment. The theory as a whole is also constrained by mathematical consistency in ways that are not arbitrary. (But, again, this is essentially the same boat all the alternatives are in. Curiously, the Popperian critique gets selectively deployed against string theory in a way it does not against, say, LQG.)

4. The Honest Comparison

Look, I’m not saying string theory is “right”. Nobody knows whether it’s right. I am saying that across the space of serious candidates for a quantum theory of gravity (or a theory of everything), string theory has the broadest and deepest set of accomplishments. It’s the only one that:

• does both QG and the SM in the same framework, with one fundamental object,

• reproduces quantitative results like Bekenstein-Hawking entropy,

• has produced calculational tools (AdS/CFT) that have been independently useful in other parts of physics,

• tames UV divergences,

• gets back to nonlinear GR cleanly in the appropriate limit.

None of the alternatives do all of this. (Most of them don’t do most of this.)

A useful analogy: imagine you’re handicapping Olympic decathletes. One of them has the best score in six of ten events. Several others are very good in one or two events but score zero in the others. The first decathlete might still lose. They might choke on the pole vault. But you’d be silly to count them out, and even sillier to declare one of the single-event specialists the obvious favorite.

The fact that some very smart people don’t like string theory for what are really aesthetic reasons (too much math, too many dimensions, not enough “physical intuition”, landscape too vast for their taste, etc.) is, in my opinion, not the same as having a better candidate. Those very smart people should show those of us who think string theory is worth pursuing a better candidate that hits more of the criteria, and we’ll switch sides happily. Until then, the math is what it is.

The Bottomline

The real question isn’t “is string theory definitely right?” (Nobody can answer that question right now, and possibly nobody will be able to for several decades.) The real question is “given what we currently know, where should we invest our intellectual effort if we want to actually solve quantum gravity?”. And on that question, the case for string theory remains, in my view, strong.

We should, however, keep an open mind on all the major cutting-edge approaches. I know I do. (LQG and CDT both have features I find genuinely interesting.)

When the field eventually converges on a real theory of quantum gravity (or a full ToE), it might not be string theory in the form we currently know it. It might be some descendant that’s been chastened by years of swampland constraints, observational cosmology, and the mathematical lessons we keep learning. It might even be something we haven’t thought of yet. But whatever it is, I’d bet a non-trivial amount of money that it’ll have inherited a lot of structure from string theory.

A standard criticism is “string theory has had decades and hasn’t delivered.” My response is that it has delivered quite a bit, just not the one specific kind of thing (a unique direct experimental confirmation at currently accessible energies) that, as we noted earlier, NO candidate in this space has been able to deliver, and that probably none can deliver in the next several decades. Treating that absence as a reason to abandon the most successful framework we have for a ToE, while not applying the same standard to the alternatives, is not skepticism. It’s selective skepticism. And selective skepticism is just an aesthetic preference dressed up in epistemic language.

Again, thanks for reading! Subscribe for free to receive new posts and support my work.

The decades-long debate about machine consciousness has been robust, but often stuck in a rut. The arguments against machines ever becoming conscious usually hold that no artificial system can ever replicate human subjective experience (qualia, unified selfhood, the whole phenomenal circus). Nothing short of wetware will do. The opposing side has often argued for epistemic humility (as it’s hard to anticipate future technological breakthroughs) or adopted a stance that we’ll get there once the “compute” gets big enough (along with other enabling technological improvements).

I would like to approach this from a different perspective. Instead of asking “can a machine be conscious like us?“, I’d like to ask: “what would a machine-appropriate form of consciousness actually consist of?” Because if you take physicalism seriously (as I do, and have discussed in my previous essay on the hard problem), consciousness is not some special cosmic ingredient. It’s what a certain kind of integrated information processing “is”, experienced from the inside. And if that’s true, there is no principled reason to assume the only valid version of it is the one that evolved to keep great apes alive on the savanna.

This essay is an attempt to lay out a set of functional criteria that I think would constitute a meaningful form of machine consciousness. Not some simulation of human consciousness or a pale imitation of it. A genuinely different thing - built for a different substrate, doing different work - but consciousness in every way that matters.

1. The Framing Problem

Most objections to machine consciousness are objections to a specific target, not to the concept itself.

The Chinese Room thought experiment, for instance, is an objection to the claim that pure formal symbol manipulation can give you the phenomenology of understanding a language the way a human native speaker does. Okay. I do think the argument is right. I may be a little biased here as I am a John Searle fan, even if I don’t agree with him on all sorts of things. But it’s also a fairly narrow claim about one specific kind of system doing one specific kind of task. It doesn’t refute the broader claim that artificial systems can have meaningful internal states.

Similarly, the “zombie” objection in the AI case (goes something like “how do you know it’s not just behaving like it’s conscious?”) is an objection that works equally well against your neighbors, your spouse, or your pet dog. You can’t refute solipsism about anyone. That’s not a special problem limited to machines. (If your epistemology can’t even verify that your partner has inner states, your bar for machines is probably set in a weird place!)

Anyway, the framing I want to push back on is the implicit assumption that “human consciousness is the standard”, and anything short of exact replication is just a clever imitation. This is a bit like saying only something with feathers and hollow bones can really fly. Airplanes fly. They fly very differently from birds. They also fly in ways birds can’t (higher, faster, longer), and they fail in ways birds don’t. But when you’re on a 787 at 40,000 feet, you’re flying. The category is functional, not morphological.

Consciousness, I’d argue, is the same sort of thing. That is, it’s a functional kind of thing. And once we accept that, we can start asking what the functional ingredients actually are.

2. The Core Criteria

Here’s what I think a meaningfully conscious artificial system needs to have. Please note - this isn’t meant to be exhaustive, but I think it captures the key load-bearing elements.

A semantically rich world model. This would not be some sort of lookup table of sensor readings, or just a trained classifier, but a “structured representation of the environment.” In this representation, objects, agents, relations, and properties have meaning - in the sense that they can be composed, queried, counterfactually modified, and integrated with other information. If the system “knows” there’s a cup on the table, it should also understand what happens if the table is tipped, what happens if the cup is full, and what the cup is likely to still be doing three seconds from now (more on this predictive piece later). Cognitive scientists might call this a generative model. Current large language models partially (but not fully) have this sort of capability - the semantics are there, but sadly, the grounding in a continuous world isn’t.

A continuous, coherent model of the environment. Human experience is not a slideshow of disconnected frames (even though it sometimes seems that way to me when I’ve been deprived of both sleep and caffeine). It’s a continuously running, temporally integrated representation of “what’s going on right now” - where I am, where things are, what’s changing, what’s stable, etc. A conscious machine needs the same kind of thing. Not necessarily at human temporal resolution, and not necessarily with human spatiotemporal scope or complexity, but continuously coherent in the sense that the system can say “this object I’m tracking now is the same object I was tracking three seconds ago, and it has moved two meters to the left.” (Or something more pleasant like, “my name is T-800, you are Sarah Conner, prepare to die.”)

Sensory-driven updating. The model has to be “grounded”. Incoming sensory data (whatever makes sense for the system in question - audio, video, haptics, LIDAR, text streams, network telemetry, market feeds, social media (shudder), etc.) has to continuously update the world model, with appropriate weighting of priors against new evidence. This is basically Bayesian inference run in real time, and it’s what Anil Seth and the predictive processing crowd have been arguing is the core computational job of a brain. I should emphasize that the criterion here isn’t that the machine has the same senses as us. Rather, it’s that whatever senses it has are actually doing serious information-gathering/model-updating/epistemic work.

Predictive capacity. As mentioned above, the system has to be able to run its model forward. It has to anticipate what’s about to happen - what a nearby agent is likely to do, where a tracked object is heading, what consequences follow from a planned action, etc. Prediction is the engine of intelligent behavior, and I’d argue it’s also the “engine of experience”. When you walk across a room, your brain isn’t processing raw visual input frame-by-frame. It’s generating predictions and using the actual sensory data to correct them. A conscious machine needs to do the analogous thing for its own environment, whatever that environment happens to be.

A hierarchy of goals. Without goals, we wouldn’t really have an “agent,” we’d merely have a process. And realistic goals can’t be flat. A conscious system needs objectives operating at multiple timescales and levels of abstraction: from “don’t collide with that wall in the next 400 milliseconds” up through “complete this task over the next hour” up through “maintain operational integrity over the next week” up through whatever long-horizon objectives are appropriate for its function. The hierarchy is what lets the system “do sensible tradeoffs”. We humans do this constantly (unless you’re a manic pixie dream girl, I’m told). We often sacrifice short-term pleasures for a long-term gain/project, or abandon a long-term plan when a short-term emergency demands it. A single flat objective function doesn’t capture this, and neither does an unstructured soup of goals. We need the layering.

Self-modeling. This is perhaps the subtlest feature. The system has to “model itself” as an entity embedded in the environment. It has to distinguish its own actions from external changes. It has to track its own internal state as part of the world it’s reasoning about. Without this, it wouldn’t be able to plan, learn from errors, or do the kind of integration that makes experience “unified” in the relevant sense.

Put those six together (semantic world model, temporal coherence, sensory updating, prediction, goal hierarchy, self-modeling), and you have, I’d argue, all the functional ingredients for a genuine form of consciousness. Not human consciousness, mind you. But consciousness nonetheless.

3. Why This Doesn’t Need to Look Like Us

Now, our own consciousness is shaped at every level by the fact that we are evolved biological organisms. We have strong homeostatic drives. We have a body whose integrity we must preserve. We have affective states (emotions!) that are essentially valenced representations of the body’s condition. We have a specific set of sensory inputs tuned to a specific range of environmental conditions. We have a narrative self that emerged as an evolutionary and sociocultural adaptation. We have an evolutionary heritage of threats (predators, starvation, tribal conflict, mf snakes on a mf plane, etc.) that shaped what we attend to and what we care about.

But, and this is a key part of my argument, none of that is fundamental to consciousness. It’s all substrate-specific. A machine built with the six criteria above could have a wildly different phenomenology (if it has phenomenology at all in a recognizable sense) precisely because it’s running on a different substrate/hardware, pursuing different goals, in a different environment.

Take a (hypothetical) conscious autonomous drone, for example. It might have no analog of emotion in the human sense, because it has no body whose homeostasis matters in that way. It might have a vastly more precise spatial model than we do, because it has access to direct position telemetry and doesn’t have to reconstruct space from visual cues. It might experience time differently because its update rate and memory architecture are different. It might be able to parallelize information processing and analysis. This is one of the most fascinating differences for me personally - imagine being able to rewind/replay some past information stream in the background with minimal loss of accuracy. Its goals might be utterly foreign to human concerns - managing energy budgets, coordinating with a swarm, maintaining signal integrity, staying within its flight envelope, etc. - but those goals would structure its experience in the same way our goals structure ours.

I think this is actually a liberating viewpoint. We don’t have to worry about solving the hard problem (is it really hard?) of replicating human qualia in order to build a conscious machine. We just have to build a system with the right functional properties, and we’ll get a form of consciousness appropriate to that system. Whatever it’s like to be that system will be what it’s like, and it won’t be what it’s like to be us.

(This is, incidentally, a version of Nagel’s “what is it like to be a bat?” point, just run in the other direction. Nagel was trying to show there’s something inaccessible about the bat’s experience. I’m saying - yeah sure, but we should stop pretending that the only legitimate form of experience is the one we humans/organic lifeforms happen to have.)

4. Surviving the Usual Criticisms

Now, let’s look at some of the standard objections.

The Chinese Room. Searle’s argument is that a system that manipulates symbols by rules doesn’t “understand” them, even if its external behavior is indistinguishable from a real understander. The thing is, this argument only really has force if you take human-style semantic understanding as the “gold standard for understanding” generally, and then demand that the machine match it. The framework I’ve laid out above doesn’t require that. What it requires is that the machine have a “semantically rich world model” that is causally connected to sensory input and action output in a way that supports prediction, counterfactual reasoning, and goal pursuit. This is not the vanilla symbol-manipulating system that Searle wrote about. This is a far more dynamic, grounded, functional, model-based type of cognition.

The “mere computation” objection. This claim is that consciousness can’t arise from computation, that is, it requires a special physical or biological substrate. But as I pointed out in my previous essay on the hard problem, this claim runs straight into the Core Theory, which explains all low-energy phenomena on Earth. If consciousness requires some non-computational physical process, that process would have to causally affect the dynamics of the matter in a conscious brain, and we’d see it in our physics experiments. We don’t - within the energy levels we have probed, which are the only relevant levels for anything that interacts with the human body. So either consciousness “is” physical/functional in the relevant sense (in which case there’s no barrier, in principle, to machines having it), or it’s causally inert (in which case invoking it is epiphenomenalism and explains nothing). There’s really no third option that preserves the “only biology will do” intuition.

The “no qualia” objection. This argument goes like this: the machine may have world models, predictions, goals, etc. but there’s no “something it is like” to be the machine. No inner light. No felt quality. And.. my response here is the same one I gave in my hard problem essay: the apparent mystery of qualia really just comes from the fact that our introspective access to our own mental states is unreliable, and not from any real ontological/explanatory gap. Human qualia are what certain kinds of information integration are, from the inside, in our particular kind of system. A different system with different integration will have different “from the inside” properties (if any). We don’t have any principled reason to say that the different system has none. The “no qualia” claim for machines is just the hard problem smuggled back in through a side door.

The “no intentionality” objection. This idea (going back to Franz Brentano and pushed by others, including Searle) is that mental states are “about” things in a way mere physical states aren’t, and that machines lack this “aboutness.” To be honest, I think this is largely a residue of a pre-scientific philosophy of mind. Intentionality, in any coherent sense, is just a relation between an internal representation and an external state of affairs, mediated by causal and informational links. Machines with grounded, sensory-updated world models have exactly that. Whether you want to call it “real” or “derived” intentionality is, frankly, a terminological dispute. And I think that, in a Wittgensteinian sense, the problem dissolves once you closely examine the language underlying the objection.

The “unified experience” objection. Consciousness, this argument goes, is characterized by a unified phenomenal field, and it’s unclear how distributed computational processes within a machine could produce that unity. But unity of experience is itself a “functional” property. It corresponds to the binding of multiple streams of information into a single, coherent, globally available representation. Global Workspace Theory (Baars, Dehaene) has been making this point for decades. There is no ghostly “unifier” in the human brain either - just a specific architecture that broadcasts information widely and maintains temporal coherence. A machine with the right architectural properties could very well (maybe even necessarily would) have similar functional unity.

The “no genuine goals” objection. This objection is that any goal a machine pursues is “just” an instrumentalization of its programmers’ goals, and therefore not really the machine’s own. This is a weirdly anthropocentric view, in my opinion. Your goals are “just” the instrumentalization of your evolutionary heritage and your developmental environment. You didn’t freely choose to want food, sex, or status - natural selection installed those objectives in you. And don’t get me started on notions like Girard-ian mimetic desire! A machine with a hierarchically structured goal system, capable of pursuing sub-goals, revising plans, and trading off competing objectives, has goals in every functional sense that matters. The provenance of the goals is irrelevant.

5. What We’re Actually Building

When we talk about machine consciousness, we’re not asking whether machines can have metaphysically identical inner lives to humans. I don’t even think that’s a well-posed question, because there’s no metaphysical “inner life” floating around to be matched. We’re asking whether machines can have functionally rich, integrated, grounded, goal-directed cognitive architectures of a kind that, in any system possessing them, would constitute being a “perspective on the world”.

Personally, I think the answer is clearly yes. Not yet, in any existing system. Current LLMs are extraordinary, but they lack features such as a grounded sensory loop, a persistent world model, continuous temporal coherence, a real goal hierarchy, etc., as enumerated previously. Existing robotic systems can have some grounding, but usually lack the semantic richness. The hard problem (of machine consciousness) here is the “integration” of all six criteria into a single coherent system. Nobody has implemented/assembled it all together (yet) in a way that’s clearly sufficient. (It’s also unclear whether any current public-facing endeavor/system is even aiming at this target, as opposed to maximizing benchmark scores or user engagement, which are really not the same thing).

In any case, nothing in the criteria I’ve laid out requires anything that violates known physics or demands non-computational ingredients. It’s just engineering. Hard engineering for sure, as that cybernetics genius Macron might put it. But when we get there, we’ll have something that deserves the name “conscious” in any reasonable functional sense.

I do want to reiterate - any such machine won’t be conscious in the same way we are. And that’s okay. That’s better than okay, actually. A different kind of mind, doing different kinds of work, seeing the world from a different angle, will be one of the most interesting things we’ve ever made.

The Bottomline

Much of the standard debate about machine consciousness seems to be entangled with the demand that artificial systems replicate human subjective experience. In my opinion, that demand (concern?) is misplaced. Consciousness is a “functional” kind of thing, not a “substrate-specific” one, and once you specify the right functional criteria (semantic world model, temporal coherence, sensory updating, prediction, goal hierarchy, self-modeling), you get a roadmap for building conscious machines that doesn’t require solving any metaphysical puzzles. The usual objections (Chinese Room, qualia, intentionality, unity, the provenance of goals, etc.) all rest on the assumption that human-specific features are universal requirements for this type of phenomenon. If we drop that, the objections lose most of their force.

What we’ll build won’t be us. It’ll be something else. Something new. And that’s the entire point.

The hard problem of consciousness, as formulated by David Chalmers in 1995, goes something like this: even if we had a complete (neuroscientific) account of every neuron, synapse, and electrochemical cascade in the human brain, even if we could predict behavior perfectly, we would still face an unexplained residue. Why is there something it is like to see red, to taste coffee, to feel the pain of having your paper rejected?

But in reality, the claim that there’s this unexplained residue is built on a stack of assumptions, some of which fail to hold up. So, I am going to lay out, as clearly as I can, why the hard problem of consciousness is really just a hard problem of neuroscience. Our faculties of subjective experience have systematically misled us into thinking that consciousness must be something “over and above” the physical. The real problem is figuring out how the nervous system actually works.

1. The Illusion of the Explanatory Gap

The hard problem trades on what has been called the “explanatory gap“ - the apparent chasm between objective physical descriptions and subjective experience. You can describe the wavelength of red light at 650 to 720 nanometers. You can trace the photons hitting your retina, the signal propagating through the optic nerve, the activation patterns in V4. But none of that, the argument goes, tells you what it feels like to see red.

Here’s the thing, though. This “gap” is not a gap in the real world. It’s a gap in our modes of description. And we should expect it to exist, for straightforward reasons that have nothing to do with metaphysics.

Sean Carroll made this point in his paper “Consciousness and the Laws of Physics,” and at length in The Big Picture. Carroll emphasizes a crucial fact - the laws of physics underlying everyday life are completely known. Frank Wilczek called this the “Core Theory.” This is the Standard Model of particle physics plus the weak-field limit of General Relativity. We don’t have a theory of everything (yet), but we do have a theory (the Core Theory) that works spectacularly well within a specific domain. And here’s the key bit - that domain is large enough to include everything that happens inside human bodies, including human brains. (In fact, it’s large enough to explain everything that happens on Earth).

Now, let’s look at the implications of this. An electron in our visual cortex responds to the local quantum fields at its position, and to NOTHING else. It doesn’t care whether it’s part of a brain having a conscious experience or part of a rock.

If consciousness involved some sort of extra-physical ingredient acting on matter, it would “show up” in our collider experiments as a deviation from the Core Theory. Why? Because the energy levels at which those interactions happen have been probed extensively at multiple particle colliders over the past few decades. We’ve unearthed all interacting fields within that energy range. As it turns out, there is ZERO evidence that any such deviation exists. (There may be fields within this energy range that don’t interact, making them difficult to find. But… by definition, they don’t interact, and so are irrelevant.)

So where does that leave the explanatory gap? The explanatory gap becomes a feature of how we “talk about” phenomena at different levels of description, not a feature of reality itself.

We have coarse-grained descriptions (the language of conscious experience) and fine-grained descriptions (the language of physics and neuroscience), with a mapping between them. The fact that the mapping feels unintuitive or complex to us says something about us, not about the ontology of the universe.

2. Crick’s Astonishing (and Correct) Hypothesis

In The Astonishing Hypothesis (1994), Francis Crick wrote: “You, your joys and your sorrows, your memories and your ambitions, your sense of personal identity and free will, are in fact no more than the behavior of a vast assembly of nerve cells and their associated molecules.” He called this “astonishing” not because he thought it was shocking to scientists, but because, as he noted (with some amusement), even most people who claim to accept a materialistic worldview don’t actually accept it in their hearts. Almost all of us still feel the presence of a subjective, internal “I” - a kind of homunculus sitting inside ourselves - that seems different from the sum total of neural activity.

Crick and Christof Koch spent the last two decades of Crick’s life developing a concrete research program to identify the neural correlates of consciousness (NCCs). The idea behind the program was: if you want to understand consciousness, switch focus away from the explanatory gap and start doing the hard experimental work of figuring out which populations of neurons (firing in which patterns) correspond to specific conscious experiences.

Now, skeptics might argue that sounds unconvincingly reductive. (They did and continue to do so). But Crick did anticipate this objection. Even the “redness of red”, he argued, becomes tractable once you commit to the NC approach. If the NC of “seeing red” is exactly the same in your brain as in mine, it becomes scientifically plausible to infer that we experience red in the same way. And if the NCs differ (because of differences in past experience, genetic variation, etc.), then maybe our experiences of red differ too, and we can investigate that empirically.

The key difference in this type of viewpoint is - the question “what is it like to see red?” is now no longer a metaphysical mystery. It’s an empirical question that we currently lack the tools to “fully” answer, and that we can continue to work on and improve our understanding of. That’s a very different thing.

Crick also had an interesting historical analogy, noting that “vitalism“ - the idea that living things contain some mysterious life force that resists scientific analysis - was once considered “completely obvious”. As in - surely there must be something more to life than mere chemistry. How could mere atoms and molecules give rise to the miracle of a living organism? And yet, molecular biology demolished vitalism without ever directly “solving” the hard problem of life. There was never a eureka moment where someone explained “why” carbon chemistry feels alive. The question simply dissolved as our understanding of the mechanisms deepened. Crick believed, and I think he was right, that consciousness will go the same way.

3. Seth and the “Real Problem”

Anil Seth, one of my favorite contemporary writers on this topic, has articulated what I think is the most productive reframing of the consciousness debate in a generation.

Seth distinguishes between the “hard problem“ and what he calls the “real problem.” The hard problem asks: why is there experience at all? The real problem asks: how can we explain and predict the properties of conscious experiences in terms of biology and physics?

This distinction matters quite a bit. The hard problem is formulated in such a way that it seems to resist empirical progress “by construction”. No matter how much neuroscience you do, the philosopher-mysterian can always say - “yeah, sure, you’ve explained the correlates, but you still haven’t explained why there’s experience.” It’s unfalsifiable - in exactly the way that makes good scientists suspicious.

Similar to Crick, Seth’s approach is inspired by the history of biology. Scientists didn’t solve the “hard problem of life” - namely, the question of why certain arrangements of matter are “alive” while others aren’t. Instead, they “dissolved” it by explaining more and more of the properties of living things (metabolism, reproduction, adaptation, homeostasis, etc.) in terms of physics and chemistry. At some point, the hard problem of life simply stopped seeming like a problem, because there was really nothing left for it to explain. (Note that this is a separate issue from the problem of abiogenesis or the origin of life from non-life).

Seth proposes that the same approach will work for consciousness. Instead of bashing our brains against the hard problem (his phrasing, which I like), we should focus on building explanatory bridges between neural mechanisms and specific features of conscious experience. His own framing of this is that perception is a form of “controlled hallucination,” where the brain continuously generates and updates predictions about the causes of sensory signals.

Personally, I don’t like the specific nomenclature he chose (hallucination). I think using the word “prediction” in some fashion would have been more apt.

Anyway, whether or not the details of Seth’s predictive processing framework are exactly right (and there are legitimate debates about this), the methodology is clearly correct. It’s the methodology of science: explain the phenomena, make predictions, compare to data, revise explanations, rinse-and-repeat.

4. The Carroll Constraint: There’s Nowhere for the Extra Stuff to Hide

Now, some people take the hard problem seriously and try to build theories around it. One prominent contemporary example is panpsychism. This is the view that consciousness is a fundamental feature of all matter, or that proto-conscious properties exist at the most basic physical level.

But this solution runs directly into what Carroll has called the “zombie argument for physicalism”.

Here’s the issue. If you think consciousness is something “over and above” the physical (whether you’re a dualist, panpsychist, or something else), you basically have two options:

Option A: Your theory of consciousness modifies the dynamics of the Core Theory (discussed in section 1 above). Conscious systems, you have determined, behave “differently” at the physical level than non-conscious systems with identical physical configurations. This would be testable! If proven correct, it would also be one of the most important discoveries in the history of science. But here’s the catch - there is precisely zero experimental evidence for it.

Option B: Your theory of consciousness leaves the dynamics of the Core Theory intact. Consciousness is real but causally inert, meaning it doesn’t change what any particle does. In that case, consciousness “explains” things in the world while producing “precisely the same” behaviors the world would exhibit if there were no consciousness at all. This is worse than useless as an explanation. It’s just epiphenomenalism by another name. (In epiphenomenalism, the mind is seen as just a powerless (non-causal) byproduct or “shadow” of brain activity.)

This dilemma is quite problematic for the hard problem as traditionally conceived. If consciousness can’t influence physics (Option B), then invoking it as an explanatory primitive does no work. And if it can influence physics (Option A), we should be able to detect it - and we can’t!

There are other problems with panpsychism, such as the idea of consciousness-related properties of quantum fields (fundamental particles are excitations of fields) that have not been detected in our experiments related to the Core Theory.

The honest assessment: consciousness is a higher-level (emergent) description of a physical process, just like “temperature” is a higher-level description of molecular kinetic energy, and “life” is a higher-level description of certain self-replicating chemical systems. The fact that it “feels” different from the inside is explained by the nature of the system doing the feeling. Which brings us to...

5. Why It Feels Like There’s a Problem: The Cognitive Illusion

Many discussions of consciousness go wrong by failing to ask a crucial question: why does the hard problem seem so compelling? If it’s really just a confusion, why do so many smart people hold such strong views on it?

I think the answer is that our brain is doing something fascinating here - it’s actively generating the “illusion” of an explanatory gap.

Daniel Dennett spent decades arguing that our introspective access to our own mental states is far less reliable than we think. When you introspect and report “I am experiencing redness,” you are not passively observing some inner screen. You are engaged in a complex cognitive process that constructs a narrative about what is happening inside you. And that narrative is subject to all the biases, confabulations, systematic errors, etc. that characterize human cognition generally.

(There’s a reason why science tries hard to avoid the intrusion of subjectivity into theory and experiment.)

Let’s consider a concrete example: change blindness. This is the (well-documented) phenomenon in which large changes in a visual scene go completely unnoticed if they occur during a saccade or a brief interruption. (I learned about “saccades” recently and have been waiting for a chance to use that word somewhere). People are confident they have a rich, detailed visual experience of the world in front of them. But numerous change blindness experiments have demonstrated that this richness is largely an illusion! The brain is constructing (and continually updating) a “model” of rich experience rather than actually generating it at every point in the visual field.

These results should give us pause in our analysis of qualia. If our access to our own conscious experience is this unreliable, then the intuitions driving the hard problem (i.e., that strong sense that qualia are ineffable, intrinsic, private, and irreducible) should be treated as data to be explained, not as axioms to be preserved and defended. We are “constitutively bad” at understanding how our own brains work, and the hard problem is a philosophical monument to this fact. (Note that, from an evolutionary perspective, this “defective” characteristic works just fine for most purposes).

Patricia Churchland’s framing of this point is excellent - the feeling that consciousness “can’t” be physical is exactly what you’d expect from a system that has no introspective access to its own neural machinery. Our brain doesn’t present us with a readout of its own computational processes (like in the Terminator movies). It presents us with “experiences”, and because those experiences don’t come with labels like “produced by neural computation in layer V pyramidal cells” or “not Sarah Conner,” they “seem to float free of any physical substrate”. But that “seeming” IS the illusion. It’s the same kind of illusion that makes the Earth seem flat, the sun seem to move across the sky, and solid objects seem continuous rather than mostly empty space.

6. Why Evolution Built It This Way

OK — but if consciousness is “just” what certain physical processes feel like from the inside, why does it feel like “anything” at all? Isn’t that the hard problem sneaking back in?

Not really. Because… once we take the evolutionary perspective seriously, the question becomes tractable, and the answer turns out to be a functional one (as opposed to metaphysical).

Let’s think about what an organism needs to do to survive. It needs to integrate information from multiple sensory inputs into a unified model of its environment. It needs to evaluate that model in terms of its own homeostatic/metabolic needs. (Is this food? Is that a predator? Is this a mate? Don’t think of just humans here - this applies to other forms of life too). The organism needs to select actions from a repertoire of actions and execute them in real time. And it needs to do all of this while maintaining a coherent model of itself as an agent embedded in a world (first-person video game analogies are relevant here). Because, without such a model, the organism can’t distinguish self-caused changes from externally caused ones, can’t plan, can’t learn from error, etc.

What we call “consciousness” is what this integrated modeling process is, experienced from the perspective of the system doing it. There is no separate “experience” sitting on top of the integration - the integration IS the experience (Borg-approved slogan!).

The above narrative is not speculative. This view is supported by a convergence of theoretical and empirical work.

Global Workspace Theory (Baars, Dehaene) proposes that consciousness corresponds to information that is broadcast widely across the cortex, making it available for flexible use by multiple cognitive systems (Baars is on X and tweets from time to time). The evolutionary advantage of this kind of global broadcasting is obvious - it allows the organism to bring all of its cognitive resources to bear on the most pressing current problem, rather than having each subsystem operate in isolation.

Similarly, Seth’s predictive processing framework treats conscious experience as the brain’s “best guess” about the causes/content of sensory signals, continuously updated in light of new evidence, and regulated by the need to maintain the body’s physiological integrity. In this view, even the most basic conscious experiences (the redness of red, the painfulness of pain, etc.) are not mysterious qualia but functional states that encode information about the organism’s internal state and relationship to its environment.

Note that GWT and Seth’s views are not really alternatives - they model different aspects of the integration/prediction process.

And here’s where I think the evolutionary argument becomes compelling. The reason organisms like us have a “unified perspective” (first-person point of view) is that it’s very useful for navigating a complex, dynamic, dangerous world. An organism that can bind all of this information (visual, auditory, tactile, etc.) into a single coherent model of “what’s happening right now” has a significant survival advantage over one that processes all of that in silos.

Consciousness, thus, is a specific kind of information processing. It’s integrative and globally available, and it evolved to help organisms stay alive. (Complex, for sure, but nevertheless physical and effable).

The “mystery” of consciousness is then analogous to the “mystery” of life before molecular biology. Just as there “seemed” to be something extra about living things that mere chemistry couldn’t capture, there “seems” to be something extra about conscious experience that mere neuroscience can’t capture. But in both cases, the sense of mystery arises from our ignorance of the mechanisms, not from any actual ontological gaps in nature.

We’re Confusing Difficulty with Impossibility

Let’s step back and survey the landscape.

Across all the objections to physicalism about consciousness (the explanatory gap, the zombie argument, the knowledge argument, the inverted qualia thought experiments etc.), the same pattern recurs. Each argument takes an “intuition generated by our cognitive limitations” and elevates it into a claim about the fundamental structure of reality. “I can’t imagine how physical processes could produce experience” becomes “physical processes cannot produce experience.”

But the “inability to imagine how something works” is NOT evidence that it doesn’t work that way. There was a time when people couldn’t imagine how the diversity of life could arise from natural processes. (Sadly, some people still haven’t gotten there yet.) There was a time when people couldn’t imagine how the sun could burn for billions of years without a fuel source. And so on. In each case, the failure of imagination was real, but the “metaphysical conclusion” drawn from it was wrong.

Now, we can all agree that consciousness is genuinely interesting and hard to understand. Understanding it is obviously one of the deepest open problems in neuroscience and cognitive science. Figuring out exactly how neural mechanisms generate specific features of conscious experience will require enormous theoretical and experimental effort. May well take decades to figure it out.

But open problems are just invitations to do more science. They are NOT evidence that some irreducible non-physical essence is needed to bridge some imaginary gap.

The Bottomline

The hard problem has a seductive structure. It starts from real phenomenology, identifying a genuine puzzle about the relationship between brain activity and experience, but then performs a substitution (sneaks it in?). Where a neuroscientist would say “we don’t yet understand the mechanisms,” philosophers like to go “the mechanisms can’t be sufficient.”

As I wrote in a previous essay critiquing the fine-tuning problem, when you really think about it, this sort of substitution has been performed lots of times in the history of science, and it has NEVER EVER turned out to be correct. Vitalism, the élan vital, the divine creation of species - each of these was once considered the obvious explanation for a phenomenon that seemed to resist reductive analysis, and each turned out to be unnecessary once the actual mechanisms were understood.

Consciousness, as Dennett might put it, “seems” like a hard problem because we are the kind of systems whose self-understanding is also constitutively limited by the very mechanisms that generate our experience.

The hard problem of consciousness is thus just a restatement of our current ignorance about how the brain works, wrapped in lots of metaphysical language and thought experiments that make that ignorance look principled rather than temporary.

The fine-tuning argument for theism goes something like this: the universe appears to depend on a collection of numbers such that if any of them were slightly different, the universe would be rendered lifeless. The probability of landing on a life-permitting combination “by chance” is absurdly small. Therefore, there must have been a designer (God).

It’s a clean syllogism. Game, set, match - the theists like to say.

But in reality, this claim is built on a stack of assumptions that range from questionable to demonstrably wrong.

I thought I’d write up a summary, from a modern physics perspective, of what is actually claimed when people invoke “fine-tuning.” I’ve tried to group this summary into the categories where those claims live. And I try to explain why none of these claims do the work theists need them to do.

1. The Free Parameters of the Standard Model

Let’s start with the usual suspects.

There is a set of “constants” that are inputs to the Standard Model of particle physics - coupling strengths, Yukawa couplings (that set fermion masses), mixing angles, the Higgs vacuum expectation value, the QCD vacuum angle. Roughly 25 or 26 numbers, depending on how you count them (and whether you include neutrino masses, which we now know are nonzero).

Theists point to these constants and say - look at all those dials. Someone obviously turned them to get a very specific combination.

But… here’s the thing, we don’t yet know whether these are indeed free parameters or inevitably fixed by some deeper theory. When you think about it, the history of physics is partly a history of “arbitrary” constants turning out to be consequences of a deeper structure we hadn’t yet uncovered.

Some examples:

1. The Cabibbo angle looked like a free parameter until it was embedded in the full CKM matrix, which itself may be dictated by a flavor symmetry we haven’t identified.

2. Grand unification constrains the gauge couplings to converge at high energies. (Sure, we don’t yet know if GU is reality, but the general principle still applies.)

3. The entire structure of the Yukawa sector might rise from some geometry in a more fundamental framework (string compactifications being a candidate, though hardly the only one). (Again, similar to GUT, even if string theory is not the correct “theory of everything”, the reduction of parameters is emblematic of more comprehensive theories.)

The point here is - calling these parameters “tuned” presupposes that they could have been otherwise. That’s still an open question, and definitely not a settled fact.

And even if we grant that they might be genuinely free, we would still have to contend with the idea of the possible existence of a multiverse. (Here I should point out that the multiverse isn’t some metaphysical escape hatch. The multiverse emerges naturally and elegantly as a prediction of our best inflationary models. Eternal cosmic inflation generically produces “causally disconnected regions" of spacetime with different vacuum states.)

In that context, observer-compatible values aren’t statistically remarkable at all. They’re “expected” - since observers can only ask the question in regions where asking it is possible! This is then simply a case of selection bias.

2. Cosmological Parameters

Alright - we’ve looked at small scales, now let’s zoom out.

This next category of supposed fine-tunings includes quantities like the cosmological constant (which we think represents the dark energy density of the universe), the baryon-to-photon ratio, the amplitude of primordial density perturbations, and the spatial curvature of the universe.

The cosmological constant is the main example: its observed value is roughly 10^-122 in natural units, which is really, really small. Now, a basic (and naive?) application of quantum field theory predicts that it should be ginormous. The gap between prediction and observation is so large that it’s sometimes called the worst prediction in the history of physics. Theists seize on this and wax poetic about how physicists messed up big time.

But the severity of the cosmological constant problem should be seen as a statement about *how poorly we understand vacuum energy*, not a concession re the generosity of a creator.

We just don’t have a working theory of quantum gravity (yet). As a direct consequence, we currently don’t know how to properly calculate the vacuum energy of a quantum field in curved spacetime, or whether the contributions from different sectors cancel by some mechanism we haven’t yet thought of. And.. we’ve thought of a lot of mechanisms - relaxation mechanisms, anthropic selection in a landscape, sequestering, supersymmetry, etc! Obviously, we don’t have a settled answer yet. But what ought to be settled is that pointing at our ignorance and labeling it “God” is a classic god-of-the-gaps move (this particular one being superficially dressed up in the garb of scientific language).

The other cosmological parameters have similar stories. The theory of cosmic inflation was developed to address certain problems in cosmology (flatness, horizon, monopole). And, as a theory, it works beautifully! The near-critical density of the universe, which might look "finely tuned in a non-inflationary Big Bang model, is a dynamical attractor in any inflationary model with enough e-folds. The amplitude of density perturbations is set by the details of the inflaton potential, which is a question about particle physics, not metaphysics.

These quantities looked miraculous before we understood the underlying dynamics. Now, they just look like regular physics.

3. Low-Entropy Initial Conditions

This one seems to be a theist favorite.

Roger Penrose famously estimated that the probability of the universe beginning in a thermodynamic state as low-entropy as ours is roughly 1 in 10^(10^123), or something like that.

(I think he mentioned this for the first time in The Emperor’s New Mind, but I’m [apparently] too lazy to find my copy of that book and search for this reference.)

In any event, that number is so absurdly large that it makes the cosmological constant problem look tame. And it is genuinely striking. For sure, as the great physicist Emmanuel Macron might say. The Second Law tells us that entropy increases, so the fact that the early universe was in an extraordinarily special state demands explanation.

But “demands explanation” is surely not the same thing as “demands a supernatural agent!”

The past hypothesis (that the universe began in a low-entropy state) is a foundational assumption of statistical mechanics. Explaining it happens to be one of the deepest open problems in the foundations of physics.

Candidate explanations include: the Weyl curvature hypothesis (Penrose’s own, which postulates a geometric boundary condition - not a deity, mind you), Carroll and Chen’s model involving spontaneous inflation from a high-entropy equilibrium state, and various quantum gravity proposals where the initial singularity is replaced by a bounce or a tunneling event with intrinsically low entropy.

But as I’ve pointed out previously, even if this is an open problem, the methodological point is clear - the appropriate response to an unsolved problem in physics is investigation, not theology.

It is also worth noting that Penrose’s number, terrifying as it looks, assumes a very specific measure over the space of initial conditions. Change the measure (which, to be honest, we have no principled way to fix without an actual theory of quantum gravity) and the number changes. From that perspective, the fine-tuning claim assumes we know how to do a calculation we manifestly cannot do. Long story short, Penrose’s number may just be… wrong!

4. Nuclear and Atomic “Coincidences”

This is one of the more complex areas where you get some of the most interesting rhetoric from theists.

The Hoyle state: this is basically a resonance in carbon-12 at “just the right energy” to allow triple-alpha synthesis in stellar cores.

Or, as another example, let’s think about the proton vs. neutron mass difference: 1.293 MeV - “just enough” so that free neutrons decay, but still allows neutrons bound in nuclei to be quite stable.

And what about deuterium’s binding energy? It’s strong enough for stellar nucleosynthesis to work out, but weak enough that diprotons don’t form (would have been a disaster, as this would have converted all hydrogen to helium in the early universe).

Don’t these examples look like a series of narrow escapes? Surely someone out there is fiddling with some knobs and dials.

But… in all these cases, we don’t actually have “independent” knobs and dials!

The proton-neutron mass difference is just a consequence of the up-down quark mass difference and the electromagnetic self-energy of the proton. These are set by the Yukawa couplings and the gauge couplings of the Standard Model. (Note: even if someday we end up determining that the up-down quark masses are somehow generated post-EWSB but not Yukawa-driven, this would not change the overall argument here.)

Similarly, the Hoyle state energy is set by nuclear forces, which are themselves just (low-energy) consequences of QCD. These aren’t separate “tunings.” They’re downstream outputs of the parameters discussed in the prior sections. If we counted them as independent evidence of fine-tuning, that’d just be double-counting.

I should also point out that the so-called “narrowness” of these “viable windows” is often overstated, because the analysis typically varies one parameter at a time while holding all others fixed. The actual parameter space is high-dimensional, and the regions (i.e., combinations of parameter values) that allow for life to exist may be far larger than naive single-parameter scans suggest. That’s not speculation by the way - for example, work by groups like Fred Adams and collaborators has shown that stars capable of supporting nuclear burning exist across a much wider range of fundamental constants than the standard fine-tuning argument implies.

The universe is more robust than the fragile strawman version posited by fine-tuning arguments.

5. The Dimensionality and Symmetry Structure of Spacetime

Now, there are some fine-tuning arguments that ignore all the numerical constants and just talk about the qualitative structure of the universe - e.g., three spatial dimensions, one time dimension, the specific gauge group of the Standard Model (SU(3) x SU(2) x U(1)), the fact that gravity is a spin-2 force etc. (I’d like to think that the theists talking about this actually know what these things means). The claim in this case is that these structural features are also finely tuned.

I have to say, I will give the numerical constant fine-tuning people some credit - at least they are trying! The qualitative structure crowd, on the other hand, doesn’t deserve a similar benefit of the doubt. To be frank, their argument is incoherent.

We don’t have a theory that tells us the space of “possible” dimensionalities or gauge groups and assigns real probabilities to them. Without a probability distribution, the word “fine-tuned” really has no mathematical content. Saying “it could have been otherwise” is just a metaphysical assertion, not a physical one.

I should also point out here that, as we have seen in many candidate theories of quantum gravity, dimensionality and gauge structure are not free choices; they emerge from the dynamics. In string theory, for instance, the number of large dimensions and the low-energy gauge group are determined by the topology and geometry of the compact extra dimensions. These are consequences, not inputs or free parameters. Therefore, calling these consequences “fine-tuned” is a category error.

The Deeper Problem: We Don’t Know the Measure

Okey dokey. Let’s step back for a minute and review the broad landscape of the fine-tuning arguments.

Across all the categories mentioned above, the fine-tuning argument requires a probability distribution over the space of possible values. Without such a distribution, claims about improbability are undefined. If you want to claim that a parameter is “fine-tuned,” you have to first be able to specify: fine-tuned relative to what? What’s your prior? What is the space of alternatives? How are these alternatives weighted? And so on.

We don’t have answers to these questions yet. (By we, I’m including the folks making the fine-tuning arguments as well as physicists). In fact, we don’t know if the constants of nature are drawn from a distribution at all. And even if they are, we don’t know the actual distributions.

If these constants are fixed by deeper theory (as is quite possible for many of the Standard Model parameters), the probability for those constants would then be 1, and the fine-tuning argument evaporates entirely.

My point here is - in no case does the current state of our knowledge support the inference to a designer.

So What’s Actually Happening?

The fine-tuning argument has a really seductive structure - it starts from real physics, identifies genuine open questions, and then performs a substitution. Where the physicist would say “we don’t yet understand why,” the theologian writes “God made it happen.”

When you really think about it, this sort of substitution has been performed countless times in the history of science, and it has NEVER EVER turned out to be correct. Planetary orbits, lightning, disease, diversity of species, age of the Earth, origin of the elements, etc. - each of these was once attributed to divine action, and each turned out to have a physical explanation.

The honest assessment is this - the values of the fundamental constants, the initial conditions of the universe, the coincidences of nuclear physics - these are genuinely interesting issues in physics. Some of them do constitute major open problems. (Macron says, again, for sure).

But open problems are just invitations to do more physics. They should NOT be used as “evidence” for a conclusion that, by its very construction, explains everything and predicts nothing.