Philosophical zombies

A philosophical zombie — sometimes abbreviated as “p-zombie” or simply “zombie” in the technical literature — is a hypothetical being that is molecule-for-molecule identical to a conscious human being, indistinguishable in physical structure, behavior, and functional organization, yet wholly devoid of subjective experience. A zombie that stubs its toe will yelp, withdraw its foot, form the belief that something painful has occurred, report the pain to a doctor, and remember the incident later, all while there is nothing it is like for the zombie to undergo any of this. The lights, in the philosophical sense, are off. The thought experiment is not a contribution to horror fiction but a tool for testing whether the physical facts about a person logically determine the facts about that person’s conscious experience. If a creature physically identical to a conscious human could lack experience even in principle, then experience is not entailed by the physical, and physicalism — the view that everything is physical or fixed by the physical — is false.^{3, 5}

The zombie thought experiment occupies a distinctive place in the philosophy of mind because it tries to extract a metaphysical conclusion from a purely modal premise. It does not appeal to neuroscience or to any contingent feature of human cognition; it asks only whether a certain scenario can be coherently described. The argument therefore stands or falls on contested questions about the relationship between conceivability and possibility, the structure of our concepts of experience, and the proper interpretation of the necessary a posteriori. Since David Chalmers’s The Conscious Mind (1996), the zombie argument has been the most discussed and most contested device in the contemporary debate over consciousness, central to disputes about the hard problem of consciousness, the prospects for physicalism, and the explanatory ambitions of cognitive science.^{3, 4, 5}

Origins and terminology

The English word “zombie” entered philosophical use in the 1970s, but the underlying idea of a behaviorally and physically perfect duplicate that lacks inner experience has older roots. Descartes’s discussion of automata in the Discourse on the Method (1637) raised a structurally similar question by entertaining whether a machine could imitate human conduct so successfully that one could not tell it apart from a human, and Leibniz’s “mill” passage in the Monadology (1714) argued that even if one could enter the brain of a perceiving being and walk among its mechanical parts, one would find nothing that explained perception itself. These early sources prefigure the central intuition that mechanical and physical organization, however complete, seems to leave the phenomenal character of experience untouched.⁵

The first sustained philosophical use of a zombie-like figure in twentieth-century analytic philosophy appears in Keith Campbell’s Body and Mind (1970), which introduces the “imitation man” — a being “exactly like a man in every observable respect” whose internal physical and chemical states are identical to those of a normal human, but whose mental life consists only of beliefs and desires without any accompanying sensations. Campbell used the imitation man to argue against behaviorism and against any view that identified sensations with their behavioral or physical correlates. Robert Kirk, working independently, deployed the term “zombie” itself in two papers published in 1974: “Sentience and Behaviour,” in Mind, and “Zombies v. Materialists,” in the supplementary volume of the Proceedings of the Aristotelian Society. Kirk argued that the conceivability of beings physically and behaviorally identical to humans but lacking sentience showed that no purely physical or functional account of mind could be complete.^{1, 2, 11}

Throughout the 1970s and 1980s, related thought experiments accumulated under different labels. Ned Block’s “absent qualia” arguments and Sydney Shoemaker’s discussions of inverted and absent qualia raised the possibility of functional duplicates whose phenomenal states differed or were missing entirely. Frank Jackson’s knowledge argument (1982) and Thomas Nagel’s “What Is It Like to Be a Bat?” (1974) generated parallel pressure on physicalism from epistemic rather than modal premises. The zombie scenario was systematized and given its now-canonical form by Chalmers in The Conscious Mind (1996), which presented the zombie argument as one element of a multi-pronged case for property dualism.^{3, 5}

The thought experiment

To grasp the force of the argument, the reader is invited to consider a specific scenario. Chalmers asks the reader to imagine a being he calls his “zombie twin”: a creature molecule-for-molecule identical to himself, occupying the same spatial position in a counterfactual world that is otherwise physically indistinguishable from the actual one.

One can also imagine, as Chalmers does, a more comprehensive variant: an entire zombie world. This is a possible world physically indistinguishable from the actual world — identical in its laws of physics, its initial conditions, and the entire history of every particle — but in which no physical system anywhere is accompanied by phenomenal consciousness. The zombie world contains zombie philosophers writing zombie books on zombie philosophy of mind, zombie poets composing zombie love poems, and zombie neuroscientists locating the zombie neural correlates of the experiences that nobody in their world is having. The zombie world is intended as a vivid way of asking whether the actual world’s physical facts logically necessitate its phenomenal facts. If the zombie world is metaphysically possible — if there is some way the world could have been in which the physical truths held but the phenomenal truths did not — then the actual world’s phenomenal truths are not entailed by its physical truths.^{3, 4}

The thought experiment is sometimes confused with weaker variants that do not threaten physicalism. A behavioral zombie — a creature that merely behaves like a conscious being but has internal physical differences — is uncontroversially possible and proves nothing. A functional zombie — one that shares the high-level functional organization of a conscious being but is built from different physical materials — is more interesting but still leaves the physicalist room to maneuver. The Chalmersian zombie is a microphysical duplicate. It shares not just behavior or function but every physical property, every physical relation, and every physical law. Only this strong version puts pressure on the doctrine that the physical facts fix all the facts.^{3, 5}

The formal argument

Chalmers’s zombie argument can be rendered in several closely related forms. The simplest version takes the zombie scenario as a premise and derives the falsity of physicalism in three steps: zombies are conceivable; whatever is conceivable is metaphysically possible; therefore zombies are metaphysically possible, which suffices to refute physicalism. A more careful formulation expands this into the following premise-conclusion structure.^{3, 4}

P1. A zombie world — a world physically and functionally identical to the actual world but lacking phenomenal consciousness — is coherently conceivable.

P2. If a scenario is coherently (ideally, on reflection) conceivable, then it is metaphysically possible.

P3. If a zombie world is metaphysically possible, then phenomenal consciousness is not logically supervenient on the physical: there is some way the physical facts could have been all the same while the phenomenal facts differed.

P4. If phenomenal consciousness is not logically supervenient on the physical, then physicalism — the thesis that all facts are entailed by the physical facts — is false.

C. Therefore, physicalism is false.

The argument is formally valid: if the premises are true, the conclusion follows. This is rarely disputed. Disputes concentrate instead on the truth of the premises — especially P1 and P2, since P3 and P4 follow from standard definitions of supervenience and physicalism. P1 is challenged by those who deny that zombies are genuinely conceivable on careful reflection. P2 is challenged by those who accept zombie conceivability but deny that conceivability is a reliable guide to metaphysical possibility. The literature since 1996 can be organized largely around which premise a critic targets.^{4, 5, 15}

How major responses target the zombie argument⁵

Defense of the conceivability premise

The case for P1 is that the zombie scenario can be described without any internal contradiction. There is nothing in the concept of a microphysical duplicate that, on analytic reflection alone, rules out the absence of phenomenal experience. We can describe in full detail the firing patterns of every neuron, the chemical composition of every neurotransmitter, the input-output relations of every cognitive subsystem, and the verbal reports of the subject; nothing in these descriptions, taken individually or jointly, involves any explicit reference to experience. Phenomenal consciousness is what Chalmers calls a “further fact” over and above the structural and dynamical description that physics provides, and it is this independence of phenomenal description from physical description that secures conceivability.^{3, 16}

The defender of P1 can also appeal to the point made by Joseph Levine’s explanatory gap argument: even when we know everything about the physical correlates of an experience, an intelligible question remains about why those correlates should be accompanied by experience at all. The intelligibility of this residual question is a manifestation of zombie conceivability. If the residual question were incoherent, the zombie scenario would also be incoherent; conversely, if the residual question is coherent, the zombie scenario seems to inherit that coherence. The two stand or fall together.^{3, 16}

Chalmers strengthens P1 by distinguishing prima facie conceivability from ideal conceivability. Prima facie conceivability is the appearance of coherence on first reflection — what one finds when one merely tries to imagine a scenario without thinking very hard. Ideal conceivability is what survives unlimited rational reflection by a cognitively perfect reasoner. Some prima facie conceivable scenarios turn out to be incoherent on closer examination (one can prima facie conceive that there is a largest prime, but reflection reveals the incoherence). Chalmers’s claim is that the zombie scenario is not merely prima facie but ideally conceivable: no amount of further reflection reveals any hidden contradiction. Critics who deny ideal conceivability bear the burden of identifying the contradiction.^{4, 15}

From conceivability to possibility

The most controversial step in the zombie argument is P2 — the inference from conceivability to metaphysical possibility. Saul Kripke’s Naming and Necessity (1980) showed that some identity statements are necessary a posteriori: “water is H₂O” is necessarily true if true at all, even though its truth is established empirically rather than by reflection on concepts. One can prima facie conceive that water is not H₂O, but this conceivability does not entail any genuine metaphysical possibility. If mind–brain identities like “pain = C-fiber firing” are necessary a posteriori in the same way, then the prima facie conceivability of zombies might be no better evidence for their metaphysical possibility than the prima facie conceivability of XYZ is evidence for the metaphysical possibility of water without H₂O.^{10, 5}

Chalmers’s response invokes two-dimensional semantics, developed in detail in The Character of Consciousness (2010). Two-dimensional semantics distinguishes two intensions associated with a term: a primary intension, corresponding to the way a term picks out its referent in the actual world, and a secondary intension, corresponding to how the term applies in counterfactual worlds. For “water,” the primary intension picks out the watery stuff in any world considered as actual, while the secondary intension picks out H₂O in any world considered as counterfactual. The Kripkean a posteriori necessity arises from the gap between these two intensions: water is not H₂O is primarily but not secondarily possible.⁴

For phenomenal concepts, Chalmers argues, the gap closes. The primary and secondary intensions of a term like “pain” coincide because the way pain picks out its referent in the actual world — via the way it feels — is the same as the way it picks out its referent in any counterfactual world. There is no analogue of the watery-stuff/H₂O distinction for phenomenal concepts: pain just is the way it feels, in this world or any other. From this Chalmers concludes that for phenomenal concepts, primary conceivability does entail metaphysical possibility, and the Kripkean escape route is unavailable. The two-dimensional argument terminates in a dilemma: either physicalism is false or some form of Russellian monism — the view that the intrinsic nature of physical entities is itself proto-phenomenal — is true.^{4, 15}

Dennett and the under-imagined zombie

Daniel Dennett’s critique, developed in Consciousness Explained (1991) and sharpened in “The Unimagined Preposterousness of Zombies” (1995), targets P1 directly. Dennett denies that zombies are coherently conceivable at all. When philosophers claim to imagine a microphysical duplicate without consciousness, Dennett argues, they invariably end up imagining something less demanding than what their definitions require. They imagine a being that lacks some inessential feature — a sort of cognitive sluggishness, a gap in self-monitoring, a missing “extra” ingredient floating above the cognitive machinery — rather than a being whose cognitive machinery is genuinely identical in every respect. A creature whose functional organization fully matches that of a conscious being, Dennett insists, just is conscious; there is nothing more to consciousness than the functional organization, so subtracting consciousness while preserving the functional organization is not a coherent operation.^{6, 7}

Dennett presses this point with what he calls the “zimbo” — a zombie equipped with the kind of higher-order, recursive self-monitoring capacities that humans possess. A zimbo has beliefs about its own beliefs, can introspect, can report on its internal states, can deliberate about its deliberations. Dennett argues that no further property could be added or subtracted to make the difference between a zimbo and a conscious being, because the higher-order self-monitoring capacities exhaust what consciousness amounts to. To grant the zimbo all of these capacities and then deny it consciousness is, for Dennett, to traffic in an empty distinction. The zombie idea, he argues, “sums up, in one leaden lump, almost everything that I think is wrong with current thinking about consciousness.”⁷

Defenders of the zombie argument respond that Dennett’s objection presupposes the very functionalism that the zombie argument is intended to refute. To say that a creature with the right functional organization just is conscious is to assume that consciousness reduces to functional organization, which is the conclusion the physicalist needs to establish, not a premise that can be deployed to dismiss zombie conceivability. The zombie thought experiment asks the reader to suspend the functionalist assumption and consider whether the suspension yields a contradiction; Dennett’s reply, on this analysis, simply refuses the request.^{3, 4}

Type-B physicalism and the phenomenal concept strategy

A different family of responses concedes that zombies are conceivable but denies that conceivability entails metaphysical possibility. Chalmers labels these responses “type-B physicalism” in contrast to the “type-A physicalism” of Dennett, which insists that there is no real epistemic gap to begin with. Type-B physicalists accept that there is a genuine epistemic gap between physical and phenomenal descriptions while denying that this epistemic gap reflects an ontological gap. The leading type-B strategy is the “phenomenal concept strategy,” named by Daniel Stoljar and developed by Brian Loar, David Papineau, Christopher Hill, Brian McLaughlin, Katalin Balog, and others.^{5, 15}

Brian Loar’s “Phenomenal States” (1990) gives the strategy its canonical formulation. Loar argues that phenomenal concepts — the concepts we deploy when we think about our experiences “from the inside,” via introspection — are recognitional concepts that pick out their referents directly, without the mediation of any descriptive content. A phenomenal concept of pain is something like a mental demonstrative: it picks out pain “by acquaintance” rather than by any theoretical description. Because phenomenal concepts and physical concepts pick out the same property in radically different ways — via different cognitive mechanisms — it can seem possible for one to be instantiated without the other, even when no genuine metaphysical possibility exists. The conceptual independence is a fact about our minds, not about reality.⁸

David Papineau’s Thinking About Consciousness (2002) develops a related quotational account on which phenomenal concepts incorporate or quote actual instances of the experiences they refer to. On Papineau’s view, mind–brain identities such as “the experience of red = neural state N” are necessary a posteriori. They cannot be known by analyzing concepts because phenomenal and physical concepts present the identical underlying property in entirely different cognitive modes, and the mismatch between modes generates a feeling of contingency that does not correspond to genuine modal possibility. The zombie scenario is conceivable, on this picture, only in the same impoverished sense in which water without H₂O is conceivable: it reflects ignorance about underlying essences, not a genuine possibility.⁹

Chalmers has replied to the phenomenal concept strategy at length in “Phenomenal Concepts and the Explanatory Gap” (2007), arguing that no account of phenomenal concepts can do the work that type-B physicalism requires. Either phenomenal concepts are physically explicable — in which case the explanatory gap reappears for the concepts themselves — or they are not, in which case the type-B physicalist has covertly admitted a phenomenal residue that physicalism cannot accommodate. The exchange between Chalmers and the phenomenal concept strategists has structured much of the post-2000 literature.^{15, 5}

Kripke’s parallel modal argument

Saul Kripke’s argument against psychophysical identity in Naming and Necessity (1980) is a closely related ancestor of the zombie argument, sufficiently similar that the two are sometimes treated together in the secondary literature. Kripke observed that if “pain” and “C-fiber firing” are rigid designators — expressions that refer to the same thing in every possible world in which they refer at all — then any true identity statement involving them must be necessarily true. But the identity “pain = C-fiber firing” appears intuitively contingent: one can imagine C-fiber firing without pain, or pain without C-fiber firing, just as one can imagine zombies. Kripke argued that the standard physicalist explanation of such intuitions — that what one is really imagining is something that resembles pain or resembles C-fiber firing, rather than the genuine articles — is unavailable in the phenomenal case.¹⁰

The asymmetry, Kripke claimed, lies in the relationship between phenomenal concepts and their referents. With heat and molecular motion, what one is really imagining when one imagines “heat without molecular motion” is the sensation of heat without molecular motion, and the sensation of heat is genuinely separable from molecular motion. With pain, however, there is no analogous gap: pain just is the feeling of pain. There is nothing one could be imagining that feels exactly like pain but is something other than pain, because feeling like pain is constitutive of being pain. So the imagined contingency of psychophysical identities cannot be explained away in the standard a posteriori manner. Either the identities are genuinely contingent, in which case strict identity theory is false, or pain is not what we ordinarily take it to be.¹⁰

Kripke’s argument anticipates Chalmers’s zombie argument both in structure and in its key move. Both arguments derive an anti-physicalist conclusion from a modal premise; both depend on the claim that phenomenal concepts behave differently from natural kind concepts; both face the same family of objections from defenders of necessary a posteriori identities. The principal difference is generality: Kripke targets type identity theory specifically, while Chalmers targets all forms of physicalism that maintain logical supervenience of the phenomenal on the physical.^{10, 3, 5}

Kirk’s reversal and the epistemic-contact problem

Robert Kirk, who introduced the term “zombie” into the philosophical literature in 1974, has spent much of his subsequent career arguing against the very arguments his early papers helped to make possible. In Zombies and Consciousness (2005) and a series of related articles, Kirk maintains that although physical facts do not entail phenomenal facts a priori — one cannot derive consciousness from physics by conceptual analysis alone — they nevertheless entail them by a kind of logical or constitutive necessity that does not require a priori derivability. Kirk’s mature view is therefore neither the type-A physicalism of Dennett nor the standard type-B physicalism of Loar and Papineau, but a third position that denies zombie conceivability without identifying mind and brain in a Loarian manner.¹³

Kirk’s central argument against zombie conceivability is the “epistemic contact” or “sole-pictures” problem. Suppose that zombies are genuinely possible: there is some way the world could have been in which the physical facts were exactly as they actually are but no one had any phenomenal experience. In this scenario, what we ordinarily call our acquaintance with our own experiences must be detachable from the physical processing that we and our zombie twins share. But our acquaintance with our experiences seems to involve some causal or constitutive relation between the experiences and our cognitive processing — a relation that allows us to think about, refer to, and report our experiences. If zombies are possible, that relation either does not exist (in which case we have no genuine acquaintance with our experiences either) or floats free of the physical processing (in which case it is causally inert and cannot explain how our reports about experience track the experiences). Either disjunct, Kirk argues, is incoherent enough to undermine zombie conceivability.^{13, 5}

In his 2019 update to the Stanford Encyclopedia of Philosophy entry on zombies, Kirk summarized the state of the debate with deliberate restraint: “In spite of the fact that the arguments on both sides have become increasingly sophisticated — or perhaps because of it — they have not become more persuasive.” The remark captures a broader weariness in the literature, in which sophisticated formal apparatus accumulates without producing convergence. Kirk’s reversal also illustrates a recurring feature of the zombie debate: defenders and opponents often share substantive intuitions but disagree about which intuitions to trust at the limit.⁵

The anti-zombie argument

Keith Frankish’s “The Anti-Zombie Argument” (2007) offers a structurally elegant response that targets the zombie argument’s symmetry rather than its premises directly. Frankish observes that one can construct a parallel argument for physicalism by appealing to the conceivability of “anti-zombies”: purely physical creatures that are conscious in virtue of nothing but their physical organization. If the dualist insists that we can coherently conceive of microphysical duplicates without consciousness, the physicalist can equally insist that we can coherently conceive of microphysical duplicates whose consciousness is wholly grounded in the physical, with no further phenomenal or non-physical ingredient required. The two scenarios are conceivable by the same standards.¹²

Frankish then argues that, given the symmetry, the conceivability arguments cancel one another. If we infer that zombies are possible from their conceivability, we should equally infer that anti-zombies are possible from theirs — but anti-zombies are possible only if physicalism is true. Since the two conclusions are jointly inconsistent, at least one of the inferences from conceivability to possibility must fail, and there is no principled reason to privilege one over the other. The zombie argument therefore offers no independent reason to reject physicalism. Frankish’s argument has generated a substantial subliterature on whether the symmetry is genuine or whether dualists and physicalists can break it by appealing to asymmetries in what each side actually asks the imagination to do.^{12, 5}

Critics of the anti-zombie argument typically attempt to break the symmetry by arguing that the conceivability of an anti-zombie presupposes substantive views about the nature of consciousness that the dualist will not grant. If, for example, anti-zombie conceivability requires the assumption that physical organization suffices for consciousness, then the anti-zombie scenario merely encodes physicalism rather than supporting it from a neutral starting point. Defenders of the anti-zombie argument reply that exactly the same charge could be levelled against the zombie argument — that zombie conceivability presupposes the falsity of functionalism — and that the symmetry charge therefore reasserts itself one level up.¹²

Significance and ongoing debate

The zombie argument continues to occupy a central place in contemporary philosophy of mind because it isolates a precise question that other arguments leave implicit: whether the physical facts about a creature logically determine the facts about its conscious experience. If they do, then zombies are inconceivable in the strong sense Chalmers requires, and physicalism is at least consistent with the existence of consciousness. If they do not, then physicalism in its standard formulations is false, and some richer ontology — property dualism, Russellian monism, panpsychism, or another non-reductive view — is needed to accommodate experience. The argument does not adjudicate between these alternatives, but it sharpens the question of whether any reductive physicalism can succeed.^{3, 4}

The 2020 PhilPapers survey of professional philosophers offers a snapshot of the resulting distribution of views. Among respondents, roughly 37% described philosophical zombies as “conceivable but not metaphysically possible,” about 24% as “metaphysically possible,” and approximately 16% as “inconceivable.” The remaining respondents were undecided or selected other options. The numbers reflect a debate that has stabilized around several distinct positions without producing convergence on any one of them, and they support Kirk’s observation that increased sophistication has not yielded increased persuasion. Among philosophers of mind specifically, the debate is more polarized, with defenders and opponents producing increasingly intricate technical machinery in defense of incompatible conclusions.⁵

The argument has implications beyond the narrow question of physicalism. It bears on the prospects for artificial consciousness: if a microphysical duplicate of a human could lack experience, then so could a functional duplicate implemented in silicon, and the question of whether a sufficiently sophisticated artificial system would be conscious cannot be settled by examining its functional organization alone. It bears on the argument from consciousness in the philosophy of religion, since theistic uses of the hard problem typically depend on something like the zombie intuition to motivate the claim that physical reality cannot by itself produce experience. And it bears on the broader question of whether the methods of cognitive science can in principle close the explanatory gap that motivates the hard problem of consciousness — a question that, after fifty years of zombies in the philosophical literature, remains open.^{3, 4, 14}