One of the chief munitions in the anti-physicalist armament is the zombie. Zombies are philosophical fantasy creatures that are physically just as we are (or at least, just as our best physicalist guess says we are) but that lack phenomenal consciousness. The zombist’s claim is that the existence of zombies is a logical and possibly a metaphysical (though not an empirical) possibility; if zombies are logically possible, then it putatively follows that facts about (the presence or absence of) phenomenal consciousness are not fixed by the totality of physical facts, and thus that consciousness is over and above the physical; hence, physicalism is false.
Compare the cases of being a live pig, being liquid, or even, probably, being Michelangelo’s David. In these cases, and all others relevantly like them, if the totality of physical facts are fixed then so too are the facts about life, pigs, liquidity, and the provenance and other (non–consciousness-dependent) qualities of artworks. A “zombie” world that is physically just like this one but in which nothing is alive—containing only the walking (breathing, reproducing, metabolising, etc.) dead, as it were—is not merely unlikely but internally incoherent. The alleged possibility of philosophical zombies is thus thought to identify something special about phenomenal consciousness, and to raise a compelling problem for physicalism.
The zombist case is far from won, however, and various physicalist responses to the zombie argument have been raised, particularly since the resurgence of interest in zombie-style arguments stimulated by David Chalmers’ The Conscious Mind in 1996.Footnote 1 One particularly prevalent style of response on the part of the physicalist is to connect the logical possibility of zombies with the truth of epiphenomenalism about phenomenal consciousness. Insofar as epiphenomenalism is implausible, unpalatable, or even impossible, and insofar as the zombist claim leads to epiphenomenalism, thus far are zombie arguments rendered doubtful.Footnote 2 In its outlines, the physicalist response looks like this:
i) A pivotal premise in zombie arguments is that zombies are (in the relevant sense) possible.
ii) If zombies were possible, then phenomenal consciousness would be epiphenomenal.
iii) But phenomenal consciousness cannot be, or at least is not, epiphenomenal.
iv) Therefore zombies are not possible, and so the zombist challenge to physicalism fails.
Premise iii) can be, and has been, motivated in various ways. What all these have in common is the intuition that it is implausible or impossible to suppose that the elimination of phenomenal consciousness from the world would make no difference—that everything (including cognition) would go on just the same even if pain were not phenomenally painful, vision not phenomenally colourful, desire not phenomenally pressing, and so on. A particularly compelling instance of this general worry is what is sometimes called the Paradox of Phenomenal Judgement (after Chalmers Reference Chalmers1996): the puzzle that, since cognition ex hypothesi remains unaffected by the removal of phenomenal consciousness, our first-personal, introspective beliefs about and reports of our own states of phenomenal consciousness would—if consciousness were epiphenomenal—prima facie be completely unchanged even in the absence of their objects. That is, if consciousness is epiphenomenal, we would not and could not notice if we ourselves were zombies.
Although there is much more to be said on this score, I shall not concern myself here about the implausibility of epiphenomenal consciousness. This paper deals instead with premise ii) of the argument above.Footnote 3 What exactly is the connection between the zombie hypothesis and the epiphenomenalism of phenomenal consciousness? Does the mere logical or metaphysical possibility of zombies entail conscious epiphenomenalism? And if conscious epiphenomenalism is entailed, is what would follow the truth of epiphenomenalism in the actual world or merely the physical possibility of epiphenomenalism (i.e., its obtaining in at least one possible world physically indiscernible from the way the physicalist takes the actual world to be)? That is, must the physicalist show, to complete her argument, that conscious epiphenomenalism is impossible or merely that it is not the case? These issues, though crucial to the dialectic about zombies and phenomenal consciousness, have not been given much attention in the literature to date.
I shall argue that the relationship between the zombie hypothesis and epiphenomenalism is significantly more complicated than is usually recognised, but that when these complications are worked through the connection emerges as tight enough to be worrying for the zombist. Though not a particularly one-sided or clear-cut conclusion—and thus in that sense less interesting than it might be—I think it has the modest virtue of being plausible, and as nuanced as the multifaceted situation requires.
1. The Initial Intuition
It is often taken for granted that, modulo the odd unimportant complication, the logical/metaphysical possibility of zombies does indeed necessitate the epiphenomenalism of phenomenal consciousness.Footnote 4 Indeed, many proponents of the zombie argument—including at one time the ur-zombist David Chalmers—seem willing to accept this claim, at least provisionally, and devote their efforts to attempting to show that conscious epiphenomenalism is a bullet we should be willing to bite.Footnote 5
It is natural to think of zombies in the following way: zombies are what are left over when you subtract phenomenal consciousness from the actual world, or a relevant part of it, while leaving everything physical unchanged. To say that zombies are logically/metaphysically possible is to say that it is logically/metaphysically possible to perform this subtraction. But to allow this is precisely to allow that the presence of phenomenal consciousness is unnecessary for any physically specifiable event to occur, since ex hypothesi all these events (or states, or properties) would occur anyway even if consciousness were not present, and furthermore that the occurrence of any or all of these physical events is not causally sufficient for the presence of consciousness, since they may occur when conscious events do not. That is, consciousness is strongly epiphenomenal—it neither causes nor is caused by any physical event.
Furthermore, according to this line of thought, consciousness is actually epiphenomenal, not merely epiphenomenal in some other possible world. Bear in mind that, for the zombie hypothesis to be meaningful as a problem case for physicalism with respect to consciousness, the zombist must allow the physicalist to specify physicalism on her own terms (rather than foist upon her a specification of physicalism that she may then reject). Thus, the zombie world is to include everything that the physicalist takes to be required for physicalism to be true, including not only the distribution and properties of all the physical entities but also all the physical laws that govern the interactions of those entities. The zombist contention is that, even given everything the physicalist might wish to include in a final completed physics, it is (logically or metaphysically) possible to subtract consciousness. So a zombie world cannot legitimately be created by adjusting the causal laws (at least if we follow the standard line in taking causation—though not necessarily the events related by causal laws—to be within the domain of physics). If physical events do not cause conscious events (or vice versa) in the zombie world, then they do not do so in the physicalist’s actual world either.Footnote 6
Bear in mind also that some sort of reference to the actual world in the specification of the zombie hypothesis is non-negotiable, since it is claims about consciousness in the actual world that are under dispute. It is uncontroversial that both ‘zombies’ and epiphenomenal consciousness are possible, in the sense that there are some possible worlds in which human-like organisms are not phenomenally conscious (perhaps because their heads are filled with sawdust) or consciousness is causally inert (perhaps because it is possessed only by disembodied spirits in that world). What is at issue is whether zombies are possible given what we know, or think we know, about the physics of the actual world—i.e., in any possible world “physically indiscernible” from the actual one.
2. Two Unpersuasive Responses
It is widely conceded by physicalists that the necessitation of conscious epiphenomenalism by the logical possibility of zombies falls short of perfect entailment. However, these initial stumbling blocks do not do much to ameliorate the power of the physicalist response to the zombist.
The first caveat is that the zombie possibility leads to epiphenomenalism only if causal overdetermination is not the case. Suppose a) that zombies are possible and hence that phenomenal consciousness is non-physical, and b) that certain events have both a sufficient physical cause and an additional phenomenal one. For example, suppose that particular brain states are alone sufficient to cause pain behaviour, but that this behaviour is also caused by (non-physical) phenomenal pain sensations. Under such circumstances, of course, consciousness would not be required for everything to happen in the actual world just as it does—i.e., zombies are possible—but nevertheless phenomenal consciousness would not be epiphenomenal.
But this does little or nothing to weaken the physicalist response. First, this zombist reply depends upon overdetermination being the case—since consciousness is non-epiphenomenal only if overdetermination is true for every instance of phenomenal causation—and this is not itself a very plausible position. More fundamentally, this complication is anyway irrelevant to the force of the physicalist response. What makes conscious epiphenomenalism implausible is the dubiousness of supposing that the elimination of consciousness would make no difference at all to how things go in the physical world—that the elimination of phenomenal pain, for example, might make no difference to our pain behaviour. But this consequence—call it conscious quepiphenomenalism—still follows from the zombie possibility, even given the possibility of overdetermination. So the physicalist response to the zombist is unaffected.
The second caveat is based on an appeal to the still-mysterious nature of causation. For consciousness to be epiphenomenal it must fail to cause—but exactly what it is to be a relatum in a causal relation is a somewhat contested notion. It may be that on some understandings of the nature of causation the possibility of zombies is consistent with the non-epiphenomenalism of consciousness. Consider, for example, a ‘Humean’ account of causation that allows causal laws to be non–counterfactual-supporting regularities: on this view, the mere fact of covariation between phenomenal consciousness and physical events in the actual world is enough to constitute mental-physical causal laws and thus give consciousness causal status. This would be so even if zombies are possible: i.e., if that covariation fails to hold in other possible worlds that are physically indiscernible from this one (or from the way physicalists take this one to be).
A more sophisticated version of this kind of response is to raise the possibility of panprotopsychism,Footnote 7 though a certain amount of fancy footwork is required before it can work on behalf of the zombist. Physical descriptions arguably do not capture the intrinsic nature of the things they describe—what it is to be a photon, or to have a certain mass, or to be metallic just is to stand in certain relations to other things; to interact with other entities or conditions in particular ways. Causal laws relate things extrinsically. But what if causality itself depends somehow upon the intrinsic nature of the things that are causally related? A kind of panprotopsychism supposes that this is so, and further that the intrinsic nature of everything is protophenomenal. In this way, the physico-causal structure of the world is explained and underwritten by protophenomenality.
To this we can add the following claims. First, phenomenal consciouness is somehow agglomerated out of the intrinsic protophenomenality of brains, and hence inherits the causal relevance of the protophenomenal. From this we can say that consciousness is not epiphenomenal. Second, although the intrinsic nature of things that underpins causality is in the actual world protophenomenal, it need not be: there is at least one other species of intrinsic essence—call it haecc [“hex”]—such that all the protophenomenal could be replaced by haecc leaving the physical structure of the world unchanged. This preserves the zombie possibility.
So, it appears that there are ways of construing the nature of causality such that the zombie possibility is consistent with the non-epiphenomenality of consciousness. The problem with these kind of responses, however, is that they require the zombist to surrender the dialectical high-ground. The initial zombie argument gains its force by granting the physicalist all her (reasonable, non–question-begging) assumptions about the nature of the physical, and then showing that even given all this, the facts about phenomenal consciousness are still undetermined—that is, that no physical facts whatever are sufficient to fix the phenomenal. Among the physicalist assumptions, it is likely, will be certain assumptions about causality: that causal laws are counterfactual supporting, and that physical causation does not depend for its efficacy on the non-physical. In order to launch the kind of response described above the zombist must reject one or more of these assumptions, and thus limit the varieties of physicalism to which the zombie argument applies (or, more accurately, the varieties of physicalism for which, when the zombie attack is made, the epiphenomenalism response is unavailable).
Hence, it is not enough for the zombist simply to point out that non-standard understandings of causation are available on which consciousness need not be epiphenomenal. She must constrain the options for the physicalist by proffering positive arguments to show that causation actually is one or other of these ways. This is a heavy dialectical burden to carry, transferring the burden of proof emphatically onto the zombist, and zombists have not so far made effective sallies on this front.Footnote 8
3. A First Complication: The Contingency of Physicalism
Physicalism is a contingent doctrine: the physicalist typically concedes not only that physicalism might not have been true (i.e., that there are possible worlds—even possible worlds that closely resemble our own—in which physicalism does not hold) but also that physicalism might not even be true of the actual world (though the physicalist bet is that it is true). This is important since it affects the way in which we must specify the zombie hypothesis; and this in turn gives rise to the first more serious problem for the physicalist intent on convicting the zombist of epiphenomenalism. The normal mode is to specify a zombie world as a world physically indiscernible from our own but in which consciousness is absent. But if physicalism is not true of the actual world, then this formulation will not suffice: there may be no possible world physically indiscernible from our own but in which consciousness is absent since, for example, interactionist dualism may just possibly turn out to be true (and assuming that removing consciousness would then alter the causal chains and thus the distribution of the physical).
The proper way to think of the zombie world, thus, is as a possible world specified according to the physicalist’s best guess about the physics of the actual world but in which phenomenal consciousness is absent. And the proper role of zombie arguments is as part of a destructive dilemma for the physicalist:
a) If physics is (relevantly, including causal closure)Footnote 9 as specified by the physicalist, then physicalism must be false because of the zombie possibility.
b) If physics is not (relevantly, including causal closure) as specified by the physicalist, then physicalism is (trivially) false.
c) So physicalism is false.
Seen in this light, the zombie possibility sidesteps commitment to epiphenomenalism in the actual world. That is, if the physicalist is right about physics—such that all physical events, including human behaviour, can be given sufficient explanations in purely physical terms—then (according to the zombie argument) consciousness must in the actual world be non-physical and epiphenomenal. But, consistently with the zombie hypothesis, the physicalist might not be right about real-world physics and thus, say, dualist interactionism might be the case—and in that case, consciousness need not be epiphenomenal. That is, interactionist zombists can claim that because we are not zombies consciousness in our world is not epiphenomenal, while still arguing that the mere possibility of zombie worlds that are physically as physicalists (wrongly) believe our world to be shows that physicalism is false.Footnote 10
One drawback for the zombist, however, is that despite all this the following remains the case: the zombie possibility plus commitment to a broadly physicalist account of physics entails the epiphenomenalism (or at least quepiphenomenalism) of consciousness in the actual world. And this is a further commitment—to physicalist physics—that many, if not most, contemporary zombists will wish to make.
Secondly, the weaker conclusion still follows from the zombie possibility that conscious epiphenomenalism must be possible, even if not actual. That is: zombie worlds are only a coherent possibility on the condition that if physics were as the physicalist describes, then consciousness would be epiphenomenal. There is a possible world, W, physically rather like this one, and containing the distribution of phenomenal consciousness with which we are familiar in this world, with the important additional stipulation that W is causally closed under physics. Now, from the perspective of W, as it were, physiological zombies are possible: that is, George Bushw might have a zombie twin, even if the real Bush could not. Thus, in this sense, the possibility of physiological zombies is consistent with the non-holding of conscious epiphenomenalism in the actual world. But epiphenomenalism still holds in W. So in any world “with respect to which” physiological zombies are conceivable, consciousness in that world is epiphenomenal. (And if the actual world is such a world, then consciousness is epiphenomenal in this world.)
The degree to which this point is pressing for the zombist is open to further debate, but one thing at least is clear: it does not let her off the hook. If the zombie possibility involved a commitment to the actual epiphenomenalism of consciousness then that would pretty much be a slam-dunk for the physicalist; and any zombist who wants to combine her dualism (or neutral monism, or some other anti-physicalist position) with an endorsement of physics as the correct science of the physical continues to face this rebuttal. If the zombist dodges the commitment to epiphenomenal consciousness by, for example, denying causal closure, he or she nevertheless must concede that the zombie possibility is a coherent attack on physicalism only if phenomenal epiphenomenalism is possible. Though less evidently untenable than the doctrine of actual epiphenomenalism, the thesis of the coherence of the notion of phenomenal epiphenomenalism is a major commitment that the zombist must assume and with respect to which he or she is vulnerable. Many physicalists are prepared to argue not just that phenomenal epiphenomenalism is false, but that it is impossible—that the very notion is unintelligible. To the extent that these arguments succeed—a matter that I shall not attempt to adjudicate here—zombist appeals to the contingency of physicalism cannot avail them.
4. A Second Complication: Gap-Filling or Gap-Bridging
“An interactionist dualist can accept the possibility of zombies, by accepting the possibility of physically identical worlds in which physical causal gaps go unfilled, or are filled by something other than mental processes” (Chalmers Reference Chalmers2004, 184). This brief comment by David Chalmers, although aimed at the interactionist dualist, suggests an additional problem for the physicalist intent of convicting the zombist of epiphenomenalism. The proposal, I take it, is simply this. Suppose that consciousness is neither epiphenomenal nor quepiphenomenal—that is, its presence is causally required for the occurrence of certain physical events, such as human pain behaviour. As we have seen, this is prima facie inconsistent with the zombie possibility: if consciousness causes some physical behaviour, and since removing the cause typically also removes the effect, a world in which consciousness is absent could not be physically indiscernible from the actual world. But this is so only as long as we insist that an event B for which A is the sole, sufficient cause would not have occurred if A had not occurred (and if we leave everything else in the physical causal nexus up to that point the same). Chalmers suggests that we might abandon this assumption. Either we could simply suppose that B occurs anyway in the zombie world, uncaused; or we could remove A but replace it in the zombie world with another equivalent cause that is neither physical (since the zombie world is physically indiscernible from the target world) nor phenomenal (since the zombie world contains no consciousness).
This proposal differs from that about the nature of causality, above, in that it does not explicitly trespass into the sphere of physicalist assumptions about the physical: in the actual world, physics can (apparently) be allowed to operate just as the physicalist supposes it does in all non–question-begging respects.Footnote 11 It is thus legitimate for the anti-physicalist to introduce it after all the relevant physicalist assumptions have been granted. And, although strange sounding at first, Chalmers’ gap-bridging/gap-filling proposal seems on a certain amount of reflection neither metaphysically nor logically impossible. If gap-handling is metaphysically possible so are zombie worlds, consistently with consciousness being non-epiphenomenal. So the zombist can apparently block the physicalist attempt to foist epiphenomenalism upon her.
However, all is not quite so clear cut. Consider gap-filling: here the proposal is that conscious causes be replaced in the zombie world by new causes that are neither physical nor phenomenal. But such a proposal is schematic at best—what could the inserted causal events be if they are neither material nor spiritual? Since we have no idea at all about the nature of these events—we have no candidate substances of which they might be modes—it is hard to assess whether the possibility being raised here is real or not.
There is another difficulty as well, which is clearest if we consider gap-bridging (i.e., simply allowing the causal gaps to remain, but having the physical events around the gaps occur as usual). Here the problem is that it is not clear that we can introduce causal gaps while leaving the physical exactly the same. This is because, as suggested above, the individuation conditions for physical entities are relational: what makes a photon a photon, as far as the physical sciences are concerned, is the way it is embedded in a set of law-like causal relationships with other entities (moving at the speed of light in all frames of reference, exerting no rest mass, transferring momentum inversely proportional to its wavelength on colliding with matter, and so on). So as not to beg any questions in favour of physicalism, let us not presuppose that these causal relations need be with other physical entities. Nevertheless, the presence of causal gaps threatens to disrupt these relations in zombie worlds.
To give a simplified example, suppose that a certain kind of pyramidal cell activity causes pain sensations, and that these pain sensations in turn cause certain nerve signals to be sent to muscles. The two physical event-types in this example are, like all physical event-types, individuated by their law-like connections to other event-types—in this case, including at least one non-physical event-type. But in the zombie world, this relation no longer exists—“pyramidal cell activity” is not a cause of pain sensations, for example. This suggests that, just as a ‘photon’ which did not transfer momentum in a particular way to the particles with which it collides would not in fact be a photon, ‘pyramidal cell activity’ in the zombie world is not really pyramidal cell activity. And in that case, the ‘zombie world’ is not really a zombie world, as it is not physically indistinguishable from the target world.
In short, these considerations might be taken to suggest that it is still impossible to remove non-epiphenomenal consciousness from the world and leave the physical exactly the same, since removing consciousness changes the causal networks in which the physical is embedded—by, for example, introducing gaps—and, since physical things are individuated by their places in the causal network, changing the network inevitably changes the physical.
But now the implausibility shoe is on the other foot. Although it is certainly reasonable to think that the physical is characterized extrinsically, it is less clear that physical events, states, or entities (particularly non-fundamental ones, like neural states) are individuated by all their lawlike relations. Thus, it may be open to the zombist to deny that causing pain sensations is essential to being pyramidal cell activity.
5. Conclusion
The upshot of all of this for the zombist-physicalist debate with respect to epiphenomenalism is fairly clear, but somewhat complex. Some frequently presented responses to the attempt to connect zombies with epiphenomenalism, such as appeals to causal overdetermination or non-standard accounts of the causal relation, on inspection fail to make much impact on the physicalist worry. The contingency of physicalism is a more substantial worry, but the zombist can reorient the zombie argument as part of a destructive dilemma for the physicalist and in this fashion avoid commitment to the epiphenomenalism of consciousness in the actual world. Yet, he or she remains committed still to the coherence of conscious epiphenomenalism in a possible world where the physics is as the physicalist supposes—including, potentially, the actual world—and on this front the zombist is vulnerable. Finally, however, to the degree that the notion of causal gaps (or causal agents that are neither physical nor phenomenal) can be made plausible, it might be possible for the zombist to evade even this commitment and sever the link between the zombie hypothesis and phenomenal epiphenomenalism entirely.
This last move will be attractive to—perhaps even necessary for—the zombist, given the widespread wariness that exists, among physicalists and anti-physicalists alike, about phenomenal epiphenomenalism. Yet it has as yet been little explored, and remains highly speculative. Perhaps this is the front on which the next battle of the “zombie wars” is to be fought.Footnote 12
