Introduction

Zombies are exact physical duplicates of, what in the actual world are, phenomenally conscious human beings. However, they lack phenomenal consciousness. Chalmers’ zombie argument[1]  states that zombies are conceivable. It infers from their conceivability their metaphysical possibility[2]. It concludes that materialism is false. The precise entailment between conceivability and possibility that Chalmers argues for, is between what he calls 1-conceivability and 1-possibility. This essay argues that Chalmers’ arguments for the entailment between 1-conceivability and 1-possibility fail because they beg the question against necessitarianism[3]. Hence, the zombie argument fails.

Saul Kripke demonstrated that necessary a posteriori propositions exist. The importance of this is that, even if a proposition is not a priori true or false, it could still be necessarily true or false. Using Gendler and Hawthorn’s definition of conceivability: ‘P is conceivable iff it is not a priori that not-P’ (Gendler and Hawthorn, 2002: 32), we can see that necessary a posteriori propositions allow for a class of conceivable but impossible situations. Examples given by Kripke include ‘water is not H₂O,’ ‘gold does not have the atomic number 79’ and ‘Hesperus is not Phosphorus.’ (Kripke, 1981) The challenge that a posteriori necessities pose to the zombie argument is this: just as certain propositions like ‘water is not H₂O,’ are conceivable but impossible, perhaps zombies are also conceivable but impossible.

Chalmers makes a distinction between what he calls 1-conceivability/possibility and 2-conceivability/possibility. By making this distinction, he can, and does, argue for an entailment between 1-conceivability and 1-possibility. He does this in order to preserve the entailment between conceivability and possibility that would otherwise be undermined by Kripkean a posteriori necessities. Chalmers argues that concepts have primary and secondary intensions. The primary intension of a concept is a function from worlds considered as actual to extensions. If a world in which XYZ fills our oceans and lakes is taken to be the actual world, then water would refer to XYZ. Hence, if such a world is treated as actual, then ‘water is H₂O’ is false in that world. The secondary intension of a concept is a function from worlds considered as counterfactual to extensions, given that the actual world reference is fixed. If the XYZ world is considered as counterfactual, then water does not refer to XYZ in that world. Hence, water is not XYZ in such a world considered as counterfactual. In effect, the primary intension is a non-rigid description: as Chalmers writes, ‘we might say that the primary intension picks out the dominant clear, drinkable liquid in the oceans and lakes: or more briefly, that it picks out the watery stuff in a world.’ (Chalmers, 1996: 52) Secondary intensions, however, are fixed by their referents in the actual world. Because water is H₂O in the actual world, the secondary intension of water is H₂O. Secondary intensions are therefore rigidified descriptions. So, the primary intension of water is something like ‘watery stuff’ whilst the secondary intension of water is ‘H₂O.’ It is clearly impossible for H₂O to be other than H­₂O, but it is not clearly impossible for watery stuff to be other than H₂O. ‘Water is not H₂O’ is (so the argument goes) metaphysically possible if the primary intension of ‘water’ is being used, whilst it is impossible if the secondary intension is being used. ‘Water is not H₂O’ is therefore primarily possible (1-possible) but secondarily impossible (2-impossible). Chalmers argues, therefore, that necessary a posteriori propositions are only 2-necessary, and so do not undermine the entailment between ideal 1-conceivability and 1-possibility. This is essential for the zombie argument whose second premise states that the 1-conceivability of zombies entails the 1-possibility of zombies[4]. I label the thesis that there is such an entailment between primary conceivability and primary possibility the ‘primary entailment thesis’ (PET).

I argue that we should be sceptical of PET. If zombies are 1-possible, then it is possible for beings that are structurally identical to conscious human beings to lack consciousness. Chalmers takes such scenarios to be nomologically impossible. I argue that there is insufficient justification for believing in metaphysically possible but nomologically impossible situations.[5] It follows that we are unjustified in believing that zombies are 1-possible. The ‘CP thesis’ states that whatever is conceivable is possible (i.e. that conceivability entails possibility). I grant that zombies are conceivable[6]. Given that I also argue that they we lack justification for believing in their possibility, it follows that a central task of this essay is to demonstrate that we lack justification for believing that conceivability entails possibility, and hence that we lack justification for believing in CP[7].

Chalmers provides (broadly speaking) two arguments in support of CP. The first is an inductive argument. Chalmers notes that we have an abundance of examples of conceivable and possible propositions. Moreover, all supposed strong necessities – propositions that are 1-conceivable but neither 1 nor 2-possible - are not really strong necessities. They are either not ideally conceivable or are in fact 1-possible. (Chalmers, 2010: 166) Because there are so many conceivable and possible scenarios and no conceivable and impossible scenarios, it follows that we can reasonably infer that all conceivable scenarios are possible. As Chalmers writes, ‘why believe in CP?... there are no clear counterexamples to it. Principles linking conceivability and possibility have been widely accepted in the history of philosophy’ (Chalmers, 2010: 166) (call this argument 1). The second argument is really a cluster of arguments concerning the attractiveness of modal rationalism[8], and the unattractiveness of strong necessities (Chalmers, 2010:184-192) (call this argument 2). I think that argument 1 begs the question against necessitarianism - I address this in section 1. I think that the arguments concerning the attractiveness of modal rationalism and the unattractiveness of strong necessities are unsuccessful because there are no aesthetic grounds for favouring modal rationalism over necessitarianism – I address this in section 2.

 1.       Undermining argument 1

a)       CP Scenarios

I define ‘conceivability-possibility scenarios’ (henceforth shortened to ‘CP scenarios’) as scenarios that are conceivable and metaphysically possible. There are as many different varieties of CP scenarios as there are varieties of conceiving and varieties of possibility, and the permutations therein. For the purposes of this essay, we can focus on two. The first variety of CP scenarios are conceivable, nomically possible and metaphysically possible (N CP scenarios). The second are conceivable, nomically impossible and metaphysically possible (¬N CP scenarios). Put in these terms, the inductive argument that Chalmers’ makes is this: the great number of known CP scenarios we have provides inductive support for CP (Chalmers, 2010: 166). I believe that the set of ¬N CP scenarios may well be empty on the grounds that all nomically impossible scenarios may well be metaphysically impossible – that is, metaphysical possibility may be bounded by nomological possibility. At the very least, Chalmers does not provide sufficient reason for thinking otherwise. But in order for CP scenarios to serve as inductive support for CP as a whole (and hence for the zombie argument), we need examples both of N CP scenarios and of ¬N CP scenarios.

I will assume that all nomological possibilities are in fact metaphysical possibilities. It at least isn’t prima facie absurd to believe so, given the almost universal acceptance of such possibilities amongst philosophers. I am making this assumption in order to frame Chalmers’ argument in the best possible light – we can at least assume that there is complete inductive support for the thesis that all conceivable N scenarios[9] are possible. Given this assumption, it might seem that Chalmers has a great arsenal of CP scenarios from which he can infer all CP scenarios. But at best, we can take the great deal of N CP scenarios we have, and infer from them the metaphysical possibility of every conceivable N scenario. This inductive argument is legitimate because it does not make an inductive inference from one class of modal scenarios to a modally distinct class of scenarios. However, if necessitarianism is true, then there are no ¬N CP scenarios, and there exist as many strong necessities as there are conceivable ¬N scenarios.  You cannot justifiably argue that the fact that a given class of conceivable scenarios with a shared modal nature (of being nomically possible) is possible, that another class of conceivable scenarios with a different modal nature (of being nomically impossible) is possible. If there are no ¬N CP scenarios, then the inductive argument for the CP thesis cannot get off the ground. If there are no ¬N CP scenarios, there is no inference to be made; there is nothing to infer from. But to assume that there are ¬N CP scenarios is to beg the question against necessitarianism. To argue that necessitarianism is false because of conceivable ¬N scenarios is circular because it assumes CP is true.

The reasons we have for believing in N CP scenarios, we don’t have for believing in ¬N CP scenarios. We exist in the actual world, with its actual laws of nature; resultantly, there are no tests we can perform to validate or invalidate claims such as ‘it is possible to travel faster than light’ or ‘zombies are possible.’ If any such scenario were empirically validated, the proposition would necessarily conform to the laws of nature. N scenarios can be validated because they, or relevantly similar scenarios, can be actualised, and actualisation is proof of possibility. This is why there is reason to believe in N CP scenarios, but not in ¬N CP scenarios.

It may be objected that just because something is not empirically verifiable, this does not mean it is not possible. I agree with this objection. Some conceivable N scenarios are unverifiable after all. For example, we cannot verify the one-way speed of light, only the round-trip speed, and we may never be able to measure the exact temperature of a particular region inside the sun. However, the proposition that the one-way speed of light is identical to the round-trip speed is not just reasonable but almost certainly true. Since the alternative would necessitate a bizarre case of non-uniformity in nature, and a flagrant violation of Occam’s Razor, it can reasonably be ruled out. We could argue with strong justification that the region in question of the sun has a temperature between x and y. If the temperature of that particular region of the sun were outside this temperature range, then this would undermine a vast body of our scientific understanding. Since we at least need strong grounds for uprooting our scientific theories, we can reasonably infer that the temperature of this region of the sun lies within x and y. We therefore have good non-verificationist reasons for believing in the possibility of certain unverifiable N scenarios. However, we lack both non-verificationist and verificationist reasons for believing in the possibility of ¬N scenarios. It is therefore reasonable to believe in the possibility of certain unverifiable N scenarios, but it is unreasonable to believe in the possibility of any unverifiable ¬N scenarios.

b)      Strong Necessities

Chalmers notes that ‘both CP+ and CP-[10] are compatible with all familiar purported counterexamples to the conceivability-possibility link.’ (Chalmers, 2010: 148) It is undeniable that many supposed strong necessities are not in fact strong necessities. Although we can conceive both Goldbach’s conjecture and its negation, it’s clear that we cannot ideally conceive them (Chalmers, 2010: 145). Although we can imagine a square circle, we cannot imagine the situation in full detail and if we did, we could easily determine a contradiction (Chalmers, 2002:152-154). There are some other undeniable examples of strong necessities ‘gone wrong’ such as ‘cases of secunda facie positive conceivability without ideal positive conceivability’ (Chalmers, 2002:155). Frege’s claim that there is a set that contains all sets is conceivable even after significant reflection, but contains a contradiction which is revealed on ideal reflection. (Chalmers, 2002:155). But it is notable that what is held in common between all these clear bogus strong necessities is the fact that they are logical/mathematical contradictions. There is a contradiction in the idea of a square circle, in the idea of a false mathematical statement being true, and in the idea of a set containing all sets. In all these examples, what is being revealed is the fact that we cannot ideally conceive of contradictions.

Kripkean non-identity claims such as ‘water is not H₂O’ do not contain contradictions (when interpreted in their primary intension). As a result, these potential strong necessities are not shown to be bogus in the same way that these logical contradictions are. However, it might be granted that CP accommodates Kripkean non-identities in a different way than these logically/mathematically impossible propositions, but argued that CP accommodates Kripkean cases nonetheless (via. The primary/secondary distinction). As Chalmers writes, ‘CP accommodates these examples [Goldbach and Kripke cases] straightforwardly. With the idealization accommodating Goldbach cases and the primary/secondary distinction accommodating Kripke cases.’ (Chalmers, 2010: 166) The problem here is that the primary/secondary distinction only accommodates Kripke cases if the argument is begged against necessitarianism. If the set of possible worlds is limited to the set of nomically possible worlds (i.e. if necessitarianism is true), then given that it is presumably nomologically impossible for the set of superficial qualities associated with water to be instantiated by something which is not H₂O, then ‘water is not H₂O’ will not be 1-possible. To argue against necessitarianism by appealing to the conceivability of certain ¬N scenarios is to assume CP.

Finally, we can see the circularity of appealing to CP scenarios and the absence of strong necessities as a way of defending CP. Those who believe in the necessity of the laws of nature believe that there are no ¬N CP scenarios (which are crucial for CP). They also believe that all conceivable ¬N scenarios are strong necessities. In essence, to defend CP by appealing to CP’s successes and lack of failures is circular. Chalmers does, however, provide arguments for CP, and against strong necessities, that do not appeal to CP. These arguments are found in the next section and I believe that we should be sceptical of them.

2.    Undermining argument 2

Chalmers first considers the fact that ‘in the absence of an explanation… strong necessities will be brute and inexplicable’ (Chalmers, 2010: 184). Chalmers notes that the fundamental laws of nature are taken as primitive, and strong necessities will be more like fundamental laws of metaphysics (Chalmers, 2010: 185). My response here is to admit that primitives are an unattractive part of any theory. However, the existence of primitives is hardly sufficient to undermine the view that there are strong necessities. The view that the laws of nature are strongly necessary has its own attractive qualities and the opposite view has its own unattractive qualities. It should also be acknowledged that there may well be other brute truths - logical truths may perhaps fall into this category, and Chalmers puts the fundamental laws of nature into this category. Although there is controversy here, we can at least acknowledge that there is an open possibility that these are brute. If we cannot rule out these truths as brute, we should not be afraid to keep the brute necessity of the laws of nature an open possibility. It is important to note that Chalmers himself assumes that the fundamental laws of nature are primitive. But he hasn’t provided us any reason for believing that the primitive truth of the laws of nature is any less unattractive than the primitive necessity of the laws of nature. He must provide reason to believe that primitive metaphysical laws are more objectionable than primitive natural laws, and not merely assume it.  Finally, we cannot be certain that ¬N strong necessities are primitive. There may be an explanation as to why the laws of nature are necessary (if they are in fact necessary). If there is such an explanation, then ¬N strong necessities are not primitive. Just because we do not have an explanation does not mean that one does not exist.

If CP is true, then all logically possible worlds[11] are metaphysically possible. Chalmers mentions that a space of logically possible worlds is coherent and useful. He writes, ‘it will be tremendously useful to appeal to a wider space of logically possible worlds with different laws to help explain the hypotheses and inferences of a scientist investigating the laws of nature.’ (Chalmers 2010: 185) My response to this is one that Chalmers anticipates: ‘at this point, opponents of the CP thesis might allow that the space of logically possible worlds is coherent and useful in its own right, but they may well suggest that we also have good reason for believing in a space of separate metaphysically possible worlds. Presumably this space will be narrower than the first space.’ (Chalmers, 2010: 187) But he rejects this, saying, ‘to introduce a further primitive, restricting the space of worlds, is to introduce an unnecessary wheel.’ (Chalmers, 2010: 187) My response is this: to allow for a space of worlds beyond the nomically possible is to allow for an unnecessary expansion of our modal ontology. There are no aesthetic grounds for favouring an expanded set of possible worlds and the truth of CP, over a diminished set of worlds and the primitive necessity of natural laws. The former features an attractive entailment from conceivability to possibility, but an unattractively bloated modal ontology. The latter features an attractively minimalist modal ontology, but an unattractive primitive necessity. If anything, we should tentatively edge towards the latter option because it does not merely curtail the space of possible worlds, but the worlds that it curtails are worlds that are beyond empirical investigation.

Chalmers further argues that ‘the believer in strong necessities must embrace a modal dualism, with distinct and independent metaphysical modalities and distinct and independent spaces of worlds that answer to them.’ (Chalmers, 2010: 190) But Chalmers is still left with a modal dualism of sorts. Although Chalmers would subsume the metaphysically possible into the logically possible, he would not subsume the metaphysically possible into the nomically possible.

Finally, Chalmers writes ‘if we postulate a metaphysical modality that is independent of conceivability, the epistemology of modality becomes quite problematic… if metaphysical modality involves an independent primitive, then it becomes quite unclear why conceivability should be any guide to it at all. Why should there not be just one metaphysically possible world, or thirty-seven?’ (Chalmers, 2010: 190-191). To ask ‘why should there not be just one… or thirty-seven?’ is to presume that conceivability is the only guide to possibility. If necessitarianism is true, then we can use scientific experimentation to determine whether certain scenarios are nomologically possible. Since nomological and metaphysical possibilities are coextensive, we will have thereby enumerated a space of possible worlds. It might be granted that we will still have a means by which we determine possibility, but that conceivability is no longer a means. This is not a disaster. After all, if CP were true, then nomological and metaphysical possibilities come apart. All ¬N CP scenarios are metaphysically possible but nomologically impossible. Scientific experimentation, therefore, cannot validate them. Thus, we have two options which are both in part attractive and in part unattractive. Either conceivability is a guide to possibility but scientific experimentation isn’t, or scientific experimentation is a guide to possibility but conceivability isn’t. I think there are reasons to prefer the latter state of affairs to the former. People often disagree about what is or is not conceivable, whilst scientific hypotheses can acquire the status of being almost unarguable statements of fact. Moreover, any disagreement in science can in principle be resolved via additional attempts to acquire corroborating or falsifying data through experimentation. There is no clear analogue of an objective means through which we can resolve disputes about what is or is not conceivable. Finally, under CP, only ideal conceivability entails possibility. But there is no means by which we can know that our conceiving is ideal. However, under necessitarianism, anything that falsifies a hypothesis thereby renders it impossible. We can also reach near certainty re. a scenario’s modal nature once we have sufficient evidence, even if we have a greatly incomplete science.  

In short, there are attractive and unattractive qualities inherent both in Chalmers’ modal rationalism and in necessitarianism. There are simply no aesthetic grounds for favouring the former over the latter.

Conclusion

The zombie argument fails because we have insufficient reason to believe one of its premises: we have insufficient reason to believe that the 1-conceivability of zombies entails the 1-possibility of zombies. This is because we are unjustified in believing in the metaphysically possible but nomically impossible. Arguments for this entailment that appeal to CP scenarios fail: the necessitarian would argue that there are plenty of N CP scenarios but no ¬N CP scenarios. These scenarios are crucial for CP. Appealing to the accommodation of supposed strong necessities by CP also begs the question against necessitarianism. The necessitarian would argue that all conceivable ¬N scenarios are genuine strong necessities. CP is no more attractive than necessitarianism. Both are dualistic – CP has distinct metaphysical and nomological modalities, whilst necessitarianism has distinct nomological and logical modalities. CP might be argued to unnecessarily expand our modal ontology whilst necessitarianism might be argued to add an unnecessary primitive that restricts our modal ontology. CP uses conceivability as a guide to possibility, necessitarianism uses scientific experimentation as a guide to possibility. In short, the success of CP cannot be used as an argument for CP without circularity or begging the question against necessitarianism, whilst aesthetic considerations can be used with equal weight to attack and defend both CP and necessitarianism. Since Chalmers provides insufficient reason to accept CP over necessitarianism, he thereby provides insufficient reason to accept the premise that the 1-conceivability of zombies entails the 1-possibility of zombies. He hence provides insufficient reason to accept the zombie argument.

References

Chalmers, David (1996), The Conscious Mind, Oxford University Press

Chalmers, David, (2002), Does Conceivability Entail Possibility?, Published in T. Gendler and John Hawthorne Conceivability and Possibility, Oxford University Press

Chalmers, David, (2010), The Character of Consciousness, Oxford University press

Dennett, Daniel (1998), Brainchildren: Essays of Designing Minds, The MIT Press

Gendler, Tamar and Hawthorne John (2002), Conceivability and Possibility, Clarendon Press Oxford

Kripke, Saul (1981), Naming and Necessity, Wiley-Blackwell

Shoemaker, Sydney, (1998), Causal and Metaphysical Necessity?, Pacific Philosophical Quarterly 79