2.1 Ghosts

An initial reason to be suspicious that naturalism is an interesting, substantive philosophical thesis is that it seems to exclude so little. Apart from some non-naturalist ethicistsFootnote ³ and some avowedly theistic philosophers,Footnote ⁴ almost everyone in philosophy considers their own preferred theory to be naturalistically acceptable – for example, panpsychists, emergentists about causation, and Free-Will Libertarians will all typically insist that their theories are naturalistically acceptable. Attempts to gloss ‘naturalism’ almost always have a line to the effect that it means ‘not being committed to anything supernatural or spooky’. But of course almost nobody in philosophy these days is a card-carrying supernaturalist,Footnote ⁵ nor would any philosopher be happy to describe their own preferred theory or posits as ‘spooky’. Moreover, it is a familiar point that sometimes what initially strikes us as ‘spooky’ can later become scientific orthodoxy. Newton's force of gravity was thought to be too spooky by many of the serious scientists of his day. Non-local quantum effects are still considered too spooky to stomach by some physicists today. De Caro and Macarthur put this point neatly: ‘What is “supernatural” today … may be part of tomorrow's new science’ (De Caro & Macarthur, Reference De Caro, Macarthur, De Caro and Macarthur2022, p. 2).

The fact that the term ‘naturalism’ excludes so little is, I suggest, a symptom of a deeper problem, which is that there is no principled basis for distinguishing between naturalistic theories that are false and (false) supernatural theories. The flipside of this is that there is likewise no principled metaphysical basis for distinguishing between supernatural theories that turn out, very surprisingly, to be true and (true) naturalistic theories. As a way of dramatizing this point, consider what we should say or think if a paradigmatically supernatural theory turned out to be true. For example, suppose it turned out, to everyone's immense surprise, that ghosts are in fact real. We establish beyond reasonable doubt that ghosts exist and they sometimes, somehow exert enough of a causal influence on things that they can occasionally be observed or detected. I think it is clear that were this to happen, there would not be any point in asking whether ghosts are ‘natural’ or ‘supernatural’. They would be real – part of reality. Once we admit that they are part of reality, what further possible theoretical interest would be served by asserting or denying that they are part of ‘nature’? For example, suppose that theorist A reacts to this amazing new discovery by saying: ‘Wow, well I guess it turns out that in addition to all the animals and plants, all the planets and galaxies and atoms, nature (the “natural world”) also includes ghosts’. And suppose that theorist B disagrees: ‘No, what the discovery of ghosts shows is that not everything that exists is natural (part of nature). It turns out that reality is divided into two realms: the natural, which includes all the atoms and animals and galaxies etc. that we already accepted in our ontology, and the supernatural, which includes the ghosts.’ This sort of disagreement about whether to use the word ‘nature’ (‘natural’) to describe everything that exists or just some of what exists is, I submit, a clear example of an empty verbal dispute. Of course, we can imagine that there is a more significant/substantial disagreement between A and B – perhaps A thinks that ghosts are constituted by fundamental particles and forces that we already accept in our ontology, whereas B thinks that ghosts require accepting a fundamental new category of stuff that cannot be composed from or reduced to other stuff that we already accept. This would be a disagreement about the fundamentality and composition of something that is accepted by both sides as real, which could be a perfectly interesting and substantial disagreement to have. But notice: there is no reason to think that positing new fundamental stuff in our ontology is any less ‘naturalistic’ than explaining the phenomenon in question by appeal to stuff that we already accept. So there would be no reason to frame this latter, substantive disagreement as taking sides about ‘naturalism’. (We will return to this point in section 2.3, below.)

Conversely, consider those theorists who try to provide scientifically respectable evidence for ghosts, or ESP, or memories of past lives, etc. There are many different journals filled with papers in this genre, complete with tests of statistical significance and p-values and impressively complicated-looking graphs and diagrams. Do such theorists qualify as supernaturalists or naturalists? Again, I think it is clear that there is simply no interesting, non-arbitrary answer. These people think the phenomena in question are real and can be somehow observed or detected by us. Of course, we orthodox folk all assume that they are wrong and that such phenomena do not exist. But once it is settled that they are wrong, what possible purpose would be served by trying to decide the further question of whether their false theory was a false naturalistic theory or a false supernatural theory? Paul Horwich claimed that the minimal and most plausible form of naturalism, shorn of any inflationary metaphysical baggage, is what he calls ‘anti-supernaturalism’ – which he defines as follows:

Horwich's anti-supernaturalism does not tell us how to deal with the sorts of believers in ghosts or telepathy who treat these phenomena as having causal, spatio-temporal effects, which they try to detect and then publish journal articles about. Such people are evidently trying to adhere to something like the norms of science. Of course, we assume that they are doing bad science whilst they presumably think that the scientific orthodoxy of the day is mistaken in not recognizing the legitimacy of their ghost-detecting methods. But exactly the same could be said of legions of bad scientific theories about more reputable subjects. For example, think of the steady stream of discredited claims to have produced cold fusion, starting with Fleischman and Pons (Reference Fleischman and Pons1989). Or just think of all the pseudo-scientific forms of cancer ‘treatment’ that exist: electro-magnetic devices, special diets, injecting various unproven chemicals, etc. So-called ‘supernatural’ theories are really just theories that posit the existence of things that we sensible, orthodox folk think obviously do not exist. But this is also true of any number of discredited scientific theories. There is no deep, interesting difference between a theory that posits the existence of a new fundamental particle that we think does not really exist, a theory that posits the existence of fundamental conscious properties that we think do not really exist, a theory that posits the existence of a vital force or ‘élan vital’ in living things that we think does not really exist, a theory that posits the existence of telekinesis that we think does not really exist, a theory that posits the existence of ghosts and spirits that we think do not really exist… The only difference between these cases is one of degree – that is, how implausible we find them and perhaps how irrational or crazy we find their proponents.

2.2 God

If instead of ghosts we take the example of theistic theories that assert the existence of a god, we find the same issues playing out. I expect that for many self-described ‘naturalists’, if naturalism is to exclude anything it must exclude the existence of God. And also in the opposing theist camp it has sometimes been taken as definitional of naturalism or ‘the natural realm’ that it excludes a deity – see e.g., Plantinga (Reference Plantinga2011) or Kopersky (Reference Kopersky2017). However, there have also been some theist philosophers who insist that God is part of the natural world. Fiona Ellis defends this view in her Reference Ellis2014 book ‘God, Value and Nature’ and in various subsequent articles, as does Teed Rockwell (Reference Rockwell2017). Likewise it is the view of the distinguished rabbi and philosopher Jack J. Cohen (Reference Cohen2019) and also of the eminent theologian and scientist Arthur Peacocke (Reference Peacocke and Pennock2001). So, we have some theists who insist that God exists but is not natural and others who insist God exists and is natural. But just as with the imaginary dispute between ghost-theorist A and ghost-theorist B, if God really does exist it is hard to see what further theoretical purpose would be served by insisting that God is or is not ‘natural’. God would be real! Suppose we somehow discover or learn for sure that some monotheistic religion is correct: God created everything, God will punish or reward us after death, God can sometimes causally intervene in our affairs, etc. What difference would it make to then insist that God is ‘part of’ nature or ‘outside’ nature? What further theoretical purpose could those labels possibly serve?

2.3 Dualism

Now, of course, very often when someone claims that some phenomenon, X, can be naturalized, they mean that X can be explained by appealing only to stuff that we already uncontroversially accept in our ontology. But clearly not everything can be reductively explained in this way. Every theory or explanation relies on some primitive elements that are taken for granted. And it is an accepted part of normal scientific practice that sometimes, when we cannot explain some phenomena as arising out of stuff we already accept and understand, we need to posit new fundamental entities or forces etc. One of the best-known instances of this in the history of science was when James Clerk Maxwell, partly in response to the many failed attempts to explain magnetism and electricity as arising out of mechanical interactions of matter, introduced a set of new fundamental laws governing electromagnetism. Clearly there is nothing inherently spooky or supernatural about accepting new fundamental stuff into one's ontology that cannot be reduced to other stuff, though this is something we try to avoid most of the time for reasons of parsimony. And if we now turn to thinking about Dualist theories of the mind/consciousness, we find that David Chalmers – surely the most prominent contemporary defender of the claim that consciousness cannot be fully explained in neurophysical terms – describes his own theory as ‘naturalistic dualism’ and invokes the story of James Clerk Maxwell as a comparison:

And on the next page he justifies his use of the label ‘naturalistic’:

In contrast to Chalmers’ suggestion that we take consciousness to be one more fundamental part of nature, there have been various philosophers – e.g. Montero and Papineau (Reference Montero and Papineau2005), Wilson (Reference Wilson2006) – who take the non-fundamentality of mental phenomena to be their core principle. This via negativa strategy for defining physicalism effectively takes physicalism to be the negative existential claim that there are no fundamental mental properties. Now, to be clear, these via-negativists were talking about physicalism rather than naturalism, but we could just as easily take the via negativa route to defining naturalism. If we define naturalism in this negative fashion, then Chalmers’ naturalistic dualism will seem like a contradiction in terms, just as naturalistic theism will seem oxymoronic to people like Plantinga and Kopersky. On this view, if consciousness turns out to be an extra fundamental feature of reality, then rather than having expanded our picture of nature, this discovery would show that naturalism is false. But would it have made any real theoretical difference if Chalmers had instead chosen the label ‘non-naturalist’ dualism for his own theory? Of course not. This alternative label might suggest how very different and sui generis conscious properties are from familiar properties by, as it were, drawing a line around them and calling them ‘non-natural’ rather than allowing the natural domain to expand so as to include them. But the actual substance of his theory would be totally unchanged, which is the claim that conscious qualities must be accepted as extra primitive/fundamental elements in our best theory of reality and that cannot be reduced to stuff we already accept. Of course, many philosophers disagree with Chalmers about this and that is a perfectly good, substantial philosophical debate to have. But whether or not Chalmers’ proposal should count as naturalistic is not a substantial question, it is an empty verbal dispute.

2.4 Possible Worlds

I think that often when people try to articulate what they mean by ‘naturalism’, in the background there is the idea that ‘the natural world’ forms a single domain or realm that is spatially connected and causally complete. The idea then would be that naturalism is the thesis that everything that exists exists within this single spatially connected, causally complete domain, whereas supernaturalism involves ontological commitment to some further distinct domain – perhaps heaven, perhaps a spirit realm, perhaps a mental realm, perhaps a Platonic realm – which is (at least mostly) spatially and causally disconnected from the natural domain. It is worthwhile to make this idea explicit so we can see that it does not survive scrutiny.

Consider, for example, David Lewis's (Reference Lewis1986) somewhat notorious realism about possible worlds. According to Lewis, other possible worlds are just as concrete and real as our world, but there are no spatio-temporal or causal connections between different possible worlds. Lewis was a student of Quine and his motivation for being ontologically committed to all these concrete possible worlds was, at bottom, the Quinean motive that they are indispensable for explaining the truth and falsity of modal claims (and modal claims are indispensable to our best theories of how the world works). So is Lewis's theory a naturalistic theory or not? David Papineau asserts that it is not:

But is it really so obvious that naturalism requires commitment to the existence of at most one spatio-temporal realm? The ‘Many Worlds’ interpretation of quantum physics (Everett Reference Everett, DeWitt and Graham1955/1973) would presumably not count as a naturalistic theory by that criterion.Footnote ⁶ More importantly, what difference does it make to call Lewis's theory naturalistic or not? All that really matters is whether it is correct or not (whether it really is the best explanation for the truth/falsity of our counterfactual claims).

Or consider Strawson's thought experiment of ‘massive reduplication’ near the start of Individuals (Reference Strawson1959). Strawson imagines that the entire known universe might be exactly qualitatively duplicated. Now, Strawson did not specify that the two qualitatively identical regions of reality are spatially and causally disconnected. But let's suppose for a moment not only that Strawson's thought experiment is in fact true – there exists a perfect qualitative duplicate of our entire known universe – but that these two regions of reality are spatially and causally disconnected. Would this mean that naturalism is false? Or would it just mean that our picture of the natural realm ought to be revised and expanded? Once more I think the only sensible answer here is: who cares?! What possible theoretical interest or significance would either withholding or awarding the label ‘naturalist’ have?

Moreover, notice that those who believe in a heavenly realm or a spirit realm or a dualist mental realm, typically do not think that these are totally causally or spatially disconnected from the realm of familiar physical things, since angels or ghosts or non-physical souls are typically supposed to interact with the familiar physical realm at least on rare occasions. So it is doubtful that defining naturalism in terms of a single spatially connected, causally complete domain will exclude these paradigm cases of alleged supernatural or non-natural phenomena. For presumably, when considered in combination, the familiar physical world and, say, the alleged spirit realm would together form a single spatially connected and causally complete domain – though the connections between the two ‘realms’ may be relatively sparse.

The basic point here: the extent to which reality comes divided into distinct, disconnected (or at least partially disconnected) domains is an open question to be answered by our best theories – scientific but perhaps also philosophical. If in fact, say, Lewis is correct and reality consists of more than one disconnected ‘world’, there is no obvious reason to think of this as any more or less ‘naturalistic’ a result than if reality consists of a single spatially connected, causally closed domain. After all, if the picture of reality provided by our best scientific theories counts as ‘naturalistic’, there is no obvious reason why having more of the same sort of stuff duplicated into two or more distinct worlds should then count as supernatural or non-natural (though of course it may be ontologically profligate).

2.5 Abstracta

What about the suggestion that in addition to the concrete, causal domain there is also a non-causal realm of abstracta – numbers, propositions, truth-values, Platonic forms, perhaps also moral values or norms? Once more it is hard to see what useful theoretical motivation there could be for insisting that abstracta must be non-natural rather than allowing that ‘nature’ includes abstract objects. After all, Quine – often thought to be the arch-naturalist supreme – allowed that abstract numbers exist since he thought that our best scientific theories ineliminably quantify over them.Footnote ⁷ To take a more recent example, Matteo Morganti writes:

I sympathise with what Morganti is saying here, and agree that there is no good theoretical reason to exclude numbers from the domain of ‘the natural’. But the conclusion that I think we should really draw is that there is no good reason for including or excluding numbers from counting as part of ‘nature’. There simply is no interesting, substantial question as to: supposing abstract entities, such as numbers, really do exist, should they then count as natural or non-natural? Again, if one theorist says that abstract numbers exist but they are not part of the natural world, and another agrees that abstract numbers exist but insists that they are part of the natural world, this is surely nothing more than an empty labelling dispute. Notice that here we have a contrast between naturalism and physicalism insofar as any plausible notion of ‘the physical’ will presumably at least have to exclude abstract objects that have no location in space and time. But it is not at all obvious whether or why some notion of ‘the natural’ (or ‘nature’) should include or exclude abstracta, supposing that reality includes them. Here is the late, great Barry Stroud making this very point:

In general, one could always insist that what is real (what exists) and what is natural coincide. No matter what we accept into our ontology this just means that the realm of the natural has expanded. Or one could choose to divide reality into two subsets – and call one of them the natural and the other the supernatural or the non-natural. Whenever we posit new items in our ontology we would face this choice. But unless there is some important further theoretical reason or pay-off, either way of using the label ‘natural’ will just be an arbitrary stipulation. We have noted that many believers in ghosts or spirits or ESP etc. seek to prove that these phenomena are real using at least the rudiments of scientific method. We have seen how some theists claim that God is natural whilst others think this is a contradiction. We have seen Chalmers claim that his dualism about the mind is naturalistic. We have recalled Lewis's view that possibilia are just as flesh-and-blood concreta as we are. And in this section we have seen Morganti and Quine claim that commitment to abstract numbers is entirely naturalistic. Ghosts, gods, primitive mental qualities, possibilia, abstracta – these were supposed to be the clear-cut cases of non-natural or supernatural stuff. And yet there is no good reason why those who believe that such things really do exist could not equally well claim that they are ‘natural’ phenomena. The moral in each case, I claim, is that the question of whether these phenomena are ‘natural’ or not is entirely empty and arbitrary. The only interesting question is whether these phenomena are real – whether they exist. Once this is established, any question about naturalism immediately fades into irrelevance. So, we should just drop talk of ‘the natural’ and simply talk about what is real, what exists, what is indispensable to our best theories of how the world works. If it turns out that this includes gods, libertarian free will, primitive conscious qualities, acausal normative facts, etc., so be it. If not, not. Once more, Barry Stroud – though he is sometimes counted as a ‘Liberal Naturalist’ by those who like that label – drew the conclusion that ‘naturalism’ is just an empty slogan, nearly 30 years ago:

2.6 Reductive Explanations and Fundamentality

Now, to be clear, I am not suggesting that once we drop the idea of ‘the natural’ then our philosophical or metaphysical anxieties about the various puzzling phenomena/topics will just evaporate. On the contrary, I think that all of the standard philosophical debates where naturalism (or physicalism) is typically invoked can still perfectly well be posed and pursued, and all the genuine, substantial philosophical questions will remain, once we expunge talk of the ‘natural’.

I think it is fair to say that philosophers who are unfriendly to naturalism (as a metaphysical thesis) are very often suspicious of the overuse of reductive explanations, feeling that important or valuable phenomena are being ignored or explained away. Whereas those who are sympathetic to the idea of (metaphysical) naturalism tend to be impressed by various successful reductive explanations and the great successes of the physical sciences in explaining various phenomena by appeal to their microscopic constituents.Footnote ⁸ But it is, I suggest, plainly absurd to have a single attitude – positive or negative – towards the very idea of reductive explanations in general. There is no single attitude we ought to take to reductive explanation simpliciter. We should simply welcome those that work (the correct/true reductive explanations) and reject those that don't (the false/bad explanations). Everyone should admit that sometimes a phenomenon X turns out to be ‘nothing over and above’ phenomenon Y. But everyone should also admit that at other times, in order to explain everything that needs explaining, we have had to admit new basic, primitive stuff into our ontology. For example: you might be unimpressed by primitivism about colours – the view that there are sui generis colour properties ‘over and above’ any other properties, which cannot be reduced to, say, spectral reflectance properties. And yet you might find yourself more sympathetic to some kind of realism about abstract mathematical entities. There would be no inconsistency or incoherence in such a combination of attitudes since they are just two separate topics. One can think that mathematical abstracta are indispensable, irreducible items in our best overall theory of reality but that primitive colours are not (or vice versa).Footnote ⁹ We have to actually look at the details of the specific debates, at what needs explaining and what the alternative rival theories are, before we can make up our minds about the plausibility of reductive explanation or of accepting new fundamental primitives.

All of the familiar metaphysical/ontological debates we have looked at where the term ‘naturalism’ tends to crop up can be framed in terms of whether some phenomenon X is a brute/irreducible feature of reality or whether it can be reduced to/explained in terms of other stuff – or perhaps be eliminated altogether. And in pretty much all of these standard debates the crucial ‘pinch-point’ concerns causation – does X do causal work? Is X part of the causal structure of reality? If we think X does do causal work then the philosophical worries that tend to arise are about over-determination: we already have something else that is apparently doing all the causal work, so there is ‘no room’ for X to make a causal difference. This is what we find in the classic debates about free will and about non-reductive theories of mental states. The issue also arises for primitivism about colours – is there any ‘space’ for the alleged primitive colour properties of an object to causally affect my visual system, given that it is already being affected by how the microscopic texture of the surface modulates the wavelengths of light which are reflected into my retina? A closely related kind of issue is: if X does do extra, distinct causal work and make a causal difference, then we should be able to observe this – but we don't. This is a standard objection to strong emergentism and also to libertarian views about free will and to interactionist dualism – if these emergent/libertarian/dualist phenomena are real and make a causal difference, then we should see exceptions to micro-physical laws. But, according to philosophers like McLaughlin (Reference McLaughlin, Beckermann, Flohr and Kim1992) and Papineau (Reference Papineau1996, Reference Papineau, Gillett and Loewer2001) we have never seen any such exceptions – McLaughlin claims that ‘there is a not a scintilla of evidence’ (Reference McLaughlin, Beckermann, Flohr and Kim1992, p. 91) for novel, emergent macro-behaviour in physical systems – and in response, some philosophers have taken the robust route of replying: oh yes, there is! For example Hendry (Reference Hendry, Paoletti and Orilia2017, Reference Hendry, Gibb, Hendry and Lancaster2019) points to various kinds of quantum chemical phenomena – roughly, discontinuous changes in the organisation of molecular structures – as examples of ‘strongly emergent’ behaviour that cannot be explained by the molecule's sub-atomic constituents. Similarly, Cucu (Reference Cucu2024) supports interactionist dualism by claiming that there are certain kinds of neural events that cannot be explained by micro-physical processes.

On the other hand, if X does not do any causal work or make any causal difference, then the standard worry becomes the epistemological question: how could we ever know about X if it cannot causally affect us or impact our mental processes in any way? This is the classic worry for realism about abstracta (numbers, sets, etc.) and also for non-naturalism about moral properties.Footnote ¹⁰ It also arises for epiphenomenalism about phenomenal properties – e.g., Chalmers (Reference Chalmers, Smith and Jokic2003) ends up endorsing an unorthodox, non-causal account of the basing relation in order to explain how our beliefs about phenomenal properties can be justified despite these properties being epiphenomenal.

The point here is just that all these familiar debates really revolve around the question: must we accept X as an irreducible/primitive extra feature of reality or can X be explained by appeal to something else we are already committed to (or even just eliminated altogether)? This question produces a familiar set of dialectical options:

I suggest that when we frame ontological questions in this way, without any mention of ‘nature’ or ‘the natural’, it becomes easier to take each X on a case-by-case basis. Sometimes we can point to a real causal difference that X makes that cannot be otherwise explained, sometimes we cannot. Sometimes we do need to accept new primitive elements into our best theory/picture of reality. At other times we can successfully keep things simpler by showing that X is fully explained by/determined by stuff that we already accept the existence of. And sometimes we should just eliminate X altogether.