We can think about things that don't exist. For example, we can think about Pegasus, and Pegasus doesn't exist. Thus, there are things we can think about that don't exist. Thus, there are things that don't exist. Thus, we need an account of how there can be things that don't exist. That is, we need an account of quantificational uses of ‘There are n Fs’, and an account of uses of ‘n Fs exist’, such that (for positive n) ‘There are n F's’ does not entail ‘n Fs exist’. And we need an account of the things that don't exist: an account of truths about those things, and of what determines those truths, given that truths are exhaustively determined by what exists. Such are the large tasks attempted by this useful and insightful book, and a sketch of their motivation (which Crane presents in more detail in Chapter 1).
A natural first thought would be that the tasks are needless. We don't really think about things that don't exist. Rather, we sometimes think about nothing. Thinking that there is a largest prime is not thinking about the largest prime, for there's no such thing to be thought about. Thinking that Pegasus is a winged horse is not thinking about Pegasus, for there's no Pegasus to be thought about.
Alternatively, we might allow that we really do think about things that don't exist, on some reasonable understanding of ‘thinking about’, but deny that it follows that there are things that don't exist. For the alleged entailment seems to depend on the acceptability of two prior claims: claim (1) ‘We can think about Fs that don't exist’ entails ‘There are Fs we can think about that don't exist’; and claim (2) that ‘There are Fs we can think about that don't exist’ is a sort of conjunction, and so entails ‘There are Fs that don't exist’. And many would balk at those claims. For example, suppose that I think that there is a largest prime. And suppose that we allow that it follows that I'm thereby thinking about the largest prime. Still, it would be an implausible leap to reason in accord with claim (1), deriving that there is a largest prime that I'm thinking about. And it would be even less plausible to reason in accord with claim (2), to the conclusion that there is a largest prime. One might doubt that we should accept Crane's claims, framed in terms of things, given that we wouldn't accept more specific claims, framed in terms of a largest prime. And if we don't accept Crane's claims, then we should take the claim that there are things we can think about that don't exist as a sort of loose summary of what everyone anyway accepts: people occupy intentional states that are naturally characterised by appeal to apparent singular terms, and apparent quantificational phrases, but where the singular terms do not refer, and the matrices of the phrases are not satisfied. (It will turn out that despite initial appearances, that may ultimately be a version of Crane's positive view about things that do not exist.)
Crane works hard, in chapter 1, to deflect such sceptical responses, and in chapter 2, he offers a sketch of an account of quantificational claims like the claim expressed by ‘There is a largest prime’, on which the truth of such claims need not be existence entailing. So, he wants to allow that ‘There is a largest prime’ might be true while ‘A largest prime exists’ is false. But the problems in this area seem to run deeper than he allows. For whatever one says about existence, it would be hard to deny that the truth of ‘There is a largest prime’ requires there to be an element in the domain that satisfies the matrix ‘is a largest prime’. But there is no witness for that matrix. So, it appears that ‘There is a largest prime’ cannot be true, whether or not it entails that a largest prime exists.
Crane might respond by denying that there is a largest prime for reasons other than the non-existence of such a number. For example, he might deny it because a mathematical proof terminates in the claim that there is no largest prime, rather than in a claim to the effect that a largest prime does not exist. The question, then, would be this. Suppose that a benighted soul thought that there is a largest prime. According to the pattern of reasoning considered above, that would be grounds for allowing that they are thinking about a largest prime. And it might seem to follow that there is a largest prime that they are thinking about. And that would again run into the difficulty that there isn't.
Crane's general response to this pattern of difficulties is the following. His claim is not that when someone thinks that there is a largest prime, and so thinks about the largest prime, there is a largest prime that they are thinking about. Rather, his claim is that when someone thinks that there is a largest prime, so thinks about the largest prime, there is something – perhaps a thing – that they are thinking about. And in making that claim he needn't be, and isn't, committed to the further claim that what they are thinking about is a largest prime. But if the thing they are thinking about is not the largest prime, what is it?
Crane's proposal is that – just like the things that exist – the things we think about that don't exist have distinctive properties. So, the truths about the things we think about that don't exist go beyond undiscriminating truths like the truth that they don't exist. However, Crane doesn't accept that everything that we think about such things is true of those things. For example, the thing we think about when we think that the largest prime exists is not a largest prime, let alone an existent largest prime. Crane disagrees with those who take a more equitable view about the things we think about the non-existent. He disagrees with those who hold that many, or all, of the things we think about the non-existent are true – for example, those who hold that Pegasus is a winged horse. And he disagrees with those who hold that more or less all of the things that we think about the non-existent are not true (as on negative free logics) – for example, those who hold that it's not true that Pegasus is a mythical winged horse (chapter 3).
Crane's attempt to take an inequitable view must deal with the following difficulty. Consider the following two sentences.
(1) Pegasus is a winged horse.
(2) Pegasus is a mythical winged horse.
Crane seems to endorse the following:
A. In (1) and (2), ‘is a winged horse’ and ‘is a mythical winged horse’ work in the same way, to attribute properties to Pegasus: the properties of being a winged horse and of being a mythical winged horse, respectively. (18–23, 67)
B. (1) is false. (55)
C. (2) is true. (23)
A is not perspicuous as it stands. However, it would be natural, if not mandatory, to understand it as implying that the predication in (2) functions in the same way as the predication in (3).
(3) Pegasus is a brindle winged horse.
The obvious difficulty is that (3) entails (1). So, the natural understanding of A seems to deliver D:
D. (2) entails (1).
And now, we need only an application of a natural generalisation about entailment to deliver E:
E. If sentence (2) entails sentence (1), then if sentence (2) is true, then sentence (1) is not false.
So, we can't accept all of A–E.
Now it might seem natural to blame A. However, it's not entirely clear how to reject A consistently with a face-value treatment of the embedded predication, and consistently with retaining the claim that (2) attributes a property to Pegasus. It may be that this difficulty figures in explaining the fact that many theorists in this area adopt an equitable view of (1) and (2), so rejecting either B or C.
Crane recognises the need to speak to this difficulty, writing: ‘Being a mythological horse is not, of course, being a kind of horse’ (136. See also 68, 135). But the question raised by the difficulty is, how is this to be explained? Crucially, it must be explained in a way that sustains Crane's claim that being a mythological horse is a property possessed by things.
One natural analogy would be with toys. For just as Crane wants to hold that a mythical winged horse need not be a winged horse, it would be natural to hold that a toy winged horse need not be a winged horse. However, it also seems natural to allow that toys are things, and that, for example, a toy winged horse has the property of being a toy in much the same way that a horse has the property of being a horse. The difficulty, however, is not removed, but only generalized. For if we take this route, we now require a unified account of what is it to be a toy F and what it is to be a mythical F. Furthermore, we would have to face up to the problem that being a toy is existence entailing (59–64, 119–20), for its being true that there is a toy F entails that a toy F exists.
Obviously, there is more to be said here. I don't want to suggest that there is no way of defending Crane's collection of commitments. However, I do want to suggest that further work would be required before we are in a position properly to assess the prospects of that collection. Moreover, even if Crane were able to supply a principled way of blocking the problematic transition from (2) to (1), the question would naturally arise: why not seek to impose the blockade at an earlier stage, by providing a means of denying that it follows from the fact that someone is thinking that there is a winged horse that there is some thing about which they are thinking?
Suppose the difficulty overcome. We now require an explanation for why the things we think about that don't exist have the specific range of properties they have. And Crane seeks to provide this explanation under the constraint that such an explanation must be given only by appeal to what exists (and properties of what exists) (p.18–23). Crane proposes what he calls a reductive explanation: a piecemeal explanation of the obtaining of truths about the non-existent on the basis of truths of other kinds. Specifically, Crane aims to explain the truths about the non-existent by appeal to truths about the intentional states, episodes, and processes of existent thinkers. That is, in Crane's terms, aside from certain special cases of properties, like non-existence and self-identity, non-existent things have only representation-dependent properties (68–71).
One worry here is the following. Suppose that individual thinkers, rather than pluralities of thinkers, are the locus of representation. In that case, one might wonder what makes it so that when you think about Sherlock Holmes, and I think about Sherlock Holmes, there is a single non-existent thing that both of us think about. Crane (162–167) seems to want to deny that there is a single thing, because he denies that such identity claims about non-existents make sense. For him, then, it seems that no one thing has the property of being a famous fictional detective.
The most fundamental concern I have about Crane's project is somewhat amorphous. The concern is that the project was motivated by appeal to the fact that our ground level characterisations of people's intentional psychologies make play with embedded commitments to things that don't exist: my belief that there is a largest prime; Eliza's desire for a unicorn; Flo's fear of ghosts. We're tempted to move from those facts to a generalisation, to the effect that in each such case, people think about things that don't exist. And from there, we're tempted to the claim that there are things that don't exist but about which people nonetheless think. The latter claims, then, seem to raise distinctive puzzles, insofar as we take them to carry commitments beyond those carried by the ground level claims that they systematise. Crane's response is that the latter claims don't really go beyond the ground level claims: their truth is exhaustively explained by appeal to the truth of the ground level claims. But now it's hard to see what explanatory work the claims about things that don't exist are doing and, so, why we should be especially puzzled by those claims. One might think that the deepest puzzles in this area concern the ground level claims themselves, rather than the claims about things that, at best, systematise those ground level claims. That is, the deepest puzzles here concern our capacity to think, for example, that Pegasus flies or that the largest prime is greater than a million, given that neither Pegasus nor the largest prime exist, and not our propensity to characterise such thinking as being about things that don't exist. And it seems to me that it takes us little distance towards resolving those puzzles to be told that such thought in effect creates its own subject matter, albeit without bestowing existence on that subject matter. (Crane briefly considers some of the ground level claims in chapter 6, and the circularity worry in his Concluding Remarks.)
Although I'm sceptical about the project undertaken in the book, it represents an admirable attempt to clarify various issues surrounding thought about the non-existent. Whether or not one agrees that there are things that don't exist, or that such thought is about those things, there is much to learn from Crane's discussion, not merely about its central target, but about intentional phenomena more generally. I recommend it highly to those interested not only in the specific matters with which it deals, but also to philosophers of mind more generally.
