Draft only; please don't cite.  Looks like the footnotes
disappeared when I converted to html; I'll try to fix that soon.
"I Wish I Had Never Existed"
Curtis Brown

Both David Lewis and Roderick Chisholm have proposed that beliefs are best understood, not as relations between people and the propositions they believe, but as relations between people and the properties they "directly attribute" to themselves or "self-ascribe." If this account is correct for belief, it seems that it ought to be possible to extend it to other "propositional attitudes" such as considering and wishing. But the most straightforward way of extending the account to such other attitudes faces difficulties, some of which are discussed in a paper by Peter J. Markie. In this paper I will show how to apply the account to considering and wishing in a way that avoids such difficulties.

The account of belief favored by Lewis and Chisholm goes roughly as follows. Some of our beliefs are best understood as relations to properties rather than to propositions. My belief that there is a mouse nearby is best understood as my "self-ascribing" (Lewis’s term) or "directly attributing to myself" (Chisholm’s) the property of being near a mouse. For reasons well rehearsed in the literature, it seems hopeless to try to understand my mental state when I believe that there is a mouse nearby merely by saying what propositions I believe. (A propositional account will attribute to me such objects of belief as the proposition that the F is near a mouse, but will then seem to leave out the very important fact that I believe that I am the F.)

Are there then two kinds of belief, one of which relates people to properties and one of which relates them instead to propositions? Lewis and Chisholm argue instead for a unified theory which subsumes propositional belief under self-ascription of properties. The device which makes this possible is simple: when I believe the proposition that P, Lewis and Chisholm say that I self-ascribe the property of being such that P. Markie’s criticisms center on this reduction of propositional attitudes toward properties.

Markie’s criticisms are not, properly speaking, objections to the account of belief just sketched. Instead they are criticisms of a proposed extension of the account to the attitude of entertaining or considering a proposition. On this proposed extension one considers that one is F if one directly considers that one has the property of being F, and one considers that P if one directly considers that one has the property of being such that P.

For convenience, let us express properties by expressions of the form <x if F> and propositions by expressions of the form [P]. Then on Lewis’s and Chisholm's view I believe [P] if I self-ascribe <x is such that P>, and on Markie’s proposed extension I consider [P] if I directly consider <x is such that P>.

Of Markie’s three related objections the third is, or can be made to be, the most troubling; I will focus my attention on it. The objection is that sometimes when we consider counterfactual but possible propositions, the property account must analyze this as the direct consideration of impossible properties. Let us consider a slight adaptation of Markie’s example:

Descartes considers [Descartes does not exist & Smith is wise]. The property theory must analyze this as a relation to a property; on Markie’s extension of Chisholm’s account the analysis will go something like this: Descartes directly considers <x does not exist and x is such that Smith is wise>. But it seems that the property here is one that nothing could have in any possible world, since it seems that one must exist in order to be such that Smith is wise. (The impossibility would be even clearer if for ‘x is such that’ we put ‘x inhabits a world in which,’ as Lewis does.)

The account given by this version of the property theory seems clearly incorrect. It is not just that it analyzes considering a possible proposition in terms of directly considering an impossible property. That by itself might be tolerable. But the analysis simply seems to get Descartes’ state of mind wrong. He is not considering being in a situation in which he does not exist and Smith is wise.


Has the property account been refuted? Markie has left the property theorist considerable room to maneuver. The property theorist might, for instance, allow nonexistent objects as values of his variables. <x does not exist and x is such that Smith is wise> would then be a property which things could possess provided only that they were nonexistent, as perhaps <x does not exist and x is such that Conan Doyle wrote stories about him> is a property possessed by Sherlock Holmes. But it would be better to meet the objection without commitment to nonexistent objects. And anyway the difficulty that the property does not capture Descartes’ mental state would remain. Surely when I consider a situation in which I do not exist and Smith is wise I am not considering being a nonexistent object which has the property of being such that Smith is wise.

The property theorist might also say, echoing Lewis on belief, that it is not straightforward to determine, from ordinary consideration ascriptions, what properties someone is directly considering. He would then owe an account of what property one does directly consider when one considers that one does not exist and Smith is wise.

But the property theorist will do best to simply deny that there is any such thing as "direct consideration." Markie’s objection does not show that the property account of belief and desire is mistaken, only that it cannot be extended in Markie’s natural way to the attitude of entertaining a proposition. Lewis, it should be noted, is cautious about this. He writes that his account applies to belief and desire and anything amounting to combinations thereof, and adds that he is unsure whether "anything is left out --perhaps some ill-understood attitudes of imagining, conceiving, contemplating, or entertaining a thought" (p. 145).

Perhaps, as Lewis here hints may be the case, considering a proposition cannot be analyzed in terms of more basic attitudes. But even if it can, it would be a mistake to extend the property analysis to it in Markie’s straightforward way. I would like to suggest a rough and tentative account of what it is to consider or entertain a proposition; if this account is approximately correct, then the property theory does apply to entertaining, but in an indirect way which does not give rise to Markie's objection.

I suggest that entertaining or considering a proposition is a generic attitude, an attitude one has to proposition if one has any of several more specific attitudes toward it. What am I doing if I entertain the thought that the moon is made of green cheese? I may believe the proposition that the moon is made of green cheese (‘M’ for short). Or I may believe not-M. Or I may wonder whether M is true (roughly: believe that I do not know whether M and desire that I either know that M or know that not-M).

Again, it may simply have struck me that the moon could be made of green cheese; that is, I may believe that possibly M. Or I may be wondering what would happen if the moon were made of green cheese; for instance, I may wonder whether, if M were the case, the moon would collapse.

In any of these cases, I suggest, I entertain the proposition that M. But, contrary to a traditional view, entertaining is not one component of belief, disbelief, wondering whether, etc. It is not something I could do independently of doing one or more of these other things. It is nothing more than the generic notion of doing at least one of these other things. (This claim needs some modification. Entertaining a proposition seems to require a conscious awareness of it, while belief need not. But I shall ignore this complication.)

Thus to say ‘Joe entertains the proposition that P’ is to say roughly: Joe believes P or Joe believes not-P or Joe wonders whether P or Joe believes that possibly P or Joe believes that necessarily not-P or Joe wonders whether if it were the case that P, it would be the case that Q.

Now, what about entertaining the proposition that one does not exist? This is not a proposition one could believe or wonder about. So in the typical case entertaining this proposition will consist in believing it to be false but possible, together perhaps with wondering what would happen if it were true. And the property account has no trouble with any of these. Nor, then does the account have trouble with Descartes’ entertaining the proposition that he does not exist and Smith is wise. The central element in Descartes’ entertaining this proposition will typically be his self ascribing <x is such that possibly (x does not exist and Smith is wise)>, a property one can very easily have.


Markie’s objection, we have seen, can be regarded as refuting his own extension of the property theory rather than the theory itself. But it suggests a related difficulty which may strike closer to home. Mel, a melancholy soul, wishes he had never existed. How can the property theory analyze this wish?

On the face of it at least, wishes seem like a kind of desire, and the theory is supposed to apply directly to beliefs and desires. But then the theory must analyze this wish as the desire to have a certain property, the property of never having existed, or perhaps as the desire to inhabit a certain world, a world in which one has never existed. Fairly clearly these analyses get the wish wrong.

We might wonder whether this is a genuine problem at all. The property theory says that wanting is always wanting to have a certain property. But it does not matter if the property Mell wants is one which he cannot in fact have. Lewis writes: "I suppose I might want to be a poached egg. ...Would I then want to inhabit one of the worlds where I am a poached egg? That’s not it. I take it there are no such worlds. ...Bit if the object is a property, it is nonempty. It is a property that plenty of poached eggs actually have" (p. 146). Can we say something similar about the desire never to have existed? It is not the desire to inhabit a world in which one has never existed. (There are such worlds, but one cannot inhabit them.) Is it then simply the desire to have a certain property, the property of nonexistence? But we cannot say of this property that it is "a property that plenty of nonexistent things actually have" unless we have a theory of nonexistent objects, and we have been trying to avoid that.

Again, we might simply take the bull by the horns and insist that Mel is genuinely muddled, that one can only wish never to have existed by confusedly imagining oneself leading a happy quasi-existence in the wished-for situation. But this seems no more persuasive than arguments that one can only fear death if one confusedly imagines oneself suffering afterward.

The proper response, I suggest, is to deny that wishing is a kind of desiring. The property theory applies directly to belief and desire, but wishing is to be construed as a complex state analyzable in terms of belief and desire. As in the case of entertaining a proposition, the property account applies--but only indirectly, in virtue of applying to belief and desire. Let us see if this response can be made at all plausible.

There is a broad distinction between two kinds of pro attitude. I will use the terms ‘wish’ and ‘want’ to indicate this distinction, although I doubt that in ordinary language they capture quite the distinction I have in mind. I can want a proposition P, in my sense, only if I think P may come about. I want to go to the game tomorrow, I want my parents to visit me, I want the sunset to be beautiful tonight. But I cannot want something in this sense if I believe that it cannot come true. I can’t want to run at the speed of sound, or to have one less sibling, or never to have existed. Nevertheless I can wish I were going to run at the speed of sound, or that I had one less sibling, or that I had never existed.

It is clear that wishing is a more sophisticated attitude than wanting. You can tell more or less what I want by observing what I try to get. Wants are action-guiding in a way wishes are not. Wishing is more cognitive than wanting. I can have desires I do not think about, even desires that I am wholly unaware of, but I cannot have wishes that I do not think about. Wants are typically dispositional, but wishes are always occurrent.

I suggest, tentatively and with considerable hesitation, that wishing is more like belief than like desire. This no doubt sounds paradoxical, but perhaps I can ease the apparent paradox at least a little.

Desires stand in a very close relation to corresponding beliefs. If I want to go to the store, then typically I believe that I want to go to the store, that going to the store will satisfy my interests, that going to the store would be good for me, etc. If I would rather eat cake than peas, then typically I believe that I prefer eating cake to eating peas, that eating cake will better satisfy my interests than eating peas, etc.

In cases like these, the belief that I want to go to the store or that I prefer cake to peas seems clearly separable from the desire or preference itself. Probably I can have such desires and preferences as these without having any beliefs about them, and certainly I can have mistaken beliefs about my preferences. In the latter case my behavior can at least potentially reveal my error, as when I believe that I want to get a job but find myself doing everything I can to avoid it, or when I believe that I prefer chocolate to vanilla but invariably choose vanilla.

In the case of wishes, however, there is no dispositional or volitional element at all. I may wish I hadn’t made that clumsy remark yesterday, but this will not lead me to try to bring it about that I did not make the remark yesterday.

The tie between wishing and belief seems much closer than the tie between desire and belief. In the first place, it seems impossible that one should wish one had never existed without believing that one wished this, or at least believing something related, such as that one’s actual interests would be better satisfied if one had never existed. What could such a beliefless wish be like? How could it be manifested?

But second, it seems not only that wishing has to involve belief, but also that it involves nothing else. When Mel believes that he would prefer never to have existed, it seems that there is no independently existing preference which this belief is about. Some evidence for this is the fact that he could never be shown to be mistaken is such a belief. There is no way for his behavior to reveal that, despite his belief, he does not really wish he had never existed. (Perhaps, belying his name, he shows a rare zest for life. Is that disconfirming evidence? Not if, given that he was born, he very much prefers to go on living, but thinks it would have been better if he had never been born. The last of seven children, he may wish that for the last hundred years all parents including his own had limited themselves to two children each.) It may perhaps be possible to show that his beliefs would be less coherent in some way if they did not include the belief that he wishes he had never existed. But this does not show that there is an independent mental state for his belief to be about it in the way that there is for his belief he prefers chocolate.

I suggest, then, that if one believes that one has a certain wish, and this belief coheres reasonably well with one’s other beliefs, then one does have the wish. Having the relevant belief is simply all there is to it. Thinking one has the wish and having it are one.

If this is correct, then Mel’s wish never to have existed poses no problem for the property theory. Mel’s wish consists simply of the belief that he prefers a world in which he never existed to the actual. Thus Mel self-ascribes something like <x prefers a world in which x never existed to the actual world>. And this self-ascription is related to others, for example his self-ascription of <x is such that if x had never existed x’s actual interests would be better satisfied>.


I have indicated the line I think the property theorist should take with respect to Markie’s objection and also with respect to the wish never to have existed. But I have been careful not to attribute this response to Chisholm or Lewis. Am I, then, the property theorist in question?

Not quite. Chisholm and Lewis take the line that belief and desire are at bottom attitudes toward properties rather than propositions. Lewis goes so far as to say that de re beliefs are not, except in the special case of belief de se, beliefs at all. He says this on the grounds that belief ascriptions should characterize states of the head, while de re "belief" ascriptions do more.

On my view beliefs are not "in the head" and are literally propositional attitudes. We have the beliefs we do partly in virtue of states of the head and partly in virtue of the character of our environment. But I agree with Lewis and Chisholm that the relevant states of the head are best characterized by properties rather than propositions. I am just not willing to call self-ascription of properties "belief." It is rather, to use Stich’s helpful term, a "subdoxastic state."

Similarly, I take it that desire is genuinely a propositional attitude. Like our beliefs, our desires are due to more than the state of our heads. The volitional state of one’s head is best characterized as a relation one bears to certain properties, a relation we might call "proto-desire"; one’s desires proper are then propositions which one desires partly in virtue of one’s proto-desires and partly in virtue of one’s beliefs. The rules which determine what we desire, given our proto-desires, may turn out to be rather complex.

Thus, given an apparently propositional attitude, I do not need to say what property it is really an attitude toward. It really is an attitude toward the proposition it appears to be an attitude toward. But I still face the task of saying in virtue of what self-ascriptions and proto-desires one has the propositional attitudes one does; the wish never to have existed provides an obstacle to the successful completion of this task; and my account of what "the property theorist" might say indicates how I would hope to remove the obstacle.