From Aaron Mon Oct 16 21:38:24 BST 1995 Newsgroups: comp.ai.philosophy Distribution: world References: <1995Oct12.034521.11034@media.mit.edu> Subject: "Exists" isn't a predicate. (Was please help me with Descarte) bill@nsma.arizona.edu (Bill Skaggs) writes: > Date: 14 Oct 1995 23:09:13 GMT > .... > > minsky@media.mit.edu (Marvin Minsky) writes: (only half seriously I assume) > > Of course, because Descartes surely doesn't really believe any of > > those proofs. So, the more proofs the better. > .... (and more seriously) > > (Personally, I don't believe that the predicate "exists" is > > itself sound. I've posted my argument on this group before, but > > I'll be glad to email it to you if you're interested.) (Bill commented) > As I understand it, this claim (i.e., "existence is not a predicate") > was defended at great length by Kant in his "Critique of Pure Reason". yes. Which reminds me, when a student I was once in a seminar taken by J.L.Austin in which he invited us to consider what it could mean for existence to be a predicate. I think he said (in his inimitable style) something like: "Perhaps existence is like breathing, only quieter?" Anyhow, I think Frege was one of Austin's heroes (he translated one of Frege's books on foundations of arithmetic), and Frege's work on predicate calculus showed one interpretation of "exists" in which it definitely is NOT a predicate: it's a second order function from predicates to truth-values. Or, if you like, a second-order predicate, which cannot possibly be satisfied by things like unicorns, or cyclic groups or electrons, or gods, but only by concepts of such things (i.e. predicates). Roughly, "X exists", on that interpretation, means "the concept of X has at least one instance". (However, this formulation is an open invitation to professional quibblers about circular definitions.) I've occasionally wondered whether one could re-introduce the notion of something like existence, not as a predicate applicable to individuals but as a relation between individuals and possible worlds. I.e. individual x exists in world w1 but not in world w2. (Minsky exists in this world, but not in a possible world in which his grandparents never met.) But there are technical problems in defining identity of individuals across possible worlds: how much has to be different before it's not the same individual but someone else...? There's no well-defined general answer: so the relation exists(x, w) can't in general have a simple true/false value. (Though we often assume it is true when we reason about counterfactuals - "If Fred had arrived earlier he would have caught the train", presupposes that it would still have been Fred, in that world.) Another application for something like fuzzy logic (fuzzy modal logic???) Or is there something more subtle going on: the truth or falsity is determined by a broader context which defines the POINT of the question (why we are discussing what would have happened if he had arrived earlier)? Given that point, perhaps the truth value is sometimes determinate and non-fuzzy. So, what sorts of points can have that power? Just a rambling thought before I get back to work. Aaron From Aaron Wed Oct 18 19:12:45 BST 1995 Newsgroups: comp.ai.philosophy References: <1995Oct12.034521.11034@media.mit.edu> <45ufrg$au9@percy.cs.bham.ac.uk> Subject: Re: "Exists" isn't a predicate. (Was please help me with Descarte) Peter, Thanks for your comments. peru@cogsci.ed.ac.uk (Peter Ruhrberg) writes: > Date: Tue, 17 Oct 1995 12:27:18 GMT > Organization: Centre for Cognitive Science, Edinburgh, UK in response to my ramblings about existence as a relation between an individual and a possible world: > In article <45ufrg$au9@percy.cs.bham.ac.uk> A.Sloman@cs.bham.ac.uk (Aaron Sloman) writes: > > > >I've occasionally wondered whether one could re-introduce the notion > >of something like existence, not as a predicate applicable to > >individuals but as a relation between individuals and possible > >worlds. I.e. individual x exists in world w1 but not in world w2. > >(Minsky exists in this world, but not in a possible world in which > >his grandparents never met.) > > > >But there are technical problems in defining identity of individuals > >across possible worlds: how much has to be different before it's not > >the same individual but someone else...? > > > >There's no well-defined general answer: so the relation > > exists(x, w) > >can't in general have a simple true/false value. > > > >(Though we often assume it is true when we reason about > >counterfactuals - "If Fred had arrived earlier he would have caught > >the train", presupposes that it would still have been Fred, in that > >world.) Peter remarks [PR] > I think you're understating the last point. Whenever we talk about > some other possible world we presuppose the existemce of certain > individuals in that world, and do not worry about how to indentify > them "over there". I agree that that's what we DO. Whether we are entitled to in all cases is another matter. Some things are pretty unproblematic If Fred had arrived earlier he would have caught the train. that's because the (counterfactual) antecedent does not make much difference to who Fred is. His arriving earlier at the station is exactly the sort of possibility he himself might have contemplated before setting out from home. There are other things which make a bigger difference: If Fred had studied physics instead of philosophy he would have been a Nobel prize winner. Again, we can consider Fred, in his youth, contemplating the two options, and he might, in a different possible world, have chosen physics, even though he actually chose philosophy. Some people might quibble as to whether the Nobel prize winning physicist would have been the same Fred as is now an impoverished professor of philosophy. Most people would probably regard the quibble as pointless nit-picking: the bodily continuity with the early Fred back to the decision point is enough for us to say it's the same person. If Fred had been born five weeks earlier, his birthday would have been on new year's day. Again, we can think of the foetus which was delivered one February morning as possibly having been provoked to emerge earlier, in a different world, e.g. using some drug to induce labour (the sort of thing that can be used in cases where it looks as if continuing to the normal nine month term will be dangerous for some reason). In that case you can say, it would have been the same individual. Of course, there might have been significant changes in the development of the individual had he not remained in the womb and instead been fed and generally cared for using high technology. For example, development of the brain might have been slightly different, leading to a different sort of personality, different kind of intelligence, or whatever. (Being deprived of breast feeding might have interfered with normal ways of coming to understand the structure of space -- which seems to me to be one of the major unnoticed features of breast feeding.) Would it have been the same Fred? At this point I hear some people saying: Yes definitely: there's sufficient bodily continuity to justify saying that it would have been the same individual, despite the earlier birth. And I can hear others saying Definitely not: it's a different person because those differences of circumstances of birth would certainly (via all the feedback loops in personal development) lead to so many differences that we could not count him the same person (though perhaps it would be the same organism). and I can hear others saying It's a non-question. Our notion of identify was never designed to cope with such cases and asking whether it would have been the same person is like asking whether it's the same time at the centre of the moon as it is in London, when the moon is over London. Now consider this case: If Fred had been born two years earlier, he would now be entitled to a pension. Well clearly if Fred is aged 63, and pension rights start at age 65, then anyone born two years earlier than Fred would be entitled to a pension. But could FRED have been born two years earlier and been entitled to a pension? How could it have been Fred, if he wasn't even conceived till a year and a quarter later. At that point one can start considering possible worlds in which the premature ovum the eventually led to Fred had been selected from Fred's mother's womb and some portion of Fred's father's anatomy had been operated on in such a way as to accelerate the development of the the sperm which actually produced Fred. If such tinkering were possible, so as to produce conception two years earlier than would normally have happend, would that justify our saying that the baby born two years earlier was the same person, in a different world? There are some people (like Peter?) who think these questions have very definite answers, and perhaps all we have to do is analyze them carefully to find the answers. E.g. Kenneth Colby wrote, in <460r14$4lf@oahu.cs.ucla.edu> [KMC] > Date: 17 Oct 1995 11:01:08 -0700 > Organization: UCLA Computer Science Dept. > Existence seems to be the only thing that does not come in degrees. This sounds like Peter's point. If we consider a whole range of possible situations in which someone was born earlier than Fred, ranging from a few minutes earlier to a hundred years earlier, then Ken is presumably saying that in every one of those possible situations Fred either definitely existed or definitely did not. (Of course, I am not disputing that in each world there's a set of objects that definitely exist: I am asking whether it's always definitely true or definitly false that one of those objects is Fred.) It makes sense to talk about all sorts of different possible worlds more or less closely related to the actual world, and only in some of them is there a clear, definite answer to whether Fred (who exists in our wold) exists in them. Another range of cases occurs when we consider a world in which an extra division of the fertilized cell that led to Fred produces a pair of identical twins. Would one of them be Fred? [PR] > ....You seem the take the "distant planet" view of > possible worlds, which Kripke criticised so long ago. I don't know (or have possibly forgotten) this work. I was not consciously thinking of anything to do with distant planets, just different possible histories of our universe, in which you can consider a whole lot of intermediate cases between those in which Fred clearly exists and those in which he clearly doesn't (though other individuals with some of his features do). [PR] > ..If we take > transworld identity and transworld (non) existence as basic facts, ^^^^^^^^^^^^^^ Then we are making totally unjustified assumptions. How can anything so complicated be "basic"? What do you mean by "basic"? (Presumably not "dogmatically asserted with no reason"). I can imagine that there might be some basic building blocks of the universe which definitely do or do not exist in all physically possible histories of the world. But it is not obvious to me that anything as complex as a person has that feature. If you believe, like some philosopher/theologians (Descartes???, Leibniz)), in souls as primitive, indivisible entities, then you would have that view of souls. But I have no reason to believe in any such thing. [PR ...] > rather than to be determined by looking at shared properties (which > would have to be non circular) I see no incoherence in the exist(x,w) > idea. Did I say it was incoherent? An idea is not incoherent just because it breaks down in some contexts. It may be very coherent in others. > Where are the "technical problems"? I hope I have answered this adequately above. Aaron