Discussion of anything presupposes its being real or possibly real enough to discuss. — Mww

Except when that's clearly false? You and I, discussing whether the Bermuda Triangle is a thing, with a mysterious ship- and plane-eating property, cannot be assuming that it is real: that is the question we are addressing.

(1) Is it possible that the Bermuda Triangle is a real thing? Is the idea consistent with the laws of nature as we understand them, for instance? Is there some suitably naturalist explanation for the disappearance of ships and planes thereabouts?
(1a) Our understanding of nature may be correct to the extent of ruling out Bermuda Triangles.
(1b) Our understanding of nature may be incorrect at least in ruling out Bermuda Triangles.

(2) If the existence of a Bermuda Triangle is consistent with our understanding of nature, or if our understanding of nature incorrectly rules it out, then the question remains whether we live in a world that has a Bermuda Triangle. It may be possible and thus real somewhere, just not here.

But now consider the actual Bermuda Triangle, marked off as a region of the Atlantic ocean through which ships and planes pass, and within which ships and planes are lost at roughly the same rate as any similarly heavily trafficked coastal region anywhere in the world.

With that in mind, to say that the Bermuda Triangle is not real, is to say that there is not something to be explained, but nothing, there being no statistical anomaly in need of explanation. The set of things to be explained exists but is empty.

the conception of universals is prior to the apprehension of particulars. — Metaphysician Undercover

But there are very good reasons people think it goes the other way.

For most people, for most concepts, acquaintance with instances of the concept precede, in time, the possession of the concept, and exposure to those particulars is instrumental in acquiring the universal they fall under. That's the argument from ontogeny: you are acquainted with moving, barking, licking particulars before you know that they are dogs. And there is a related argument from phylogeny: modern humans have a great many concepts that they were taught, often through the use of exemplars, but it stands to reason that not every human being was taught: there must have been at least one person who passed from not having to having a concept unaided. In essence, we imagine that person somehow teaching themselves a concept through the use of exemplars, and we imagine that process proceeding as we do when analyzing a population of objects, looking for commonalities.

In thinking about this thread, I was reminded of the Sesame Street approach to teaching about classes, an approach presumably backed by research, probably the most famous educational bit in Sesame Street:



(The irony of this song, "One of these things is not like the others. One of these things doesn't belong," in a show teaching inclusion and tolerance, was not lost on the makers of the show, and the bit was largely retired in favor of "three of these things go together," which is not much of an improvement.)

What's of interest here is that resemblance is not only relative, but comparative: resemblance is a three-way relation, a given object resembles another more, or less, than it resembles a third.

uncertainty avoidance — Isaac

Most people avoid tense situations. Repo man spends his life getting into tense situations.

I see. Static bad; dynamic good.

Serves me right for needlessly poking Isaac.

It is a point of interest how much you can do with a preference for order and predictability — any order, however arbitrary.

Sellars has that just-so story in "Philosophy and the Scientific Image of Man" in which he derives the idea of natural law from the observation that some of the persons in nature (old man river, old man mountain, that sort of thing) are set in their ways, the way people get, and thus predictable, the way some people are. (Big Lake is freezing over again, like he always does this time of year.) He suggests we recognized the efficacy of habit first and derived the idea of mechanical determination from that. (A sort of corollary to the 'theory' that we derive the idea of force from our own efficacious action.)

DishBrain is able to identify habits or tendencies in the "ball" and to develop matching habits or tendencies or propensities. For what purpose? In an earlier age, we might have heard this described as a manifestation of the death drive, the will to become mechanical, but maybe Freud was on the right track in seeing life as paradoxically trying always to reduce irritation and excitation, or to predict it well enough that it ceases to be experienced as surprise. (See, @Isaac, I do listen. Did you know you're a closet Freudian?)

Insofar as the practice of philosophy is largely a sort of conversation, there are obvious analogies to dance.

The issue of imaginary numbers is different though. It is an issue of there being two distinct conventions, yet each convention is correct in its own field of application. In the one case there is no square root of a negative number, in the other case there is. — Metaphysician Undercover

I don't think so. It remains true that negatives do not have *real* square roots, and that's the same as saying that if your domain is discourse is restricted to real numbers they have *no* square roots. The complex plane is a perfectly natural extension of the real line.

It is how the negative are conceived to relate to the positive, that creates the problem, i.e. it is not a straight forward inversion due to the role that zero plays. — Metaphysician Undercover

Not following this at all.

This is a terrible idea.

Chess is illuminating because it presents questions that may be decidable in principle but are not, for humans, in practice. The alternating reliance on calculation and heuristics, with the goal of grounding a decision under uncertainty, is very reminiscent of philosophy, which rarely gives opportunities for decisive arguments and must content itself with persuasion. And you still calculate whenever you can.

I usually don't see what we do here on the forum as competition. — T Clark

But don't forget that chess is also cooperative. Takes two to play a game.

Given a single ball being dropped into a Dalton Box, can you tell me where the ball will finish? — Banno

I will read. Shouldn't have neglected her.

In the meantime, and acknowledging that I may be putting my foot in it, this is a slightly bizarre way to talk about Galton boxes, the point of which is that even if chance is real, and not just a consequence of our non-omniscience, nature rather makes a point of capturing chance and turning it to the creation of order. The path of any single ball on a Galton box is at least effectively, for us, random, if not genuinely random, but the result of thousands of balls flooding onto the board is a perfectly predictable gaussian distribution. The actual shape of the distribution will vary from run to run, and the amount of variance is also predictable. Nature seems to believe in statistics.

<offtopic>

I'll consider adding "useful for us to believe" to the OP I want to write about "context dependent" and "purpose relative."

Maybe I'll save it for the one about "from our human perspective."

</offtopic>

For a start, the brain cells did not "learn to play pong", they just avoided "a chaotic stream of white noise". — Banno

Same thing. (insert appropriate smiley)

Still, not bad for @apokrisis, @Isaac, and anyone else in the free energy camp.

Okay, I think I get how you're thinking now. Your idea is that we start with the phenomenon of resemblance, explain that in terms of predication, then explain predication in terms of universals. You want to cut off the last step, but you keep saying it means taking resemblance as more fundamental: it means no such thing; you're still explaining resemblance using predication, you just want to take predication as primitive.

You can do that, but you need to argue for it, and you ought to quit talking about resemblance, which is apparently of no interest to you.

So you intend to keep "is red", for instance, so that we get to say A resembles B because they are both red. That just makes the resemblance relation entirely derivative of predication: predicate F of two objects, and poof they resemble each other. If I say that predication is constituted by the instantiation of universals, you have exactly the same resemblance relation, and its existence is no challenge at all to the universals account of predication. Resemblance is merely a consumer of predication, not a producer.

Then forming your collection by using that predicate presumes you have access to a whole machinery of predicates, membership, and classes. I thought we weren't going to do that, but ground our use of predicates in the resemblance of things to each other.

Instead you're just analyzing one sort of predicate (color) in terms of others (microstructure). Nothing at all to do with resemblance.

I find all this bollocks about a 'nation's right to exist' really sickening. — Isaac

I do not believe Ukrainians are fighting for an abstraction like this, do you? Even to say "self determination" instead is just shorthand for saying they want their families, homes, businesses, friends, libraries, parks, opera houses, and, you know, etc., not to exist only at the mercy of a large group of armed people who don't even live there. It couldn't be less abstract for them.

It is unavoidably abstract for us, but we can still understand, at least intellectually, why they are fighting, and call that "what they're fighting for." I'll also say that I'm betting a lot of Ukrainians are grateful there was already a state apparatus in place, and an armed forces, else they would absolutely be at the mercy of any armed group, whether a foreign government's army or criminals and outlaws. Part of the point of the state, and worth preserving even though it can be abused, as Russia is doing.

both predicates — litewave

That was my point. You're supposed to be grounding the use of predicates, aren't you? Or was your intention all along to ground some kinds of predicates in other kinds?

I just identify a universal with a resemblance relation and thus simplify the metaphysical picture — litewave

I know what you say you're doing. It's just not what you're doing. Unless I've missed something.

According to set theory, any mathematical universal can be instantiated as a collection. — litewave

It doesn't matter much in practice, but of course we *don't* have the axiom of comprehension because of frickin' Russell and his damned paradox.

Did you mean something else?

That's extraordinary. — Manuel

Right? I believe he said he just doesn't picture the drawing before doing it, but that working on a drawing is otherwise straightforward. Still...

It seems that I could in principle define a part of the ball that constitutes the ball's particular red color. That part would be a subcollection in the ball, a subcollection whose structure interacts with light in such a way that it reflects certain wavelengths of light, — litewave

Is there a difference in principle between a simple looking predicate like "is red" and a complicated looking predicate like "whose structure interacts with light in such a way that it reflects certain wavelengths of light"? I don't see how you allow yourself the latter if you can't allow yourself the former.

Anyway, there's no trace here of your proposed resemblance relation. You're just reducing gross properties to microstructure. You don't need resemblance to do that.

Just curious, that’s all. — Mww

See, that's what threw me off.

I would not be surprised if there were cases of people who lacked the capacity to visualize such elementary figures — Manuel

That part is just fact, I understand, as there are people who have no visual imagination. I heard an interview with one such person who works as a professional animator.

Have you come up with anything in that respect....from scratch? — Mww

You mean something no one else has? I don't know. The odds are against it, of course. But it's more fun, for me anyway. And it means that when I reach for an argument it's because it persuades me, not because I recall that Hume said it, even if I learned it from Hume and have forgotten that I did. (When you do mathematics, it's not generally important where you got the argument, but that it works.) It also means I have the experience of reading works of philosophy and finding on the page thoughts I have already had; having already worked through something, knowing some of its ins and outs, provides a framework for seeing what another thinker does with it.

Should I actually be defending thinking for myself here? Or were you making some point about the conceptual scheme I ought to admit I'm stuck with?

I can also define a collection by enumerating its members — litewave

That's a good point, but is it any use? If there's no criterion for membership, then the class you create is arbitrary, isn't it?

Over in the thread about causation, I found myself talking about approximations, and noting that we begin with noisy data and idealize it as a mathematical formula that we call an approximation; and if we can do that, we can go the other way and see the data as approximating the function.

How do we learn to do such things? If you draw a line on a chalkboard and say, "Imagine this is a straight line," what do you do if your audience asks, "What do you mean?" Once we know the trick, once we have the knack of idealizing, we can do this sort of thing all day long. But what if that trick is still opaque to you? I wonder if there are brain lesion studies on people who are unable to abstract in this way, or if there is a cognitive disorder that impedes this ability. It seems like you would get the sort of conversation you get with someone hyper-literal.

I suppose I should add, I wasn't just presenting empiricism in disguise; I was really just trying to see how I could come up with properties "from scratch". Looking back, what I did strikes me as an empiricist move, but it wouldn't surprise me to find it elsewhere. If it's one of the obvious ways to go, people have gone that way.

I like actually doing philosophy more than I like exegesis.

The Old Guys. — Mww

C'est moi.

I used to have a pet theory, also somewhere between logical and psychological, that generality is not a matter of classification but a type of procedure. Example: you want to talk about triangles in general, what you imagine or what you draw on the whiteboard are going to be approximate actual triangles, the edges and angles having particular values; you can treat that as a concrete triangle such that you might just measure to determine these values, or you can treat the object as general, or abstract, meaning that in your reasoning you are careful *not* to rely on these particular values. You carefully ignore them. And so in doing geometry we get to just stipulate (and indicate with those little hash marks) that these edges or those angles are equal, without giving a thought to the actual values the representation, being a concrete object, has.

Supposing that causation stands in need of explanation misunderstands that causation is fundamental to explanation. — Banno

Okay, right, that's why I bristled at saying something like "taking a causal stance" toward the world is something we do because it is useful. The word "useful" is being asked to do something here that it can't.

.....take a ball and you imaginatively delete its location.....
— Srap Tasmaner

Just curious. Where did you get the idea for doing this? — Mww

It's sort of the way empiricists like Hume talk. (@Manuel reads the early moderns a lot, so he could point out what a travesty of empiricism this is.)

It is also literally how I treat generating equivalence classes in everyday cases, choosing what to ignore, but there I've got a predicate machinery I don't intend to question.

It was intended as a simple, bone-headed account of abstraction, just to have a starting point. I assumed the main issue would be that you have to individuate properties in order to bracket them, which means you have to have universals to create a universal. (And now it sounds like Sellars's argument in EPM.) Never even got that far.

Srap demonstrates how this is not an acceptable starting place. — Metaphysician Undercover

If you say so. (Thanks for the notes on the ancients, btw.)

I don't think I demonstrated anything. I suspect the argument I gave is junk, but it had, for me, the desired effect of showing that the problem is not so simple as we might pre-theoretically think. Ask a non-philosopher friend and they'll probably sound like empiricists: concepts come from us "noticing patterns", "seeing what's in common", all this sort of thing. Maybe a "thank you, Darwin."

Abstract, "pure mathematics" shows that we dream up universal principles (axioms) first, from the imagination, or they come to us intuitively, then we try to force the particulars of specific circumstances to be consistent with the universals. — Metaphysician Undercover

That's at least in the neighborhood of Sellars's argument and the impasse I expected to reach, that empiricism from a blank slate can't actually get started.

Not perfectly clear to me what the status of that argument is though. I was promising to look at logic not psychology, and while this isn't empirical psychology, there's something unavoidably psychological here. I have thought it might be a matter of being unable to state the empiricist position coherently, so again a matter of logic, but now it looks like the logic at stake is somewhat transcendental. Well, no big surprise since Sellars was profoundly Kantian, but I didn't intend to drag that in. I'd rather there was a clear way to avoid this whole line of argumentation...

Actually, I would say that the partial particular, for example the particular redness of this ball, is a concrete part of the concrete whole (this ball). A concrete object is a collection of other concrete objects and there are various overlapping collections inside this collection. In the case of this ball, one of those overlapping collections is a particular red color because the structure of that collection is such that it reflects certain wavelengths of incoming light. — litewave

So I would say that the particular properties of a concrete object are overlapping parts (collections) of that object; their existences are mutually dependent on each other and the existence of the object as a whole is dependent on its parts. — litewave

I see what you're trying to say, but you can't say "part" because parts are concrete rather than abstract exactly in the sense that they can exist independently. (That much I learned from @Andrew M's explanation of hylomorphism.) And you really shouldn't be saying "collection" because that's a soft word for "class" and you precisely can't have classes without universals or predicates to define them. Clearly you're hoping to get structure — which is crucial, particulars aren't bags of properties — out of how the various collections are arranged.

Stepping back, this begins to sound like breaking down an object into its fundamental particles and then reassembling it, down the chain through chemistry to quarks and then back up again. We assume such a thing is possible in principle, I guess, but the argument for special sciences has always been that on the way back up, you have no way to know where you are and what you're building, so the particulars of interest are gone forever, leaving just an undifferentiated sea of particles.

Nicaragua — jorndoe

Oh Nicaragua! I remember when you were cool...

The other examples of Vietnam and Afghanistan, as far as I know, did not resemble this one in that sanctions of this scale, followed by constant coverage of a humiliating retreat right after annexation, were put into play. — Manuel

So you're saying Putin has more at stake than other major powers have had when failing to conquer little countries they thought they could steamroll, so while maybe the US half-assed it in Vietnam, Putin will really go all in and get the job done? (I don't know how many Americans still feel this way, but something like that used to be a common opinion of our conduct of the war in Vietnam, usually blaming protesting hippies for Washington pulling its punches. I believe Kissinger wanted to nuke Hanoi, so maybe there's something to that.)

I get your position, I think. When I heard that Russia had invaded, I just assumed Ukraine didn't stand a chance. When that starts to look wrong, there's a fallback fear that Putin, for all sorts of reasons, will not accept defeat, will do whatever it takes to achieve at the very least a strong enough military position that he can demand whatever he wants at negotiations, should he decide not to bother trying to subdue the entire country. (I mean, he can't occupy Ukraine, that was never an option. But puppet government and some invited guests in Russian uniforms, you figure that's what he was shooting for.) His fate, we fear, is irrevocably tied to success in Ukraine. His ships lie burned on the shore; it's win or die, and that makes him a more dangerous foe than the US in Vietnam or the USSR in Afghanistan. That may be. It is a serious concern. Didn't Sun Tzu counsel always leaving your enemy an escape route precisely so they wouldn't fight to the death?

In this scientific age, it seems like the obvious way to take "useful" here is to say it's an approximation, and it's cheap. It's interesting that we can get by with what may be a pretty drastic simplification of what's going on with us and our surroundings; what we idealize as cause-and-effect would be a pattern that events show a tendency to instantiate — reversing things, so that the data seem to approximate the mathematical formula that idealizes and approximates the data. Whatever's going on there, whatever relates those two, is real.

But this is just shifting ground again. We know what it means to say an approximation is useful: when we rely on an approximation, we know our actions will produce a result that's near what's predicted and predictably varying from it. But that very description relies on the idea of an action producing a result, and I find it hard to imagine there will be any way of putting this that doesn't smuggle in causality somewhere. Point being that explaining our reliance on the idea of causation as something we do because it's useful may be no explanation at all. (We rely on causation because we rely on causation.) But it won't fly as description either because it deliberately obscures the place of causality in our thinking.

Right. And I assume we're talking about this logically, not psychologically. An account of, I guess, properties, rather than how we come to learn them, or think them up, and so on.

A theory of universals presumably has something to offer, is supposed to explain how something works or what something means. I haven't thought about universals in a long time (having gotten accustomed to predicates and such) but it would appear to be in the neighborhood of where you started, two numerically distinct objects both being red, for example.

We have to first see how this is a problem, right? The objects are distinct. Anything you pick out to describe a concrete object is a bit of that object, is that object minus almost everything about it except some particular aspect you've chosen. (Passing over how we do that, for the moment anyway.) At least, that's how I presume abstraction works.

So you take a ball and you imaginatively delete its location, its mass, the texture of its surface, everything but the light it radiates and you call that its color. (This is no good, of course, since it needs to be a propensity or a disposition, but we don't know whether we need to bother yet.) We do the same thing with some other object, maybe a car. If this is how we 'create', as it were, colors of things, by taking a particular and leaving out everything else, we still end up with just a partial particular.

What you get is still two numerically distinct abstract objects rather than concrete ones, yes? No matter how similar the abstract objects are, they are distinct. What could possibly entitle us to say that they are in any sense the same thing?

Now we might think — identity of indiscernibles to the rescue! And now that we come to it, how did we imagine the sort of partial particular I described being a numerically distinct entity? It's not, after all; it's only an aspect of a 'genuine' concrete entity. Not even a part of it, but something that, obviously it seems, cannot exist on its own, but only as an aspect of something concrete.

No problem; we knew that as soon as we said we were creating an abstract object (the red of this ball) from a concrete object (this ball). But if it's no real objection that these things can't exist on their own, then we can't rely on their individual existence to underwrite their being numerically distinct. Maybe abstract objects can be numerically distinct, but if they can it's not the way regular concrete objects are.

Which leaves us where? We want to head toward saying these two abstract objects are the same or similar, but now it's not even clear in what sense they are objects at all, or whether they can be distinguished in order to be compared or identified. If these are objects, it's not clear what use they can be to us. We're in a muddle.

Abstraction looked so straightforward, but it seems to leave us nowhere.

So what do you think? We want an account of properties of objects, and we expect to be able to say that two objects have the same property, or that they have properties that are similar but not identical. But how do we fix our account of properties? Was my account of abstraction all wrong? Or are we in a better position than it seems?

So the question is (1) whether resemblance, or similarity in some respect, or something like that, gets us everything we want from universals, and (2) whether such an account is coherent, non-circular, doesn't need to smuggle in universals somewhere to work.

Yes?

But I think you're saying less than you think you are.

Usefulness admits of comparison: a thimble is useful for emptying a swimming pool, but not as useful as a giant shop-vac on wheels. For that matter, almost anything can be used to empty a swimming pool. A pencil will hold onto a few drops each time you dip it in.

So is this grammar optimal in some way? Do we use it because it is more useful than alternatives? Are some of those alternatives truly awful, like emptying a swimming pool with a piece of paper?

If the claim is only that the grammar is usable, that's a pretty low bar to clear. Thimbles, pencils and scraps of paper clear that bar for emptying swimming pools.

Or is the idea of usefulness just there to point at the purpose for which we use this bit of cognitive-linguistic technology? To point out that the grammar is being used?

Afghanistan was nowhere near the level of importance to the USSR as Ukraine is now. — Manuel

Probably true, at least in the sense that Putin would like to straight up annex the entire country, which is to say, restore it to its rightful status as part of Russia. He doesn't consider its current status as a sovereign nation legitimate, does he? (I may be misinformed.)

But then the US was in Afghanistan a long damn time, and in Vietnam a long damn time. And there was no military victory for the US at the end of either engagement. The US had reasons, and Russia presumably has reasons something like those for what they're doing. (Geopolitics. Perception. Hubris. Etc.) The question then is whether it's those reasons or the historical-territorial stuff that counts more. If it's not the homeland stuff, then we have clear examples of a small nation fighting for its life overcoming those sorts of reasons. (In fact, I think Putin have the greenlight when he saw Biden accept the humiliation of leaving Afghanistan in disarray.)

with no further justification than its usefulness — Banno

Is this usefulness just brute fact, or can we hope to explain why this grammar is useful?

If Anscombe addresses that, you can just point at her again.

Hmmm. Is there going to be any way to flesh out the idea of using something to do something that doesn't rely on causality?

I'm not quite following.

Is the idea to drop the idea of instantiation?

But what are you going to do with universals if not instantiate them?

If that model has issues you want to avoid, then you just go for predicates (which you're already keeping) and their extensions.

The idea of *starting* from resemblance and building everything from that might be worth pursuing. (@bongo fury has some ideas about how to police the borders.)

But then you won't start out talking about universals.

in the end we're doing things with words. The relevant context reveals what we're doing, not just the words or sentences themselves. — Andrew M

I just can't get around the idea that in most, but not all, cases we use the words we do because they're the right ones. I don't think a linguistics that is all pragmatics with no syntax or semantics is a real option.

Which, in turn, would seem to relate identity and convention to purpose. — Andrew M

I wouldn't deny that there are choices we make, sometimes implicitly, which enable us to enact our purpose; I just don't think that makes our purpose constitutive of the objects we interact with. I think they have to be there, as they are, for us to have the options we do, among which we select the one that aligns with our purpose. If you can sometimes sort papers by author and sometimes by keyword, depending on your purpose at the moment, it's because they have authors and keywords. If they didn't, these wouldn't be options for you.

Each and every property of a particular must be understood as essential to that particular, that's what makes a particular a unique individual, distinct from every other particular. This is what the law of identity recognizes. — Metaphysician Undercover

So now we're back to @Isaac's teapot and the missing screw. In that discussion, the question was only about successfully referring to a particular that (might or) might not possess a property you believe (or don't believe) it does. I think it's plain that you can; for some cases, I'm leaning on the causal theory of names, and for others on how demonstratives work: you can clearly demand someone get "that" off your kitchen table even when you know very little about what "that" is. Exactly how that works may be unclear; that it works, I believe, is not. (We may come back to the double-bind theory of reference eventually.)

Here, we might start with the question of whether "being on my kitchen table" is a property of the object in question. It can be expressed as a predicate, as I've just done, but we could just as well express the situation as my kitchen table having the property of "having that on it," assuming again that "that" will manage to refer to the object. Or we could define a two-place predicate "on" such that "on" is true of an ordered pair <that, my kitchen table>. For either of the one-place predicates (of that, or of the table), I would be asking you to make something that is true of one of them false; for the two-place predicate, I would be asking you to make something that is true of the two of them false.

There are a couple ways to take that: I described "on" as a relation not just of two objects but of two objects in a particular order, so that on(table, thing) was already false, but on(thing, table) was true and I want it to be false. on(table, thing) and on(thing, table) describe different states of the world; in this case, the demand to make whichever is true false needn't concern itself with the order, because context will take care of that. But if I asked you to put that thing on the table, my demand would not be satisfied by you putting the table on that thing. So if we want on/2 to carry the same meaning across different uses, we can't rely on context in that way, and have to build in the required order. How do we do that?

Do we say that "on" takes three objects, the two from before and a third that specifies the order? If so, the third would look something like this: "1 = thing, 2 = table". Such a list can be presented in any order, so we don't have a regress, only a rule about each natural number up to the arity of the predicate being used, so this is a genuine option. But our new on/3 takes two concrete objects and a third which, whatever it is, is not like that. I say "whatever it is," because the semantics of the ordering list are unclear at this point: are those objects in the list, or expressions referring to objects? I guess either would do, but we're still building in a lot of other stuff, some of which looks suspiciously abstract, so we could just give in and have "on" take a single abstract object which is the ordered pair <thing, table>.

If we do that, my asking you to get that thing off my kitchen table would be asking you to make "on" false of the ordered pair <thing, table>. No properties of the concrete objects will change when you do so. We've added a step, so that "on" is not true of the concrete objects themselves, but of an ordered pair of the objects, which is a bit of a surprise. I'm not sure how much that should count against the scheme.

Can we do something similar with other cases? For instance, if my bike tire is flat, is it a different object once it's inflated, or is it just a different arrangement of tire and air, the tire itself never changing? (In this case, we may or may not have any specific batch of air in mind.) But then what would we say about the shape of the tire, that surely changes when it's inflated? If anything is a property of an object, surely its shape is. But I make different shapes when I sit and when I stand — does that make me a different person? What all of these examples have in common is that there are at least two different times considered: the tire is never flat and inflated at the same time, I am never sitting and standing at the same time, and so on. So a first attempt at distinguishing what is essential to an object from what is accidental is, naturally, distinguishing what is constant or invariant about it, what does not change from one time to another, and what does or can change from one time to another. Essential is what is time-less, and accidental is what is time-dependent. The same dog barks at one time and not at another.

But Isaac's screw-missing teapot raises a batch of familiar problems: evidently material constitution is not a great candidate for the timeless identity of an object. If we replace the missing screw with another of the same size, we have the Teapot of Theseus: is it the same teapot after as before the installation of the new screw? (It's considerations like this, if memory serves, that drove Peter van Inwagen to conclude that inanimate objects lack identity altogether, and thus do not, strictly speaking, exist.) One solution offered, in a sort of conventionalist spirit, is that this is all a collective fiction: there are no things with identities that we come along afterward and refer to; rather, our various acts of reference, intended and accepted by us as such, and our deeming these acts successful, is all there really is here. Thus, the slight oddity of Russell's account of definite descriptions — that they involves implicit existence claims — is vindicated, because indeed we are asking others to accept , at least for the duration of this exchange, what amounts to a stipulation that there is a dog when we say "the dog is barking."

The conventionalist account doesn't automatically undermine a distinction between essential and accidental properties, of course; you could take it as simply falsifying all claims of essence, or you could conversely take essence as whatever we tacitly agree it is. We generally count me as being the same person sitting or standing, and since that's all there is, that's enough.

But there's an odd wrinkle to all this. If I, like Isaac's teapot, do have an identity, then a proper semantics of me would require everyone to speak of me as if I do, and we would expect the corpus of attempted references to me to roughly, and only roughly, follow this requirement. That means the conventionalist will argue that our broad agreement in how to talk is just that, and nothing more; while the identitarian will argue that our broad agreement is a consequence of there being objects with identities. The conventionalist would seem to have parsimony on their side, and can allow or disallow the hypothesis of concrete self-identical objects as their mood dictates; but the base position is that it is more perspicuous to venture only that we say what we say. The object-identitarian offers a theory that explains why we talk the way we do, and the conventionalist can just say he doesn't need one.

That means there are two overlapping arguments here: on the one hand, the conventionalist can keep poking holes in whatever theory of object identity the other side comes up, because he needs no such theory anyway, and may even think no such theory is possible; on the other hand, the object-identitarian has to come up with a theory that works and show that it is needed, which means he also has to find some flaw in the conventionalist account of our referential speech acts — not for the sake of his theory but to show that some theory is even needed. What's not clear in any of this is how the evidence is to be handled: I'll venture that most people's pre-theoretical intuition is that we talk the way we do because things are the way they are, and that our talking the way we do is in fact evidence that things are the way we say they are.

But we have those pesky scientific refutations of how we talk: sunrise, solidity, and so on. That doesn't show that how we talk is never evidence of how things are, but it does show that it isn't always such evidence. On the other hand, the conventionalist can shift from the claim that how we talk is only evidence of how we talk, and nothing more only for methodological reasons, to a claim that how we talk is only we how talk — now meaning our agreement is precisely evidence that there is nothing more.

If that were true, it would not only deny the object-identitarian what was counted pre-theoretically as evidence but change the character of what's to be explained by any such theory. If the mean girls call you a loser, that's just a thing they say: the truth-value of their statement matters to you, but not to them; what matters to them is producing some effect, of hurting your feelings. That's the sense in which it is "just something they say." But not only can you not conclude from someone saying something that it must not have a truth-value, in this case the effect is only produced if you assume that it does, and they assume that you will assume that it does. If they know you will discount what they say as being just mean-girl noise, or just noise period, there's no reason for them to say it. The conventionalist can retreat again and say that the hurt feelings are known inductively to follow utterances of "loser," and that's all the mean girls need. That might actually be true! But you have to show that such an account really will extend to cover all language use. This situation is so simple that I think what we're really seeing is not exactly language at all but something more like dominance signaling that happens to use language because, well, there it is; we tend to use words even when what we're doing is really nothing more than growling articulately.

Cannonball Adderley, Somethin' Else

what "trading intensions for extensions" means — Janus

Just that classical logic can't deal with propositions of the form "It is possible that you pick a red marble," but can happily deal with propositions like "There is a red marble in the set."