might be interested — fdrake

Will be reading back through this latest run of posts and maybe commenting. Threw in something else in the meantime.

I can say "there are 66 coins in the jar" and that claim can be true even if I haven't counted the coins in the jar and even if nobody knows how many coins are in the jar. — Michael

You seem to be missing the point that @Metaphysician Undercover is an anti-realist, and his account of truth is some version of verificationism. (For meaning, it is sufficient that the coins can be counted, but for truth, it must actually have been done.)

Thus, as a self-organising system, we must, by definition, have internal states, and boundary states (and there must exist external states). Without these three states we cannot say that there is a system at all, we cannot define it from 'not-system' without defining a boundary and (as far as data is concerned) that boundary must be Markov boundary if the internal network is any more complex than a single ring of nodes. — Isaac

I can't imagine disagreeing with any of this.

But it is also evident, to me at least, that our language and how we conceive mentality does not match up, in any simple way, with this description. Now what?

One option is to say, well, we've moved on. Our languages are the ossification of a folk psychology that we know better than now. Of course we'll have trouble expressing this new view of things in the terms of the old paradigm. You can even soften the pitch a little by claiming only that the new view is different, rather than less wrong, but the hope is still that it is a more fruitful paradigm for inquiry. There must be reason to switch, and problems with the old paradigm provide plenty of motivation there.

But I think it's not that simple. There is an almost irresistible temptation to identify mentality with the internal states of such a system -- a sort of "what else could it be?" But much of the last fifty years of philosophy in the English-speaking world has been devoted to showing that this identification is mistaken. This cluster of issues became important precisely because of the promise of early work in artificial intelligence, generative linguistics, and brain science -- everything that would become cognitive science -- and the realization of some philosophers that we might be able to say we had finally found the mind, and it is the brain. Endless debate ensued, some of which continues, and some of which seems very old-fashioned now. But in the meantime, other philosophers noticed that our mental concepts and all of our language, in fact, seem not to respect this apparently natural identification of the mental with the internal. I find those arguments pretty persuasive.

There's no problem for science here. If you tell a scientist that what he's investigating turns out not to be X, he can shrug and go on investigating whatever it is he is investigating. Whether it's X is not really his concern. He may have been considering writing a popular piece for Scientific American explaining how his work changes our understanding of X, and now he can just not do that and spend his time in the lab instead. (Occasionally a scientist will decide that showing up philosophers is part of the job.)

But there's still plenty for philosophy to worry about. Science, so far, may have failed to straight up solve issues of mentality for us, but it has perhaps sharpened (at least indirectly) what those issues are.

What I think we need to be careful about, is thinking the mismatch between a particular scientific model, on the one hand, and a philosophical one, on the other, indicates that one has not sufficiently slurped up the other yet, but it will. It's that "if all you have is a hammer" thing.

All of which is to say that knowledge, for instance, is not a relation that holds or fails to hold between the internal and inferred external data nodes of a self-organizing system. Apples and oranges.

Since the world is all that is the case, it is also a collective story. — Banno

'According to the Bible' or 'Fred says that' are not restricting modifiers; they do not pass through the truth-functional connectives. 'Fred says that not P' and 'Not: Fred says that P' are independent: both, either, or neither might be true. If worlds were like stories or story-tellers, there would indeed be room for worlds according to which contradictions are true. The sad truth about the prevarications of these worlds would not itself be contradictory. But worlds, as I understand them, are not like stories or story-tellers. They are like this world; and this world is no story, not even a true story. — David Lewis, On the Plurality of Worlds

@Isaac

Before you post "pragmatism" and count that as a job well done, plan on explaining exactly how pragmatism answers any of the questions I asked, or shows the questions to be ill-conceived. Jobs to be done, purposes, free-energy gradients, surprise minimization -- all part of the model, after all. You don't get to have your cake and eat it too, not even by saying that pragmatism entitles you to the impossible.

Because the hidden states the world is a collective model of may be modelled imperfectly. — Isaac

I

If we have modeled imperfectly some detail of the hidden states, but we never encounter evidence that would encourage us to update our model, were we wrong?

Another question: can our model be properly said to supervene upon the hidden states? That is, can there be a change in our model without a change in the ("underlying") hidden states?

If the answer is "no," if our model is not so tightly coupled to the hidden states as that, what is the source of that relative freedom? And if our model is then, to some undetermined degree, independent of the hidden states, what entitles us to describe changes to our model as updates rather than just changes, which could, for all we know, be arbitrary, or, if not arbitrary, free?

There is nothing, it seems, that we can point to as "evidence" that is outside the model, not even surprise; surprise is not a fact, but part of our model of ourselves.

II

There's an impressive set of studies showing just how constructed our visual perception of the world is, the ones with the flashing lights. Put some people in a dark room facing a screen or a wall and flash a sequence of lights in just the right way and people will report seeing a single light moving, say, left to right. Even better, if you arrange the lights as you would to go around a small obstacle, people will report actually seeing the obstacle -- or at least report that there was "something" there that the light had to go around. That latter result shows just how much "filling in" we do from our priors, as you might say, about how the world works.

But this study requires carefully controlled circumstances. To determine the speed at which to flash the lights and how far apart to space them, no doubt experiments were needed. I doubt they nailed it the very first time, and there's a range -- I don't know how big -- outside of which the illusion of a single moving light would not hold. Similarly, there must be no other sensible information about the space where the light "detours," else people would report that the light behaved as if something were there but there wasn't.

Outside the lab, none of those restrictions apply. The simulacrum we are said to inhabit is so detailed that we can test it however we like. We can prove to our satisfaction that a tree before us is not a plastic model by cutting into it and seeing the rings, the xylem and phloem, all that. We can study a bit of the wood under a microscope and see more, even under an electron microscope if we choose, we can "touch" individual molecules of water in the tree sample. And we can do this sort of thing anywhere to any degree we are capable.

If a map reproduces every last detail of the territory, and does so not with ink on paper, but using the same materials, for all we know, as the territory, then the map is in fact a perfect duplicate of the territory, not a map at all, and to find your way around the so-called "map" is exactly the same process as finding your way around the territory.

What becomes questionable is the claim that the "map" is not the territory but only a map, and the positing of a "genuine" territory out there, somewhere, that the "map" we wander around in is a copy of. That will surely strike most residents of the "map" as an article of faith. Anything can count as evidence for it, and nothing can count as evidence for it.

III

I don't think it will quite do to answer that "data underdetermines theory." What "data" there is, is not just theory-laden; it is crushed under the weight of the theory it's carrying on its back. It could, for all we know, be 100% theory.

You want to call your view a sort of realism because you maintain there is "something" outside our Markov blanket. Is that "something" similar to the non-existent "something" that the non-moving light did not actually detour around?

If this is realism, it is indistinguishable from idealism, if only in some suitably circumspect Kantian sense of idealism.

IV

We seem to have a sort of antinomy here. On the one hand, we claim to know only our conception of the world, loosely enough coupled to it that it can deviate from the world's supposed true state. But (1) nothing entitles us to make any claim that there is such a true state, or to make any claim about how close our conception is to it, and (2) our conception is so complete that it qualifies as itself a world of the sort we claim only to have a conception of.

We have a model that is, for all we know, 100% mistaken, and at the same time, for all we know, all there is and no model at all.

But there will be no one to judge him to be dead, therefore he will not be dead,

Oh, and there will be no one to judge him to be alive, therefore he will not be alive.

No matter, because there will neither be nor not be a universe that includes or does not include him either not alive or not dead.

Your turn. Philosophy is fun!

No.

Since we've been trafficking in this thread lately with more than accounts of truth, or even phenomena that are quite nearby, like reference and knowledge, but in what amount to complete theories (or sketches of theories) of mental life, I thought I'd share a comment of Herbert Simon's, in defense of the sort of computational model he helped develop in the 50s and 60s: our mental life is indeed complex, but it is not complex because we are complex -- the mechanisms of mental activity are relatively simple -- but our environments are quite complex.

It may be that the history of research in artificial intelligence refutes that suggestion -- I'm in no position to say -- but it is a pregnant thought as we imagine modeling our mental lives, or at least a reminder to give some thought to the source of its evident complexity.

Here's some chitchat about one of his cases, since I can't help it. There are lots of big problems with Hazlett's account, not least his use of Grice.

(1) Alice knew that stealing is a crime.
(2) Alice knew that stealing isn't a crime.
(3) Alice didn't know that stealing is a crime.
(4) Alice wasn't aware that stealing is a crime.

Hazlett points out that both (1) and (3) implicate (either imply or entail) that stealing is a crime, supporting, he thinks, the case that "stealing is a crime" is a (conversational, i.e., non-conventional?) implicature of (1). But what it supports, if anything, is that "stealing is a crime" is a presupposition of (1) and (3). Presupposition is not the same thing as conversational implicature. (On Strawson's account, roughly, where "The present king of France is bald" and "The present king of France is not bald" both presuppose that there is a present king of France.)

Every philosophy neophyte learns to distinguish (2) from (3), and that (2) is not the negation of (1), but at the same time learns that (3) is ambiguous between (2) and (4). The version of (3) that aligns with (2) could be expanded to

(3') Alice didn't know that stealing is a crime because it is isn't.

Now we're in the territory of something else that might look like conversational (rather than conventional) implicature, because this looks like cancellation, just as one might say

(5) I haven't stopped beating my spouse, because I never started beating my spouse.

But similarly, we might say

(6) The present king of France is not bald, because there is no present king of France.

And that's Russell's account, which disambiguates the scope as

(7) It is not the case that the present king of France is bald, because there is no present king of France.

But Russell's account is not based on conversational, non-conventional implicature, but simply entailment. On Russell's account,

(8) The present king of France is bald.

has the logical form

(9) There is a unique entity such that it is the present king of France, and that entity is bald.

Can we apply a similar analysis to Alice's knowledge of the criminality of theft? On the one hand, (3) could have the form:

(10) Stealing is a crime but Alice didn't know that.

or

(11) Stealing is not a crime, so Alice could not know that it is, and therefore did not know that it is.

((Or, "what's more, she didn't know," etc. There are options here.))

(10) makes a simple claim about Alice's epistemic state. (11) makes a claim about what Alice's epistemic state could or could not possibly be, and then infers what it was. Both make simple claims about the criminality of theft, which allow us to negate them by negating that claim, without reference to Alice, as with Russell's analysis.

(10) is noncommittal on whether knowledge entails truth, as it simply states two facts, one about stealing and one about Alice; (11) is not only consistent with a claim that knowledge entails truth, but relies on it.

Where does that leave the question of conversational implicature?

Grice claims that conversational implicature is "triggered" by an apparent violation of a maxim of conversation, which suggests that what you mean by uttering p must be different from the plain meaning of p, in order to preserve the assumption that you are cooperative (and not after all violating a maxim).

It does seem that the most natural reading of

(3) Alice didn't know that stealing is a crime.

is

(10) Stealing is a crime but Alice didn't know that.

rather than (11), and if you mean (11), you need to say so explicitly. Why should that be? And is this indication of how you expect (3) to be understood a case of implicature?

One reason (10) might be the more natural reading is because we expect the clause governed by "know" to be true or to be asserted to be true, so it is surprising bordering on misuse to place after "know" a proposition you assert to be false, just because you intend also to deny that this is a case of knowledge, precisely because its object is false. To speak in such a way would be a rhetorical flourish. ("I know no such thing, because it is not so!")

There may be other points in favor of (10): it is simpler, and more to the point, suggesting compliance with other maxims to be relevant and concise. But what we're looking for, as evidence of implicature, is apparent maxim violation, not compliance.

I'm also tempted to wonder whether (10) is more natural because it is "common knowledge" that stealing is a crime, but that's not (to my memory) part of Grice's account.

I haven't resolved the implicature issue but I still see nothing to support "knows" not being factive.

+++

To clarify: the presupposition analysis relies on a pair of entailments, not implicature; neither of those entails that Alice knows something that is not the case.

(10) says stealing is wrong and she doesn't know it; (11) is perhaps most simply taken as the negation of (2):

(12) Alice did not know that stealing is not a crime.

But then we have ambiguity again, so that's no help, hence (11).

******

Actually there's no need to stress over (11) and its relation to (10). (Or about implicature, since his usage has other issues anyway.)

What Hazlett is interested in is the straightforward (10), because then we have both Kp → p and ~Kp → p. That's the point of his argument. That's supposed to undercut the unique entailment from Kp to p. But that's because he gets there by (10), rather than (11), which doesn't even lead there.

And (10) interprets "Alice didn't know that stealing is a crime" as "Stealing is a crime, and Alice didn't know that," which of course entails that stealing is a crime.

The issue here is how we justify the (10) interpretation of (3). We would not treat all content this way; we would not, for instance, render

(B1) Harry thinks today is Sunday.

as

(B2) Today is Sunday and Harry thinks that.

Why not?

The simplest answer is that "believes" is not factive, but "knows" is. It allows us to rewrite

(K1) S knows that p

as

(K2) p and S knows that.

Another Russellian move would be to look at the scope: we're taking

(3) Alice didn't know that stealing is a crime.

as

(3') It is not the case that: Alice knew that stealing is a crime.

That means we have all of (1) embedded, and it's form should come out that same as before, without negation in front of it. If that's as above, we have a negated conjunction, and our ambiguity is a matter of which conjunct is negated.

(3'') Not both (i) stealing is a crime, and (ii) Alice knew that.

And, again, that analysis only comes off if we have the rewrite rule (K2).

There are responses to the paper; I have three queued up that aren't buying it.

Should probably be pushed off to another thread if people want to get into this.

Just to play devil's advocate: The Myth of Factive Verbs. — Michael

Nice find, reading it now.

Even though “knows” is, according to Hazlett, not a factive verb, even Hazlett accepts that knowledge itself is a state that can only obtain if its content is true.

We'll see how it goes. I've been using "factive" as a shorthand for this:

The inference rule

     Kp ⊢ p

Allows you to conclude p from Kp — Srap Tasmaner

I would rather take the inference rule as primary and say that our usage of "know" mostly, though imperfectly, follows that -- that this is the nature of knowledge -- rather than saying the inference rule rests on an analysis of how we use the word "knows." (But there's a whole mess there on the relation of logic to the ordinary words we use for reasoning.)

Hazlett says

I'm suggesting, in other words, a divorce for the linguistic theory of knowledge attributions and traditional epistemology.

And I might be okay with that. Still reading.

the initial metaphysical question cannot be answered with empirical examples. — Mww

This is a thing.

patterns of association in language mirror patterns of association in environments; the histories of the two get intertwined through the mirroring relationship. — fdrake

And I almost asked if you felt a little queasy when you reached for words like "tracking" and "mirroring," but it turns out you had something quite specific in mind.

I found your post really interesting but couldn't help feeling -- sorry -- that it was old wine in a new bottle. That is, same problem in new language that doesn't have the apparent baggage of the old, but must if it's to do what we want -- so if universally accepted among philosophers, would lead to sixty years of debate about what mirroring is and whether it's a real thing, as a sequel to the debate over reference. That's not a substantive reply so I didn't -- though now I have!

I think @Mww had a gut reaction near mine, that this is just not what a solution to the question at hand must look like, and that's why I felt it must be a restatement of the problem instead of a solution.

Issues I am alive to in what I'm writing:

(1) Not all questions get answers. Some questions are ill-conceived and attempts to answer them, no matter how circumspect, are doomed to fail. (So, above, "what we want" might be something we shouldn't want, or we only think that's what we want but it isn't, etc.)

(2) There is a difference between a problem-and-proposed-solutions approach, and a model-building approach. Model-builders claim, in part, that the problem can only be a problem within a given -- which may mean, presumed -- model.

(3) One can claim, not quite to the converse, that a model is a framework for presenting and clarifying a problem; problem first, then model. That's one, more or less happy, way of taking "within."

But there are two sides here, and while they agree that a problem can only be presented within a framework, the other side -- model first -- has the option of claiming that a problem "within" a model (or framework) can also be taken as a problem for the model, an indication there is something wrong with it. In that case, the solution is always a new model, even if that model is merely an extension or outgrowth of the old one. Correct models -- Zeus's models -- do not have problems.

This is kinda what the progress of science looks like sometimes, this iterative (and cumulative, ratcheting) re-modeling structured around eliminating each generation's problems in the next generation. (Eliminating in a way consistent with the evidence, not just defining away. Why are these variable related but not those? --- Oh my god! If you rotate the axes, you can see that ..., and that must mean we were actually measuring ..., and so on.)

(4) And that question, of the fidelity and effectiveness of a model, looks shockingly like the substantive issue under discussion. Enter @Mww with his (?) reminder that there are metaphysical stakes here.

(5) Minor issue. There are differences in intellectual temperament that make your posts difficult for me sometimes. ("You" = @fdrake.) You're more "synthetical" and speculative; I'm more "analytical" and -- what's an opposite for "speculative"? Evidence-focused rather than theory-focused? Even with a post like this, I can't help including a folksy example. (Thought maybe I hadn't, but nope, it's right there, end of (3).) Apo said once that I was "too concrete." Analytical me can't ever use words like "enmeshed" or "intertwined" without feeling like I'm cheating. "Enmeshed," to me, is a weasel word -- but it's a perfectly legitimate placeholder when you're model-building! ("These are intimately related, I just can't specify how yet.")

*

That's enough. I really want to start all over with this reference and truth stuff, but we'll see. Nothing I've posted so far has gone anywhere.

Right, right. I forgot she added a step.

'Warsaw was bombed in World War II' is asserted in the Bible iff Warsaw was bombed in World War II.

I really expected something like this:

'Warsaw was bombed in World War II' is true iff 'Warsaw was bombed in World War II' is asserted in the Bible.

Now I feel like I've misunderstood something.

I'm also no longer sure what I had mind when I wrote this:

there is no other conceivable way to do so — Srap Tasmaner

The things that you're liable
To read in the Bible
It ain't necessarily so

Yes, that's how I'm looking at it. And not only does it fix the extension of true, there is no other conceivable way to do so. That's why it must be a consequence of any substantive theory of truth.

For us, a lot of the interesting stuff is on the intensional side, modal contexts, propositional attitudes, all that business.

And there is still plenty of room for a metaphysical theory of what makes true sentences true, because this is not such a theory but only a semantics of true -- and the semantics of true is, for model-theoretic truth-conditional Montague-style semantics, trivial.

This is from page 4 of a classic textbook on formal semantics:

1. Truth-Conditional Semantics

A truth-conditional theory of semantics is one which adheres to the following dictum: To know the meaning of a (declarative) sentence is to know what the world would have to be like for the sentence to be true. Put another way, to give the meaning of a sentence is to specify its truth conditions, i.e., to give necessary and sufficient conditions for the truth of that sentence. (Note, by the way, that we are using "true" to indicate something like "corresponding to the way the world is." We are thus implicitly adopting a correspondence theory of truth.) — Dowty, Wall, and Peters (italics in original)

And that's the difference between doing philosophy and doing linguistics, I guess.

'X' doesn't refer at all, it's a type of action that gets a job done — Isaac

Here's a link to a post of mine about this. If you clink on that link, it takes you right to what I said. In this context, we could say it refers to what I said. Following that link is how you get the job done of finding out what I said.

So "p" and "'p' is true" have the same extension but might have a different intension? — Michael

Well, something like that has always been the complaint about purely extensional semantics. From "The Meaning of 'Meaning'," which just happens to be on another tab in my browser:

the timeworn example of the two terms "creature with a kidney" and "creature with a heart" does show that two terms can have the same extension and yet differ in intension.

So you're saying that these are equivalent?

1. "p" is true iff p
2. "'p' is true" means "p" — Michael

If you take take means as has the same extension as, then yes. Otherwise, no, or depends.

Are these equivalent?

1. "p" is true iff p
2. "p" is a true sentence iff p
3. "p" is a sentence iff p — Michael

They can't all be true at the same time, because the use of "sentence" in (2) conflicts with its use in (3), doesn't it?

My point was that any function that assigns to every sentence the same truth-value it has already, is equivalent to what we've been writing as "is true," and there can only be one such function. Am I missing something?

Is "p" is foo iff p the semantics of "is foo"? — Michael

You just read the Tarski, right? I haven't done that in years, so you're better placed than I.

I think, yes, that is the semantics of "is foo." It says, in plain English, that whatever the truth conditions of p are, those are the truth conditions of 'p' is foo, and vice versa. And it's also obvious that any such predicate "is foo" is equivalent to "is true," that there is a unique identity function on truth-values, and thus a unique identity function on truth conditions.

Honestly, though, I'm out of my depth here. I know little formal semantics.

if you did state those premises, "if you know it then it is true" — Metaphysician Undercover

You mean if I wrote something like this?

$\ \ \ \ \ Kp\ \vdash\ p$

Like stating that kind of premise? Or would you prefer something like this?

$\ \ \ \ \ \forall p(\exists x Kxp\ \to\ p)$

But then, honestly, I'm not sure what there is to talk about if your position is that one can know things that are not so, see things that are not there, remember things that did not happen, and regret doing things you did not do.

I'll let you have the last word.

I lean toward (2), but I just don't know enough to say.

I keep thinking there's something of interest there in truth as a sort of identity function. Have you noticed that it works for anything you might count as a truth-value? It works for "unknown," it works for "likely" or "probably," even for numerical probabilities. Whatever you plug in for the truth-value of p, that's the truth-value of p is true. If you think of logic as a sort of algebra, that makes the is-true operator (rather than predicate) kind of interesting.

The conclusion, that today is Joe's birthday does not follow necessarily from the premise "I remember that today is Joe's birthday", because there is no premise to relate "I remember", to what "is" — Metaphysician Undercover

But the connection is right there: the conclusion is the object of the propositional attitude.

You cannot know what is not so. You cannot see what is not there. You cannot remember what did not happen. You cannot regret doing what you did not do.

Every failure you imagine of claims like these are cases where you are simply wrong -- you think it's so but it isn't, you think it's there but it isn't, you think it happened but it didn't, you think you did but you didn't. When you're right, what you are right about is a fact.

My saying that I know, or that you know, or that someone else knows, is of course no guarantee. So what? Logic doesn't guarantee the truth of what you say, but connects one truth to another.

That's all we're doing here. There's nothing particularly subtle about it.

The proposition "I remember that today is Joe's birthday", does not necessitate the conclusion that today is Joe's birthday, without the added premise that my memory is infallible. — Metaphysician Undercover

Yes it does. No claim that someone's memory is infallible is needed to support the claim that, in this case, it is accurate. Neither does modus ponens require p to be a necessary truth. (Which would just make the exercise pointless.)

Holism in some form follows if one accepts the Tarski's idea that a theory of truth must generate a sentence of the form "S" is true IFF X for every sentence of the object language. — Banno

I'm not following this. I think of holism as indicating that the members of the set are not independent in some respect, in this case truth. Isn't the construction of T-sentences a one-by-one affair?

if you understand how to construct a T-sentences of any sentence in the language, then you understand that language. — Banno

Gotcha. I can never remember quite how he puts this. But this is quite mechanical isn't it? I could construct T-sentences for any collection of sentences of whatever language, whether I understand it or them or not. What am I missing?

I think I agree, though I'm slightly unclear on what the first part of your post is saying. (The explanation of T-sentences in English.)

I think there are three possible answers:
(1) Semantics in terms of truth conditions, and no analysis of "is true" is possible because of circularity.
(2) Semantics in terms of truth conditions, and the T-schema is the semantics of "is true". That's it; that's all it can be.
(3) Something besides truth conditions.

natural language semantics — Moliere

And what would those be?

"Forms of life" is a phrase I try not to use because I don't feel like I really understand it too well — Moliere

You can hardly be faulted for that. It's a linguistic gesture that seems to have no propositional content.

Would it help if I called gesture linguistic in a broad sense, whereas "kettle" is linguistic in a narrow sense? — Moliere

Probably a bad idea. Language is a real thing, a specific thing. Not every form of communication, for instance, is language. Neither is every form of intentional or referential action linguistic.

Gesture, for instance may in some cases be propositional without being linguistic. That's messy, but I don't have anything riding on it. What does matter is that there is a well-known linguistic use of gesture in sign language. If you just define all gesture up front as linguistic, you miss what distinguishes sign language from pointing or waving.

me picking up the kettle is also already linguistic. — Moliere

That just strikes me as clearly false. I understand the point you're making, but lately on this forum people making that point use the phrase "forms of life" more often than they use "language-games" to try to mitigate its implausibility.

Language is embedded in the body; gesture is often as important as the written word in determining the meaning of a sentence. — Moliere

Language, in obvious ways, supervenes on the body and on gesture. No fine motor control, no speech, no writing. Can you say the same thing the other way? Obviously not.

"In the beginning was the word" is false.

It's not the case that we just define every sentence in our language and the truth of every sentence follows from those definitions, so it must be that something which isn't our language plays an essential role. — Michael

A side note, possibly off track.

One way I've seen Davidson's program described is that he aims at explaining not what a given sentence means or what makes it true, but more fundamentally at speaker's competence; hence the claim that if you understand all the T-sentences of a language, then you understand that language. (That set of T-sentences is a theory of meaning for that language. Wittgenstein makes similarly holistic noises, but the point here is vaguely against compositionality, I think.)

But here's a question people might be inclined to answer very differently: if you understand all the T-sentences of a language, do you also understand a world? Or maybe even the world? Either answer is interesting.

Maybe Davidson alludes to this somewhere, I wouldn't know.

In what way 'successful'? — Isaac

I don't think this thread can or should accommodate a digression on names (though maybe we're going to end up there anyway).

I will, though, point you back to Oliver Sacks's dad and the glass on the table. The essence of that joke is the conflict between dad's descriptivist theory of his reference to the glass,** and little Oliver's more causal, externalist theory of his dad's reference. That should clarify the difference at least. Is it a coincidence that Oliver is capable of retrieving the intended glass but not even Zeus could act on Oliver's dad's theory?

(I'm cheating a little, but in a way warranted by our examples, because definite descriptions are ambiguous between picking out an unknown object that uniquely answers to the description, and picking out a known object by what amounts to the construction of a nonce name. Here, I'm relying on the latter in treating "the kettle" and "the glass on the table" as essentially names. It may well be I have forgotten too much about this and am passing over some important distinction there.)

** Or, I should probably say, his pretense of holding such a view.

And it is true that ordinarily English speakers don't have the material conditional in mind. — TonesInDeepFreeze

Oh I'd have to go look, but, if memory serves, Grice defended material implication as a faithful representation of conditional reasoning in natural languages and did not join any campaign (not even Strawson's) either to reform logic or to abandon natural language for more precise pastures. And in my view, if it was Grice's view, it deserves deep consideration.

If we have no way of knowing for sure whether what we honestly believe "is true" or not, then what good is the "propositional attitude"? It cannot be an acceptable logical principle, to allow us to draw any valid conclusions. — Metaphysician Undercover

Well, I mean, it's exactly that, and nothing else.

The inference rule

$\ \ \ \ \ Kp\ \vdash\ p$

Allows you to conclude p from Kp, but doesn't tell you whether Kp is true. It is indeed just a logical principle along the lines of modus ponens, which also can't tell you that your premises are true. Does that make modus ponens useless?

I'm not seeing the link to it being sufficient for the analysis of truth. — Isaac

Wasn't meant to be. This round of the conversation was my attempt to clarify the merry-go-round several of you were on that seemed to me to come down to a problem about reference rather than a problem about truth. I still think that. The endless back and forth was about whether the kettle is a linguistic object.

Speaking of truth:

in order to check the truth of "the kettle is boiling" — Isaac

we cannot asses the truth value of any statement about it by correspondence. — Isaac

I suppose I could be wrong, but this is not how I understand things.

A substantive theory of truth would be a metaphysical theory that explains why true statements are true and false statements are not true. It's an account of what constitutes the truth of a proposition. Or maybe what makes a proposition true. Or in virtue of what a proposition is true. And so on.

It doesn't tell you how to check whether a statement is true; it doesn't tell you how to assess the truth-value of a statement, so perforce not even a correspondence theory, if anyone has one of those, would tell you that you assess the truth-value of a statement by checking to see if it corresponds to something or other. What a correspondence theory would say is that if a proposition is true, it is true because it corresponds to 'the facts' or whatever.

I would be wary of thinking that the book suggests that natural languages lie down so easily that we can just read off its sentences always unambiguously into formal sentences. — TonesInDeepFreeze

Absolutely.

For example, I would be surprised if the authors held that "if then" is always in English the material conditional. — TonesInDeepFreeze

I don't happen to remember, but I would presume there's a way around needing to make such a claim. You can stipulate that your account applies to one way of using a word or a phrase, though there may be others, and still claim to have given an account of a subset of English usage, even if that subset doesn't take words as the joints it's carving at.

I agree that the book is very careful indeed in how it states things and formulates things. I always recommend the book. — TonesInDeepFreeze

I found it really fascinating.

I originally taught myself logic out of Quine's Methods of Logic, maybe the 3rd or 4th edition, and though he's meticulous, it's meant to be more accessible than the mathematical logic book. He has that breezy style and a lingering sort of logical positivist disdain for the unscientific, which English certainly is, so you're made to feel you're learning how to think more scientifically.

Kalish and Montague doesn't feel like that at all. It's a theory of what you were actually doing some of the time. There are very precise rules about what English words go where in the schemata, because it's intended to apply to, not replace English. So my memory of it is, anyway.

Is my remembering it the criterion for saying that I knew it all along? — Janus

Yes.

What if I never remember something I once knew, but have forgotten, is it then the case that I nonetheless know it? — Janus

Who can say? We do of course lose knowledge.

And look none of this is transparent to us. You can rack your brains trying to remember something, conclude that you've well and truly forgotten, and then an hour and half later it pops into your head. So it goes.

So, you seem to be saying that if I remember or regret something, then that something is a fact, and that even though it seems like I might remember or regret something, if that thing is not a fact, then I am not remembering or regretting it, but merely think I am remembering or regretting? — Janus

Yes, and I think obviously so, if you just think about what you're saying.

"Steve can see that Mark is uncomfortable," if true, entails that Mark is uncomfortable. If it's false, we've got nothing: maybe Mark is uncomfortable, maybe not. But it can't be true without Mark in fact being uncomfortable.

If it can truly be said of me that I remembered that you don't like strawberries, then it must be the case that you don't like strawberries, There are multiple ways for this to go wrong. I could be thinking of someone else, so it's false that I'm remembering something about you; I could be lying, claiming to remember this when I'm just guessing, so false again; and of course if you do like strawberries then there's no way I could remember that you don't -- I can only be under the mistaken impression that you don't, so again no true memory.

First, I can't speak to the later revision.

I've tried reading Montague, but it's pretty demanding, so he's still an aspiration for me. I do know a little about his views, in a general way, and his place in the history of logic and formal semantics. So I worked through a good chunk of the logic book, to see how he (and Kalish) handled classical logic, and there's a very different vibe to it from many logic textbooks.

Now I could be wrong, but this was my impression. Most logic textbooks go something like this: hey, you youngsters reason, everybody does, but I can show you a better way to do that, one that isn't hampered by the messiness and ambiguity of a language like English; we're going to show you a kind of language made just for reasoning; you'll recognize some of it from what you've been trying to do in English, and we'll show you how to take those groping attempts at reasoning in a medium not really suited for it, and instead do it in our nice clean system. (This is a vaguely Fregean conception, I guess.)

That is not what Montague seems to be up to at all. He and Kalish are obsessively precise about how the familiar English form relates to their notation -- something many logic textbooks swish by with some inadequate handwaving. (If memory serves, this is one of the things that Peter King will pillory a book for, playing it too loose with what exactly logical schemata are supposed to be, etc.)

And I think that's so because Kalish and Montague are not offering an alternative to reasoning in English -- a formal language you would translate some English into -- but an account of how the logical constants and quantifiers in English actually work. To put it plainly, I think this book presents something a lot more like a formal semantics of the logical constants and quantifiers. The result may look similar to what other textbooks are up to, because there is a formalism, but the relation of the formalism to the natural language English is quite different, and I think you can tell that it's different in the way the book is written.

So that's what I meant by "they weren't kidding." This book is not about a formal system someone invented that you might find a useful alternative to English; this book is a formal account of a subset of English, the words we use in connection with reasoning.

Does that make sense? Am I way off base?

What about all the times you end up being wrong? I remember that today is Joe's birthday, but that turns out to be wrong. — Metaphysician Undercover

Then you didn't. Nobody's talking about infallibility here. You thought you did, you could've sworn it was today, whatever. But "I remember that I put my keys on the table," if true, entails that I did. No more than any other sort of statement, propositional attitude reports cannot vouch for their own truth.

Could I be said to have known it all along? — Janus

Yes.

The main thing is to recognize when propositional attitude verbs are factive. If I remember that today is Joe's birthday, then today is Joe's birthday. When I see that a package has been delivered, a package has in fact been delivered. If I regret leaving my car window down, it's down.