Substitute any sentence you like for P. — Banno

You're right, you're right -- forgot for a moment that this is just a schema, and it includes the quotes to produce a name for the substituted sentence --- since "... is true" needs a referring expression, which P isn't. It's just a place-holder, not a name, not even a variable.

is illformed. — Banno

Oh you're right! Reached for quotes to group, so that it's

P ↔ (P is true)

instead of

(P ↔ P) is true.

But that's not what quotes are for.

Wasn't making a point about the order, duh, but, as I said, about your quotes around P in

P ≡ "P" is true — Banno

That's not what you mean. Here ' "P" ' is a name for ' P ', which is a name for a proposition.

That part you obviously agree with, since you passed over it in silence.

P ≡ "P" is true — Banno

P ↔ "P is true"

You've got your quotes in the wrong place. P is already a name.

P ↔ True(P)

I'm just here to help.

I would say that if I had forgotten my mother's name temporarily, then for that temporal period, I did not know her name, even if I could be said to have the potential to know it, since it would likely come to me soon enough. — Janus

I would call the passage from ignorance to knowledge learning. You learned your mother's name from her or from someone else who knew it. On your usage, by remembering you would learn your mother's name (again) from someone (yourself) who doesn't know it.

I think what you call a "potential to know" is what the rest of us call trying to remember something you do know. The idea that you might be able to remember something you do not know, is puzzling.

Did you come up with this usage of "know" yourself?

I chose the alternate example, whether the fence is wood or brick, as better suited to the task in hand; it's more obviously not just a question of opinion. — Banno

I like that -- especially the word "opinion" there -- but I doubt I'll continue. Far as I can tell, people were only ((or at least mostly)) reading what I wrote to see what conclusion I reached so they could agree or disagree with it. I mean, sure, philosophy traffics in abstractions, but I really hoped I could engage people at the level of a concrete scenario we could look at closely together. But I seem to be the only one interested in such a procedure.

I commend adopting a strategy that shows the public nature of justification. — Banno

I don't share your allergy to all things mental.

Why should possession of knowledge be a static unchanging thing? — Janus

If you mean, why do I think knowledge is, at least relatively, persistent --- I'm not quite sure what to say. I could say (a) it's part of our concept of knowledge for it to be persistent (not my favorite argument) or (b) there's an embarrassment of evidence that knowledge persists, for varying durations, certainly, but it's not ephemeral like perception; and maybe (a) derives from (b).

Are you a citizen only when you're showing your passport? Do you know how to ride a bike only while you're actually on a bike? Do you know your mother's name only when you're using it in a sentence?

In the example I gave -- which of course isn't quite knowledge of where Tim is but knowledge of where he said he was going -- I came to know what he said when I heard him say it; it's least committal I guess to say that I then inferred his intentions, and made further inferences about where he'd be later, and so on. Then I forget. Then I remember. For the latter, I would have to learn what Tim said from my memory of what he said, in order for me to create a new instance of knowing what he said.

Okay, that's interesting, and we could talk about how remembering and hearing in the first place might be compromised in similar or different ways, both episodes being theory-laden, both to some degree confabulations, whatever you'd like to say there.

Except, remember that by stipulation I don't know what he said, so what am I remembering? If I recreate his words from something, what is that something? I don't mean that as question for neuroscientists; it can obviously be that too, but for us, it needs to be something that's capable of engendering knowledge. That's the whole point of this, to say that there are these separate instances of knowledge and I create a new one when I need it. How do I do that?

To my reckoning, neither account treats knowledge as a "first class mental state", it's derivative of belief. — fdrake

Probably so. Honestly, it's like no one is convinced there's any such thing anymore.

For what it's worth, I wasn't thinking about knowledge when I started up my little model; I thought it was going to be more behavioral -- people, things, sentences, but I had been thinking about knowledge a lot, so what seemed natural to me eventually turned out to be more stuff about knowledge.

I'd love to see a similar sort of toy model that's beliefs all the way down, and doesn't include knowledge anywhere. What does that look like?

Perhaps, depending upon how we want to construe belief, we could say that we stopped believing Tim was going to Josh's because we couldn't remember, but upon remembering (recalling the script, the line, due to whatever it is that made us believe that) we do believe that -- while we knew it the entire time (there has to be some way we have a memory, after all -- I don't want to deny memory, only modify the picture we're using a bit). — Moliere

I think this is fine, probably. Maybe we could fill in some details even here, but maybe it's unnecessary. This is close to what I took@fdrake to be saying, that knowledge might have some mechanism that allows recreation of the belief on demand, more or less.

I do want to add that it was not my intention to contest the neuroscience of memory, or the idea that memories are, to some degree, confabulations, recreations, and so on. I assume the research on that is sound, and I think it accords readily with experiences most of us have had. It's the extent of the finding, rather than the finding itself, that might be a little surprising, but there you go. It is what it is.

That does make knowledge -- as distinct from memory -- a little tricky, because knowledge is obviously persistent in some sense, even if that sense is transformed into "presenting consistently" or something. It's not like we only just discovered that we can misremember things we know, so there's reason to think the concept of knowledge ought to be able to survive our improved understanding of memory.

I really had no idea we would end up so focused on memory. Honestly hadn't occurred to me that memory would be taken as a sort of proxy for the persistence of knowledge. So this is really interesting.

On the other hand, I did note along the way that one reason for making a model, like a map, is to improve access to the knowledge you've acquired. You noted something similar in the institutional memory of the sciences and academia at large.

This might also be the place to say that I wondered if my toy model would end up functioning in the Republic’s man-writ-large way. (I didn't explicitly design it for that, but not so that it couldn't be either.)

As I described things, we might make a model in language precisely to improve access to our knowledge, but now it looks like access to the model might be hampered by the very same problem it was designed (hypothetically) to solve: namely, that access to the model is in some sense inherently unreliable because memory, including memory of the model, is unreliable.

That's very nice. It looks like it really undermines the motivation for such linguistic models. As I said, I had no idea we might be headed here, but this is the sort of result I hoped for. (Though I expected it to be less general: if you can't remember what color Pat's house, you can't remember what color your model says it is -- something like that.)

But what if this wrong because overbroad, mainly. Maybe the point of a model is precisely that it involves a type of access that is more reliable? For instance, there's that early work of Herbert Simon and others on the memories of chess players: shown a position with pieces randomly placed, strong players (masters) do no better than anyone else at reconstructing them; shown a position from a real game, they do dramatically better because they break down the position into meaningful chunks and assemble those. The random position is harder to model efficiently, and the position modeling that masters do seems to enhance access (masters remember many, many patterns, and use them in modeling a given position). So the argument might fail if this is another point of modeling, to enhance access and make it more reliable. Both cases of memory, but not the same kind.

Chess masters know a lot, standard development patterns, openings, endgame techniques, middlegame themes, on and on and on. I just can't imagine "giving up" the entire category of knowledge. I don't know how we could understand chess performance without it.

I feel bad now that I took the bait, but in for a penny, in for a pound...

I've almost directly quoted you with the 'just look' aspect and you've at the very least been pointing in the direction of knowledge being obtainable via our empirical investigations. — Isaac

Sure, you're quoting me, but since you don't understand the context of anything I say, what's the point?

The whole point of this exercise was to provide a way of making the differences in approaches precise enough and explicit enough that we could actually discuss those differences, instead of going round and round on the same crap. To get there, I say things that may not represent my position, but are more like bringing out a position that the setup I did, shows is a possible position. The intent, again, was just to be clear enough that problems would be clear or could be made clear. @Banno's not onboard with much that I've said but that's fine; as I told him, if my model has assumptions that suck, we should get to see exactly where and how it fails. That would be a win, in my book.

The same goes for the setup. Tried to make it just explicit enough to criticize. But I have to say something, so I did the best I could to get things started.

I don't think of philosophical discussion as a contest of wills. YMMV.

You were given just such an option with...

Our models are projective, anticipatory. Models change our interactions with our world and thus are thus reciprocally changed by the world they modify.
— Joshs

...that models are anticipatory, not recollective. That models predict and enact those predictions, not collect and curate passive data. You've rejected that approach. — Isaac

Not really.

That's all very 30,000-feet for my purposes. In this context, that's just a lot of handwaving. Show me exactly what that looks like, if not in my toy model then in another. I offered @Joshs the same invitation. (Maybe he answered and I missed it; I'll look again.)

Or don't. If you'd rather argue about whether something is anticipatory or recollective, have at it. Not what I'm after.

Because each of you seem quite strongly realist about worldly objects, no enacted constructions for you guys, if you want to know what colour the house is, you just look. Anyone saying it's blue is wrong because it's white, etc. — Isaac

Okay, now, that's an appalling mischaracterization of what's going on here.

I'm responsible for this current round of the discussion going in the direction it has, and from the beginning I left open the possibility that the explanation for "Pat's house is white" counting as true is just that this is what people by and large say, that there is an implicit convention, no more. "Just look" was offered, by @Banno if memory serves, as another thing people do that has bearing on the question. It hasn't been accepted as some official argument settler, certainly not by me. I have in fact tried to bring the discussion right up to the point where there is such direct disagreement over a purported fact, and I have been reluctant to describe this simplistically as one person saying something true and the other false. I have tried to be scrupulous about this, while still pushing the conversation toward such questions being unavoidable. (If you've given 1000 4-year-olds the wug test, how many of them answered "wugs"? Don't know? Why not? Oh yeah -- the only way to know is to actually go and look at the data.)

I have also described the process of model building as beginning with collecting some data, going and checking the layout of Pat's neighborhood, but only because I don't know how else model building might be done. I have noted that the procedures I described do not guarantee fidelity in the model, and that this could matter when it is put to use. I tried to lay this all out in just enough detail that anyone could find something to criticize. I've been trying not to hide my assumptions, but point them out, even where I can find no option but to rely on them. We are capable of collecting data aren't we? Or should we quit bothering since it's all enactively constructed anyway...

@Srap Tasmaner has Pat's house as white. Let's say it seems green to me, and it seems grey to you. No amount of agreement between us regarding what colour Pat's house seems to us to be is capable (under a hard realist assumption) of yielding facts about what colour Pat's house actually is. It's immune to our agreement about the colour it seems to us to be. — Isaac

Gee, this sounds rather like the scenario I was asking for input about. And you seem to be providing some sort of account here, of roughly the sort I asked for. And you know all this how exactly? Have you done research to determine whether this is so? Did you check wikipedia? Or did you sit in your armchair and reason your way to these conclusions?

In the latter kinds of cases I would say the information is there, but access to it is not, and I would not count such a condition as knowing. To count as knowing, I would say it is necessary to have the appropriate information; in other words to know that you know. — Janus

"Where's Tim?" "Dunno. Wait --- he said he was going over to Josh's."

Did I switch from knowing Tim was going to Josh's, maybe for a few hours, to not knowing for a moment or two, and then to knowing it again? I don't think so.

Knowledge you have no access to whatsoever sounds sketchy, I agree, but according to the movies there's hypnosis and therapy. Not the most important case. Knowledge you have imperfect access to is so common, the examples pile up easily. Keeping a grocery list in your head, you might easily recall all but one of the items you intended to buy, and you have to really think to get the last. Again, I can't see describing that as knowing, then not knowing, and then knowing again. You know the whole time, but have trouble remembering, simple as that. And we do, a great deal of the time, readily recall what we know, as needed.

For real arguments against the requirement that to know you must also know you know, see Williamson, Knowledge and Its Limits (which I've only just started reading).

For me, then, to know is to be certain — Janus

I've just never found this compelling. I always immediately think of cases where people are as confident as they can imagine being, what they would naturally describe as "certain," and they're wrong, or cases where someone nurses unwarranted doubts about knowing what they do indeed know.

It always seems to me that certainty is just a different thing that may or may not accompany knowledge. I suppose we might say that if you know that p, you're entitled to be certain that p, and probably even certain that you know that p, but being entitled to judge or to feel (whichever version we're using) is just not the same as in fact judging or feeling.

I think there are straightforward, persuasive counterexamples to the idea that you can't be certain of anything, but the first ones that leap to mind are backwards. Do you know the population of the county where you live? I don't know mine. In fact, I'm absolutely certain I don't know mine.

Hang your head in shame, Andrew.

It strikes me that these ideas are not in direct conflict. This is because it could be the case that a continual behavioural disposition comes equipped with the ability to recreate the state of mind and action to exhibit what is believed as a transitory state. — fdrake

Even if you want to say, as I've been inclined to lately, that knowledge is not a kind of belief but a "first class" mental state in its own right, distinct from belief -- which is enough to keep our positions from conflicting -- we may still want to say that knowledge entails belief. (I'm undecided, but I see the appeal.) If S knows p, then S believes p -- and that can be true even if you don't analyze knowledge as belief + some other stuff.

Which in terms of psychology might come out as you describe -- and we might experience knowledge roughly this way.

Not that I'm ready to plump for knowledge as a disposition to entertain particular beliefs, but that might be the psychology.

on the whole I think our psychologies are such that we don't hold onto beliefs. We don't check them and put them into our box of knowledge. We let go of beliefs as fast as we hold onto them and upon needing them again we re-create them, and they are re-created in light of us speaking to someone. — Moliere

I think this is exactly where I disagree.

It's become clear to me that the key ingredient in the model is knowledge. Step 1 in building a model is, what do we know?

Knowledge is precisely that belief-like state that persists over time without being recreated, reimagined, or re-experienced. We have imperfect access to the knowledge we possess, and we can lose knowledge, but the knowledge we possess we possess continuously.

it'd be important to make explicit that truth and knowledge are not mental — Moliere

Yeah that's exactly the issue between us. Truth is slightly to one side here, but yes indeed knowledge is a mental state.

That's a big discussion, but I'm happy that we've landed on a very specific point of disagreement. That's just the sort of thing I was hoping for.

Guy gets sent to prison and his first night inside he hears guys up and down the cell-block calling out a number now and then, followed by scattered chuckling from the other cons. He asks his cellmate what's going on.

"Well, some of us have been in here so long, we've heard all of each other's jokes, so we numbered 'em. That's what you're hearing."

Guy says, "That's pretty interesting. Can I try it?" When his cellmate nods, he calls out "47."

Crickets.

"Geez, am I in trouble? Are new guys not allowed?"

"Nah, you told it wrong."

It was a post for my own amusement. — Isaac

Why do I feel like you may have argued somewhere that all off our posts are for our own respective amusement...

Maybe you're about to, and I've time-slipped again. Hmmmm.

Just joining in this new trend of quoting one's self rather than one's actual interlocutors. — Isaac

Don't be so snooty. I did it to show the links between posts that were always intended to be linked.



Wonderful!

I have lots to say about the lots you said, but it'll be a little while.

I'm not sure where we are — Srap Tasmaner

At the very beginning, I compared linguistic models to other sorts. If Pat's house is white, when you build a scale model of Pat's neighborhood, you paint Pat's house white ((that is, the model of Pat's house!)); when you build a linguistic model of Pat's neighborhood, you include in that model the sentence "Pat's house is white."

There, we're talking about decisions the model-builder makes. If you apply a particular color to your scale-model of Pat's house, what justifies your choice is that you know what color Pat's house is; it's the same with including "Pat's house is white" in your linguistic model. If you're not sure, when it comes time to paint or to pick your predicate, you can go and look, or ask someone you believe knows.

If your model, scale or linguistic, is faithful, someone who doesn't know can learn from it. You could show someone your scale model of Pat's neighborhood and point out Pat's house, and they might say, "Oh, I didn't know Pat's house was white." You can infer the state of Pat's house from a faithful model of it. If the model is very accurate, you can infer from it the exact shade of white that Pat's house is. There are probably some natural limits there; I might tell you it's not exactly the shade I used in my model, but it's close, that I couldn't exactly match the shade or didn't even try.

A static model like this is clearly a way of storing knowledge. If I need to know the layout of Pat's neighborhood for some reason, but have trouble remembering it all, I can make a model of it, encoding my knowledge to make it more accessible. I could walk around the neighborhood with pencil and paper and make myself a map, Pat's house there, left of him is so-and-so, and who's up on the corner? is that Joe's house? You needn't, at this point, write "Joe's house" on the map, but can go and check. (No, this is Miriam's house (write it down), so where's Joe's house?) You can, in this way, assemble acquirable-sized chunks of knowledge into a whole that you could not acquire in one go.

It is perhaps notable that even the process of model building is subject to failures of execution. People mistype numbers into spreadsheets with alarming regularity. I might have specifically checked the color of Pat's house, but then painted it the wrong color because the lighting in my model room is weird, or I let too much time pass before painting and got confused about what color I determined on my field trip, and so on. Someone could point out my error to me ("Hey, I thought Pat's house is white") and I could even agree with them before they point out that I painted it light blue. ("Grabbed the wrong bottle, I guess.")

The question would be whether this sort of thing really extends to linguistic models: is it really possible that you could know Cheyenne is the capital of Wyoming but store "Casper is the capital of Wyoming," not just mistakenly retrieve "Casper" or misspeak for some other reason, but store the wrong thing mistakenly. That looks really dubious to me. If your knowledge here is in linguistic form, knowing is exactly a matter of storing the right sentence; you cannot store the wrong one and still be said to know the right one.

Unless it is possible to store both, even though they're inconsistent. And that certainly happens. It's why teachers used to talk about the rule, never write the wrong answer on the blackboard -- students will sometimes remember what they saw on the board but forget that it was an example of what not to do.

So it could be that "Cheyenne is the capital of Wyoming" is what I know, and is stored as such, but "Casper is the capital of Wyoming" is something I heard someone mistakenly say once, and it's also stored as a memory, or maybe I just know that Casper is another town in Wyoming beginning with "C". I can't have stored anything about Casper if I don't know anything about Casper, even if that's only what someone said.

The whole point of a model is that it represents my knowledge; if it doesn't at least do that -- and sometimes models don't -- that's a particular sort of failing. But if your knowledge is linguistic, and your model is linguistic, there is no step of "translation" to screw up; the linguistic object you store is exactly the thing you know. (I'm a little leery of this argument, strong as it is, because we have no grounds to assume further that all knowledge is linguistic and stored in a linguistic model. That's clearly false, since we also know, remember, and recognize images, scents, textures, and so on.)

That may provide support to the no-models view, but as I noted above, we are likely also to have stored or otherwise be able to produce sentences that are inconsistent with our knowledge. And that forces us to confront issues the no-models view wanted to sidestep:

what is "Pat's house is blue"? Is it an object? Does it have, or lack, the property of being part of our model of Pat's house? We can attempt to go around these questions by saying that the users of the model simply agree to say, or not say, the sentence "Pat's house is blue," without talking about the model at all. By saying or not saying a given sentence, users of a model show that the sentence is, or is not, part of their linguistic model, without actually saying that. — Srap Tasmaner

The question is whether a sentence I am familiar with represents something I know -- and that's precisely this second-order issue of whether it's part of my model of the world or not. It is of value to me to be able to store and produce sentences that are not representations of my knowledge: it is how I know what someone else mistakenly believes; it is how I hypothesize in the absence of knowledge, and so on. But that means I may not always be certain whether a sentence I have to hand is part of my total knowledge or not.

The majority of disagreements are not quibbles about the facts, they are quibbles about the lenses through which we view the world. — fdrake

Quickly, this is probably right, but for my purpose here it's facts that matter, if facts are going to be how we talk about truth. As I understand your hierarchy, differences at any of the three levels may present as a disagreement over facts or truth, but the disagreement must be resolved at the level at which it originates, so only disagreements that are simply about facts are resolvable at the level of facts.

That's also plausible, but at this point, I don't even know how best to characterize what a disagreement over facts is, much less resolve it, much less discern its origin. I want to try to stick to my little model a bit longer to force myself to say exactly what's going on if I can, rather than take anything for granted.

On your first post, the first part of an argument against people who disagree each having their own model:

Before we get to that, I want to fill out this:

we may have no choice but to give an account of how (a) the model I use, (b) the sentences I utter, and (c) the occasions upon which I utter them, are related. There are multiple possible explanations for the utterance of a sentence not in the model. — Srap Tasmaner

What I had in mind was this: it is plausible that any individual has only imperfect access to the model of the world they've been working on, and they only imperfectly "translate" it into an utterance. In the case at hand, there are at least these possiblities:

• I may know perfectly well what color Pat's house is, but forgot for the moment, or misremembered;
• I may have only known that Pat's house is the same color as Joe's, which I know to be white, but have failed to make the inference that Pat's is white;
• I may not have recognized that this is an occasion for using my Pat's-house knowledge -- maybe I misheard "Pat's" as "Srap's";
• I may have simply misspoken, perhaps because I was just a moment ago thinking of something black and was primed to say "black" instead of "white".

In addition, if we presume the two speakers share a model, it's reasonable to expect they would actually only be familiar with non-identical proper subsets of the community-wide model. I may know that Pat's house is white, and that his front door is white, and assume that the back door is likewise white, while those that have seen it know it to be grey; I possess slightly less knowledge of Pat's house than some do, but I can readily extend my acquaintance with the shared model by being informed or seeing the back door for myself.

That should add at least one more possibility for those who have a single model disagreeing, that one of them knows and the other assumed or guessed or made a valid but unsound inference, etc., because he wasn't familiar with a part of their model that the other is. (Or maybe neither of them actually know and they're both bullshitting.)

On your account, what people say is presented as a perfect reflection of the model they are using, and that's tantamount to simply identifying the model with what they say.

On my account, differences in what we say are inconclusive evidence that we have different models. There may be other reasons (as above) why on this occasion we didn't end up saying the same thing. And this is so because, differences aside, what any one person says is an imperfect reflection of the model they use.

If that's so, it's hard right off to say whether an occasion of disagreement indicates two models or one in use. You've presented -- at least, along the way somewhere else -- the argument for there being one. That was also more or less @fdrake's reading of Davidson, in part. I'll have to think a while about what, in my test-bed here, multiple models would look like and whether we can tell the difference between that and a single one. --- Should probably say here more clearly: above I suggested there is community-constructed model that it is something like the union of all the models actually in use by individuals, each of whom is familiar with only a proper subset of that union; I'm inclined to consider that another access issue and say individuals familiar with different subsets of a single model share just one, but I'd be open to arguments that these should be considered different, if consistent, models. I'm not sure it much matters what you say here.

And then there's your main point, that the argument for no models runs through a single shared model just being unnecessary, that the only conceivable use for the model talk in the first place was if competing models were in the offing. If we all have the same one, we don't need that one and can just all have the same nothing.

I'm inclined to pause here and wonder whether the model, even if singular, is doing work that just the raw corpus of utterances can't. For instance, I can say that I deviated in speech from my model because of a priming effect, or misremembering, or misunderstanding the context. ("Oh Pat's house. Yeah, it's white.") What does the no-models account say? Most of the time I say one thing, but on this occasion I say something else, and --- and what? Why did I deviate? It seems to me the idea of a model gives you at least a start on dispositions to speak in certain ways, dispositions that are not absolute guarantees. But on the no-models view, I just say stuff, and what I "believe" is represented by whatever I said most recently or whatever I say most commonly, or who knows what.

And perhaps now that I've dropped the B-word, we should look a little again at what the word "model" was doing for me. It is frankly representational -- I don't know how else to take "model." If we do develop such models of the world, and happen to use language as a medium for doing so -- no doubt necause of its considerable efficiency and portability compared to other media -- then, while language is the medium of the model, I need not use it only for producing speech. It can be simply how I store a considerable portion of my knowledge, and my knowledge I can rely on in doing many more things than speaking. I can also use it to store hypotheses, possible but uncertain extensions of my knowledge, which I can act on to confirm or disconfirm, and so on.

If there is no model, but only my speech behavior, then to do any of these things in which I rely on my linguistic knowledge, I must, presumably, speak to myself about them. Now I talk to myself a lot, but I don't have to form the sentence "Cheyenne is the capital of Wyoming," much less speak it, even sotto voce, to remember that it is. Do we perhaps engage in silent and unconscious speech in order to retrieve the facts we know?

That begins to look a bit like a "language of thought," which, oddly, is where my use of language as model medium seems to be headed. It's natural to talk about at least some of our knowledge being stored linguistically only because so much of it is acquired linguistically or is intrinsically linguistic. "Cheyenne" and "Wyoming" are after all names, related in certain ways, which, in this case, are in part purely matters of convention and thus linguistic. My knowledge that Cheyenne is the capital of Wyoming has no option but to be a bit of linguistic knowledge.

But the issue that arises next is obvious: I have considerable knowledge of my native language which I rely on in order to speak it. If that knowledge is not stored linguistically, how can I possibly speak my native language? How could I ever "whisper" to myself, even unconsciously, that Cheyenne is the capital of Wyoming, if I cannot call on my knowledge of English to do that, because I cannot conceivably remind myself linguistically how to speak?

Some of those arguments may not be very good, I dunno. I'm not sure where we are, now, but at least there's now something in the neighborhood of an argument for my initial assumption, that we use language as a medium with which to build a model of the world, which was unargued for to start with.

I hope we're not quite there yet, but if we are at the point where none of the not-really-disagreeing explanations work, then we may be forced to say that one of our two speakers has said something false, although at the moment we don't know which one. There are worse solutions than, as both @Banno and Herodotus said, going and looking for yourself. As things are in my little test-bed, the model is in part a matter of convenience, and I'm still in a position to compare it directly to what it is a model of -- I can test at least some of it in the most direct way imaginable.

This is already covering a lot of ground, so I'll stop, but there ought to be more on what's happened here, whether I had an idiosyncratic and inaccurate model, and so on. But it looks like it's getting much harder here, so I wonder if we can take a step back and simplify things again.

I can be more specific.

Can you offer a no-models account of disagreement, while I'm working (you know, for money) and mulling over what to say next, and others are thinking about -- and maybe even posting! -- their own accounts of disagreement?

Oh good lord no!

I mean, suit yourself, if you just want to see how it plays out.

But I wasn't planning on doing it all myself. I'm not dribbling out something I've already got all of. I just wanted to do some setup people might agree to so disagreements could be clearer and a way to resolve them might be possible. Was hoping others would be pitching in once I got that out of the way.

I'm wondering if instead you might address the limits of the notion of language as a model. — Banno

Yes and no.

No, in the sense that I'm not trying to build a complete model of language use, just enough to clarify questions about truth.

Yes, in that, if there are problems with such a partial model of language, we should land clearly and unambiguously right on top of them.

The question stays, what is the nature of such a disagreement? About what do these two people disagree? — Srap Tasmaner

Actually thinking now I shouldn't have thrown in the second question. We don't absolutely need it yet -- that is, we don't need to pose this as a question these two people might be expected to answer. This way of posing the question is ambiguous between our description of what's going on and what they might think is going on. The latter is interesting, but I think we can wait.

We have one person saying "Pat's house is white," and the other saying, "Pat's house is black." (Definitive colors.) What are we to make of that?

I said we know immediately that at least one of them is wrong. Why? Presumably because we have accepted the limitation on models that they be consistent, which here means that there can be no model they might rely on that includes both "Pat's house is white" and "Pat's house is black."

Is that reasonable? If it is, we may have no choice but to give an account of how (a) the model I use, (b) the sentences I utter, and (c) the occasions upon which I utter them, are related. There are multiple possible explanations for the utterance of a sentence not in the model.

The issue here is not, how do we flesh out our account of language, because this isn't intended to be a complete account. It's that the only definite path toward truth we have found so far is an account of being wrong (which we hope will be useful). So far we've only established that one of these guys is wrong, but we don't know what that really means. For instance, it needn't mean diverging from the previously accepted model; it could be the divergent sentence is a correction, and wrong is staying with the given model.

I get that you would prefer another approach. Maybe I would too -- I'm undecided.

But this thread has overwhelmingly been about redundancy and T-sentences. All I've done is provide a sort of test-bed that I hope will clarify that conversation. I have already indirectly described both, and I believe most readers here recognize that.

I get that you think this entire approach, and most of this thread is wrong-headed. I hoped one of the virtues of my presentation would be that it is explicit enough, without becoming pedantic, that disagreement with the model could be tied to something I said explicitly. Not just, "here you shouldn't do that but this other thing," but "if you do that, here's the problem you won't be able to solve."

Can you point to something like that? These posts have a very specific purpose, and it's not to provide evidence of whether I think something you approve of.

If I say "Pat's house is green," and you say, "Pat's house is aqua," can we still be considered users of the same model? Do we have different models, or do we disagree about which sentence is part of the model? — Srap Tasmaner

I think disagreement is a natural way into the question of truth; maybe neither of us is right but at least one of us is wrong. (Those colors don't contrast so clearly as I wanted, sorry.)

Is that enough, to get at what being wrong is, and being right is not that? We're accustomed to doing this the other way because of the asymmetry: there's one way to be right but an infinite number of ways to be wrong.

This is an odd case, because it's clearly true (if I had picked better colors) that we can't both be right, even before we give any more substance to what being right is. Without disagreement, you're forced to give an account of being right directly, I think.

Do we need an account of how disagreement is possible? I'd like to assume, to begin with, that we don't, and that even a very minimal sort of disagreement, like one of us misspeaking, will be good enough. We'll find out.

So I'm going to proceed from a scenario like this, to start with, and nothing else, however this situation arose. The question stays, what is the nature of such a disagreement? About what do these two people disagree?

utterly sterile and un-insightful in grappling with why and how humans actually use language. — Joshs

Probably. The principal assumption here (since we're headed for truth) is that language can be used as a medium for making models of the world; if model making is interesting, that would make language interesting.

There are no words or sentences outside o f their actual context use — Joshs

That's literally false, for obvious reasons.

and in their use a word does not point at an object, it creates the object in that it produces a new sense of meaning. — Joshs

Not sure how producing a sense creates an object, but whatever that means it is far more controversial than I was going for.

If you think there is no sense whatsoever in which language can be used as a medium for modeling the world, I won't be saying much that makes sense to you.

But without more I can't judge. — Moliere

What me? I just assumed someone else would pick it up from here...

a good way into the role truth - or like concepts such as accuracy and felicity — fdrake

Yeah that's all I'm going for. Language is other things too, but I'm waiting to see if anything else makes a difference to this discussion.

Back in a little while.

I think that's just about enough setup to begin talking about truth.

I thought something like a simple model of language would be more useful than going round and round about what existing idioms mean. It was intended to be uncontroversial, which turns out to be as much as is uncontroversial of something you might call a kind of functionalism. I'm letting the word "model" do a lot of the work, which some people may not like. I haven't, for instance, tied linguistic behavior to anything more, occasions of utterance, what utterance might imply, anything like that.

Any strenuous objections so far?

the only account we have so far of whether "Pat's house is blue" is part of the model, is precisely the users of the model agreeing to say it. — Srap Tasmaner

Which raises new questions.

(1) Are we really either entitled or required to say there is a model here at all, or are we really only talking about what people agree to say and not say?

If you argue first that being a user of a model just is saying certain things not others, and nothing else, you can quickly reach the conclusion that the model itself is unnecessary.

(2) So what does being a user of the model amount to? If I say "Pat's house is green," and you say, "Pat's house is aqua," can we still be considered users of the same model? Do we have different models, or do we disagree about which sentence is part of the model?

The no-models account seems to have nothing to say here at all: we just say different things; if there are reasons for that, they will come from elsewhere (perhaps even a causal account).

But what would make "My house is blue" part of a model of my house? How can we decide that? Is "My house is green" also part of such a model? Why not? What about "Joe's house is blue"? Is that also part of a model of my house? What about "Joe's car is green"? — Srap Tasmaner

There are two natural options here.

One is a matter of agreement, among the users of the linguistic model, to say "Pat's house is blue" is part of our shared model of Pat's house, or to say it isn't. But we've slipped in new problems and possibly new assumptions: what is "Pat's house is blue"? Is it an object? Does it have, or lack, the property of being part of our model of Pat's house? We can attempt to go around these questions by saying that the users of the model simply agree to say, or not say, the sentence "Pat's house is blue," without talking about the model at all. By saying or not saying a given sentence, users of a model show that the sentence is, or is not, part of their linguistic model, without actually saying that. The latter is still implied, though, and this fact makes certain sorts of sentences ridiculous or puzzling.

The other option is to focus on the model, rather than our use of the model, to devise a systematic way of relating sentences like "Pat's house is blue" and "Joe's car is green." We want to have the kind of model that, by including a sentence like "Pat's house is blue," excludes sentences like "Pat's house is green," "Pat's house is chartreuse," and so on. We also haven't given up entirely on properties, but we want to put aside the question of whether they are objects with names. We still want to say that "Pat's house is blue" means that Pat's house has the property of being blue, and that "Joe's house is blue" means it has that same property, in some sense or other.

There is some motivation for using both approaches. We would like the users of a model, who have agreed to say "Pat's house is blue," also to agree not to say "Pat's house is vermilion," but agreement by itself provides no obvious guarantee that they will do so. On the other hand, the only account we have so far of whether "Pat's house is blue" is part of the model, is precisely the users of the model agreeing to say it. The model can allow only one of "Pat's house is blue" and "Pat's house is cornflower," without telling you which one.

Objects can have names. The name of an object is a linguistic object, which can be used as a representative, a stand-in, for the object (which may or may not be itself linguistic) in linguistic contexts. If you build a small model of your neighborhood, the model of the house in which you live is not a name for your house, though it functions in this context similarly to how a name functions in a linguistic context.

To say something of the state or circumstances of an object, we need more than just names of objects in our language, just as to show the color of your house, you paint the model the same color, to show its doors and windows, you make small versions of those in the, ahem, corresponding places on the model, and to show where your house is in relation to streets, trees, and other houses, you place models of those in the appropriate places.

How we do that in language is controversial. We could say -- in some sense, following Plato -- that there are also objects that are properties of objects, and these sorts of objects can also have names, and so we can conjoin the name of an object and the name of a property, perhaps in a special way, to show that the object has that property, to say that it does. You paint the model of your house blue to show that your house is blue; you say "My house is blue" to show, in language, that it is blue, to model in language its state of possessing the property of being blue.

There are objections to treating properties as themselves objects, objections very important to some philosophers. Does it make any difference? You could still model your house in language by saying things like "My house is blue" even without considering "blue" a name of anything. But what justifies the "is blue" part of the sentence? The presence of "my house" is justified by being a name for my house. If "blue" is not a name for anything, what justifies including it in a sentence which is part of a linguistic model of your house? If your house possesses a certain property, and a name for that property is "blue", we are justified; but if not?

At this point, for some philosophers, suspicion begins to fall on the dominant role of naming here: what we are about here is modeling things in language, and naming is a part of that, but only a part; it's not "in charge", and perhaps shouldn't even be treated as the paradigmatic case of linguistic modeling. The first question, they say, should be whether "My house is blue" is part of a linguistic model of my house, not whether "blue" is a name of anything, not even whether "my house" is a name of anything.

But what would make "My house is blue" part of a model of my house? How can we decide that? Is "My house is green" also part of such a model? Why not? What about "Joe's house is blue"? Is that also part of a model of my house? What about "Joe's car is green"?

[ Off to work. ]

You'll have to join the dots. — Isaac

Nothing complicated or subtle. I assumed concepts include at the barest minimum class membership and exclusion: if I have a concept of jabberwocky, then I'm in a position to say, rightly or wrongly, something is or isn't a jabberwocky. Then I can behave toward it in the way I believe appropriate to jabberwockies. I have to have criteria I rely on to reach a decision regarding an entity about whether it's a jabberwocky or not. Those criteria might be characteristics of the thing, but might be as simple as me believing that you possess such criteria even though I don't, and just asking you and trusting your judgment. But that's pretty weak, and doesn't allow me to have my own jabberwocky-specific dispositions.

That all sounds very old-fashioned. I'm sure there are problems there that need fixing. But it's a starting point, and I think something a lot like that should be a consequence of a better theory of concepts.

So described, concepts sound like predicates, and for some cases that's right. But for a long time I've been uncomfortable with the way classical logic is constructed, which treats all sorts of classification as predication of a completely generic x. I think quantification in natural languages is almost always implicitly restricted, so the logical form of "My dog is barking" is not "There is something such that it is a dog and it is mine and it is barking," but, for a start at least, "There is a member of the class my dogs such that, it is barking." I think we handle sortals quite differently from predicates. An entity that is barking might not be. Some entities that are mine might not be. An entity that is a dog is always a dog, and couldn't be, for instance, a lamp of mine that is now on or off.

(Note that's not a defense of ordinary usage against logic, but a claim that classical logic worked fine for mathematics where quantification is usually restricted but has always been an uncomfortable fit for natural languages where the restrictions on quantification tend to be implicit. Modal logics might get me a lot of what I want, dunno.)

That's my beef with classical logic, and it turns out to be relevant here, not just because jabberwocky is a sortal rather than a predicate, but because you're also erasing all the different ways we might reach for to describe entities and calling them all behaviours, and then even identifying the entity itself as a bundle of behaviours. It's behaviours all the way down, with no agents anywhere.

Which means all we ever do now is describe behaviours, and bundles of behaviours, and that makes them the new entities of unrestricted quantification. Which, you know, fine, but I'm going to be uncomfortable.

What we're talking about, at root, is what it is to be an entity at all. — Isaac

Well, it's not like philosophy has never been here before. I just find this

Not only are we nothing but soup without behaviour, but behaving (acting against the gradient of entropy) is what we are. We are units of anti-entropic behaviour. — Isaac

a bit of an odd halfway house between ontology and physics. I can totally see the appeal, in a unity-of-science way, of something like this, but you're starting with a lot of conceptual apparatus about entropy and the laws of thermodynamics and all that, and then using that to explain the being of entities. Even @apokrisis (who has a related big story) doesn't try to do that, but starts from a more fundamental metaphysics and then gets the physics out of that, eventuating in the universe of medium sized dry goods.

This is the same conversation we were having about truth. In essence, my claim is that you're cheating, but you don't know it. I'm not convinced there is a coherent account of belief that doesn't rely on knowledge as a separate category, and that implies a genuine category of truth distinct from people's opinions. Since you want to deny just that, you have to smuggle it in. Same sort of thing here: you want to explain being in terms of physics, but that's backwards, so you'll have to smuggle in all sorts of stuff physics needs and not acknowledge it.

One issue that's come up recently in this thread is the extent to which we might project the structure of our thought, particularly its linguistic structure, onto reality. I don't want to get into that here, but note that this is an odd area for scientists, because in everyday work they take an instrumental view -- we've got our models, we don't pretend reality is actually like the model, that's not even the point, the only question is how well the model works, we're pragmatists, and if we think about it at all what we think is that we should be self-consciously agnostic about what's really out there, it doesn't change the work anyway. That position only becomes untenable when working in really fundamental areas. (There's a similar situation in mathematics, where people not working in fundamentals take a whole lot of stuff for granted. Fundamentals is almost a separate field.) It gets harder to take the instrumental view, model over here, thing modeled over there. It gets harder to know what even counts as justification or evidence anymore if you're not even sure what the nature of your model is. (There are physicists who believe fundamental physics has been wandering off course for a while now, but even if they're wrong, that's a genuine possibility.) ----- Point being, you come along, a methodological behaviourist, and tell me, in essence, that it turns out your methodology is literal fact, that it's not just a matter of modeling entities in terms of their behaviour, but that entities just literally are their behavior. Now maybe you're right, and you were terribly lucky to have chosen a methodology that turns out not to be a research strategy but a factual description of the universe -- or maybe, just maybe, you're projecting the structure of your thought onto reality.

Well them, I might be wrong... — Banno

Heh. Saw what you did there.

So I don't understand the question. — Banno

I was just hoping you would be more precise. As it stands, your position is that everything we do and say kinda goes together, and I don't know what use that's supposed to be. Not that I would claim it doesn't all kinda go together, but maybe there is something specific we can say now & then.

I'm not sure we are not saying the same thing, but arguing the expressions used. — Banno

I don't have any statistics on this, but I think a safer bet would be that I disagree with everything you say.

I only asked about what you said. Thinking about it on the way home from work, I think I have some idea what you meant, but I'm not going to guess. If you don't want to clarify what you meant, I will live with the disappointment.

I'd still be happy to have some answers about talk of kettles.

I've no idea - that's your phrasing.


Your asking me to explain your own terminology, a terminology I think doesn't work. — Banno

What?

The exchange is right there. None of this my phrasing.

There's nothing in reality that is internal nor external; there's just the stuff we talk about. — Banno

@fdrake mentioned "'access to exterior reality'," and he put scare quotes on it.

I know roughly what he was trying to get at it, but I'm not pressing him for details because it was a broad, speculative post. You seemed to be making a specific point but I don't know what it is.

There's nothing in reality that is internal nor external; there's just the stuff we talk about.
— Banno

What kind of opposition is that?
— Srap Tasmaner

I choose not to talk about the stuff we can't talk about... — Banno

Good for you. What does that have to do with whether anything is internal or external?

There's nothing in reality that is internal nor external; there's just the stuff we talk about. — Banno

What kind of opposition is that?

You could have finished "there's just the stuff that is," or "there's just the stuff we say is inside or outside," but you end up here: x isn't internal or external; x is something we talk about. How is that not just a non sequitur?