Do you know of an instance of Habermas bringing the charge of performative contradiction against Wittgenstein? — Paine

I don’t. If anything, Habermas seems sympathetic to the later Wittgenstein (less so to the Tractatus). But there’s a lot I don’t know about Habermas.

“Performative contradiction”: I suppose it depends on how seriously you take the idea that LW was not doing philosophy in, e.g., the PI. The passage you quote begins “When we do philosophy . . .” and the meaning of the sentence implies that LW himself, in this “performance,” isn’t doing philosophy, but rather commenting on it or criticizing it. Is this really tenable?

. . . and a general reply to this conversation: In his excellent book The Logic of Reflection, Julian Roberts reads LW as asserting in the Tractatus that “the structures of truth and certainty are only very inadequately rendered in natural language,” quoting 4.0002: “Language disguises the thought.” But by the time of the Remarks on the Foundation of Mathematics, “Wittgenstein questioned whether there was any such thing as a ‛fact’ independent of the system we used to describe it. Arithmetic, for example, taught us to see particular sorts of ‛fact’.” So language, and the various language-games, would trump logic or structure.

I think LW is open to the charge of what Habermas calls “performative contradiction,” in that he seems to be privileging a particular language-game in his later writing to discuss, criticize, and relativize language-games. It doesn’t matter whether we call this game “philosophy” or not. The question rather is whether such a critical perspective carries its own warrant, so to speak, or whether it is merely another way (among many) of seeing “particular sorts of ‛facts’.”

↪Jamal That makes sense to me -- The purposes of philosophy differ between them. — Moliere

Makes sense to me as well, though I think there is indeed a parallel, which you've pointed out.

Very interesting. Do you think the later Wittgenstein was in sympathy with the idea that reason can be self-reflective, or at any rate can reflect critically upon the forms of understanding? I’m not sure how to read Wittgenstein on this. In the Tractatus, I think LW is saying that such a critical project would be just "metaphysics". But can a “form of life” include a more generous scope for philosophical language that abstracts from experience (or "my world") to question itself?

So, this version of the P-Region has the same sort of multiple realizability that B-Minimal properties do, but at the cost of becoming virtually the same thing. — Count Timothy von Icarus

Yes, this is helpful. I agree.

The way I understand it, we are talking about the actual spatio-temporal region involved in producing M given any one actual instance of M. — Count Timothy von Icarus

This seems to be key. Is an "actual" S-T region one that can only change in certain ways? What I mean is, adding to and subtracting from the mass of P-Reg(P) is doable, in our world. It's a manipulation we can perform on the same region, or at least that's how it seems sensible to think about it. But can we change the physical composition of the substance, without replacing the "actual" example with a different one -- one of the "physical ensembles"? In short, are changes in composition allowed in theory, or does the idea of an "actual" S-T region automatically shut down any such theoretical manipulations?

I think this question makes sense, but we may be losing sight of why multiple realizability matters in the first place. What we want is an explanation of supervenience that preserves a kind of rough-and-ready causality (no M without P) but doesn't commit us to a S-T causality that is unique, that can only happen once. Any sphere will have the property of roundness. The roundness of SphereA isn't caused in space and time -- you don't start with the sphere at T1 and then get the roundness at T2 -- yet it is a property that supervenes on a specified physical (or mathematical, if you prefer) arrangement. So it ought to be multiply realizable. And then the question gets interesting if we go on to ask, "Are minds and brains like this? Is consciousness a property of the brain? Does that mean that a given thought is multiply realizable? How about an entire mind?" etc.

I figured. Just wanted to make sure I understood you, thanks!

Agreed. The skeptical position is almost always about the limits of knowledge, not a declaration about what does or doesn't exist. And it tends to equate knowledge with certainty, as you say -- a much easier target for doubt.

One thing I'd add: You say we might agree with the skeptic that

(b) proof and absolute certainty are chimeras in epistemology — Jamal

but aren't they only chimeras in reference to the external/empirical world? I think you can be a Humean skeptic while reserving a place for genuine analytic knowledge. For Hume, relations of ideas, which would include math and its proofs, are not problematic, because they can be known by reason alone, requiring no reliance on experience.

Love it! Can we see his picture? Well, not "his" picture, if by that you mean Noumenal Former Fred . . .

Of course the Kantian analysis can be played for laughs. The question, if we want to be serious, remains whether some kind of intersubjective agreement is a sensible way of describing human "objective" knowledge.

OK, I thought you were referring to brain states (P) giving rise to mental states (M). The P ("exterior" object) to M (perception of it) is different, I think. For the latter case, it's trivially true that all the Ps have to be unique, if there is more than one perception. (This is assuming that P must include a temporal dimension.). And we could generously allow that the M states are identical across perceptions, though this is by no means certain. But how does this bear on the multiple-realizability question? Is there even such a question with regard to this kind of example? We could talk about "multiple realizability" of perceptions of an apple, but this won't tell us what we want to know about the relationship between physical brain states and the mental events.

Perhaps, to make the two kinds of examples clearer, we could eliminate the "exterior" perception entirely and instead ask, "Is your thought of 7 + 5 = 12 realizable in my mind as well? If so, is this because our respective neuronal arrangements are/must be identical?"

I don't want to lose sight of this exchange between Jamal and Hanover. Hanover is uneasy with the idea of an “intersubjective” way-station between individual subjectivity and some kind of “divine” objectivity which presumably would include knowledge of the noumena. I think the distinction Jamal is making is quite innocent, and conforms well with our experience and practice. I see a dog but for some reason have doubts. So I invite my neighbors to have a look. Together, we agree that I’m not deceived. Further doubts could resolved by a zoologist, post mortem, I suppose. But the point is that my increasing confidence in my subjective judgment never approaches statements about “reality” of the sort that concerns Hanover. We could all be deceived about the “reality” of the dog if you want to limit the use of “reality” to “what the noumena reveal”. But this isn’t how humans (other than philosophers!) operate, and Kant surely wasn’t doubting this intersubjective version of reality. On the contrary, his entire project is to make sense of it, to understand what would have to be the case in order for it to exist.

As for the question about mistaken or invalid reasoning, this seems to me an argument in favor of intersubjectivity. It’s precisely by engaging in rational discussion with others that we’re able to correct our mistakes. Could we all be simultaneously mistaken? Sure, but only more rational investigation will tell. And same point as above: None of this back-and-forth around possible mistakes has any bearing on a “reality” that would put us in direct contact with Kant’s noumena.

No, I think your OP is right. I suppose it depends on whether we can really imagine that "identical B-minimal properties do not entail identical P-regions". This degree of abstraction requires that B-minimal properties could be successfully identified without specifying P-regions, which seems doubtful. What is one of these "properties" supposed to be, anyway, if it isn't defined by its location in space-time? But I guess I'm just calling into question the intelligibility of the whole multiple-realization model. It may not depend on which way we go on B-minimal properties, but rather on whether functionalism, broadly interpreted, can make sense of the physical/mental world at all. Time will tell . . .

superveniance famously fails to ground such ontological dependence — Count Timothy von Icarus

Could you refer us to a paper or discussion about this?

It seems like the two are mutually dependent such that changes in M entail changes in P AND neither current states of M nor P determine future states of M or P. — Count Timothy von Icarus

This is an important implication, and I'm glad you brought it out. If one is looking for a different, non-supervenience-based model of how the physical and the mental are related, some kind of mutual dependence is a natural substitute. But of course this is very hard to describe or understand with our current (pathetically limited) knowledge about what consciousness is, and whether downward causation (mental to physical) is even possible.

B-Minimal properties remove the possibility of multiple realizability, perhaps they also obviate the need for it? — Count Timothy von Icarus

Yes. If you're not a friend of multiple realizability, this is probably the way to go.

That seems to be the way I'm reading Kant, taking him in the direction of Husserl (even at the risk of reading too much back into Kant). — Jamal

I also find this a promising direction. Has anyone ever written a piece called "Kant: The First Phenomenologist"? :smile: At any rate, it's proved impossible to do phenomenology without more or less constant reference back to Kant.

The question about the status of noumena is hard to resolve. Kant tells us, "Doubtless, indeed, there are intelligible entities [these would be the noumena] corresponding to the sensible entities; there may also be intelligible entities to which our sensible faculty of intuition has no relation whatsoever; but our concepts of understanding, being mere forms of thought for our sensible intuition, could not in the least apply to them" [B308-9]

In a way, you could argue that the whole history of Kantian interpretation turns on what he means by "corresponding." (I don't know German, so I don't know whether understanding the German word would help.) Does he mean "corresponding" in the way that a map corresponds to the territory -- a kind of picture? Does he mean "corresponding" in the sense of cause and effect -- the cause (noumenon) might not resemble the effect in the slightest, but still correspond one-to-one as cause and effect? Or is it the correspondence between "concept of understanding" and "Husserlian intentional object"? This latter type of correspondence is itself difficult to articulate, even if we imagine Husserl admitting noumena to his philosophy. But I mean something like the way a theorem creates reasons for assent to its truth, rather than "causing" us, ineluctably, to believe it.

Sorry, this may be shedding more shade than light. But the questions are endlessly interesting, and important. And your original point about not denigrating scientific or phenomenal knowledge on supposedly Kantian grounds is surely right.

In the Critique of Judgment Kant uses the term 'objective' to mean 'disinterested'. A valid judgment of taste is subjective, universal, and not based on concepts. To put it somewhat paradoxically, objectivity is universal subjectivity. — Fooloso4

Another way to think about this, using terminology I don't believe was available to Kant: Objectivity would be universal intersubjectivity. We can theoretically have universal agreement on phenomenal facts, like the cat's whiskers. This avoids the charge that my belief in my cat's whiskers is "merely subjective," while not going so far as to claim that I've achieved the "view from nowhere."

Okay, I can at least start with this . . .

Concerning Finn & Bueno: as I said, a wonderful paper, full of insight. I’m particularly grateful for the four-part counterclaim to quantifier variance around which they structure the paper, because you can then use those four issues as a kind of checklist for any defense of QV. That will be part of another post I’ll write, but for now I want to consider a different question.

Finn & Bueno write that Ǝ “invariably has the function of ranging over the domain and signaling that some, rather than none, of its members satisfy the relevant formula. Yet the quantifier-variance theorist requires Ǝ to have multiple meanings. . . . This raises the issue of how the meaning of a quantifier can differ, and what the other meanings could be. And it is this issue that we tackle, arguing that one cannot make sense of variation in quantificational apparatus in the way the the quantifier-variance theorist demands.”

I think there’s a subtle but crucial equivocation going on here, around the term “meaning”. Consider this from the Sider paper referenced above: Sider also wants to know what these “candidate meanings” could be, but he lays out the question differently. “Understand a ‛candidate meaning’ henceforth as an assignment of meaning to each sentence of the quantificational language in question, where the assigned meanings are assumed to determine, at the least, truth conditions. ‛Candidate meanings’ here are located in the first instance at the level of the sentence; subsentential expressions (like quantifiers)[my itals] can be thought of as having meaning insofar as they contribute to the meanings of the sentences that contain them.”

If Sider means “can be thought of as having meaning only insofar as they contribute to the meanings of the sentences” (which I believe he does), then we have an important distinction. It would be possible, on this view, for the meanings of sentences containing quantifiers to vary according to one’s chosen L, while the quantifiers themselves do not vary. They still get used only one way, the way Finn & Bueno think they must. We would thus fulfill the requirement that Ǝ always has to mean what it ought to mean in well-formed logical expressions. But there’s still room for “quantifier variance” if the meaning resides not at the level of the quantifier but, as Sider suggests, at the level of the sentence.

An example might be helpful. I say “numbers exist”; you say “numbers do not exist”. Each of us would have to use Ǝ to formulate our position in Logicalese. What I’m arguing is that we’re each going to use Ǝ the same way, as we state our respective contradictory positions. The difference in our statements is not at the subsentential, quantifier level. We have no quarrel about "variation in quantificational apparatus." We differ on what exists, not on the use of the quantifier.

Is this still quantifier variance? I say yes, in spirit if not in name. It sharpens the question of multiple ontologies rather than dismissing it. Granted, I’m also suggesting that the term “quantifier variance” is perhaps poorly chosen, since it does seem to imply that it’s the meaning of the quantifier per se, rather than any sentence formed using it, that can change. But the reason why someone would want to posit QV is unaffected. The question never was “Can we find multiple meanings for Ǝ (or ‛&’ or ‛→’ or any of the other operators)?” Rather, what Hirsch is interested in is the question, “Can sentences about existence (which logicians express using Ǝ) change their meanings based on what criteria the speaker is using for existence? Can people talk past each other because their sentences, as a result, mean different things? If so, is there one privileged or distinguished way we ought to write these sentences in order to capture something true about the structure of the world?” If we accept ontological pluralism, then the last question (usually) gets a “no,” but all those many ontologies will still be expressed with well-behaved, consistent operators, satisfying Finn & Bueno. (And yes, I agree with them and with Sider that logical pluralism is untenable as an argument for QV.)

This analysis overlaps with another problem I want to raise about the entire debate, concerning whether ‛Ǝ’ is uniquely troublesome in that it’s used to refer to both a quantifier and a predicate. But I’ll save it and invite comment on this question of equivocation on “meaning”. To summarize: Is it the quantifier whose meaning changes, or the sentences in which the (unchanged) quantifier occurs? And if the latter, is it still QV?

Funny, I was just sitting down to start a reply. I thought the article was brilliant, in about a dozen ways. Enormously helpful in clarifying the issues, especially when read side by side with the Sider paper. I also think there are a couple of points they missed which I want to try to articulate. I'm a slow (re)reader and even slower writer. I'll be back with something in a day or two, hopefully.

Agree about the profound implications. It's all part of the same discussion we visited back with Davidson et al.

Thanks for appreciating the thread!

Thanks, Banno. The paper looks right on target. Sider also has a lot of arguments against quantifier variance. I'll read it carefully and reply. But just looking at objections i - iv, I agree that (iv) raises issues but perhaps should be phrased, "QV appears not to be compatible with charitable translation, and thus requires a defense of its internal consistency." Soften it, in other words. And I think (i) is largely where the debate between Hirschians and Siderians takes place. Sider says the quantifiers have to be unrestricted. But let me read and cogitate . . .

Alisdair MacIntyre, Elizabeth Anscombe, and others would say that the words "moral", "virtue", "obligation", and similar ethical terms no longer describe what Plato and Aristotle meant, nor would they understand what we mean. The entire project of contemporary ethics would be as if "the notion 'criminal' were to remain when criminal law and criminal courts had been abolished and forgotten," in Anscombe's words. This idea of modern ethics as a degenerate, vestigial grasping at what the Greeks meant is even stronger in MacIntyre, and characterizes a lot of the resurgence of virtue ethics. An interesting question is whether it's possible to return to a previous paradigm.

:smirk: Happens to the best of us!

I'm not familiar with the term "scientific miracle." Probably I should be, but could you explain it for me? Thanks.

Two basic alternatives are:

1. Our understanding of physical laws was incomplete.
2. It simply happened. No other explanation behind it. — Philosophim

Concerning 1., the further assumption you need is that physical laws, were we to understand them completely, would explain the purported miracle. I'm not sure what the warrant for this would be. Compare to mathematics: No one can say with certainty whether numbers are "real" or "natural" or "simply happen" (whether in the Platonic world or human brains). Is this because we lack the necessary physical laws to explain them? I think this is doubtful.

So the point is that -- unless you want to call numbers and other abstractions "miracles" -- you need to show why particular miracle X requires a physical explanation. Otherwise I think we're going in a circle.

Indeed. The physicalist dream is that one day we'll have robust reductive explanations, and that these will make sense of any and all phenomena, including people's social behavior. I think this is wildly unlikely, but I count reductionism as an important argument (or perhaps "prop" is a better word) for physicalism because it appeals to one of the most basic human intuitions -- that to understand big things, you need to know what they're made of.

I’m not a physicalist, but you’re asking what I think the strongest arguments for physicalism are. Currently, there are two: The very successful use of scientific method in the West, and reductionist arguments as possible explanations of seemingly non-physical phenomena.

How strong are they, in fact? By now, my reply may be familiar: They’re only as strong as the accuracy of our understanding of the terms “physical,” “mental,” and “emergence,” and probably a few other key terms as well. I believe we’re like children, playing with conceptual building blocks that look increasingly unlikely to correspond to anything foundational in reality. I would bet $1,000 (not that I’ll be around to collect!) that in, say, 500 years, our “arguments for physicalism” will be quaint artifacts of an era before science and philosophy and religion made up their differences and presented a unified world-view. And then there’s the next 500 years – oh boy!

This would make a fascinating thread -- invite people to describe, as best they can, what their personal "stream" is actually like.

One of our most important living philosophers, David Chalmers, has published a recent book that deals with this, and many similar questions about techno-philosophy. It's called Reality + -- I highly recommend it.

And this is good, sensible place to leave it. I have trouble with the idea of "objective consciousness" but it may be just the terminology; now, thanks to your careful explanation, I at least have a better sense of what you mean. Still doubtful, but time will tell. And I didn't know there were people without a stream of consciousness running on in their heads! This is how I've always pictured non-human animals: Maybe not zero stream of con., but very little. I look at my cat and think, I bet it's real quiet in there!

I think the problem is something like this: You want to say that “Consciousness can only be identified through behaviors” and also “Therefore, anything with certain specified behaviors is conscious.” I’m not persuaded by the idea that “being alive” consists of behaviors, but let’s grant it. The argument is still shaky. The fact that (at the moment) we can only identify consciousness through behaviors doesn’t mean that all things that exhibit those behaviors must be conscious. Compare: Some Xs are Y; a is an X; therefore a is Y. This doesn’t follow.

Here’s another way to think about it. You’ve said you don’t like speculating about the future, but if consciousness is truly a scientific problem, as we both believe it is, then at some future point we’re going to know a lot more about it. Let’s imagine that someday we’ll be able to say the following: “Consciousness (C) is caused by (X + Y + Z), and only by (X + Y + Z), and is necessarily so caused.” So, in determining a particular case, we could say, “C iff (X + Y + Z); ~(X + Y + Z); therefore, ~C”. This would give us objective criteria to ascertain consciousness for any given entity. It wouldn’t rely on either behavior or subjective reports.

Now, lest you think I’m deliberately practicing sleight of hand, let me point out that this happy state of affairs is only true if it turns out that X and Y and Z are both objective and unproblematically causal. This may not be the case; we are currently clueless about what gives rise to consciousness. But if it is true, then the hard problem will have been solved. We will know what causes consciousness, and why this is necessarily so. Wouldn’t it be prudent, then, to assume that our current reliance on behavioral markers to identify consciousness is an unfortunate crutch, and that there is no important connection between the two? After all, we know that behaviors don’t cause consciousness, but something does. When we learn what that something is, we may be able to abandon functional “explanations” entirely.

A final thought: Perhaps all you’re saying is that AIs and robots and other artifacts might be conscious, for all we know. To me, that’s unobjectionable, though unlikely. It’s only when we start saying things like “Joe AI is conscious, which we know because of its behaviors,” that anti-functionalists like me get aroused.

Since we cannot know the subjective experience of anything, we can only go by behavior. — Philosophim

I don’t think that’s right. We also place a lot of emphasis on what kind of thing it is. If you ask a non-philosophical friend what the major difference is between an AI program (or even the most sophisticated robot) and a human, the friend is likely to reply, “Humans are alive and hardware isn’t.” Being alive is not a behavior, it’s a state or condition. This allows us to say things like, “I don’t care how ‛lifelike’ the behavior of X is, the fact remains that it’s not alive.” I’m suggesting that we may wind up saying something similar about consciousness.

if we have an AI that ticks all the behaviors of consciousness, we cannot claim that it does, or does not have a subjective experience. Its impossible for us to know. Since we cannot objectively evaluate a subjective experience, all we can do to measure consciousness is through another being's behavior. — Philosophim

The first two sentences are fine: We can guess, but we can’t know for sure, whether an AI is conscious or not. But the third sentence is a non sequitur. How does it follow that behavior would be a measure of consciousness under these circumstances? It’s just the old functionalist argument, which assumes the conclusion by stating that consciousness is measured by behavior.

So if an AI is conscious, its subjective experience is that of a non-biological being, not a biological being. — Philosophim

No doubt. But the question is whether this is even possible.

The controversy centers on whether part of the meaning of the word “cat” is indeed that a cat is an animal,
— J

That seem to be like saying how do we know that "12" represents the integer twelve and not a plate of brownies crushed on the floor? — PL Olcott

Well, no. The analogous question with "12" and the brownies would be, "How do we know that 'cat' represents those furry critters we like so much and not [insert wildly unlikely referent]?" We know this because we know how to use the word "cat", just as we know how to use "12". But in neither case is there some further, purportedly analytical fact about animals or integers. That, at any rate, would be how Kripke and others (including me, most of the time) would argue it.

In fairness, the whole analogy is probably questionable anyway, since the status of numbers is so hotly debated. It's not even clear that they do refer. And Kant, famously, thought arithmetic consisted of synthetic a priori truths, arguing that "12" is not contained in the concept of "7 + 5".

Are you sure? That's a VERY important part of the hard problem. — Philosophim

Quite right, and we’re miles (and decades) away from being sure about any of this. While I can’t know what the subjective experience of a given something is, it seems probable that most things don’t have any. I assume you agree with this. So we’re just trying to draw the most likely line as to consciousness. You say with some assurance that AI programs already have limited consciousness. Is there any evidence for this beyond their behaviors? A purely functionalist argument can’t resolve this, since it begs the question.

Not quite sure why the hard problem rules out denying consciousness to computers at some future date, or why you describe the hard problem as “true.” It’s certainly true that it’s a problem, and that it’s hard, but I don’t think Chalmers or anyone else (except maybe McGinn and company) means to say that it’s by definition irresolvable.

Do I think that any non-living thing can be conscious? No, I’m strongly inclined, on the evidence, to believe that consciousness is exclusively a biological property. How surprised would I be if this turned out to be wrong? Fairly surprised (and fascinated), but again, we have almost literally no idea what we’re talking about. Let’s check back in 2123!

The controversy centers on whether part of the meaning of the word “cat” is indeed that a cat is an animal, or whether words like “cat” are rigid designators, using Kripke’s terminology. Suppose “water” is a rigid designator in all possible worlds, regardless of whether it’s composed of H2O in all of these worlds. So is it analytically true that “water” is H2O? Different philosophers have different takes on this. I’m suggesting that the same questions apply to “cat” and “animal”. We don’t seem to need the concept of “animal” to refer to cats, or recognize them, or talk about them. As it happens, all cats are animals, just as all water is H2O – in our world. But this, according to Kripke, is an "a posteriori necessary identity," and hence not analytic. We can imagine a different world in which this isn’t so, yet presumably we’d still recognize cats using other criteria – furriness, friendliness, general cattish appearance, etc.

The Stanford EP article on “Natural Kinds” gives a good account of the pros and cons of viewing terms like “cat” as rigid designators, and how this might relate to questions of analyticity.

Does that mean the file isn't a set of physical 1's and zeros on my hard drive? Of course it is. Its obvious. — Philosophim

And that is the key difference between a computer and a human. For a computer, there's nothing more the file could be. It isn't "like anything" to be a computer. But we have a different experience, which gives rise to all of the problems discussed on this thread.

This may be one of those fundamental philosophical problems that simply call up different basic intuitions. I certainly don't think that physicalists are less intelligent than I am, or wrong in some obvious, silly way. I hope they feel the same way about me! It just seems like a brute fact (to me) that what I do when I interpret a story about Sherlock Holmes is completely different from what a computer does when it realizes a program. And functionalism has had a very hard time since its heyday in the late 20th century.

All of it is physical. — Philosophim

What we want to understand is how "a physical thought" could "describe a fictional brilliant detective." Referring to Popper again, he would say that any World 2 event -- that is, a particular thought-event that is the product of a particular brain at a particular time -- is merely a thought of Sherlock Holmes. It isn't SH himself. In order to have SH, we have to move to World 3, where objects of thought are named. Even if we allow that "a physical thought" isn't question-begging (it seems so to me), we still have to explain how an idea that depends on no particular brain for its instantiation can be called physical.

Right, that's the problem. It's a kind of reductio ad absurdum, because we know that SH can't be all those things, since he doesn't exist in the material world in the same way that you or I do. So that would seem to suggest that consciousness, if it's a physical process, can create phenomena that aren't physical. This is part of what makes the hard problem so hard.

Well said, and Nagel is great on this.

I too think that consciousness is likely a physical (specifically, biological) phenomenon, but we're being awfully sloppy here, in our talk about what "makes" a physical thing. Consider: Is Sherlock Holmes a physical thing? Everything that could possibly be said to comprise him is physical, but what about SH himself? I find it bizarre and counter-intuitive to say that SH, and any other World 3* phenomenon, must be called physical simply because a physical system produces it.

Again, I don't think this applies to consciousness. I just want to be careful about assuming physicalist truisms.


*From Popper: products or objects of thought, as separated from any given instance of that thought.

My concern, reading this, is more about what's allowed in than what's left out. Are you wanting analytic "truth" to refer, arbitrarily, to whatever may be stipulated as true in a given language (plus the derivable expressions)? If instead of stipulating "Cats are animals", I stipulate "No cat is an animal" as one of my axioms, does this create any problems? I think it has to, especially since you also speak about this as "knowledge." But knowledge of what? Perhaps this is why the analytic status of a statement like "Cats are animals" is controversial. (On my view, it isn't analytic at all.)

Or perhaps I'm not understanding what you want BOAK to do -- what its purpose is.

Thank you for directing us to these essays. I'm reading Simpson now . . . and coincidentally, the group biography "Metaphysical Animals" which focuses on Anscombe, among other women philosophers.

He's a bit like Hume, who demolishes the claim that logic or reason establishes causal powers or causal laws, and then turns to psychology to fill the gap. I'm doubtful about this, because it seems to reduce the issues to causality or subjectivity. Which misrepresents what's going on, I think. — Ludwig V

Yes, well put. By introducing the idea of reasons, or reason as a faculty, rather than causes or preferences, we enter a different way of thinking, one in which (as your next sentence shows) we can start to use concepts like "better," "more accurate," etc.