Thursday, February 06, 2014

Knowledge Without Belief and the Knowledge Norm of Assertion

Philosophers sometimes say that knowledge is a norm of assertion -- that a person should assert only what she knows. Since knowing some proposition P is usually taken to imply believing that same proposition P, commitment to a knowledge norm of assertion is generally thought to imply commitment to a belief norm of assertion: A person should assert only what she believes.

What happens, however, if one accepts, as I do, that knowledge that P does not imply belief that P? Can the belief norm be violated as long as the knowledge norm is satisfied? Bracketing, if we can, pragmatic and contextualist concerns (which I normally take quite seriously), is it acceptable to assert something one knows but does not believe?

I'm inclined to think it is.

Consider my favorite case of knowledge without belief (or at least without full, determinate belief), the prejudiced professor case:

Juliet, let’s suppose, is a philosophy professor who racially identifies as white. She has critically examined the literature on racial differences in intelligence, and she finds the case for racial equality compelling. She is prepared to argue coherently, sincerely, and vehemently for equality of intelligence and has argued the point repeatedly in the past. And yet Juliet is systematically racist in most of her spontaneous reactions, her unguarded behavior, and her judgments about particular cases. When she gazes out on class the first day of each term, she can’t help but think that some students look brighter than others – and to her, the black students never look bright. When a black student makes an insightful comment or submits an excellent essay, she feels more surprise than she would were a white or Asian student to do so, even though her black students make insightful comments and submit excellent essays at the same rate as do the others. And so on.
I am inclined to say, in such a case (assuming the details are fleshed out in plausible ways) that Juliet knows that all the races are equally intelligent, but her belief state is muddy and in-betweenish; it's not determinately correct to say that she believes it. Such in-between cases of belief require a more nuanced treatment than is permitted by straightforward ascription or denial of the belief. (I often analogize here to in-betweenish cases of having a personality trait like courage or extraversion.) Juliet determinately knows but does not determinately believe.

You might not accept this description of the case. My view about it is distinctly in the philosophical minority. However, suppose you grant my description. Is Juliet justified in asserting that all the races are equally intelligent, despite her not determinately believing that to be the case?

I'm inclined to think so. She has the evidence, she's taken her stand, she does not err when she asserts the proposition in debate, even if she cannot bring herself quite to live in a way consistent with determinately believing it to be so. However she is inclined spontaneously to respond to the world, the egalitarian proposition reflects her best, most informed judgment. Assertability in this sense tracks knowledge better than it tracks belief. She can properly assert despite not determinately believing.

Objection: "Moore's paradox" is the strangeness of saying things like "It's raining but I don't believe that it's raining". One might object to the above that I now seem to be committed to the assertability of Moore-paradoxical sentences like

(1.) All the races are intellectually equal but I don't (determinately) believe that they are.
Reply: I grant that (1) is not properly assertable in most contexts. Rather, what is properly assertable on my view is something like:
(2.) All the races are intellectually equal, but I accept Schwitzgebel's dispositional approach to belief and it is true in the terms of that theory that I do not determinately believe that all the races are intellectually equal.
The non-assertability of (1) flows from the fact that my dispositional approach to belief is not the standard conception of belief. If my view about belief were to become the standard view, then (1) would become assertable.


Nick Byrd said...

I am inclined to agree, so I have no substantive criticism. I wonder if an example about one's belief about induction would also work (and be less riddled with moral content).

It would go something like this:

So-and-so believes that inductive inferences are [bad/unjustified/etc.] but so-and-so relies on inductive inferences in almost all of his conscious and pre-conscious decisions. Does so-and-so really believe that inductive inferences are [bad/unjustified/etc.]?

Thanks, as always, for posting interesting stuff!

Scott Bakker said...

With Juliette, you're transforming 'belief' into something bigger than she can report, while maintaining that 'knowledge' does fall within her capacity to report.

But this actually seems to cut against the notion that belief is something bigger than she report, given that she can, in this instance, report her TRUE belief, her knowledge, but can only indirectly indicate her false belief.

So why not attribute two distinct dispositions to believe, one reportable, one not. It's only the theoretical construct of 'in-between belief' that makes this example odd, isn't it?

Eric Schwitzgebel said...

Nick: That's a nice example! I've been thinking a lot recently about the psychology of radical skepticism. Right up my alley.

Eric Schwitzgebel said...

Scott: My own view (which might be the minority) is that it is less awkward to say of someone that she's in an in-between state of belief (such that it's not quite right to say she believes but also not quite right to say she fails to believe), than to say that she has two directly contradictory beliefs (such that she believes both P and not-P). Both are a little strange to say but neither is hopelessly strange, so the question is what is the best way to shape our language to model to the situation?

I think the in-between approach has at least one major advantage over the believe P and not-P approach, which is that it naturally handles all the various and graded ways our dispositions can splinter, while the P & not-P struggles much more to accommodate the variety of splintering. I think that what tends to be behind the move to model with directly contradictory beliefs is a misguided assumption that one must always determinately believe or fail to believe -- an assumption that doesn't work well with functionalist/dispositionalist models of belief that seem to admit in principle of vague cases because match to any plausible functional/dispositional belief profile can admit of degrees.

Wesley Buckwalter said...

Hey Eric, really great to see another cool defense of the knowledge norm of assertion! For my part, I've always thought that agents like Juliet (and agents in similar cases used by Jennifer Lackey actually do have knowledge and beliefs in the relevant sense. But granting your view of belief and setting that aside, I had a different question to ask, having to do with the intuitions for the assertion.

Almost everyone agrees that people should sometimes assert things that they don’t know, or even things they know to be false such as lying to protect someone, given some sense of “should”. The crucial thing here is that the sense of "should" is should qua assertion, rather than other senses like moral, prudential, legal, etc. Certainly Juliet should assert that all the races are equal (despite, even, having any prior mental states in-between or otherwise!). This is to say, it seemed like maybe one possible interpretation of the intuition about what a person, especially in the role of a professor should assert regarding prejudice is that our intuitions are keying in on one of these other senses of "should" for instance moral responsibility. Do you think this is possibly driving assertability intuitions, and/or perhaps there are other cases that don't readily give rise to such senses?

Michel X. said...

I haven't much of anything to add, but I wonder if an appeal to fiction might not yield a cleaner example. So, for instance, I know that Othello murders Desdemona, even though I (obviously) don't believe it. And that certainly seems assertible.

Marco Devillers said...

"Do I contradict myself? Very well, then I contradict myself, I am large, I contain multitudes." - Walt Whitman

The use of 'she' in this context actually reads somewhat stereotypical. I also read somewhere on the Internet that a self-proclaimed feminist advocates the idea that women use paraconsistent logic. Which is silly because it is a) grossly stereotypical, abject even, and b) men have lots of paraconsistent ideas too.

Maybe you should use something modal? Sometimes, in some context, people tend to believe P, and in another context, people believe ~P? Your example suggests it's in that particular case of the female professor a dichotomy between the cerebral and primitive brain. Of course, there are countless examples of that.

Nice read.

Marco Devillers said...

The interesting thing about dissecting a woman's call to paraconsistency is that a woman implicitly conveys the association with 'la vida loca' latin behavior. A reflection on passionate living and lovemaking; in essence the feminist says she likes to fuck.

So I can only imagine lots of women in a audience listening to this speak, fervently agreeing in a 'fifty shades of grey' kind-of manner. And they all signal just one thing: we like to fuck.

Talking about modal behavior, the next moment some of them will fervently deny their own sexual behavior.

People. Sheesh.

Scott Bakker said...

Eric: "I think the in-between approach has at least one major advantage over the believe P and not-P approach, which is that it naturally handles all the various and graded ways our dispositions can splinter, while the P & not-P struggles much more to accommodate the variety of splintering. I think that what tends to be behind the move to model with directly contradictory beliefs is a misguided assumption that one must always determinately believe or fail to believe -- an assumption that doesn't work well with functionalist/dispositionalist models of belief that seem to admit in principle of vague cases because match to any plausible functional/dispositional belief profile can admit of degrees."

I'm very sympathetic to your motives: I'm just wondering if you aren't letting your apparatus, which captures certain wrinkles the classical apparatus cannot, occlude yet another dimension of variability, and so undermining you instrumental motive for your original complication.

In this case, the question is one of how to marry your account to our intuitions regarding knowledge claims - and you have to admit, the idea of knowing something you don't believe is quite counterintuitive! Dialectically, anyway, this leaves you vulnerable to the claims of the superiority of an apparatus using 'aliefs,' for instance.

So why not look at this as an opportunity to revise your apparatus to capture this hanging inferential thread (or even more generally, the way your model feeds into the larger web of epistemic concepts)? Perhaps talk about 'epistemic dispositions' independent of the cartoon of belief, something that allows you to map knowledge claims across varying typical/atypical topographies of epistemic dispositions. So Juliette knows the races are equal, even though explicit/implicit skew in her 'epistemic topography' reveals that she doesn't believe this is the case.

Just pulled that out of my ass. Anyway, for me, 'belief' is so granular (and so heuristically geared to specific kinds of first-order problem solving) as to be more of a liability than anything when it comes to tackling these intricacies. We need a radically new comic strip!

If you get a chance check out Dehaene's discussion of belief (p95 and following) in his Consciousness and the Brain, where he contrasts the probabilistic nature of subliminal belief and the absolutist tendencies of consciousness. Fascinating stuff! and suggestive of a variety of ways to reconceptualize 'belief.'

Marco Devillers said...

Ah well. Guess I went overboard on that one. But I really don't like feminism these days. Roughly, I think it used to be equal rights, which was a good thing.

But it seems to have dropped down to asserting feminine stereotypes these days. Bloody awful.

Charles T. Wolverton said...

Eric -

The assumption that "knowing P" doesn't imply "believing P" precludes a JTB-like "knowing". So, given your interest in dispositional approaches, I assume you have in mind a concept of "knowing" that goes something like this:

In some contexts, Juliet has a behavioral disposition to assert P and is capable of justifying that assertion - ie, she also has dispositions to provide reasons for the assertion if challenged. Assuming the reasons are convincing to a relevant community, Juliet is warranted in asserting that she "knows P".

However, in other contexts she exhibits behavior that appears to be inconsistent with the assertion. One could collect dispositions that underlie such a behavior and label the collection a "belief" as you propose. And since no concept of "belief" plays a role in the described disposition-based concept of knowing, there would be no conflict between asserting "I know but don't believe". But the value of reintroducing "belief" in that modified form isn't obvious to me. On the contrary, my sense is that "belief" may be a word that has become unnecessary - even undesirable - in situations where at least some basic vocabulary from brain science is admissible.

Note: I composed this before reading Scott's last comment, which is along similar lines. I go a bit further in suggesting elimination of "belief" rather than reconceptualization of it, motivated to some extent by the seeming difficulty in reconceptualizing "belief" - which raises suspicions that while useful in casual conversation, the word may not have a role in more formal discourse.

Callan S. said...

Seems rather the same as Juliet has read all about how smoking gives you cancer, prematurely ages your skin, causes blood clots and many other things - and Juliet has tried to quit but relapsed into smoking three times now.

Or does that seem too base?

a person should assert only what she knows.

I'll use that as my excuse for having writers block, then!

David Duffy said...

Armstrong 1970 says:

It seems obvious that it is possible for a person to hold contradictory beliefs simultaneously. I may believe p, believe q and then discover that q is logically equivalent to ~p. In doing so, I discover that I have been holding contradictory beliefs. In
such a case the contradiction was not an obvious one. But the human mind is a large and an untidy place, and it is quite easy to hold beliefs that are obviously contradictory, but which we fail to bring together, so that the contradiction is not noticed. (Anybody who has written a philosophy book will know how difficult it is to avoid contradicting oneself in the course of it. Such contradictory utterances could very well be expressions of
simultaneously held beliefs. For the honour of philosophers, let us hope they sometimes are.) The problematic case is that of being currently aware of holding two obviously contradictory
beliefs. But even here the situation seems a possible one, although it must include our recognition that we are in an irrational state. We can continue to believe a proposition,
although recognizing that the arguments against it are overwhelming and convincing (vide Tertullian).

Eric Schwitzgebel said...

Thanks for the very interesting comments, folks! Sorry I lost tabs on this thread for a few days.

Wesley: I'm inclined to think that the Juliet case is not one in which the "should" is only moral or pragmatic or due to role, etc. -- though I see how the case could invite that interpretation. A cleaner case would be nice! Part of the problem is that psychologically realistic K-without-B cases often either involve partial forgetfulness or are loaded up with the agent's values. Hmmm. How about the self-deceived husband case? Or the geocentrism case? Here my intuitions are more wobbly, but I feel that the wobbliness in those cases is for the types of contextual and pragmatic reasons you describe. In the Juliet case, maybe because the pragmatic reasons are in alignment, I can more comfortably hang onto my hunch that she should assert what she knows not what she believes. But I'm inclined to think that the hunch survives changes to the scenario (e.g., she is not a prof, she is speaking only to a peer, it's not race but people with green vs yellow shirts, it's not intelligence but skill at backgammon...).

Eric Schwitzgebel said...

Michel: Maybe, but fiction has such its own load of issues in philosophy of language! And arguably I both know and believe that in Othello Othello murders Desdemona and I both know and believe that there is no real Othello who did any such thing.

Eric Schwitzgebel said...

Marco: I have a very similar sexism case with Ralph the sexist philosophy professor (in my paper "Self-Ignorance"), and a somewhat similar elitism case with Piotr the history professor (in my paper "Knowing Your Own Beliefs"). In the spirit of equality, I didn't want to make all by bigots males. But for whatever reason, the Juliet case has received the most discussion. Perhaps because it was the first one I published, and then I backed it up with a bit of x-phi. I guess I also like the case a little better than Ralph because my hunch is that it's slightly easier to see her as complex and conflicted because of her gender role in the profession.

Eric Schwitzgebel said...

Scott: I think I can more or less agree with all of that, although I think the idea of knowing without believing is less unintuitive than most philosophers seem to think. As you say, the topography complex and our concepts granular, and the fundamental issue is how to deal with that.

And thanks for the tip on Dehaene!

Eric Schwitzgebel said...

Charles: I am not opposed to that in general principle, but I don't think the neuroscience and cognitive psychology are there yet (if they ever will be) -- and I also think that there would be consequences for how we conceive of ourselves as thinkers that have moral as well as epistemic significance which would need to be carefully considered.

Eric Schwitzgebel said...

Callan: I think that example works too, probably. But "addiction" sounds like a possible excuser consistent with genuinely believing as well as knowing that smoking is dangerous. For this reason, addiction cases aren't as dialectically effective, I think.

Eric Schwitzgebel said...

David: Armstrong's whole conception here works on the assumption of something like the "belief box" model of the mind, in which proposition are stored. It's much harder to see how this would work out on a thoroughly dispositional approach to belief. (I explore some of these issues in my essay "A Dispositional Approach to Attitudes: Thinking Outside of the Belief Box".)

Wren said...

Instead of "it's raining but I don't believe it is raining."

how about: "it's dark but I don't believe it is dark."

The "I" in the second case is a blind person. Seems to match your Juliet example a bit better and helps us get closer to reconciliation. Belief being knowledge one can't help but act on.

Finally, I think it is interesting that in the "raining" example, the first clause uses the contraction (more common when telling the truth?) while the second clause doesn't. So maybe the second clause is a lie?

Callan S. said...

Hi Eric,

But if someone touches a hotplate, they don't touch it again - they believe the hotplate burns. If we really believe (as well as think) smoking is bad, why do we touch the hotplate again and again?

What you're saying suggests even deeper levels of belief, where we can think smoking is bad, believe it's bad - but at an even deeper belief, we think it's good!! To be touched!

I'm pretty sure the racism example would involve the same thing. But on the other hand it probably does communicate better than the smoking example.

Eric Schwitzgebel said...

Wren: The blind person case is an interesting example, although I'd want to see more detail. Not sure about the contraction issue. It seems that it can be done either way, but still maybe I've unwittingly revealed something in my choice!

Eric Schwitzgebel said...

Callan: Maybe that is the best way to think about the smoking case, but there are issues of addiction that make it a bit different from the Juliet case and that make me less confident about the analysis.

Eric Schwitzgebel said...

Callan: Maybe that is the best way to think about the smoking case, but there are issues of addiction that make it a bit different from the Juliet case and that make me less confident about the analysis.

Wren said...

Eric and Callan:

Knowledge, belief and assertion are all conscious acts but can be affected by the subconscious. Hence the subconscious choice between contraction or not is thought to be a "tell" for lying. I don't think there's evidence for that, really.

Does the subconscious always know the truth? Yes for hotplates, no for smoking, no for darkness (except for those interesting cases of blindsight..)

G. Randolph Mayes said...

Eric, I think I agree almost completely with you, but (as seems often the case for some reason :) I prefer a different vocabulary for expressing your point. I would say that Juliette believes her equality thesis, but she does not feel it. I think philosophers, especially analytic philosophers, are traditionally too dismissive and patronizing about the locution "I feel that x." We can be intellectually convinced by the evidence for something without being able to incorporate into the emotional architecture that runs our prediction/surprise network.(Which is what you are saying, I think.) I think we can know things without believing them for other reasons (animal knowledge) but I kind of like the idea that knowledge requires that we both feel and believe a proposition to be true. I posted about this in DR fairly recently in case anyone is interested.

Eric Schwitzgebel said...

Thanks for the link, Randy. I agree with what you say above and in the blog post, apart from terminology. But "belief" is such a crucial piece of terminology in philosophy of mind and epistemology that I think it's really important to reserve it for the belief-like phenomenon that is most important to our cognition -- and that important thing, I think, is the whole suite of dispositions that Juliet only partly has rather than the more intellectual aspects that she more fully has.

G. Randolph Mayes said...

Interesting, but gosh I believe (not sure about feel :) that my analysis might do a better job of this than yours. Belief, in our epistemological tradition, is an intellectual state, and the fact that J. would sincerely defend the eguality thesis, check that bubble on an anonymous survey, or whatever, is explained by the fact that she believes it. But the fact that she doesn't feel it explains why she is nevertheless surprised when a black student makes an astute comment in class. The mistake of our tradition, I think, is that we have tended to want belief to explain too much. Doing so allows us to make plausible, but what I regard as generally specious arguments that people don't really believe certain things, because they engage in such and such behavior. e.g., You don't really believe in heaven, because if you did, you would not grieve for someone when they die. You don't really believe it is wrong to flirt with a colleague, because if you did, you wouldn't do it. When we make the same point along the lines of the knowledge/belief distinction, I think we lose the explanatory value, because knowing doesn't seem to be a state that explains individual behavior any differently than believing.

Eric Schwitzgebel said...

Randy, I agree that those considerations go the other way. Given that we agree on the substance, the question is whether to tweak the word "belief" to a different use than its majority use in the tradition, or to give the tradition "belief" and try to shift the tradition's attention to other types of states like how one "feels". Both are uphill battles!

I do think that there is a minority tradition that uses "belief" similarly to me, so it's not a radical innovation. You cite those "specious arguments", but on my view they're not necessarily specious, though they are probably overstated at best (perhaps the person is in an "in-between" state, not a flat state of failing to believe). And HH Price, and some aspects of Ryle. And Blake Myers-Schulz and I in a paper published last year find results that suggest that a fair number of undergrads will ascribe Juliet knowledge but not belief.

G. Randolph Mayes said...

Right. Agreed on all accounts. I call them specious because they just feel wrong to me. But a sufficiently nuanced and graded sense of belief would attenuate the conclusionz to a degree I'd be comfortable with I'm sure.

Steve said...

It struck me during a conversation about math that I know 0.9 bar is equivalent to 1, and, that I and other non mathematicians wouldn't at least initially believe.

I would think that entailment and the closure principle would make something clearer.

By virtue of knowing (1) mathematicians know math, and (2) the proposition "0.9 bar is equivalent to 1" is a mathematical certainty, yet (3) knowing such a proposition exists in a specific context and one they may be incomplete, (4) the particulars may be mistaken, and (5) may natural intuitions against it ... these lead me to confidently assert P (from (1) and (2) yet have warranted disbelief via (3), (4) and (5) and not just that knowledge doesn't entail belief. One may in fact have warrant to assert P and doubt it, both coherently.