Knowledge Intuition Fight Club

Negotiation Breakdown

Originally uploaded by Pete Mandik.

In “Wanted: An Actual Argument for the Knowledge Intuition” I pitted KI (”One cannot know what it is like to have an experience of a certain type unless one has had an experience of that type”) against bold claim BC (”No one has ever given an argument for the Knowledge Intuition”). All commentators provided much useful feedback. Some commentators questioned whether KI was something anyone would find plausible or relevant and other commentators provided some interesting counterexamples to BC.

Now for round two.

In this corner, in the red trunks, we have:

(KI+): There exists at least one type of experience such that one cannot know what it is like to have it unless one has had an experience of that or some relevantly similar type.

And in the other corner, in the black and white trunks, we have:

(BC+): No one has ever given an argument for KI+.

1. Don’t talk about Knowledge Intuition Fight Club
2. Single-premise existential generalizations are not welcome.
3. Arguments for KI+ that have premises concerning phenomenal concepts, phenomenal beliefs, and their brethren should specify what those things are in ways such that assertions of their existence aren’t simply different ways of stating KI+.

15 Responses to “Knowledge Intuition Fight Club”

  1. Nick Treanor says:

    Pete, I’m curious to know whether you think it is important that these issues are framed as issues of knowledge rather than as issues of belief. I mentioned this in one of the earlier comment threads but want to return to it. Are you interested in whether:

    “there exists at least one type of experience such that one cannot have a true belief concerning what it is like to have it unless one has had an experience….”

    or in whether:

    “there exists at least one type of experience such that one cannot know what it is like to have it….”

    I gather that what you are interested in is the former question, not the latter. For instance, suppose someone said “no, you can’t _know_ what it is like to have such and such experience without having it, because a condition on knowledge is that one’s belief that p is causally connected in the right kind of way, etc., to the state of affairs that makes p true”. I assume that this kind of response would be beside the point, as far as you are concerned. Am I right about this?

  2. Pete Mandik says:

    Hi Nick,

    I think you raise some really crucial points. My hunch is that what matters most is framing the issue in terms of knowledge. Of course, as you pretty much suggest, there are at least two ways that one might fail to have the knowledge in question: one might fail to have the relevant belief or one might have the belief but fail to have the relevant warrant. Since my grip on epistemology is much weaker than my grip on the philosophy of mind, I’m pretty familiar with the sorts of moves one might make in discussing the belief option. This might explain the appearance that I’m most interested in that option. But I’m interested in both.

    Re: the warrant option. I assume the KWIL is knowledge concerning types and a warrant-based defense of KI+ along the lines you suggest would need to supply reasons for believing that there is at least one case of knowledge of types that cannot be had unless one had some kind of causal relation to past or present tokens of that type. I don’t think that’s true of all types, so I’d be curious to see why that’s supposed to be true of any types.

  3. Nick Treanor says:

    Hi Pete,

    you’re very kind to find the point important. I agree with you that both questions are interesting (assuming that they are two different questions, that is, that knowledge is not mere true belief). The argument or argument area is always called the “knowledge argument”, but I gather that this is only in a loose sense, that those thinking about the question aren’t really taking pains to distinguish knowledge from true belief. (It’s not just that Mary final knows what red is like, it’s that she finally has a true belief about what it is like. It’s not just, so the story goes, that she is in a position to know what red is like, it’s that she’s in a position to have a true belief about what red is like.)

    Regarding your second point about the warrant option, I hadn’t meant to suggest that a causal connection of a certain kind to an actual experience is required for epistemic warrant….I don’t really have an opinion on that, since, like you, epistemology is not my central area. I mentioned it only as an example of a way in which one might think someone has a true belief about what an experience is like while failing to know what the experience is like.

    Okay, how about another question: do you think someone could have a _false_ belief about what it was like to experience red? (This question isn’t: Could they have a false belief about the class of colors to which the English word ‘red’ applies?) I’m not sure that this is possible, and am interested in hearing what you think.

    (I hope this doesn’t count as hijacking the thread….I think this stuff is quite connected to the issues you’re exploring.)

  4. Pete Mandik says:

    Hi Nick,

    The question of false BWILs is pretty interesting. I’d say they’re possible. I wonder why they wouldn’t be. Are you supposing BWILs to have some form like a=a whereby the negation is not rationally believable?

  5. Nick Treanor says:

    Hi Pete,

    I was thinking false BWIL might not be possible for something like the following reason: Suppose the belief in question concerns what it is like to see red. If the belief is false, then what grounds the fact that it is a belief about red?

    Or to put it another way:

    Suppose Smith’s belief concerning what it is like to see red is false because Smith believes that what it is like to see red is x, where x is in fact what it is like to see blue. What makes Smith’s belief a false belief about WIL to see red rather than a true belief about WIL to see blue?

    By analogy: would it be possible for someone to have the false belief “Bill Clinton is a mineral kind, found only on Mars, formed via such and such process, and known about only via information sent back to earth by space probe”. Quite plausibly no — whatever that belief is, it is not a belief about Bill Clinton.

  6. Pete Mandik says:

    Hi Nick,

    The Clinton example makes it sound like this isn’t going to be a problem restricted to BWILs. I’m reminded of an example of Quine concerning two physicists disagreeing on whether nutrinos have mass. Are they referring to two different particle types, one of which is massless, or are they referring to the same type and disagreeing about whether it has mass? Quine urges that such cases are simply indeterminate and I’m inclined to agree.

    Now, one might respond to this and say even though there are some unclear cases there may be clear ones as well. So if you and I agree with the majority of popular beliefs about Bill Clinton and disagree only on whether he went to Yale, then we have a case in which at least one of us has a false belief about Bill Clinton and not a true belief about some fictional character who shares his name.

    Returning to BWILs, it seems that for at least some cases false BWILs are possible. I think it is possible and perhaps common for people to believe falsely that what it is like to feel thirsty is like having a dry throat. And this can happen even to people who have had each on several occasions. More examples WIL confusions might be easy to come by: pleasure vs. relief; embarassment vs. anger; tasting vs. smelling.

    Returning to BWILs about phenomenal red, how about this one: Consider someone momentarily being confused about whether what it is like to experience red is more like experiencing orange than green. Couldn’t a person at least temporarily hold the false belief that phenomenal red is more like phenomenal green than phenomenal orange? Later they realize that they were confused, and say, ah yes, it is more like orange. Would this example even make sense if it were utterly indeterminate whether they were thinking about phenomenal red in the first place?

  7. Pete Mandik says:

    [Due to problems with the web site, Torin Alter submitted the following via email.]

    How about an argument against KI+? The argument is based on Dennett’s Swamp Mary (from his paper in Phenomenal Knowledge and Phenomenal Concepts, which should be out in about 2 weeks). Swamp Mary is molecule-for-molecule just like Mary at time t, except she was created by a cosmic accident. At time t, Mary has already seen red but has returned to her room; she isn’t at t seeing (or imagining) red. Here’s the argument:

    1. At t, Mary knows what it’s like to see red.
    2. If at t Mary knows what it’s like to see red, then Swamp Mary knows what it’s like to see red.
    3. So, Swamp Mary knows what it’s like to see red.
    4. Swamp Mary has never had an experience of seeing red, pink, or anything similar.
    5. Swamp Mary is possible.
    6. So, KI+ is false.

    Also, I think it’s worth noting that the phrase “the knowledge intuition” is sometimes used for a different claim (see the Stoljar/Nagasawa intro to There’s Something About Mary). This is the claim that (roughly put) there are truths about experience that cannot be a priori deduced from the complete physical truth. No one could reasonably deny that arguments have been given for that! One is based on the Mary case; another is based on the conceivability of zombies and inverted spectrum cases. I hope it’s clear that one can consistently accept the claim about non-deducibility and deny KI+.

    Finally, a question for Pete: suppose KI+ is false (as I believe). This supposition doesn’t affect the knowledge argument, which is based on the non-deducibility claim, not KI+. So, what do you think follows from denying KI+?

  8. Nick Treanor says:

    Torin, this kind of argument was considered in an earlier post (I think Dave Chalmers first brought it up), and it seems to be a good one. However, one might doubt that _knowledge_ supervenes in the way the argument supposes, even if _mere belief_ does, which would make premise 2 false. (I think that most accounts of knowledge would have to say that premise 2 is false.)

  9. Pete Mandik says:

    Hi Torin,

    Thank you for the comments.

    Here are some responses in no particular order.

    I’m persuaded that KI+ is false and I find Swamp Mary type arguments pretty convincing. Other considerations come from empirical work on the structure and content of concepts and I’m finishing a paper on this right now (“The Neurophilosophy of Subjectivity”). I should note, though, that I think Nick raises a pretty good challenge in his point about differentiating knowledge from belief. I’m not sure how to meet Nick’s challenge.

    I think you raise a fair point about the use of the label “knowledge intuition” elsewhere in the literature. But allow me to pick nit. A lot of what Stoljar and Nagasawa discuss in their Intro is the claim that no amount of physical knowledge suffices for phenomenal knowledge (Call this SNKI). A failure of a priori deducibility may be one interpretation or explanation of the failure of sufficiency, but it seems open that there are other interpretations or explanations.

    What follows from the falsity of KI+? A few things. It would undermine most motivations I’m aware of for believing certain proposals for what phenomenal concepts consist in. Regarding the knowledge argument, if KI+ is false, then I wonder what reasons there are for believing SNKI? I would have thought that the main reason anyone believed in SNKI is because they bought KI+. I would have thought that the main reason anyone believed that physical knowledge cannot suffice for knowing what it is like to have such-and-such experience is because they believe that when it comes to experience, one has to have it to know it. I take it you would disagree that the failure of KI+ casts doubt on SNKI. Is this correct?

  10. Eric Thomson says:

    Dennett’s new thought experiment is wonderful: an all-too-rare push into an interesting novel direction in the qualia debates (he discusses swamp mary in Sweet Dreams, Ch 5). Too bad he is so loathe to engage is such fanciful thought experiments (before bringing up swamp Mary, he says, “Suppressing my gag reflex and my giggle reflex, here she is…”.

    My question is, what will a qualia-phile’s intuition be about KWIL? To KWIL to experience redness, do you have to have an occurrent experience of redness, or can I KWIL to experience red even when it is not an occurrent experience? While I think most would say it doesn’t have to be occurrent for you to KWIL, Dennett then undercuts many forms of the knowledge argument.

    OTOH, someone like Dretske, an externalist about the contents of consciousness, will say that Swamp Mary doesn’t have experiences of anything until the contents are fixed via interactions with a world. But I think Dretske isn’t what I would call a qualiaphile because for him qualia can be analyzed without remainder into informational/causal/functional terms.

  11. Pete Mandik says:


    The head-world interactions that Dretske cares are secured by evolution. Swamp Mary is a normally evolved human being who is able to see red. She just hasn’t seen red yet. The lightning bolt doesn’t bring her into existence. It puts her into a state just like the state Mary would have been in after (but not during) a red experience. No Drestkean objection can rule out that Swamp Mary’s new mental state represents red.

  12. Eric Thomson says:

    Good point, Pete (in Dretske-speak, from Naturalizing the Mind, they have systemic versus acquired indicator functions). Even better for Dennett, then. Impressive.

    I wonder how many brilliant philosophers could come up with such great thought experiments, but don’t because they are constitutionally averse to such things.

  13. Torin Alter says:

    Hi Pete and Nick,

    Thanks for the reactions and sorry to be slow to response. First, concessions:

    -right, I misdescribed swamp mary. Only her brain state is caused by a cosmic accident. I might add, though, that Unger in 1966 had a set of cases that are similar to swamp mary. One even involved a cosmic accident.

    -right, the deducibility interpretation of the knowledge intuition is only one interpretation. But it’s the one that’s most relevant to the knowledge argument, I think.

    Okay, some more brief thoughts:

    Could someone please clarify why warrant/justification would be lacking in the cases under discussion? If it’s because of the cosmic accident part, then why not switch to neurosurgery cases? Is there some other reason?

    Pete: the answer to your question in #3 is yes. The Mary case seems to support SNKI. The (ideal, primary) conceivability of zombies definitely supports SNKI. But neither argument relies on KI+.

    What are the proposals about p-concepts that ~KI+ refutes? Are they included in the ones Unger discusses in his 1966 paper, or more recent proposals?

    Btw, I didn’t mean to suggest that ~KI+ is uninteresting. On the contrary, I’m finishing up a paper on a closely related topic (”deviant phenomenal knowledge”). All comments would be welcome. Here’s the url:

    Again, sorry for my slow blogging.

  14. Pete Mandik says:

    Hi Torin,

    Thanks for the link to your paper. I haven’t finished reading it yet, but so far I’m enjoying it. Perhaps I’ll whip up some comments that will be of use to you.

    I’ll leave it to Nick to clarify the point about warrant, I really have no further ideas about the remark.

    Re: what p-concepts proposals ~KI+ would threatened, people I had in mind were Papineau, Chalmers, and Tye. I wasn’t at all aware of the Unger paper.

    Here’s my (perhaps not fully-baked) take on the relationship between deducibility, “deviant phenomenal knowledge”, and the knowledge argument: If deviant phenomenal knowledge is indeed possible, then there’s no logical barrier to pre-release Mary getting into a state of knowing what it is like. Even if she cannot deductively infer WIL from her physical knowledge (itself a questionable supposition), there may nonetheless be various ways (other than surgery and swamp-lightning) to get herself in the appropriate state, ways that merit being described as “figuring out” what it is like. Depending on various empirical proposals regarding concept acquisition and cognitive architecture, there may be all sorts of epistemic entailments from the physical to the phenomenal.

  15. Torin Alter says:

    Hi Pete,

    I just read your Dec. 7 entry on your blog. (I’d have checked sooner but I’ve been in grading hell, from which I will emerge later today.) Thanks.

    I’d appreciate any comments on my paper (from you or others). Here’s the url again:

    The paper’s late for a volume called “The Case for Qualia” (ed. E. Wright, MIT). I got an extension until Christmas to revise it, so the sooner the better. Glad you’re enjoying the part you’ve gotten to.

    The Unger article is terrific. It’s written a bit combatively, but it’s clear and compelling. It doesn’t read like a 40-year old paper.

    I don’t quite see how ~KI+ creates problems for Papineau and Chalmers (or for any version of the “quotational” or constitution view), as long as we distinguish direct and standing p. concepts. Tye’s account may be threatened by ~KI+; at least, some of his remarks in his 1995 book seem inconsistent with ~KI+. Even so, I’m not sure how deep this cuts. I don’t see why Tye couldn’t modify his view accordingly.

    I’m interested in your ideas about how Mary could (perhaps non-deductively) figure out what color experiences are like from within the room. I’m also interested in how the empirical proposals you allude to might affect this. Perhaps this will come out in comments on my paper, but in any case I’m interested. Any more you can say about this I’m sure I’ll find interesting.

    Here’s the dilemma for the sort of moves you may have in mind. If the states Mary puts herself in–the ones *from which* she deduces what it’s like–involve some sort of color phenomenology, then she’s not really figuring out the phenomenology from only the discursive information conveyed by the science lectures. In that case, she’s cheating (i.e., this irrelevant to the claim that she can’t reason from the complete physical truth to what it’s like to see red). But if the states she puts herself in don’t involve color phenomenology, then it’s hard to see how she will be able to figure out the phenomenology. Anyway, that’s the challenge.