Wanted: An Actual Argument for the Knowledge Intuition


The Gates of Central Park

Originally uploaded by Pete Mandik.

The Knowledge Intuition is what makes so much of the philosophy of mind go ’round. The Knowledge Intuition is this:

(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.

Consider, now, the following Bold Claim:

(BC): No one has ever given an argument for the Knowledge Intuition.

Before considering the truth of BC, let’s talk a bit about its alleged boldness. On the one hand, maybe it isn’t so bold. The Knowledge Intuition is an intuition after all, and nowadays calling something an intuition is a gentle way of admitting to yourself and others that you don’t have an argument for this thing you really want to believe anyway. On the other hand, however, BC is a non-analytic negative existential and those are notoriously difficult to prove short of an exhaustive search of the entirety of creation. And this leads us now to the question of the truth of BC…

Maybe someone somewhere has an argument for KI and thus a counterexample to BC. And maybe they could post it here on Brain Hammer as a comment to this post. I’m willing to be pretty liberal as to what counts as an argument for KI. Maybe it would be entailed by some theory of consciousness that has independent support. Or maybe someone has a weird argument for how, even though KI, strictly speaking, is supported by no argument, it merits acceptance due to the health-conducive effects of believing in it. Or something. Note, however, that mere assertions of the alleged intuitiveness or obviousness of KI are not welcome.

31 Responses to “Wanted: An Actual Argument for the Knowledge Intuition”

  1. Pete,
    Can you try to explicate further what would count as knowledge of what is it like to have experience of certain type?

    The idea I get by reading BC is that it would require that one needs to be able to imagine what is it like to have experience of certain type (without previously having that experience). Would this be fair formulation, or you had something other in mind by “knowledge of what is it like to have an experience of certain type”? (e.g. maybe possibility to be able to form judgments about that type of experience)

  2. Pete Mandik says:

    Tanasije,

    Good questions, thanks.

    I’m happy to leave it pretty open as to what the relevant notions of knowledge are in “knowledge of what it is like to have an experience of a certain type”. I am thus happy to make it quite easy for the lover of KI to come up with an argument. Maybe the knowledge in question is a kind of know how or ability and thus involves the ability of being able to imagine something or other. Or maybe the knowledge in question is propositional knowledge and further has belief as an analytic component so that knowing what it is like requires having a justified or reliably formed true belief that what it is like is such-and-such.

    So defeating BC ought to be pretty easy. All that is needed is a non-question-begging argument that in some sense or other of “knowledge”, “experience”, and “what it is like” one needs to actually have an experience of a certain type in order to know what it is like to have an experience of that type.

  3. Paul Gowder says:

    This is less an argument and more a gesture in the direction of an argument, but what about something following from the subjective, first-person character of experience? Like (purely a stab in the dark):

    p1: All experiences have subjective aspects (or even “an experience is distinguished from other kinds of knowledge by its subjective aspects”) (qualia?)
    p2: To know what an experience is like is to know its subjective aspects (sort of follows from p1?)
    p3: Subjective aspects of experience can’t be communicated, or at least, can’t be fully communicated, through language (sort of follows from their subjectivity?)
    p4: Language and experience are the only ways to gain knowledge, in the relevant sense, of a thing. (ruling out introspection I suppose, but I’m sure that someone who knows more epistemology and/or philosophy of mind than me can defend that)
    c: experience is the only way to gain knowledge of experience.

  4. Nick Treanor says:

    Pete, as a minor point, I think your wording of the KI might be a bit strong, at least if types are finely individuated. Think of the unseen shade of blue: a proponent of the knowledge argument wouldn’t have to hold that it is impossible for someone who hasn’t seen some particular shade of blue to know what it would be like to see that shade of blue. (She could maintain this, of course, but it doesn’t seem she is required too.) The KI seems to be something more like: a person cannot know what it is like to have an experience of a certain type unless she has had an experience of that type or of a relevantly similar type (where lord knows exactly how that is to be cashed out).

  5. Chase Wrenn says:

    An Argument for KI

    1. ‘I know what it is like to have an experience of kind K’ means, “I have veridical episodic memories that include experiences of kind K.”
    2. It is impossible to have a veridical episodic memory that includes an experience of kind K unless you have had an experience of kind K.
    3. So, ‘I know what it is like to have an experience of kind K’ is true only if the speaker has had an experience of kind K.
    4. So, a person cannot know what it is like to have an experience of kind K unless she has had an experience of kind K.

    A note on premise 2:
    If one has an episodic memory that includes an experience of kind K, but one has never had such an experience, then one’s memory is not *veridical*. So, 2 must be true.

    Another note on premises 1 and 2:
    You might weaken 1 by taking out the veridicality requirement, and take it out of 2 as well. 2 would still be defensible. If I have an episodic memory of having had an experience of kind K, then the episodic nature of the memory requires that such an experience actually be part of it, even if the rest of the content of the memory is false. So, I can’t have an episodic memory of having had an experience of kind K unless I have had an experience of kind K, though the experience might not be distinct from the memory.

  6. How about argument by example?

    The example: Blind people don’t know what it’s like to see red. The best explanation for this is that they have never experienced red. If so, this offers some support for the generalization that in general people don’t know what it’s like to have an experience unless they’ve had an experience of that type.

    Options: (1.) Insist that blind people do know what it’s like to see red. (2.) Deny that the best explanation of this is that they lack experience of red. (3.) Deny that this instance supports the generalization.

    Good luck!

  7. djc says:

    I think that KI is the sort of claim that is less often asserted than meta-asserted: that is, people assert that other people assert it! Personally I think it is false (at least when “KWIL” is interpreted in the most common way). Most obviously for missing-shade-of-blue reasons, as Nick suggests, but also for brain surgery reasons (take someone with KWIL without currently having the experience, and put someone into a relevantly similar internal state e.g. via brain surgery). Correspondingly I think it’s false that KI is what makes “so much of the philosophy of mind go round”. Certainly e.g. the knowledge argument doesn’t depend on this claim.

  8. Nick Treanor says:

    I think it’s possible to build on Eric’s argument, where the inductive ground is our own experience. The following seems true, for instance:

    (x) (x is knowledge I have of what it is like to undergo a certain experience > x is knowledge I gained by undergoing the experience in question)

    This might be taken as offering reason to think I couldn’t gain knowledge of what it is like to undergo a certain experience without having an experience of that type. (The idea is to invoke a premise similar to Eric’s number 2, to the effect that the best explanation of why all the knowledge I have of what it is like to undergo an experience is knowledge I gained by undergoing the experience is a modal claim to the effect that this is the only way to gain such knowledge.) I confess however that I find such arguments quite unsatisfying.

  9. Pete Mandik says:

    Hi Paul,

    I think I would count that as an argument for KI as opposed to a mere assertion of KI. However, I suspect that once one tries to defend its premises, or even explicate what some of its key terms mean, we’d see that we’re in a tight little circle that has the assumption of KI as a major link. Take, for instance, “subjective character of experience”. Does it mean something other than “knowable only by having experiences with said character”?

  10. Pete Mandik says:

    Heya Chase,

    I’d count that as an argument for KI but I’m not prepared yet to count it as something I fully understand. I’m a bit puzzled by this stuff about inclusion and also what you think “episodic memories” are.

    I think so-called episodic memories are just memories of events. And of course, I can only have veridical memories of events that actually happened. So, for example, I can only have a veridical memory of having pet a puppy on my lunch break if I have in fact pet a puppy on my lunch break. But this doesn’t look like anything remotely near the sort of thing that would make the philosophy of mind go round.

    This leads me to think that whatever you mean by “include” in premise 1 couldn’t be captured by simply replacing it with “are of” or “are about”. So maybe by “include” you means something like “include”? Like, as you say, “actually be a part of”? Unless you are talking about temporal parts of 4-d spacetime worms this strikes me as a weird thing to say about episodic memory, since I guess it would entail, among other things, that every time I episodically remember having an orgasm I have another orgasm.

  11. Pete Mandik says:

    Hi Eric,

    I think that’s pretty nice and it’s hard to deny that the example lends some support to the generalization. But a lot more needs to be done before we’ve come anywhere near establishing a pretty strong modal claim. One will need to address, for example, just how significant potentially relevant counterexamples are (like that I can know what a 19-sided figure looks like even though I’ve never seen one).

  12. Pete Mandik says:

    Hi Dave,

    I’m somewhat surprised by your surgery remark. Do you think KWIL could be surgically implanted into your zombie twin?

    Re: KI, KA and the World. I’m sure you’re right that one could run the knowledge argument while denying the KI as I’ve formulated it, but I’d still bet that a lot of the philosophical uptake of the argument is due to acceptance of KI or something pretty darn close to it.

  13. Pete Mandik says:

    Hi Nick,

    Re: your first comment on types. Points taken. Especially important is your closing parenthetical remark. More on this below.

    Re: your second comment, on Eric’s argument. That universally quantified conditional you offer may seem true to you, but it doesn’t work for me. As I said in response to Eric, I can know what it’s like to see a 19-sided figure without having seen one. Even granting its truth, I share your lack of satisfaction with such arguments. I’d add that I think its pretty rare that someone’s reason for believing something is simply an induction, and I think it’s even rarer when the belief in question is a modal claim. As for inferences to the best explanation, I don’t think the best explanation of anything is ever something smaller than a pretty big theory. I worry that sometimes people think of inferring to the best explanation as something you can do in a single step of a proof. (Not that I’m accusing you or Eric of doing that!)

  14. Nick Treanor says:

    Pete, I’m not convinced your claim about what it is like to see a 19-sided figure is a counterexample. Here’s what is worrying me:

    First, are you really confident that you know what it’s like to see a 19-sided figure? To me this is a bit like saying: “I know what it’s like to see a hen with 48 speckles”. Well, not really. I know what it’s like to see a hen with three speckles, or six. But 48?. (Or consider Descartes discussion of the difference, or lack thereof, between ‘mentally seeing’ a chiliagon and a myriagon.)

    Second, setting aside the first worry, suppose you do in fact know what it is like to see a 19-sided figure. One might think you know this by an act of mental imagination that is a genuine case of ‘experiencing what it is like to see a 19-sided figure.’ (You picture the figure to yourself, as it were.) You might object: “but that’s not really _seeing_, I’m managed to know what it is like to have an experience of seeing a 19-sided figure without actually having an experience of seeing a 19-sided figure.” But that is only an objection to the KI if you sneak too much into the meaning of the word ‘experience’ in setting up the definition. For instance, could someone defeat the KI just by saying: “surely you know what it is like to have an experience of seeing an object that looks exactly like a horse but is made of straw, even though you have never had an experience of seeing an object that looks exactly like a horse but is made of straw.” This is no objection because although I have not had an experience of an object that looks exactly like a horse but is made of straw, I have had an experience that is relevantly similar — I’ve experienced horses. Similarly, can’t a proponent of the KI say: you’ve never experienced a physical, 19-sided figure, but you have had an experience that is relevantly similar — you’ve ‘pictured to yourself’, as it were, such a figure, and that experience is qualitatively similar to the experience you would have if you saw a 19-sided figure (and, moreover, your judgment that you know what it is like to see a 19-sided figure is based on your taking it to be the case that the experience you have had is relevantly similar.)

    (as with my earlier comment, lord knows how relevantly similar is to be analysed)

  15. Chase Wrenn says:

    Pete-

    Yes. By ‘include’ I mean “include.” What I have in mind is that episodic memories are (retrievable by means of) the reactivation, offline, of perceptual representations that were activated in the past. The argument I made claims that a person who says “I know what it is like to have an experience of kind K” means that she is able to generate offline a copy of an experience of that kind.

    This doesn’t mean that episodic memories of orgasms are orgasms. It does mean that having an episodic memory of an orgasm involves reactivating the mental represenations involved in the original event. The offline representation isn’t as (dare I say it?) vivid as the original, but that’s all right. I’m happy saying that I only *sort of* know what it’s like to have an orgasm when I’m not actually having one. I’m also happy saying that remembering an orgasm feels a lot like having an orgasm.

  16. djc says:

    Pete — No, I don’t think a one could give a (paradigmatic) zombie KWIL via brain surgery. That claim is irrelevant to the counterexample to KI, though, which involved a normal human being.

  17. Pete Mandik says:

    Dave, I’m still a bit puzzled. To spell this out a bit more…

    I gather that you don’t think that zombies have KWIL because their beliefs are not appropriately connected to qualia which, in turn, is due to their not having any qualia.

    Suppose that pre-red Mary has a (non-zombie) doppleganger that gets surgically altered to have KWIL by being altered to physically resemble post-red Mary without herself (the doppleganger) having had any red qualia. I’m assuming that post-red Mary and post-surgical Mary are alike with respect to their beliefs except for the fact that the post-surgical doppleganger’s beliefs are not appropriately connected to red qualia (since the doppleganger lacks red qualia).

    I hope this spells out why I don’t see how zombies can lack KWIL while there can be surgically induced counterexamples to KI. Where have I gone wrong?

  18. Pete Mandik says:

    Chase,

    Using Chase-ish lingo, we can construct the following claim:

    (CW): One cannot generate off-line a replica of an experience of a certain type unless one has had an on-line experience of that type.

    Two questions:
    (1) Are there any actual arguments for CW?
    (2) Isn’t CW pretty much just KI translated into Chase-ish lingo?

  19. djc says:

    Pete — I think it’s plausible that there’s some dependence of KWIL on experience, so that a being that lacks experience entirely couldn’t have KWIL (perhaps because some experience is required to have phenomenal concepts). But that’s a far weaker claim than KI. Well, I suppose it is an instance of KI where the type in question is “consciousness” — but it doesn’t require KI for specific types such as phenomenal redness, which I take it is what you have in mind, and is what post-surgical Mary is a counterexample too. So I don’t think one can move from the possibility of post-surgical Mary with KWIL to the possibility of a zombie with KWIL. Any argument that tried to do so would require some further strong premises that I’d certainly reject.

  20. Pete Mandik says:

    Nick,

    Your point about the straw horse is well taken. For such reasons, I wouldn’t offer a chilliagon type of example as a compelling counter-example (although, I have surveyed my undergraduates and they are confident that they know what it would be like to see a chilliagon even though they’ve never seen one). Chilliagons surely look just like circles to normal observers and however one resolves the experience-typing problem, visual experiences of circles and visual experiences of chilliagons are of the same type.

    Let’s reduce the number of sides of the polygon in question, then. Unlike chilliagons, polygons with different numbers of sides but with sides less than ten are usually visually distinct, and thus give us some relevant examples of the type differences we need. Suppose Mary had seen polygons with sides of 3, 4, 5, and 6 and also polygons of sides 8, 9, and 10. But no 7-sided figures. Is she going to be surprised when she sees her first septagon?

    I doubt adeherents of KI are going to say she’ll be surprised and I bet lots of them will say the sort of thing you suggest: While no actual septagons ever triggered a septagon experience in Mary prior to this day, through imagination she was able to induce in herself a septagon experience. They will say further that one can only know what it is like to have septagon experiences if one has actually had septagon experiences (and imagining seeing septagons counts as having a septagon experiences).

    Of course, if someone believed in KI, then they would say all of those things. But what actual arguments are there in support of those things?

  21. Nick says:

    Pete,

    suppose someone claims to know what it is like to experience a septagon, and also claims never to have seen a real septagon. The KI adherent will say the person in question has imagined seeing one, and that this experience is relevantly similar to (or is) the experiencing of a septagon. Would you want to challenge the KI adherent on one or both of these points:

    1. the subject in question imagined seeing a septagon
    2. imagining seeing a septagon is relevantly similar to (or is) experiencing a septagon

  22. Pete Mandik says:

    Hi Nick,

    I think this is a nice way of focusing the discussion.

    Re 1 and 2, I think both are questionable, but I’m happy to accept 2. I challenge 1.

  23. Chase Wrenn says:

    Answers to your questions, Pete:

    1. CW is analytic, since there can be no replicas without originals. (GHOST IN THE SHELL: STAND ALONE COMPLEX to the contrary notwithstanding.) That’s the argument.

    Below, I’ll describe how to argue for KI without CW.

    2. CW isn’t equivalent to KI. It implies KI with the further premise that ‘I know what it’s like to …’ means ‘I can token a replica of an experience of …’. You might say THAT’S just KI in new clothes, and you might be right.

  24. Nick says:

    Pete,

    okay, I had thought you might find 1 more plausible than 2.

    Two questions (I’m making this up on the spot here, so bear with me if these lines of argument goes nowhere):

    1. Suppose someone says: “Cool, so you know what it is like to have septagon-experience. Tell me about it.” Do you think you could report your knowledge without imagining a septagon?

    2. Suppose someone says: “Cool. Now imagine, in the sense of ‘picture to yourself’, a septagon. What is the difference between being in the state of imagining a septagon and being in the state of merely knowing what it is like to have a septagon-experience (without actually having such an experience)?” How do you answer?

    (I’m a bit unhappy with this example, because it is not clear to me that there’s anything it’s like to have septagon experience, since septagons could presumably be of a zillion different shapes and sizes. Let’s suppose that we’re talking about a regular septagon.)

  25. It seems to me that the notion of “relevantly similar type” that Nick had mentioned (or something similar) needs to be cached out, as without that the whole notion seems to become ridiculous. Here are some examples:
    Would I notice distinction between 19 sided regular polygon and 20 sided regular polygon? I might not notice any difference between both, even if they are both before me. With training I might learn to differentiate both, but then we would have “what it is like to have an experience of 19 sided polygon after undergoing specific training to recognize them” vs. “what it is like to have an experience of 19 sided polygon without such training”. Then rotation would probably affect how we experience the polygon (e.g. if we experience it as resting on one of its sides, or on one of its angles). Further the polygon can come in lot of colors, can be seen at different distance, and so on.
    If there is no further specification to KI, one can also talk about what it is like to have an experience of 19 sided polygon when one has had eggs and bacon for breakfast.

  26. Thanks for your reply! I agree some nuance is needed here. To be plausible the principle must be restricted to avoid the missing-shade-of-blue and 19-sided polygon cases. I’d say the answer hinges on a broad reading of “type” at the end of KI.

  27. Pete Mandik says:

    Chase,

    “replica” was poor word choice on my part. I meant to say “GITS:SAC-replica”

    Another thing: I’d like to talk about orgasms again. The view you are entertaining seems to commit one to holding that knowing what it is like to have an orgasm at time t entails having at time t an occurrent conscious mental state that, while it may or may not be an orgasm, has some things in common with orgasms. Besides raising probelms with the “something in common” part of the claim, one can question the “occurent mental state” part of the claim. Consider:

    One frequently can know that P at time t without at time t entertaining any conscious occurrent mental states with the propositional content that P. Consider: you’ve known for a long time, including a moment just 30 seconds ago, call it t, the following fact: Bechtel was my advisor. But until now you probably didn’t have an occurrent conscious mental state with the content that Bechtel was my advisor. Whatever knowing what it is like to have an orgasm is, it shares this in common with knowing who my advisor was: you don’t have to be currently consciously entertaining anything about the thing that you know to know it.

  28. Pete Mandik says:

    Nick,

    These are useful questions. Here are my answers:

    1. Yes

    2. I’m happy to suppose that imagination is off-line perception. So, in perceiving a puppy, there’s a puppy before my and it causes some stuff to light up in my head that carries information about the puppy (call that stuff “sensations”) which in turn trigger in me an application of some concepts like a puppy concept. Imagining puppies involves using my puppy concept to trigger some of those sensations without there being an actual puppy in the vicinity. Mental states that involve just the concepts without any online or offline triggering of sensations include knowledge and standing beliefs.

    2. cont’d. States of imagining are occurent conscious mental states. Consciously perceiving that P is also an occurrent conscious mental state. Knowing at time t that P infrequently involves having at time t any occurrent conscious states with the content that P. Consider: you’ve known your own phone number for a long time. But you are only now consciously thinking that your phone number is blankity-blank.

    2. cont’d even further: knowing what it’s like is similar. I’ve known what it’s like to taste vegemite for many years, but up until now, I hadn’t been thinking about it much or imagining it. Whatever knowing what it is like consists in, it’s what knowledge in general consists in: having a bunch of concepts and being able to apply them in appropriate ways when needed. Having the knowledge does not consist in the current offline activation of related sensations.

  29. Nick Treanor says:

    Pete,

    thanks for getting back to me on the questions. I quite agree that to know something (my phone number, or what vegemite tastes like) one need not be consciously thinking of the knowledge content. I suppose this fact alone is all that’s needed to defeat the KI, for the reasons Dave noted — it seems reasonable to think that knowledge of what it is like supervenes on some physical property, and a subject could be brought to have that property via surgery, without ever having had the experience. (I’m not sure this would count as _knowledge_, rather than mere _true belief_, but that is a complicated epistemic question that might not be important to the phil of mind issues.)

  30. patrick says:

    i think the (ki) seems to be so ’spooky’ because no one has been able to prove it = (bc).

    but how to prove an intuition?? perhaps to really PROVE something you have to ‘reduce’ it in an analytical way like ‘2+2=4′ or ‘water is h2o’ or ‘if a then b’.

    given the case that the (ki) can really be proved, it cannot be longer an intuition or a non-reductive knowledge or a quale or a raw feel, it has to have the property to be ‘translatable’ into the language that proofs use, namely the language of science (in a broad way). this language works with functions, structures, and so on.

    thus if you can prove (bc) the (ki) it (the ki) cannot be any longer an intuition or a non-reductive item of the world.

  31. [...] 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. [...]