Remembering, Gettier, and Fake Barns

I’ve argued in a recent paper (see here or here) that remembering that p entails knowing that p.  If this is so, then remembering that p should not be present in Gettier cases, which is what we find.  Here is an example I use:

“I take my students by a field where I know that there are no sheep, but there is a sheep-shaped rock that, from the road, looks exactly like a real sheep.  I also know that there is a lone sheep on the far side of the field, well out of sight.  I see that one of my students has caught her eye on the sheep-shaped rock, and I hear her say to herself, ‘What a happy looking sheep in that field!’ I snicker and say quietly to myself, ‘She doesn’t know that there’s a sheep in that field.’ Later, when we arrive at our destination, I ask, ‘Does anybody remember whether there were any sheep in the field we drove by?’ The same student says, ‘Yeah, I remember that there was a sheep in the field.’ I judge her and say quietly to myself, ‘She doesn’t remember that there was a sheep in the field.’ In this scenario, my final statement seems true, despite the fact that she has a justified, true belief that there was a sheep in the field.  It is by sheer accident that she has a true belief.”

I argue this against claims/arguments made by Sven Bernecker in his recent book (2010, p. 74), who thinks both that remembering does not entail knowing and also that one can remember in Gettier cases.

I’m interested in whether we remember in Ginet’s old fake barn cases.  Suppose I am in fake barn country and I happen to only walk by the one real barn in that country.  Later, I have the belief that that was a nice barn.  Do I remember that that was a nice barn?

I’m a little more inclined to say “yes” than in my above fake sheep Gettier case.  But I’m also inclined to think that I also knew that that was a nice barn.  Recently, probably due to protests by philosophers like Keith DeRose, Ruth Millikan, and others, there has been more skepticism about the lack of knowledge in fake barn cases.

Personally, I’m mixed, and my intuitions go both ways.  Here’s my hope.  Those who have the intuition that there is no knowledge will also have the intuition that there is no remembering.  And those who have the intuition that there is knowledge will also have the intuition that there is remembering.  I think that is so for me.  I ask Certain Doubts readers, is that the case for you?

(Thanks to Dustin Locke for first raising the fake barn version of the case to me.)


Remembering, Gettier, and Fake Barns — 21 Comments

  1. Here’s my first reaction. (For the record, I’m on Sven’s, rather than on your, side that remembering that p doesn’t entail knowing that p. I also think that people in fake barn country don’t know that there are barns.) I think it is true that the person remembers seeing a barn but that if she were to say “I remember seeing that barn” she’d say something at least misleading because the implicature is that she knew she saw a barn, and she didn’t’ know any such thing. Yet another reason we shouldn’t pay too much attention to what it would be natural to say when we construct philosophical theories. (I’ve been saying this a lot lately!).

    • thanks for the comment, Tom. Just to clarify, I want to distinguish:
      1) S remembers that p
      2) S remembers X (where ‘X’ denotes an event).

      These have different truth conditions. The example I use in my paper is that my students, on their test, might remember _that Socrates was Plato’s teacher_, but they will never remember _Socrates’ being Plato’s teacher_. For statements like (2) to be true, S must’ve experienced the event in some relevant way.

      “I remember seeing that barn” would be more like (2) than (1). Do you think that the person remembers that there was a barn? I hope you will say “no”, since you think that the person doesn’t know that there was a barn.

      Also, out of curiosity, do you think my student remembers _that there was a sheep_? I think we both agree that she doesn’t know that there was a sheep.

      • Andrew,

        I like your distinction between 1 and 2. And I do think that in your example, it is even more plausible to say that the person remembers the barn as opposed to remembering that there was a red barn. But I guess I want to say that she remembers both. If she is thinking about the ride she took in fake barn country (when she doesn’t know she was in fake barn country) and she thinks “That barn was red” when she really did see a red barn, I think she is remembering seeing a red barn even though she never knew she was seeing a red barn. Again, if she utters in a standard context “I remember seeing a red barn” I think she says something infelicitous or maybe even false (because I think at least conversational implicature indicates that an utterance of “I remember that p” indicates that the speaker knows that p. But I take epistemology to be the enterprise of the evaluation of beliefs (or maybe propositions) for a person and I don’t see much of a connection between that and whatever rules there may be for felicitous assertion.

        • Tom,
          I am quite open to agreeing that Sally remembered seeing a red barn. I also think that first-person cases are tricky because of conversational implicatures, as you mentioned. I think that what is more reliable are third-person cases. And our disagreement comes down to whether remembering-that-p is present, or whether

          “Sally remembered that there was a red barn”

          When I’m in the mood to think that Sally didn’t know that there was a red barn, I’m also inclined to think that Sally also didn’t remember that there was a red barn. That is where we differ.

          But I want to try to get my foot in the door. What do you think about whether my student remembered that there was a sheep? There, I even used a 3rd-person utterance, and I have a clearer intuition that the student didn’t remember so.

          • Andrew, thanks for the dialogue and pushing me on this. I’m starting a book on this stuff that I plan to write most of next year on sabbatical so it is good for me to be thinking about this now. That said, I’m really busy and so probably won’t be able to do justice to this here.

            I’m not sure this is defensible but here is aline I’m really tempted by: one can remember with either of the two relata you distinguished above only if one’s current memory is appropriately causally connected with the relevant thing (object or fact, I’d suppose) or one knows. So in your case, your student doesn’t remember the sheep. If she had seen the sheep but now has a defeater of some type, but still recalls the sheep (or that it was white) then she remembers. In the second case that the speaker at the APA session discussed–Sven’s historical belief example–I found myself torn. I don’t think remembering entails knowledge but my intuitions weren’t clearly with Sven (although they weren’t clearly with the speaker either). Once I changed the case to being grounded in episodic memory, it was clear to me that remembering was happening without knowledge.

    • AM wrote: “I argue this against claims/arguments made by Sven Bernecker in his recent book (2010, p. 74), who thinks both that remembering does not entail knowing and also that one can remember in Gettier cases.”

      “Moreover, Bernecker shows that memory doesn’t require identity, but only similarity, of past and present attitudes and contents. The notion of content similarity is explicated in terms of the entailment relation.”

      The old idea of memory was that it worked like a snapshot and that the details of that memory were fairly static. The new idea of memory is that it is broken down into core content and distributed throughout different areas of the brain. So that we remember something, we are creating a new memory each time which is reconstituted with different emphasis and content. So that our memory of an event of say High School graduation is not the same of a picture taken of the event. The details in one recalled memory, M_1 will likely be similar to memory M_2 but not identical. Not identical means evidence to base beliefs is not distributed among all the generated memory copies. The (perhaps justified) beliefs based on the evidence found in different versions of the original memory event can vary as much as when a group of eyewitnesses report to police what really happened at a bank robbery; usually similar, not identical, and sometimes contradictory. My intuition about knowledge is that it should have a firmer foundation than memory is capable of providing, since one doesn’t know which memory version has the blurry (only approximate) details and where the discrepancies will occur. I don’t think your disagreement with Bernecker hits this nail squarely on the head.

      • Hi Stephen,
        Thanks for the note. You said, “I don’t think your disagreement with Bernecker hits this nail squarely on the head.”

        It’s because our disagreement is a different topic from your nail. So it naturally won’t hit it. There’re lots of interesting memory issues, including some raised by your comment.

        • “I’ve argued in a recent paper (see here or here) that remembering that p entails knowing that p.”

          SH: Does the claim above require that the claim below, that the evidence thesis is false, for consistency?

          “At the end of section 2, I consider the primary thesis of my paper – that the evidence thesis is false – to be successfully defended.”

          SH: I’m trying to get a handle on the depth of your disagreement with Bernecker, who wrote,

          “Internalism and externalism are reconciled at the price of creating a Gettier problem at the level of “reflective” or second-order knowledge. The general lesson to be learned from the critique of virtue perspectivism is that internalism and externalism cannot be combined by bifurcating justification and knowledge into an object-level and a meta-level and assigning externalism and internalism to different levels.”

          • Hi Stephen,
            Whoa, are you quoting from my “Knowing Without Evidence” paper? Well, I do think that the evidence thesis, as defined there, is false. I don’t see any interesting connections between the evidence thesis and the thesis that remembering that p entails knowing that p.

            As for the depth of our disagreement, see my paper, “Remembering Entails Knowing”, available here:

            There are a number of claims and arguments we disagree on. I think we are both inclined toward externalism though.

  2. Hi Andrew (and Thomas),

    Andrew, I really regret that I couldn’t have stayed longer at the APA as I had hoped to talk to you about some of these issues.

    I think that ‘S remembers that’ stands or falls with ‘S knows that’. A few points that I plan on defending elsewhere in much more detail.

    First point. I don’t think we can set issues about the way we use language aside. Consider these two hypotheses:

    H1: S sees that p only if S knows that p.
    H2: S remembers that p only if S knows that p.

    What do we take defenders of H1 and H2 to be defending, precisely? I wouldn’t defend H1 as a theory about what’s involved in vision. I take the truth in H1 to be about ascriptions of the form ‘S sees that p’. The truth in H1 is that this is what English speakers use to report knowledge and its source. The same holds true for H2. I don’t take the truth of H2 to be some important or surprising fact about memory, exactly, I take the truth about H2 to pertain to something about the language–we say that ‘S remembers that p’ to specify a ground or source for knowledge. This matters for two reasons. Think about H1. One source of resistance to it in the literature comes from people who have the intuition that subjects will stand in the same visual relations to their surroundings in the good case and cases of environmental luck. That’s true, but if the truth contained in H1 isn’t really about visual relations but is instead about how language works, these intuitions are the wrong things to bring to bear when trying to decide whether H1 is true. The same holds for H2.

    Second point. The linguistic data might not be decisive, but it’s at least one data point. One point where I think Sven and I disagree (and I think this might be true of Thomas and myself, but I don’t know yet) is on what the data is. I take the following ascription to be problematic:

    (a) She remembered that her mother was born in June, but she never knew when her mother was born.

    I have a very simple explanation as to why this is problematic: H2 is true.

    At this point, those who wish to deny H2 can either say that they don’t have the intuition that (a) is problematic or they can try to explain that intuition away.

    Third point. I think people need to be very careful in how they report their intuitions. Just as we can know facts, persons, places, etc., we can remember facts, persons, places, etc. The issue isn’t about whether someone remembers a barn, remembers a dance, remembers a city, etc. The issue is about whether someone remembers that something is so and whether they can remember whether something is so without knowing that it is so. I think some intuitions that seem to cause trouble for H2 only do so because people aren’t as careful as they should be about describing their cases. (This comes up a lot in the literature on seeing. People gin up intuitions to cause trouble for H1 by describing cases in which someone sees an object and takes that as evidence that they can see that something is true. I’d look at Craig French’s papers on this.)

    Fourth point. I _think_ people agree to the following: in cases of intervening luck where the subject believes p and p happens to be true, the subject will not see that p, remember that p, etc… Intervening luck doesn’t just defeat knowledge, it prevents us from having factive attitudes towards p. If that’s right, they then have to say whether they think there are other cases of remembering that p without knowing that p and provide some understanding of what’s involved in remembering that p that allows for the possibility of remembering that p without knowing that p. It’s hard to see what this account might be.

    The account cannot be this: to have factive attitude FA towards p, one must have a presentational attitude towards p in a p-world. That won’t rule out the cases of accidental connection that we have in veritic luck cases. It has to be that the fact and attitude are related ‘in the appropriate’ way. It’s sort of hard to see how it could be related ‘in the appropriate’ way without putting the subject in a position to know.

    Fifth point. People assume too readily, I think, that there are ways of being in touch with the facts cognitively that doesn’t involve something belief like. I think there’s a reason why seeing that p isn’t really a matter of standing in a visual relation to a fact. First, you might think that facts have constituents that are general and vision always relates us to the particular. Second, you might think that vision relates us to things in our surroundings that have locations. Facts don’t have that. Are things different with remembering that something is so? It’s hard to say. My guess is that it’s only by means of the intellect that we ever get in touch with the facts and that this requires something belief-like. The critics of H2 can either challenge this point or explain how belief gets involved in remembering that p. I think that some people want remembering that p to be something that can explain why you believe p, so if that’s the way they want to go, they need to help us see how we can be in touch with the facts in a non-accidental way without believing these facts to obtain. I seriously doubt that they can do this, but someone might offer an account.

    Sixth point. Why think that the difference between intervening luck and environmental luck is deep? I sort of suspect that there’s no real difference here at all. When people try to explain the difference, they’ll talk about something that gets ‘in between’ a person and a fact. The trouble with this is that there’s nothing that could get ‘in between’ a person and a fact if this is understood spatially as facts don’t have spatial location. Sometimes people say that the difference has to do with a false representation. The thought seems to be that intervening luck cases involve a false representation and environmental luck cases do not. This doesn’t seem very persuasive. It’s not clear that there _has_ to be a false representation in the paradigm cases of intervening luck. It’s not clear that there _has_ to be no false representation in the cases of environmental luck. My guess is that the impression that these cases differ in some interesting way is a mistake. I could be wrong about that, but I haven’t yet seen a helpful way of distinguishing the cases.

    • wow, thanks for taking the time to write all that, Clayton! I’ll only comment on the parts I had something to say:

      On your (a), you seem to be attacking more the past-knowledge condition on remembering. The old theory (I think defended by Norman Malcolm?) was that

      S remembers that p at t2 IFF
      1) S knows that p at t2
      2) S knew that p at earlier time t1
      3) S’s knowledge that p at t2 and t1 are appropriately causally connected.

      I focused my paper on condition (1), and your (a) seems to apply more to condition (2). I’m not sure what I think about (2), although I know Bernecker attacks it.

      FWIW, Bernecker does deal with first person attributions “I remember that p but don’t know that p” in his book but explains away the intuition clash by way of conversational implicature. In my paper, I use third person cases like

      b) ‘Jordan remembered that Kristina’s birthday was on Monday, but he didn’t know that it was on Monday.’

      In conversation at the APA this past weekend, asked Bernecker what he thought of the third person cases, and he wasn’t convinced.

      See my reply to Tom, right above. Glad we agree!

  3. Hi Andrew,

    I don’t know if I buy Malcolm’s view. I don’t see why I couldn’t remember that p without having known p at some prior moment. What matters, I take it, to H2, is not whether someone has known p over some stretch of time but whether someone’s current knowledge that p has its source in memory. I think Peter Hacker has some helpful stuff on memory that’s worth looking at in his book, _The Intellectual Powers_ (or something along those lines).

    • Clayton,
      Right, I’m skeptical of it too. I think we’d need something more like my (b) to support H2.

      There’s a rich literature on memory in the 60s and 70s that seems to have been much forgotten (pun!). They explored potential examples and counterexamples to the various conditions in Malcolm’s theory. Don Locke’s short book “Memory” explores this stuff pretty well.

  4. I want to flesh out an argument briefly hinted at earlier. I’d be curious to know where people like Tom or Sven thinks the argument goes wrong.
    P1. If S remembers that p doesn’t entail that S knows p, there has to be a condition necessary for propositional knowledge that isn’t necessary for propositional memory.
    P2. The obvious candidates are these: belief, absence of environmental luck, absence of intervening luck.
    P3. It cannot be the absence of intervening luck because intervening luck prevents us from remembering that p. [Why? Think back to the Gettiered up Nozick case from above. The mere fact that the subject has a true belief in that case doesn’t mean that the subject can have factive attitudes with p as their object.]
    P4. It cannot be the belief condition. [Why? It seems very odd that someone could both remember that p and believe only p’s negation. (e.g., She thinks that it rained yesterday? Yes, she does, but she remembers that it didn’t rain yesterday). It also seems very odd that someone can be said to both remember that p if we deny that they are right about p. (e.g., Doesn’t she realize that it rained yesterday? No, she has no idea whether it rained yesterday. She merely remembers that it rained yesterday).]
    P5. It cannot be the environmental luck condition. [Why? This isn’t conclusive, but here’s a stab. If the subject was in an environmental luck case, she wouldn’t have seen that p. So, even if she had various sorts of experiences at the time, memory couldn’t keep her in touch with the fact that p by virtue of the way that experiences put her in touch with the fact that p. And if that’s so, it’s not as if a further belief would have put her in touch with the facts at the time. In the relevant cases, it’s only possible for memory to put a person in touch with the relevant facts if there’s another route by which they’d be put in touch with the facts, but there’s no such route. Memory cannot _create_ such routes, it can only preserve them.]
    C. Since there’s no obvious candidate for the necessary condition on propositional knowledge that isn’t met for propositional memory, we should conclude that propositional memory requires a set of conditions sufficient for propositional knowledge.

    Obviously, parts of this are hand wavy. I’d be curious to know where, precisely, critics of H2 think the argument goes wrong.

    • quick clarification question, Clayton. Intervening luck is the sort of luck in the sheep case, right? And environmental luck is the sort of luck in the fake barn case?

    • Thanks for the interaction on this, Clayton. I don’t think we’ve met, but I’m old and (ironically) my memory sucks so perhaps we have. If not, I hope our paths cross soon.

      My first response to your argument is to deny P4, and for the standard reason that dates back to Martin and Deutscher. One might have a clear memory impression that expresses P, but have lots of misleading evidence that not-P, and so come to believe that not-P, and take one’s memory to be unreliable. (E.g., Mary remembers clearly cousin Bill being at her 6th birthday party, but all of her friends and family tell her he wasn’t there and she comes to believe them.) In such a case, the person remembers P but doesn’t believe it.

      My second response is to wonder why you haven’t ruled out the justification condition. Take Mary’s case: she’s not justified in believing that cousin Bill was at her party even though she remembers that he was there. (And even though I’m not a fan of taking ordinary talk with much philosophical seriousness, doesn’t that previous sentence sound fine to you once your understand the context? )

      Again thanks for the fun and I apologize in advance if I don’t keep up!

  5. That’s right. I think they’re both supposed to be kinds of veritic luck and both thought to be epistemically malignant. I’m not quite so sure I think there are really two kinds of luck here, but that’s for another day.

  6. I am a bit unclear on your opinions on what I am going to suggest so any responses would be appreciated. I think the point that memory is not knowledge can be made much simpler without going into thought experiments. I’d also be very interested to hear your opinions on this.

    I am going to work on the basic premise that if S thinks that p is a case of knowledge then S is going to assent to it (because it is rational to believe in what you take to be true, admittedly this is somewhat similar to Kant’s claims in Groundwork I don’t know if this is a shared intuition). If we view knowledge as involving the cognitive attitude of assent, we can look at memory/perception and say do we assent to all our memorial beliefs? Presumably, if memory is just knowledge then we would assent to all our memories. For me, the answer is not that obvious that I do assent to all my memories and memories have a degree of doxastic voluntariness that is missing from knowledge (when you believe p is knowledge). Perception is probably the easiest example to start with to push the claim that I don’t constantly assent. So, I am walking down a corridor and I see someone who I think is my friend, but I’m not certain. Here, I am not really sure of anything and I am not really making any claim to knowledge. I have seen something, but I don’t know what. (Or to make the point even simpler, I think I have seen something out of the corner of my eye but I’m not really sure if I have seen something or if I am just tired and need to go to bed). Now consider a case where someone asks me the date of a party and I say, from memory, the party is on Friday and then I question this and say, is this really Friday? I check and find out the party is actually Saturday. If memory is just knowledge, then it is hard to explain how we can have (a) false information and (b) why we chose to distrust our own memory and the memory of other people. I would be interested to know what you think about this. (If this argument works, it seems much simpler than using barn cases etc.)

    Maybe the weakness in the argument outlined that what is it exactly that I am distrusting? Is it memory or something else. If we are going to work purely on intuitions/phenomenology then how are we to respond to cases where people say “I could have sworn that this was the case” it seems to them be a memory but if you are maintaining memory = knowledge you’d have to deny this and tell them they are mistaken and appeal to some theory of pseudo-memories.

    Connected to the above, what I think makes this difficult to answer is that there is no obvious hallmark to where the content of a claim comes from. What I could be take to be a case of false remembering could actually just be a case of my imagination and p is not really a memorial belief. (This is why I find it difficult discussing memory in this context. It is not obvious if something is a memory or imagination and this ambiguity can affect the examples we use in assessing theories. Consider all the debate on Hume distinguishing things through vividness, most consider a distinction made through these qualities as untenable. How do we know this is a memory we are talking about?)

    Also, I find it interesting how you view what it is exactly memory gives us. What seems to me to be the intuitive way of memory is that it involves two stages. Memory gives us information, just like perception does, and then we take an attitude towards it. To me, it’s not obvious a memory belief is simultaneously information/content and knowledge and do you think maybe this is an assumption in your arguments? If there is a two-stage view of memory, then memory could both be instances of knowledge and not instances of knowledge – it just depends on the individual memory belief than memory as a faculty. (This also links with a point by Hacker and Furlong that our trust of memory as a faculty is a single judgement and our trust of memorial beliefs p, q, etc. can occur in different degrees and whether we should trust memory per se and memory belief p is a different question.)

  7. Hey Tom,
    In your last comment to me, you said, “here is aline I’m really tempted by: one can remember with either of the two relata you distinguished above only if one’s current memory is appropriately causally connected with the relevant thing (object or fact, I’d suppose) or one knows.”

    So, it might be that

    S remembers that p only if S is appropriately causally connected to the fact that p.

    But note the same sorts of difficulties for causal theories of knowledge rear their heads here. A child might remember that 4×5=20 but not be causally connected to that fact. (See how similar knowledge and remembering are? See???)

    On whether there must be a causal connection for remembering events (S remembers event X only if S is causally connected to X), well, a referee made a an objection in my paper. Here’s how I replied in footnote 2:

    “A referee for Synthese raises the interesting point that one can remember 4’s being the square root of 16 without having been in contact with either number. Taking a definite stand on how to deal with this point is not necessary for my defense of RK. However, here are a few responses I could make. I could restrict the point about contact to only events occurring in space and time. I could deny that I remember 4’s being the square root of 16; it does sound a little odd to say. I could also say that I am in contact with these numbers in virtue of grasping them. Unfortunately, I am not entirely satisfied with any of these responses; fortunately, it does not matter for the purposes of this paper.”

    I’m still not sure what to say here, but I do like the causal condition for remembering events.

    • Andrew,

      Thanks for your reply. Notice that the line I said I was tempted by is a disjunction: if S remembers that p then either S is appropriately causally related to the fact that makes p true OR S knows that p. So the examples you give won’t be a problem for this position.

Leave a Reply

Your email address will not be published. Required fields are marked *