Epistemic closure and folk epistemology

Consider this case:

When Maxwell arrives at work in the morning, he always parks in one of two spots: C8 or D8. Half the time he parks in C8, and half the time he parks in D8. Today Maxwell parked in C8. It’s lunchtime at work. Maxwell and his assistant are up in the archives room searching for a particular document. Maxwell says, “I might have left the document in my car.” The assistant asks, “Mr. Maxwell, is your car parked in space C8? It’s not unheard of for cars to be stolen.” Maxwell thinks carefully for a moment and then responds, “No, my car has not been stolen. It is parked in C8.”

Which of the following options best describes Maxwell?

  1. He knows that his car is parked in C8. And he knows that his car has not been stolen.
  2. He does not know that his car is parked in C8. But he does know that his car has not been stolen.
  3. He knows that his car is parked in C8. But he does not know that his car has not been stolen.
  4. He does not know that his car is parked in C8. And he does not know that his car has not been stolen.

The epistemic closure principle says, roughly, that if one knows that P, and one knows that if P then Q, and one infers Q, then one knows Q. Some philosophers, most notably Robert Nozick and Fred Dretske, reject the closure principle. However, many epistemologists have claimed that rejecting closure is extremely counterintuitive and radically revisionary. Related to these claims, philosophers have also claimed that conjunctive assertions suggesting a violation of closure are “abominable” and “repugnant.”

So if conventional wisdom in epistemology is correct, then when people consider the question about Maxwell above, the intuitively best answer will not be option 3. Instead, option 3 should seem absurd.

However, as reported in a paper forthcoming in Philosophers’ Imprint, when I tested this case, it turned out that option 3 was viewed as the best option: nearly two-thirds of participants selected it.

best option


We see a similar pattern if we just ask people whether (A) Maxwell knows that his car is parked in the lot, and (B) Maxwell knows that his car has not been stolen. Roughly 80% of people agree with A, while only about 35% of people agree with B.

In light of these results, it seems that closure-denying conjunctions don’t actually strike people as absurd. Moreover, it is highly doubtful that rejecting the epistemic closure principle actually is revisionary.



Epistemic closure and folk epistemology — 36 Comments

  1. Does this experiment control for the possibility that subjects are simply reading the prompts as “Maxwell *knew* that his car was parked in C8, but doesn’t know now that it’s not been stolen”? I.e., he accurately remembers where he parked it, but he doesn’t know where it is now?

    • Hi, Ben. The proposed interpretation of responses requires people to (1) interpret “knows that his car is parked in C8” as “knew that his car was parked in C8,” while (2) interpreting “does not know that his car has not been stolen” to mean what it says. Why would they do that?

      • Which of the following options best describes Maxwell?

        1) He knows that his car is parked in C8. And he knows that his car has not been stolen.
        2) He does not know that his car is parked in C8. But he does know that his car has not been stolen.
        3) He knows that his car is parked in C8. But he does not know that his car has not been stolen.
        4) He does not know that his car is parked in C8. And he does not know that his car has not been stolen.


        Hi, Ben. The proposed interpretation of responses requires people to (1) interpret “knows that his car is parked in C8″ as “knew that his car was parked in C8,” while interpreting “does not know that his car has not been stolen” to mean what it says. Why would they do that?


        SH: There is major problem with how the test is worded, I would call it equivocation/ignorance and what I mean by that is it conflates belief into knowledge and doesn’t allow the correct response by the people offering their opinions. They are forced into choosing the “best” answer allowed which means treating a very likely fact but still a belief, as a fact or known.

        Reply 1) He knew in the past tense that his car was parked in C8. Humans don’t know the future. Their predictions involve varying levels of confidence in the predictions and as such are beliefs. Time transforms that which is known into what which is believed. Maxwell hasn’t had the information he needs to make to make an accurate assessment about his car updated when he decides four hours later. Four hours later, he is making a belief prediction based on what he knew four hours earlier which was something he knew then.

        This actually happened to me. Somebody tipsy came out and got in a car that was his car’s make and model, and the key fit. He drove off and about five minutes later, it dawned on him he was in the wrong car. He turned around and parked the car in the same parking space. The actual owner of the car is correctly described as believing his car is parked outside and not stolen. When he comes outside and sees his car, he believes that he knows that the car was not (technically)stolen. If he had come out five minutes earlier he would know that his car was missing and presumably stolen. It likely humans can know things about the past, but they cannot predict future events from the pas accurately enough to call it knowledge. What they have are less likely to very likely beliefs.

        Reply 2) Using the reasoning above, I don’t see how this could happen. Maybe Maxwell was going to lend the car to his wife briefly. Even if his wife does borrow the car and it’s missing, that doesn’t preclude the car from having been stolen and returned before his wife showed up to borrow it.

        Reply 3) He knows that his car was parked in C8, but he doesn’t know that it is parked there now, it could have been towed and not stolen. I think you need current knowledge of an event to use the word know, he doesn’t have that. He has a strong belief.

        Reply 4) That only leaves reply 4 as something that he knows. He knows the status of his car in uncertain because he doesn’t have enough information to use the word know in the now, which is four hours later.

        Reply 1) This should read ‘ He believes that his car is parked in C8. And he believes that his car has not been stolen.’ I think the first three replies introduce counterfactual realities which do not coincide with our own physical reality. Reply 1, isn’t true in our universe, the correct word to use in our universe is belief. I can’t imagine what people thought they could figure out from such a sloppy test which skews the possible answers. The word “know” should not be used to describe predicting the future.

      • One possibility is that because “He knows that his car is parked in C8” is presented before “But he does not know that his car has not been stolen” in the questionnaire, it is easy for careless subjects (as most subjects probably were) to think that the event in the former belief occurred prior to that in the latter, that is, to think that the first knowledge is about his act of parking the car, instead of the current where-about of the car.
        To control for this effect, we should probably at least add something like “still” and “now” into the first sentence: “He knows that his car is still parked in C8 now.”
        Or even better, we could ask subjects to evaluate three instead of two statements, thus eliminating any room for ambiguity: “He knows/does not know that he parked his car in C8. He knows/does not know that his car has not been stolen. He knows/does not know that his car is still in C8 now.”

        • Hi, Yao. Yes, it’s worth considering the possibility that people interpreted the first sentence as pertaining to the early part of the story and the second sentence as pertaining to the later part of the story. However, the results from the between-subjects study using the same story rules this out (Experiment 1 in the paper). In that study, no one was presented with both knowledge attributions (i.e. “Maxwell knows that his car is parked in C8” and “Maxwell knows that his car has not been stolen”) and the knowledge attribution that did appear was randomized along with comprehension questions. So there would be no reason for people to systematically map the two knowledge attributions (is parked/not stolen) onto different parts of the story in a certain order.

          Also, what’s your evidence that “most subjects probably were careless”? People passed the comprehension questions at really high rates.

      • That’s a fair question, John. But here’s another way of reading it which gets the same point but doesn’t involve reading the same word two different ways. Maybe subjects are reading it as “Maxwell knows that his car *was parked* in C8, but doesn’t know that it hasn’t been stolen.” Know is now used in the same sense in both clauses, but this sentence doesn’t violate closure. Sure, “is parked” is very different from “was parked,” but the issue here is that student readers are sloppy. When developmental psychologists study epistemic concepts in young children, they control for these kinds of misinterpretations all the time. And you could do it very easily by adding an option to the list of questions. My preliminary bet is that if readers clearly saw the difference between “is parked” on one line, and “was parked on the other,” they’d pick “was parked” as the obvious description of the situation.

        • Hi, Ben. Thanks for your further remarks on this. A few thoughts in reply:

          (1) To clarify, I’m not sampling a student population. The population consists of workers aged roughly 18-75. The mean and median age tends to be in the early- to mid-30s. For more detail, check out the following paper: Paolacci, G., & Chandler, J. (2014). Inside the Turk: Understanding Mechanical Turk as a Participant Pool. Current Directions in Psychological Science 23(3), 184-188.

          (2) There is no evidence that the participants were sloppy. To the contrary, they correctly answered comprehension questions at impressively high rates. More generally, my experience is that Mturk workers are very conscientious.

          (3) Which developmental papers include control conditions to rule out that people misunderstand “is” to mean “was”?

          (4) If people cannot be trusted to understand that “is” means “is,” then I submit there is no chance that they understand and accept a closure principle of the sort epistemologists have been interested in.

          (5) I’ve now run a follow-up study that addresses the basic concern without adding yet more options. I’ll report the results in a follow-up post, but for now I’ll just say that the same basic pattern of non-closure for inferential belief emerged.

    • 3) “He knows that his car is parked in C8. But he does not know that his car has not been stolen.”

      The problem with this answer is the use of the word “is”.
      He _knows_ that his car _was_ parked in C8, but true to reality, he _believes_ that the car _is_ parked, which is causally hours later, and therefore unpredictable, hence a belief.

  2. Hi, Stephen. That is all interesting but it cannot explain the similar results observed with different tasks and experimental designs. For example, no one is forced to choose anything in a “check all that apply” task. Also, people know all kinds of things about the future. Such knowledge is utterly routine.

    • Hello John,

      I haven’t examined those results as you have. But do those questions have a mix of the words, knowing and belief replies, to chose from because that provides a full picture of reality? Only using one or the other word in choices will bias the test. So do the results you refer to include both words in the answers-> in the check all that apply?

      I was careful to distinguish knowing historical facts from the possibility of predicting a chain of causal events that move into the future. I may know my birthday is January 6, as a historical fact and that will fix the same historical fact into the future as long as I live. That’s more like knowing logical truths from using definitions or axioms.

      Maxwell could well know where he parked his car in the morning, and that is a historical fact. But hours later he has a speculation or belief. Because the question of where his car is parked does not depend on a historical fact, but on a chain of causal events which transpire outside of his awareness. Maxwell’s situation is not in the category of “all kinds of things people can know about the future”. He doesn’t know if the car is in C8, he believes it because he predicts it as the most likely situation, before a physical inspection. Maxwell can’t know where his car is without a physical confirmation because this kernel of knowing isn’t available to him to arrive at from his earlier historical knowledge.

      The point I’m making is embedded in the dispute about identifying Causation that goes back to Hume and Kant at least. It hasn’t been resolved with counterfactuals either. If you want to call knowing a very, very, very, degree of probability about the actual state of physical reality, one can choose to do that. But Maxwell’s situation isn’t like that, he harbors a belief based on uncritical thinking in any of his first three replies.

      There are limitations, probabilistic quantum theory and a proof that not all knowledge can be computed by even the most powerful supercomputer (not Super-Turing). Perhaps only two or three major philosophers adumbrate the limitations of (ultimate) knowledge. It will be hard for you to provide an example of something you know (without any possibility at all of being mistaken) of something that will happen five years from now. You are not even certain you will wake up in the morning.

      Perhaps you may suggest that you know that it is certain we will age and die. I believe that. I do not believe in God but I don’t know that there is no God. About 40+% of the US population believes in God, and a good portion of them believe in a literal interpretation of the Bible. They believe that a man was taken to heaven without dying. We can’t know that is false. So a large portion of the populace will disagree that you know we will age and die; and there is a possibility of medical advancement you don’t know about. The only correct answer is Reply 4, if you know enough about reality. 🙂

  3. Hi, Stephen. I think you’re overlooking how rarefied and theoretical the issues you’re raising are. Knowledge is perfectly compatible with the chance of error and with not being certain. Whatever the limitations of “ultimate knowledge” are, they don’t resemble the limitations of knowledge.

    PS: lots of us can and do know that no man was taken to heaven without dying.

    • Hi John, who wrote,
      “I think you’re overlooking how rarefied and theoretical the issues you’re raising are. Knowledge is perfectly compatible with the chance of error and with not being certain.”

      SH: Well, I would say somewhat yes to that, but I think I’m talking about an application of the Gettier Problem. I’ll try to keep this reply short, quoting from Plato.Stanford,

      “Above, we noted that one role of the justification is to rule out lucky guesses as cases of knowledge. A lesson of the Gettier problem is that it appears that even true beliefs that are justified can nevertheless be epistemically lucky in a way inconsistent with knowledge.”

      SH: When Maxwell pronounces that he _knows_ his car is in C8, is that a JTB? Suppose he departs the archives and checks its location. If the car is not there, then he issued a Justified False Belief earlier. Since the car could have been towed, stolen, or a relative borrowed it, and since that’s unpredictable for Maxwell, why isn’t ok to think that the uncertainty involved is another word for luck? Or, why doesn’t the Gettier principle apply to Maxwell’s scenario as well as barns etc? Don’t the Gettier cases contain a hybrid mixture of true and false (beliefs) expectations and luck?

      Maxwell had some good reasons to expect his car to be in C8, but that it was actually there, involved some luck, and since it didn’t have to be there, he didn’t know it was there. I think you use the word “know” more liberally than I do. Who says they know what the weather will be? But they might say they know what time they will arrive at work. Also you may not esteem Timothy Williamson. If you think I’ve made a crumby argument, you can be honest.

      • Hey, Stephen. Just to be clear, Maxwell doesn’t ever say he “knows” these things. He just says the things. Then people are asked to judge whether he knows them. As for the issue of luck, people are definitely sensitive to some forms of luck that they view as inconsistent with knowing. However, with respect to knowing that his car is parked in C8, this does not seem to be one of those cases. That said, I’m not passing judgment on your argument that this is not a case of knowledge. I’d only say that, if that is the correct conclusion, it goes against how the case tends to strike people. (Btw, while I’m not sure what prompted the point about TW, I’ll note that I hold him in very high regard.)

        • Hello John, Thanks for your reply. You wrote,
          “Just to be clear, Maxwell doesn’t ever say he “knows” these things. He just says the things. Then people are asked to judge whether he knows them.”

          SH: I don’t agree the distinction matters on the basis of what you wrote earlier.

          Maxwell says, “I might have left the document in my car.” The assistant asks, “Mr. Maxwell, is your car parked in space C8? It’s not unheard of for cars to be stolen.” Maxwell thinks carefully for a moment and then responds, “No, my car has not been stolen. It is parked in C8.”

          SH: The paragraph above has a meaning which is a function of a person’s reading comprehension. The assistant’s question to Maxwell means: Are you certain that the car is still parked in C8 or do you _believe_ that it is parked in C8 (because it could be stolen). The meaning of Maxwell’s reply expresses knowing, and it doesn’t matter whether he uses the word _know_ or not.

          I think that at least 90% of the readers here if queried would agree that Maxwell’s reply to the assistant is a deliberated (thinks carefully) claim for/of knowledge. Maybe 10% or less would think that the meaning of the background description is Maxwell’s _belief_ claim. Now the people who take the survey, are asked to judge whether or not Maxwell’s answer is partly or fully justified, or not at all=Reply 4, and their answers are generated by what it means to know using their own ideas.

          I think knowledge requires necessary truth, and surmised that TW disagrees. I don’t think you agree? I didn’t mean you don’t respect him; but doubt his view? Reading that into your stance could well be a stretch. Your topic is interesting!

          • Hi, Stephen. That’s a bold claim! I’m somewhat sympathetic with the basic proposal here, though. After all, by making the assertion, Maxwell represents himself as knowing. Nevertheless, I assume there must still be a difference between asserting “P” versus “I know that P.” For, by the same token, wouldn’t asserting “I know that P” count as expressing second-order knowledge, namely, “I know that I know that P”?

            I do not think that knowledge requires necessary truth.

          • Hi John. Suppose you ask me, “What do you know will happen tomorrow”. And I think about it and reply, “The sun will rise.” Or, I reply, “I know the sun will rise.” I think these both mean the same thing, (knowing) because the verb know is embedded in the reply imputed (right verb?) from the previous question. I don’t think the first is only an assertion, while only the second expresses knowing. I think this because the dialog is filtered through the lens of context. I was trying to convey this idea in my last post.

            “The assistant’s question to Maxwell means: Are you certain[know] that the car is still parked in C8 or do you _believe_ that it is parked in C8 (because it could be stolen).”
            The verb know or believe is implicit from the question -> forward into the answer. Maxwell would use the verb believe or think if he is making an assertion. “Maxwell thinks carefully” tends to imply interpreting his response toward knowing, rather than believing/asserting. Anyway, that’s how I think conversations work.
            Xphi:.. ‘They ask whether someone in a hypothetical scenario knows, or only believes, that some proposition is true.’
            I thought there should have been a mixture of know and believe as choices for the people taking the survey. In particular, “4) He does not know that his car is parked in C8. And he does not know that his car has not been stolen.”

            I think that wording is dubiously literate, and would tend to deflect choices away from that answer.

            TW: “Like many philosophers, I have long been impressed by the failure of attempts to find a correct analysis of the notion of knowledge in terms of supposedly more basic notions, such as belief, truth, and justification. One natural explanation of the failure is that knowledge has no such analysis.”

            SH: So you may not agree with necessary truth, but do you also eschew notions of “belief and justification”?

  4. There’s a pretty simple answer to this “dilemma.” People have different requirements for counting a justification as satisfactory depending on the proposition being justified (aka “contextualism”). A sufficient justification for knowing that one’s car is that one has seen it recently enough that it couldn’t have been stolen, &c. and a sufficient justification for knowing that the car is parked in C8 is having a memory of parking in that spot, &c. Just because proposition A entails proposition B, that doesn’t mean propositions A and B have the same, or even compatible, requirements for justification since communication is about fulfilling Gricean maxims of information, not logic. Giving a justification is a social act, and its standards are ethical/political, not logical.

    • Said in other words, the conditions for justified *assertion* of a statement are different than the conditions for justified assent to a proposition. Assenting to the statement that “the car is parked in C8” entails the proposition “the car is not stolen,” but the assertability conditions for the statement “the car is not stolen” are radically different because of its Gricean implications.

    • Hello, Carl. Thanks for these comments. Interestingly, one rationale Dretske gave for thinking that knowledge isn’t closed under known entailment is that reasons aren’t closed under known entailment. However, I don’t think he had in mind the sort of conversational reasons, as it were, that you’re suggesting here.

      In any case, it occurs to me that, on one way of developing the sort of approach you’ve proposed, commonsense is neither committed nor opposed to the epistemic closure principle, in any form familiar from the philosophical literature. Instead, it could be that ordinary practice is simply concerned with a different set of considerations, and it isn’t profitable to look for behavioral evidence regarding the principle’s status in ordinary practice.

      Apologies if I’m over-interpreting your comment. But I do think that something along these lines might be correct and is consistent with all my results (which is why my main conclusion is that ordinary practice is “ambivalent” about closure).

      • That’s a fair interpretation of my comments.

        I honestly don’t get the point of x-phi as philosophy (as opposed to as psychology or sociology, where it might be independently interesting without shedding any light on philosophical questions). We’ve known since the time of Socrates that the public have confused philosophical intuitions. So what’s new?

        • Hi, Carl. Two thoughts in response. First, with respect to closure, claims about pre-theoretical intuitions and judgments have been part of the philosophical debate for decades. So the application of x-phi here is really just an extension of what’s been happening already, rather than something qualitatively new. Second, and more generally, some philosophers view it as a central task of philosophy to clarify what Sellars called “the manifest image,” or our “pre-reflective orientation” and “heritage.” So the application of x-phi here is just an example of that more general philosophical project.

    • Hello Carl,

      One claim of Contextualism is that ‘there are two senses of “know”,
      strong, philosophical, (skeptic) or weak or ordinary.’

      I think that distinction is relevant to this thread. I read some Kornblith who writes fairly convincingly in favor of X-phi.

  5. Hej John, very interesting results and definitely something lovers of closure (like myself) have to find an explanation for. There’re probably several strategies that could be explored, but for now I just want to raise two questions that you might already have thought about:

    (1) It is possible that Maxwell’s car has been stolen and is also parked in C8. For depending on the definition of stealing it’s possible to steal something without actually moving it. Moreover, it’s possible to steal something and return it (intentionally or unintentionally): The absent minded thieves may take Maxwell’s car and out of sheer luck they (attempt to) hide it in C8 or bank robbers may steal Maxwell’s car to commit the crime, but also be so kind to return it to C8. I’d be more impressed by your results if the pair would have been something like
    – Maxwell’s car is parked in C8.
    – Maxwell’s car isn’t parked in C8 because thieves are driving it out of town right now.
    My idea is that it is much more repugnant to ascribe knowledge of the first proposition but not knowledge of the second proposition when the entailment is blatantly obvious. Maybe lovers of closure just need to be more explicit about which kinds of conditionals/entailments they have in mind! Do you agree that this might make a difference?

    (2) Do your results contradict the claim that closure is part of the folk concept of knowledge? Well, there is a difference between claiming that the folk accept the general closure principle and claiming that the folk judge all instances accordingly. I see no reason to think that intuitions or folk judgements are closed under known entailment: It’s possible that intuitively P and it’s known that P entails Q, yet not intuitively Q (or even intuitively not-Q, if intuitions can be inconsistent). Only if intuitions are closed, can you argue that
    (1) If closure is an intuitive part of the folk conception of knowledge, all instances of closure must be accepted intuitively.
    (2) Not all instance of closure are accepted intuitively.
    (3) Therefore: Closure isn’t part of the folk conception of knowledge.
    But maybe I’m misunderstanding your argument. Thus, my question is whether your argument rests on the premise that intuitions/folk judgements are closed and, if so, why one should accept it.

    • Hi Tim,

      (1) I agree that the results are consistent with certain very careful and potentially complicated versions of the closure principle (and/or stipulations of relevant entailments). The trade-off here is that the more subtleties built into the principle, the less likely it is that people actually even understand it, let alone have any pre-theoretical commitment to it. So it could not be said that being inconsistent with such a principle is revisionary or grates our pre-theoretical intuitions. Consistent with this, there could still be good theoretical arguments in favor of it.

      (2) I agree that it could still be a principle of folk epistemology even though people don’t conform to it in every instance. And, no, my argument does not rest on a premise about folk intuitions/judgments being closed. Instead, I’m pointing out that the patterns of attribution I found are not what we’d expect if closure was a core folk principle. There is a pretty consistent pattern whereby the one attribution (knows where the car is) is significantly higher than the other (knows that the car isn’t stolen). It seems that the population as a whole views these two attributions differently, at least in non-perceptual cases. To my mind, this suggests, though it does not entail, that the ordinary concept of knowledge does not come bundled with a closure principle. Again, there could still be evidence (further behavioral studies, theoretical arguments) leading us to conclude that, nevertheless, the ordinary concept does come bundled with the principle. But I’m aware of no such evidence.

  6. John wrote: “Hi, Stephen. Do you mean that, necessarily, knowledge requires truth? Or do you mean that knowledge requires necessary truth?”

    Sorry it took so long to reply, but I had never thought about making the distinction you offered. At this point, I don’t think the word “or” operates. I think the statements are compatible instead. I think the term “contingent truth” is the “or” to use in opposition with “necessary truth”.

    The totality of physical reality unfolding is in principle unknowable and I think that is an absolute. Human languages, including come mathematics, are infinite games which try to map or encompass the territory, which comes down to defining boundaries which I think is always relatively true. I think of “the map is not the territory” is the right concept. I associate contingent with relative. We know self-defined (language) relative truths only, which are not congruent with the actually true, precisely accurate, description/fact about the nature of reality if there were a birds eye view, or oracular source. Claims of a Platonic Realm actually informing human intuition, are like the _virtual_ lifelike properties of the Game of Life that some adherents claim as truly being alive.

    A contingent claim is a contoured/bounded mental claim, not the fact of the matter itself; physical knowledge has limitations such as the Three Body Problem. Penrose has a fairly unique slant on the Heisenburg Uncertainty Principle. He states it is impossible to know the initial positions and trajectories of the particles of physical reality, and how they collided through billions of years, in order to predict physical events. That’s why I involved the Causation quandary/debate as a scale limit to physical knowledge at the macrocosmic level. That doesn’t apply to humanly invented finite rule games such as checkers, and has little conflict with logically possible and logically impossible conclusions. Although with the Zombie argument, the boundary conditions, and how much that possible world(s) really meant about physical reality conclusions about consciousness was hotly debated. John Perry and Chalmer’s one time mentor Hofstadter, described the argument as “silly”.
    I see nothing wrong with using the word know in everyday language. I think extracting insights from that category to draw conclusion about another category with somewhat different underlying properties, is a tricky Empirical business. “The Truth is Out There.”

  7. John, two comments:

    First, there is no relation of logical consequence between the sentences in your example. It is logically possible that: Maxwell’s car is parked in C8 and Maxwell’s car is stolen. So philosophers fond of knowledge’s closure under logical consequence don’t have to be concerned about your results.

    Second, closure under logical consequence aside, I predict that you can carry out a study with two sentences A and B and a subject S such that:
    (1) the majority of participants in the study intuit that S knows that A,
    (2) the majority of participants in the study intuit that S does not know that (A or B)-and-(A or not-B).

    I also predict that you and I differ on the question whether such a result would be bad news for philosophers fond of knowledge’s closure under logical equivalence, or the case method in epistemology and other activities relying on intuitions.

    • Hi Franz,

      As I mention in the paper, this research is not concerned with closure principles that only philosophers or logicians understand. The reason for this is that such principles are not even remotely plausible candidates for principles of commonsense epistemology. A corollary of this is that rejecting such principles cannot count as revisionary. Neither are such principles required for the study of knowledge in the cognitive, social, or life sciences.

  8. Thank you for the response, John! However, now I am puzzled: as a matter of logical and commonsensical fact, the example of your study does not refute the “logical” closure principle from “logical epistemology”. If it is supposed to refute some other, commonsensical closure principle from commonsense epistemology, then this other, commonsensical closure principle has to be even more demanding than the logical closure principle, or else the two principles are not logically related at all.

    • Hi Franz,

      The principle in question is this: if one knows that P, and one knows that if P then Q, and one infers Q, then one knows Q. The example I tested involves a car being parked somewhere and its not being stolen, which in the literature is a paradigm example illustrating the alleged intuitive force of this principle. Many have claimed that we are pre-theoretically committed to this principle, but the results I observed suggest that we are not committed to it. The results do not directly refute any closure principle but instead provide evidence against widespread claims about the concept of knowledge and commonsense epistemology, which have frequently been used to reach substantive theoretical conclusions.

  9. Hi John,

    I don’t see how this principle is easier to understand than the logical closure principle, but independently of this: it remains a mystery to me why a principle about knowledge needs to be easy to understand in order for knowledge to satisfy it.

    But let’s leave this question about commonsense versus “logical” epistemology to one side.

    Even more mysterious to me is your claim that data about people’s knowledge judgements are evidence for claims about knowledge itself, as opposed to talk or thought about knowledge.

    Let’s grant for the sake of argument that intuition is for philosophy what perception is for physics, an analogy that seems to be popular among x-phi-ers.

    According to contemporary physics, what reality is like could hardly differ more from what we perceive it to be (not that I would understand any of the physics, but this much I do). Perception is the result of evolution and is a means to attaining many ends, including survival and procreation. However, finding out what reality is like at a fundamental level is not one of them.

    Similarly, intuition is the result of evolution and is certainly a means to attaining many things, including survival and procreation. But finding about what the nature of knowledge is like, and answering other philosophical questions, isn’t one of them.

    Physics made progress by focusing on theory, and ignoring the illusions generated by perception such as that the earth appears to be flat, and that the sun appears to evolve around the earth. If intuition really plays the role for philosophy that perception plays for physics, it’s time to focus on philosophical theory and ignore the illusions generated by intuition.

  10. Hi Franz,

    Could you please tell me what you’re referring to when you say that I claim that “data about people’s knowledge judgments are evidence for claims about knowledge itself, as opposed to talk or thought about knowledge”? Because in my last comment I wrote, “The results do not directly refute any closure principle but instead provide evidence against widespread claims about the concept of knowledge and commonsense epistemology.”

    Whatever the merits of your brief against intuition and x-phi of whatever sort, it’s just not applicable here. In fact, it’s far more applicable when directed at closure’s proponents, who have appealed mightily to intuition and commonsense (and not much else).

  11. Hi John, earlier you were talking about “principles of commonsense epistemology” and “the study of knowledge in the cognitive, social, or life sciences.”

    I took this to imply that you are studying knowledge itself, rather than how people talk or think about knowledge. If I am mistaken about this and you are studying people’s talking and thinking about knowledge, my points against intuitions are indeed not applicable.

  12. But then, with all due respect, John, ‘epistemology’ and ‘commonsense epistemology’ are misnomers:
    you do not study (‘logos’) knowledge (‘episteme’), but language and thought.

    • Dear Franz,

      I’m surprised and saddened to see you responding this way, leaving no doubt as to how very little respect you think is due, or to your ignorance of my research.

  13. Pingback: Philosophers’ Carnival #168 | Posts