Knowing and Knowledge I. Introduction Though the scope, limits, and conditions of human knowledge are of personal and professional interests to thinkers of all types, it is philosophers, specifically epistemologists, who are most devoted to studying knowledge. Epistemology, or theory of knowledge, is the subfield of philosophy that studies knowledge, justification or rationality, and the conditions under which agents have knowledge or justification. Philosophers, epistemologists included, have been very interested especially since the early twentieth century in using our natural language as a guide to understand truths about the world around us. The approach could be labeled a broadly commonsensist approach: assuming that most of our beliefs are true, then we should be able to use things we already believe to discover more about the world, the thought goes. Combine that with the assumption that the beliefs the truth of which we are taking for granted are reliably expressed by the words we use, and the linguistic approach is born: we can use the natural utterances of speakers to learn more about the world. Of course, there are bound to be some imperfections, the linguicist thinks. After getting her list of utterances using knows or knowledge or related terms, some tinkering will have to be done to figure out which linguistic rules govern speakers knowledge utterances. But the approach is one we should be able to count on. There are multiple ways to be a skeptic. One is to claim that a bulk of knowledge ascriptions claims about someone having knowledge are false. These ascriptions could be about some particular domain of inquiry, for instance physics ( No one knows whether or not there are preons. ) or astrology ( You do not know that you ll get that promotion just because your horoscope said so! ) Ascriptions could be about the mind-independent world ( The beliefs that there are rocks, coffee shops, and other people are very highly rational, but no one knows for sure whether any of them exist. ) There are more ways to be a skeptic, but we need not consider them. There are 1
many reasons why someone would think some kind of skepticism is true. Take the physics example, above. Someone might reasonably argue that knowledge of something 1 is had only if that person is able to directly observe that thing. Preons are not directly observable and therefore it is unknowable in principle whether or not they exist. Someone might also reasonably claim that knowledge of astrological claims is impossible because the claims are all false. Plausibly, to know some assertion requires that the assertion be true. Finally, someone might deny that knowledge of the external world is possible for a number of reasons. She might be compelled by an argument like this: if there were an evil scientist stimulating your brain in specific ways, he could cause you to have just the sensations you are having now. And because you lack any way of distinguishing between the sensations you are having right now and the sensations you would have were to you to be stimulated by the scientist, you cannot know that your sensations are caused by the objects you seem to be experiencing rather than by wires and electrical charges in the scientist s laboratory. Therefore you do not know that what you seem to see is really there. Or she could think that to know something requires being absolutely certain about it. It is possible, she might think, that what appears to be a coffee shop on the corner is actually a building façade with hired actors entering and exiting just to make it look like a busy coffee shop. (Why? Maybe it is a front for mob activity.) Finally, it would speak in favor of skepticism if it were impossible to find any principled basis for a definition of knowledge. The idea here is that we need a definition of knowledge if we are to say that anyone has it. But if we don t know what knowledge is, how can we say whether anyone has it? Gridlock. This issue will be especially salient later in the paper. For our purposes, skepticism shall be defined as follows: skepticism about X denies that knowledge ascriptions about X are generally (if ever) true. 1 I here speak of knowledge of things. In a few sentences I will mention knowledge of an assertion. I continue shifting my word choice for what the objects of knowledge are in order to remain accessible to non-philosophers. Philosophers most commonly speak of knowledge of propositions. The question naturally arises: What is a proposition? Any answer to the question is controversial. One thing that can be said is this: when I say it is raining in English and il pleut in French (which translates to it is raining ), I express exactly the same proposition. The details do not concern us here. 2
I will argue that the now-popular linguistic approach is misguided for a number of reasons. First, since there are systematic irregularities in the use of knows in English, it follows that there is no unique concept that can be identified by the word knowledge. Since the linguistic approach requires that we take our assertions at face-value, this problem suggests that the linguistic approach leads to skepticism. Second, thanks to a more global outlook, philosophers have become interested in discovering the rules behind words equivalent to knows in other languages. But doing this just exacerbates the problem: by adding more (supposedly equivalent) words that follow different rules, we simply add more concepts that are picked out by knows. We all think we know some things, and when we say this, we don t just mean to claim that we use a certain word. We mean that a certain, unique relation holds between us and some fact in the world. That the linguistic approach shows that there is no unique relation means that the linguistic approach forces us to deny that there is any such (one) thing as knowledge. That is a skeptical conclusion. Third and finally, I argue that the linguistic approach brings with it a troubling methodological problem: we do not know which knowledge claims are genuine unless we know what knowledge is, but according to the linguistic approach we do not know what knowledge is except by understanding how speakers use knowledge. I conclude by briefly arguing for a different methodology: if we are to find out what knowledge is, we won t do it just by finding out how people use words. II. Funny Things We Say With Knowledge Compare the following utterances. I (Said on Thursday) The bank closes at 5:00 pm on Saturday. I know because I just checked the website. II I don t believe I have a brain, I know it! 3
III You didn t know you were going to have a boy just because the ultrasound seemed to show it! Sure you believed you would, and you did have a boy, but ultrasounds can be wrong! IV I guess now that you mention it, you re right: I might die in a car crash on the way to the mall tomorrow. So I suppose I don t really know that I ll be at the mall. But, practically speaking, I do know it. V Yes I know I have hands! No, I don t know that I m not in the Matrix, but that s silly: I know I have hands! See! I expect that each of these sentences is plausibly like the sort of thing we hear ordinary, intelligent, competent English speakers say. I won t go into detail explaining all of the theoretical commitments presupposed by these statements, but I will identify a few. (I) supposes that we can have knowledge of the future, while (IV) denies this. (II) supposes that knowledge is possible without belief while (III) can be read as assuming the opposite. 2 (III) supposes that belief plus very good reason is insufficient for knowledge because of the possibility of error but (V) affirms knowledge while denying that a certain possibility of error needs to be eliminated. Finally, (IV) and (V) are very similar in structure, but (IV) affirms and (V) denies that there is knowledge. 3 What we see is that these ordinary sentences express different and incompatible theories of knowledge. This is a skeptical result, for it shows that no matter how we attempt to supply a theory of knowledge that makes true as many of these sentences as possible, at least some of these sentences will turn out false. According to our definition, skepticism about X says that knowledge claims about X are false. On the assumption that our set of dummy sentences are the type of 2 Timothy Williamson (2002) is without question the most influential defender of the view that knowledge is unanalyzable. By this he means to claim that knowledge cannot be given a definition because it is not made up of simpler concepts. It follows (for Williamson) that knowledge does not require belief. 3 The principle supposed by (IV) and (V) is the thesis that knowledge is closed under known entailment, often referred to as the closure principle. A standard formulation is this: if a subject S knows that P and knows that P entails Q, then S knows that Q. The principle is very widely accepted. Skeptical arguments often appeal to the principle and to our inability to eliminate certain Q, like the possibility of our being in the Matrix, to argue that we lack knowledge. Cf. Unger (1975, 7); Hawthorne (2004, 31). 4
sentences that are commonly used in ordinary English, then we are currently at risk of claiming that all sentences that express a certain theoretical commitment are false. It should be granted that when we speak, we sometimes speak with a certain degree of looseness, accepting that some things we say are, strictly speaking, false. 4 We call an opening in the mountains flat if we want to put a runway there, even if there are a few trees and several dips and hills on it. The attempted line of defense here is to indicate that even though we say we know certain things, or that certain things are flat, we do not really mean it. This is clearly a skeptical result as well. According to this proposal, not only do we not know what it is we say we know, we do not even believe what we say we know. 5 This defense started as a way of making sense of the way we speak but ended up conceding that perhaps there is no such thing as knowledge. It would appear that we are left with accepting either that we do not understand how to speak our own language in a sufficiently literal way or that we usually speak falsely on purpose when we ascribe knowledge. One might suggest another proposal. Thinking again of our dummy sentences, someone might suggest that there is a way to make all of the sentences come out true. The suggestion here is that knows expresses a different relation in some of the sentences than it expresses in other sentences. Here we understand know to be like tall or sturdy : what makes the sentence X is tall or X is sturdy true depends upon what else you are comparing X to. If while looking at a photo of Wilt Chamberlain (7 1 ) I say Michael Jordan is tall (6 6 ), you might not be inclined to agree with me. But if we are watching MJ dominate a team of seventh-graders, we will probably agree that he is tall. The proposal here is known as contextualism. The contextualist s slogan is that the truth of knowledge ascriptions depends on the context of utterance. There are as many 4 Richard Fumerton (2006, 23) illustrates this point well. After offering a few examples of ordinary but literally false statements, he states, We don t both conditionalizing everything we claim to know because we don t want to bore our audience to death. His claim is that rather than saying If A or B or C or D or then probably not Z, but otherwise, Z we just say Z as a matter of conversational appropriateness. 5 Unger (1975, 89) uses this line of thought in defense of his radical skepticism. 5
ways to fill in the details of contextualism as there are contextualists, and not all of them would agree with some of the presentation above, but the basic idea should be clear. 6 I hope you find the contextualist proposal unsettling. I do. We take knowledge and knows to be absolute terms, not relative ones. The contextualist denies this, instead claiming that English speakers are often blind to the workings of the language they speak. 7 Yet contextualists works as hard as they can to avoid the skeptical conclusion that many or most of our knowledge ascriptions are false. I take the contextualist proposal that the meaning of knows varies by context to itself be a skeptical one: did it ever occur to you that what it takes to know something varies by who is making the knowledge claim? We tend to think of knowledge, for lack of a better expression, as just one thing. It s not like tall. It is rather like being certain. If you are certain of X, then you could not be more certain of Y. It does not even make sense to say, I am certain that it is raining, but I am even more certain that I m going to the party on Saturday (unless you supply nonliteral meanings to one or both occurrences of certain so as to make the sentence make sense). Certain is a limit word: even if we are hardly ever certain of anything, we understand that certainty picks out a point beyond which there are no points. Certainty does not come in degrees. I can say that I am more certain or less certain, but what I take myself to mean when I use those expressions is that I am closer to being certain or further from being certain. We ordinarily take knowledge to be like certainty. The point of this paragraph is that even if the contextualist were to succeed in making our dummy sentences all true, it would be at an enormous cost: denying that we understand some key elements of our language. 6 Good introductions to contextualism can be found at the Stanford Encyclopedia of Philosophy and John Hawthorne (2004). Note in particular that I said the contextualist claims that knows expresses different relations in different contexts. I think that is the clearest way to understand the thesis, but at least one famous contextualist disagrees with this construal. Hawthorne, a contextualist, endorses the different relation interpretation but notes Stuart Cohen s (another contextualist) disagreement (2004, 81). 7 Hawthorne (2004, 107) calls this thesis semantic blindness. He seems to agree with me that contextualists are necessarily committed to it. 6
Some philosophers think that when I asked you to read the sentences at the beginning of this section and figure out whether you think the sentences are true, I committed a methodological blunder. A recent trend in philosophy has been committed to the thesis that any single individual or groups intuitions (more-or-less theoretically driven but off-the-cuff, unresearched answers to questions) are unreliable. What we need to do, they think, is ask lots of different people groups how they would respond to the questions we want to ask in order to find out what our intuitions really are. The trend is called Experimental Philosophy. Experimenters claim to have found plenty of statistical differences between races, socioeconomic groups, education levels, and speakers of different languages with respect to how individuals answer certain questions. They have found that people evaluate the same question differently based on the font it is written in, or the order in which ideas are presented. I will not offer a diagnosis of what the Experimental Philosophy groups have uncovered; I bring up Experimental Philosophy and its results only to point out that our problem of figuring out what theory we have in mind when considering our disorganized use of knows is actually much worse than we originally thought. 8 If we ourselves were confused, how much worse is the problem when we start asking everybody what they think? Clearly, the answers will only get more scattered. Experimental Philosophers have become even more recently interested in how speakers of other languages use words equivalent in meaning to our knows. The hypothesis is that knows is a universal term, which means that the properties expressed by our knows are expressed by some word that behaves in roughly the same way in other languages. Even facile evidence threatens this thesis, it would appear: there are two words used to translate knows in Japanese, and it appears that they are not synonyms. 9 Once we take into account all of the data for 8 Jennifer Nagel argues that even though experimental data suggests that we are as easily influenced in our responses to certain prompts as the experimentalists say, we can still appeal to intuitions by using constructed cases to elicit responses: we just have to shape our case prompts carefully. Cf. Nagel (2012). 9 Since much of this material is very new, I refer to a recent call for papers to an experimental philosophy conference that will explore some of these ideas. The call for papers itself explains much of what is going on: http://experimentalphilosophy.typepad.com/experimental_philosophy/2013/01/conference-announcement.html 7
how knows and its foreign equivalents are used, the project of finding one set of rules that accounts for everyone s concept of knowledge looks completely hopeless. As before, even if we did for consistent rules that would make the sentences using knows come out true, it would be at the cost of knows picking out one special relation between us and the world: we would have something more like the contextualist proposal that there are an incredible number of different relations expressed by knows. Neither sounds very assuring. The goal of this section has been to argue that any attempt to discover what knowledge and knowing by reading the meaning off of the words used to express the concept is simply hopeless. There is no way to go with that project that will not end up denying either that we know what we are talking about, that we have knowledge when we say we do, or that there is any such (one) thing as knowledge. Time to consider an alternative proposal and a problem it faces. III. The Problem of the Criterion In order to avoid one of the kinds of skepticism just mentioned, we must some of our types of knowledge attributions as the central cases and build our theory of knowledge around those cases. As before, this is a slight victory for the skeptic. Not all of our knowledge claims will turn out true. But if we are willing to admit that we sometimes speak falsely on purpose, it should not be too troubling. ( Yes, I know my wallet is in the car! said while my wallet is in my desk, where I left it.) Here is the problem: which cases do we select as the paradigmatic ones? Perhaps the ones that fit what I think exemplify what knowledge really is: the ones that presuppose absolute certainty and having very good reasons in support of the claim being asserted. But why those ones? The fact that I find those assumptions to accurately pick out what knowledge is really about hardly seems like a principled reason to build a theory of knowledge around certain sentences and not others. 8
This problem is a new variation on an old one. Roderick Chisholm introduced The Problem of The Criterion as a skeptical problem several decades ago. 10 The problem is this: unless we already have a definition of knowledge, we have no principled means by which to pick out which instances of putative knowledge are instances of real knowledge; but without knowing which putative instances of knowledge are instances of real knowledge, we have nothing to use to construct a definition. The basic problem is that we cannot have a definition of knowledge without instances of knowledge, and we cannot have instances of knowledge without a definition of knowledge. The linguistic approach we have been criticizing faces this problem in its own way. Rather than worrying about the (more fundamental) problem of when knowledge happens, the linguistic approach seeks to know when knowledge ascriptions are true. The question being asked concerns when a certain sentence is true. When faced with the Problem of the Criterion, the linguistic approach is particularly damned: there is no reason to favor any sentences being true over any other sentences being true. None at all. So, sure, I can arbitrarily choose the sentences that presuppose certainty and very good reasons on behalf of the subject said to have knowledge, but without some underlying, well-motivated theory of knowledge, my choice is completely arbitrary. This problem, I submit, is worse than the old Problem of the Criterion. For our new problem offers no basis at all for choosing which sentences are true. At least the old problem gave us the opportunity to argue for principles of knowledge that are not expressed by ordinary utterances. The problem is inescapable. Not a little ink has been spilt over the old Problem of the Criterion. But I think the new problem should spell the end of the project of reading off the conditions of knowledge through expressions in natural language. IV. In Search of a New Methodology 10 Chisholm (1966). 9
We have seen a few ways to run headlong into some form or other of skepticism. In order to secure the result that there is such a thing as knowledge, and that we at least sometimes have knowledge, we need a philosophical methodology that will allow for the understanding and analysis of concepts which is independent of the language used to express those concepts. One very clear and straightforward language-independent methodology is proposed by Richard Fumerton. 11 The proposal states that when any individual does philosophical analysis, that is, when some individual searches for an explicit definition that will state the conditions under which some concept applies or exists, that individual imagines different scenarios and asks himself whether he is inclined to apply the concept in that situation. The proposal is as clear and language-independent as it is egocentric. If there were some reason to think that my concept of knowledge accurately fit what knowledge really is, then Fumerton s proposal would suffice for me to run the tests he suggests and arrive at truth about what knowledge is. I hope it is not just youthful optimism that desires to discover truth about the world through philosophy. There are rather outstanding, but hopefully not insurmountable, metaphysical problems here. What does it even mean to say the thing which is knowledge? And what would it take for that thing to either be in or before my mind, or to have a mental duplicate of that thing, knowledge? These are questions I lack answers to at the moment. But I do believe that the methodological proposal is the one that philosophers must use to figure out what knowledge is and under what circumstances anybody has it. Indeed, I could even suggest (though I am hesitant to make such a bold claim) that this approach is the one epistemologists have been taking all along. There is surely substantial disagreement as to what the correct definition of knowledge is, but the existence of disagreement does not and has never been a reliable indicator that there is no fact of the matter. 11 Fumerton (2008), (1983). 10
References Chisholm, Roderick. The Theory of Knowledge, Englewood Cliffs, NJ: Prentice-Hall, (1966). Fumerton, Richard. Epistemology, Blackwell Publishing, (2006). --. The Problem of the Criterion, Oxford Handbook on Skepticism, ed. John Greco, Oxford University Press, (2008) 34-52 --. The Paradox of Analysis, Philosophy and Phenomenological Research, 43, (1983) 477-497. Hawthorne, John. Knowledge and Lotteries, Oxford: Clarendon Press, (2004). Nagel, Jennifer. Intuitions and Experiments: In Defense of the Case Method in Epistemology, Philosophy and Phenomenological Research, 85:3 (November 2012). Rysiew, Patrick. "Epistemic Contextualism", The Stanford Encyclopedia of Philosophy (Winter 2011 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/win2011/entries/contextualismepistemology/> Unger, Peter. Ignorance: A Case For Scepticism, Oxford: Clarendon Press, (1975). Williamson, Timothy. Knowledge and its Limits, New York: Oxford University Press, (2002). 11