Examining Epistemic Contextualism
Epistemic contextualism (EC) is a thesis about propositional knowledge which is intended to recognize the appeal of both skeptical and anti-skeptical intuitions about such knowledge. More precisely, EC is a semantic thesis, not an epistemological one; that is, EC is a claim about sentences of knowledge attribution such as “Jones knows that p.” [1] According to EC, whether or not knowledge attributions are true depends on the factors of a given epistemic or conversational context. In some contexts, “Jones knows that p” is true, but in others, the attribution is false because the context is different. EC is thus a context-relative position in the sense that whether or not a knowledge ascription is true is relative to the context-sensitive factors of a given epistemic situation. [2]
EC was designed to deal with the (supposed) threat of epistemic skepticism (i.e., the position that we cannot obtain propositional knowledge, either globally or locally), while granting some skeptic intuitions in a manner that seems consistent with some anti-skeptic intuitions. [3] On one hand, the skeptic might claim that since knowledge requires epistemic certainty and we lack such certainty regarding most types of propositions, we don’t know very much after all. On the other hand, the anti-skeptic denies skepticism, perhaps claiming that since we know much of what we commonly take ourselves to know, the standard for knowledge must be lower than epistemic certainty, and hence such certainty is not necessary for knowledge. The contextualist tries to respect both intuitions by seeking a via media: the standard for true knowledge attributions is relative to one’s epistemic context. In contexts that have high stakes, such as cases of serious existential encroachment [4], philosophical matters [5], or situations in which one of the dialogue partners raises the epistemic stakes by asking certain questions, making particular assertions (say, that we don’t know that we are not brains-in-vats), or expressing particular interests or intentions, the standard is high, perhaps as high as epistemic certainty. But in contexts with low stakes, such as ordinary matters of life [6], or cases in which nobody raises the standard, the standard is low, perhaps as low as minimally but fallibly justified true belief. EC is a pluralistic view about the semantics of knowledge attributions. What people mean by ‘know’ and ‘knowledge’ depends on the context in which those terms are used.
Consider an analogy. According to contextualism, whether or not a knowledge attribution is true is a fluctuating kind of thing. It is not stable and invariant. Rather, the meaning of ‘knowledge’ varies relative to situational conditions. The semantics of ‘knowledge’ is something like atmospheric pressure. There is barometric pressure, but it comes in degrees that change according to altitude. At sea level, the pressure is high: on average, it is 14.7 pounds per square inch (PSI). At higher elevations, air pressure drops in terms of PSI. Air pressure is low in the mountains, and the low pressure can cause altitude sickness.
Similarly, on the contextualist view, there is knowledge, but the term ‘knowledge’ varies according to one’s epistemic situation. In low-stakes cases, the standard for what counts as knowledge is low, and ‘knowledge’ picks out something in that situation which meets that low standard. But in high-stakes situations, the standard is high, and ‘knowledge’ picks out something in that situation which is more demanding. One thus needs something like an epistemic barometer to determine in any given case whether or not someone has knowledge. And if someone accustomed to low-stakes ordinary epistemic matters is suddenly thrown into the high-stakes circumstances of, say, philosophy or theoretical science, one might begin to feel unwell or disoriented and hence use ‘knowledge’ improperly. [7]
I find EC an intriguing position. It has some explanatory value, and I like that it is an attempt to do justice to both skeptic and anti-skeptic intuitions. But I don’t think EC has enough and the right kind of explanatory value. I am unconvinced. To start, it seems to me that EC is neither obvious nor intuitive. The theory seems quite counterintuitive. Although it makes sense of some of the ways in which we speak about knowledge, there are other plausible ways to explain such matters, and moreover, EC seems to provide awkward accounts in some respects. Hence, the burden of proof seems to be on the contextualist to make the case for EC. I don’t see that the contextualist has fulfilled the burden.
To elaborate, here are some specific objections I have.
I: What do the different versions of knowledge picked out by ‘knowledge’ in the various epistemic situations have in common? Is there some family resemblance? What factors make them all cases of knowledge? To put the question differently: in what respect does ‘knowledge’ refer to something similar across different contexts? And why are these factors sufficient to make the different things count as actual cases of knowledge? There seems not to be a clear and convincing answer to this question – at least in the literature that I have read.
II. Objection (I) raises two important concerns. First, it is unclear what ‘epistemic context’ means. What exactly are the factors that constitute an epistemic context? Some writers speak in terms of assertions, interests, purposes, background information and evidence, and discussion topics and questions raised by those involved in discussions that include knowledge attributions. Other writers speak in terms of whatever conditions are relevant or salient to the knowledge attribution at hand such that one can eliminate (there is debate about what ‘relevant’ means and about what it means to ‘eliminate’) whatever is not practically germane to the ascription as a method of verifying the truth of the ascription. And other writers speak in terms of the epistemic, psychological, or existential interests and needs of the speakers, matters of practical encroachment, etc. The question of what ‘epistemic context’ means seems to me an important one that needs to be worked through before EC can be accepted.
III. Objection (I) raises a second concern. Since EC is a semantic thesis about what ‘knowledge’ means in different contexts, it seems that EC is not an epistemological thesis. In this case, EC does not directly address the dispute between skeptics and anti-skeptics who are debating the nature of knowledge (the thing itself, not the word) and not debating what words such as ‘knowledge’ and ‘know’ mean in different contexts. EC seems guilty of changing the subject rather than addressing the dispute.
The skeptic could grant that people use ‘knowledge’ and ‘know’ differently in various contexts and still insist that knowledge itself requires epistemic certainty (i.e., roughly, infallibility given the evidence) and hence that we don’t know very much (if anything). The skeptic could explain that people use ‘knowledge’ and ‘know’ differently in various contexts for the sake of convenience and that the loose uses of ‘knowledge’ are not precisely accurate because they pick out cases in which people lack epistemic certainty and thus don’t really know what they are said to know.
On this objection, EC doesn’t entail anything about the nature of knowledge itself but is concerned only with the semantics of our epistemic vocabulary. Smith’s attribution of knowledge to Jones in “Jones knows that p” could be true if by “knows” Smith means ‘has a fallibly justified belief’ and Jones really does possess such justification. And yet it still might be the case that the skeptic is correct that knowledge — precisely speaking — requires epistemic certainty and thus that, according to the skeptic’s standard, Jones doesn’t know that p and Smith is using “knows” in a loose and imprecise manner.
The contextualist might respond that although EC is a semantic thesis, it nevertheless entails the epistemological thesis that knowledge itself varies depending on the epistemic or conversational context at hand. On this view, whatever knowledge is, it is something that is variable like atmospheric pressure. As atmospheric pressure differs depending on whether the altitude is low or high, the standard for knowledge differs depending on whether the stakes of the conversational context are low or high. This is not a thesis about the meanings of our epistemic vocabulary, as EC is, but rather a claim about what knowledge is. But this claim raises problems.
IV. For one thing, knowledge does not appear to be a gradable sort of thing. It doesn’t seem to come in degrees like atmospheric pressure, tallness, and swiftness come in degrees. If EC entails the view that knowledge itself is a degreed property, then we need a sufficient defense of this view.
Consider terms like ‘make’ and ‘fight’ and ‘morally permissible.’ Suppose that Smith and Jones are talking about matters gustatory, and Smith claims “Brown makes a good pizza.” Now, Brown in fact buys ready-made pizza at the local grocery store and merely heats it up in the oven so that it’s hot for Smith’s consumption. Smith likes the pizza, and there’s not much riding on Smith’s conversation with Jones. By “makes” Smith means only ‘heats up’ and Jones doesn’t raise the stakes of the conversation by challenging Smith’s use of “makes.” So, in this conversational context, the semantics of ‘make’ is covered by Brown’s act of heating up the ready-made pizza. But now suppose that Green and Miller are chefs at the local five-star Italian eatery. Green claims the following: “Brown serves pizza at his home, but he merely heats up food that someone else has made. He doesn’t make the pizza.” Here, Green claims that Smith doesn’t make the pizza, because in this context, ‘make’ means much more than ‘heat up.’ Italian chefs make pizza from scratch before they heat up what they’ve made. Now, here’s the difficulty: just because ‘make’ is used differently in these semantic contexts, it (arguably) doesn’t follow that what it is to make a meal is itself relative to such contexts.
Commodus (the son of Marcus Aurelius in the film Gladiator) is sparring against friends using wooden swords. Young boys are watching. They say to each other: “Look at Commodus fight with that sword! That’s amazing!” However, General Maximus Decimus Meridius and some officers under his command are also watching. Maximus overhears the young boys and says “Commodus is practicing. He’s not fighting. Fighting occurs in battle, my friends, and Commodus has never experienced even a moment of battle.” In both contexts, ‘fight’ is used. For the boys, the term refers to inconsequential sparring. For Maximus, the term ‘fight’ refers to combat in battle. Yet it doesn’t follow from these variations in semantic context that the acts that count as acts of fighting vary according to semantic context.
Why then should what counts as knowledge vary in such a manner? Why isn’t Maximus’ use of ‘fight’ the invariably correct use? He is the relevant expert, after all. The young boys are not experts.
Mary says to Jane “Don’t worry. What Mark is doing is morally permissible.” But by “morally permissible” Mary means “the act is unimportant; nobody will notice it, and if someone were to notice, he wouldn’t care, since what Mark is doing is no big deal, even if it is technically morally wrong.” Shannon and Jenn also observe Mark. Jenn says to Shannon: “Although what Mark is doing is a venial act, it is still morally impermissible. It’s morally wrong. He ought not to be doing that.” Now, in the first context, “morally permissible” means something different from its meaning in the second context. But quite arguably, it doesn’t follow from these semantic differences that moral permissibility in itself is merely relative to what people mean by the word ‘morally permissible’ in different contexts. Such a position about morality would be an extreme form of moral subjectivism which most moral philosophers would be inclined to reject, or so it seems to me.
Again: Why should what counts as knowledge vary according to conversational context?
V: For another thing, knowledge is (arguably) a mental or intentional state. To know that p is to believe that p on the basis of adequate evidence combined with p’s being true. Belief is a mental state. Hence, it seems, knowledge is (or at least involves) a mental state. How can the external (i.e., extra-mental) conditions of conversational situations — such as changes of discussion topics, the posing of stakes-raising questions, intended uses of words, and changes of environment in which one might have a discussion about some topic — generate a change of one’s internal mental state from knowledge in some circumstances to lack of knowledge (i.e., ignorance) in others although if one were to self-reflect, one might well report no change to his mental state?
Take the posing of questions: How can a conversational context have a low standard for knowledge at time t1, but at t2, the mere asking of a skeptical question raises the standard? The contextualist might reply that the question asked, since it introduces into the conversation a reason to be skeptical, raises the standard and changes the context. To my mind, this reply seems implausible, and a bit forced; it isolates artificially our conversations from the world as if conversations (and the states of belief and knowledge involved in them) were little epistemic spaces sequestered from the world and its information. It seems to me that if one of the dialogue partners raises a skeptical question, he is not introducing a skeptical consideration and a higher standard into the conversation that wasn’t already there. He is merely revealing what was there the whole time.
Consider this dialogue:
A. I know that the fluid in that opaque glass is water. I asked for water, and I saw the fluid being poured into the glass.
B. Really? Can you prove with objective certainty, without tasting it or looking at it again, that the fluid is not, say, clear lemonade that looks just like water?
A: No, I guess I can’t.
B: Shouldn’t you confirm your belief that the liquid is water, given your life-threatening allergic reactions to lemons and lemon juice?
A: I guess I should confirm, that is, if I want to avoid a life-threatening situation.
B: So, shouldn’t you also retract your knowledge claim?
A: No. I knew that the fluid was water until at 5:15 PM ET you raised the stakes of the situation by asking your questions, thereby changing the epistemic context and raising the standard for knowledge. At 5:15, before you raised the standard, I knew. Now, at 5:17, because your questions raised the standard, I don’t know. Had you not raised the standard, I would still know.
B: That seems strange. Has your internal mental state with respect to the fluid in the glass changed?
A: I guess it hasn’t.
A’s position seems wrong for several reasons. First, how can knowledge standards change so easily and quickly? Second, why does B’s asking of questions change the standard? Isn’t the standard already fixed by facts such as A’s fallible evidence and A’s allergy? And if so, why not think that extra-mental facts such as one’s evidence and background information invariantly determine the standard for knowledge? Third, how can extra-mental factors change mental states? How can the asking of questions and the stating of facts regarding allergies move one from the internal state of knowledge to the internal state of ignorance? Fourth, A’s position seems contradictory: he both knows and doesn’t know that the fluid is water, although admittedly, he knows at one time and doesn’t a moment later. Fifth, why exactly is A’s mental state before the change different from his state after the change? What precisely is the difference?
One might reply that once B raises the standard with his questions, there is no going back. The standard is now higher, and thus A cannot claim to have known that the liquid was water based on a lower standard. He now must live up to the new standard and hence must concede that his previous claim to knowledge was false. He thus avoids the apparent contradiction by admitting that he doesn’t know and didn’t know prior to B’s questions. But this reply seems problematic. First, it seems to beg the question in favor of EC. Arguably, B didn’t raise the standard by asking the questions. He merely recognized and revealed the standard via his questions. The standard was already present in that situation. B merely made the standard known to A. The standard was invariant throughout the dialogue. Second, why does raising a standard annul the previous lower standard with respect to the previous ascription of knowledge? This seems inconsistent with the contextualist position, which seeks to recognize the legitimacy of both the lower and the higher standards.
VI: The contextualist grants that the skeptic is operating with a very high and perhaps pure standard for knowledge, a standard which is legitimate in cases in which the stakes are high. But this concession seems to corroborate the skeptic’s intuition that his standard is, in fact, the precisely correct one for knowledge and that the anti-skeptic is misled by the illusion of our common language usage – a usage which motivates the false conclusion that knowledge has lower (and perhaps impure) standards in lower stakes cases. The skeptic can use the contextualist’s concession to argue that knowledge has a very high standard across all cases. (The skeptic can grant that we commonly and conveniently use ‘knowledge’ and ‘know’ in practical cases in which we really don’t know, such as when someone claims to know that the animal in the field is a sheep, although it is really a sheep-dog, or when someone claims to know that he is getting the job when in fact someone else has been given the job, or when someone claims to know that the grocery store is open and it is, in fact, open, though it might well have been closed, and the person who claimed that it is open merely got lucky. But these cases, according to the skeptic, are matters of convenience involving loose speaking. They are not correct uses, precisely speaking.)
These are just a few objections to EC. There are others. Given these problems, to my mind, EC is not a persuasive position. But I am open-minded and ready to continue reflecting on the issue.
[1] It is important to note that there are different versions of EC. For the sake of space, I won’t address the different versions.
[2] Some writers say that ‘know’ is (or at least functions like) an indexical term such as ‘here’ or ‘I.’ On this account, the meaning of ‘know’ varies according to context like the meaning of ‘I’ varies according to context. Yet the fact that we use ‘I’ differently relative to context doesn’t entail any particular metaphysical position on the nature of the self. Similarly, the fact that we use ‘know’ differently according to context doesn’t entail any particular philosophical position on the nature of knowledge.
[3] I say that this is a “supposed” threat because to assert that skepticism is a legitimate threat to knowledge seems to presuppose that we have it and hence that skepticism might take away something that we have. Such language seems to beg the question against the skeptic, who says that we don’t have knowledge in the first place (either not at all, or not with respect to particular topics) and thus there is nothing to take away. Also, if skepticism is in fact true, then that is how the world is, and hence it seems we ought to accept this fact. In this case, there wouldn’t be any threat to possessed knowledge, though perhaps there would be a threat to many persons’ beliefs and desires that we have knowledge which we don’t in fact possess.
[4] For instance: knowledge of whether or not a berry is poisonous in situations in which one must decide whether or not to eat the berry. I should note that reference to existential encroachment might be more relevant to a similar but competing view of knowledge attributions called subject-sensitive invariantism. Epistemic invariantism (EI) is the position that the truth value of sentences of knowledge attribution such as “John knows that the grocery store is open now” does not shift according to conversational context. There are different versions of EI, such as pure invariantism (which might take skeptical or non-skeptical forms), pragmatic invariantism, and subject-sensitive invariantism. In any case, it is worth noting here that some contextualists speak as though matters of encroachment are relevant factors in one’s conversational context rather than pertaining only to one’s subject-sensitive situation.
[5] For example, whether knowledge requires epistemic certainty, whether ontological naturalism is true, whether human life has a transcendent purpose, whether one has a soul, whether it needs saving, whether humans have free will, whether there is an afterlife, or whether suicide is morally acceptable, and if so in which cases, etc.
[6] E.g., whether or not the tree in my backyard exists, whether or not it’s an elm, situations in which two friends are having an informal chat about whether or not to use the tree to hang a birdhouse, etc.
[7] E.g., one might get confused. A form of rational dizziness might set in, as in the case of Meno, who, in Plato’s eponymous dialogue, complained that when encountering Socrates’ philosophical method felt as if he had been stung by a sting ray.