Skip to content

Now You Know It, Now You Don’t

Keith DeRose
Yale University

This should be fairly close to the text of this paper as it appears in the Proceedings of the Twentieth World Congress of Philosophy (Bowling Green, Ohio: Philosophy Documentation Center, 2000); Vol. V, Epistemology: 91-106.

1. Introduction: Contextualism, Invariantism, and Some Objections
2. Yourgrau’s Dialogue
3. Rejecting the Dialogues
4. Describing the Future and Describing Counterfactual Situations
5. Elusive Knowledge?
6. Lewis and “Semantic Ascent”
7. The Fallacy of Semantic Descent

1. Introduction: Contextualism, Invariantism, and Some Objections

“Contextualism” will here refer to the position that the truth-conditions knowledge-ascribing and knowledge-denying sentences (sentences of the form “S knows that P” and “S doesn’t know that P” and related variants of such sentences, henceforth, “K-sentences”) vary in certain ways according to the context in which they are uttered. What so varies is the epistemic standards that S must meet (or, in the case of a denial of knowledge, fail to meet) in order for such a statement to be true. In some contexts, “S knows that P” requires for its truth that S have a true belief that P and also be in a very strong epistemic position with respect to P, while in other contexts, the very same sentence may require for its truth, in addition to S’s having a true belief that P, only that S meet some lower epistemic standards. Thus, the contextualist will allow that one speaker can truthfully say “S knows that P”, while another speaker, in a different context where higher standards are in place, can truthfully say “S doesn’t know that P”, though both speakers are talking about the same S and the same P at the same time.

The “invariantist” — Peter Unger’s good name for one who denies contextualism(1) — will have none of this. According to her, there’s a single, invariant set of standards which, at least as far as truth-conditions go, govern the use of K-sentences regardless of the context in which they’re uttered. (Invariantists often hold that the warranted assertability conditions of K-sentences vary with context.) Thus, according to the invariantist, our two speakers can’t both be speaking a truth.

Resistance to contextualism comes in the form of many very different types of objections. My topic here is a certain group or family of related objections to contextualism that I call “Now you know it, now you don’t” objections. I responded to some such objections in my “Contextualism and Knowledge Attributions”(2) a few years back. In what follows here, I will expand on that earlier response in various ways, and, in doing so, I will discuss some aspects of David Lewis’s recent paper, “Elusive Knowledge.”(3)

Many seem to think that contextualism would counsel one to say some very odd things. If the standards for knowledge have gone up during the course of a conversation so that, while you did meet the standards that were in place earlier in the conversation, you fail to meet the standards that have been newly installed, it apparently can seem to some (in fact, to many) that, according to contextualism, you’ve lost knowledge, and should say something along the lines of “I did know earlier, but I no longer know.” That you’ve lost knowledge or should say such a thing certainly does seem absurd when the only change that’s occurred is a conversational change which seems to have no affect on how strong an epistemic position you’re in with respect to the proposition in question. Fortunately, these objections are based on a bad misunderstanding of contextualism’s commitments, for, according to contextualism, in the situations in question, you’ve lost no knowledge, and should say no such thing.

2. Yourgrau’s Dialogue

Setting aside for now the question of what, if anything, one has lost in such a situation if contextualism is true, let’s begin by considering whether the contextualist would have you say some absurd thing along the lines of “I did know, but now I don’t.” I hear such charges against contextualism often. And it has found its way into print. Palle Yourgrau, for instance, constructs the following conversation, and rightly observes that “something is amiss” in it:

Dialogue 1
A: Is that a zebra?
B: Yes, it is a zebra.
A: But can you rule out its being merely a cleverly painted mule?
B: No, I can’t.
A: So, you admit you didn’t know it was a zebra?
B: No, I did know then that it was a zebra. But after your question, I no longer know.(4)

Yourgrau’s specific target is a contextualist version of the Relevant Alternatives theory of knowledge (RA). According to RA, to know that P is roughly to have a true belief that P and to be in a position to rule out all the relevant alternatives to — or contraries of — P. Contextualist versions of RA will allow that what transpires in a conversation can affect which alternatives are relevant to knowledge attributions made in that conversation.(5) For instance, some contextualist versions of RA would have it that, in Yourgrau’s dialogue, A’s mentioning of the possibility that what B is seeing is a cleverly painted mule may make that a relevant alternative to the animal’s being a zebra. Thus, since B presumably can’t rule that alternative out, B may no longer count as knowing that the animal is a zebra in a context in which that bizarre alternative has been made relevant, though B did count asknowing that the animal is a zebra before any such bizarre alternatives were made relevant. But Yourgrau’s complaint would seem to apply to any contextualist view according to which A’s question manages to raise the standards for knowledge so that B no longer counts as knowing, whether or not the view is an RA one (according to which this raising of epistemic standards consists in an expansion of the range of alternatives that are relevant in context). The complaint is that B’s last line is absurd, but is what B should say if contextualism were true.

It’s interesting to note that, on one understanding of what’s transpiring in Yourgrau’s dialogue, what B says could well be true. This should be admitted by just about any non-skeptic, whether contextualist or not. For any non-skeptic should think that B could know that the animals he’s seeing are zebras. And A’s question really could make B cease knowing if it renders B less confident of the fact that what he’s seeing is a zebra, for B’s level of confidence clearly is relevant to the question of whether or not she knows. In an extreme case, A’s question could make B cease believing that the animal is a zebra, if, for instance, B takes the fact that A has asked the question to be evidence that it’s quite likely that cleverly painted mules are being used by the zoo. (Perhaps A has heard that, due to a zebra shortage, many zoos have taken to using cleverly painted mules in their zebra cages.)

But I think it’s clear that Yourgrau didn’t intend for his dialogue to be understood in such a way. Rather, we’re to suppose that A has no particular reason to think that the zoo is using painted mules in its zebra cage. In a spirit of idle skepticism, he’s just raising the possibility for no particular reason (other than that of engaging in some pleasant, idle skepticism), and B realizes this. Still, one might worry that, even so, B’s level of confidence is diminished, since she is now considering a possibility of error that she normally wouldn’t even think of. But to suppose that B’s level of confidence really is diminished muddies the waters considerably, for, as I noted above, that could well make what B says true, even on non-contextualist theories, and Yourgrau intends to be raising a problem specific to the contextualist. We can clear up the mud either by stipulating that B’s confidence is not diminished by what transpires in the conversation, or, perhaps more helpfully, by moving to a third-person version of Yourgrau’s dialogue. So, suppose that A and B are watching someone known to them, say, Henry, who’s at the zebra cage at the zoo. Henry is about 100 yards away from A and B, and is no party to their conversation, when this dialogue transpires:

Dialogue 2
A: Does Henry know that that’s a zebra?
B: Yes, he knows.
A: But can he rule out its being merely a cleverly painted mule?
B: No, he can’t.
A: So, you admit he didn’t know it was a zebra?
B: No, he did know then that it was a zebra. But after your question, he no longer knows.

Now, since Henry is well out of earshot, there’s no worry that his level of confidence is diminished by A’s question. B’s last line is absurd, but isn’t it precisely what contextualism would counsel B to say?

Well, for many reasons, no. But let us set aside here one way that contextualists may be able to avoid licensing B’s absurd last line in either of our two dialogues. Some contextualists may think that, although the standards for knowledge can vary according to conversational context, it’s not all that easy to change the standards, and, in particular, that the mere mention of a bizarre alternative like painted mules is not sufficient to render that alternative relevant. I actually have a lot of sympathy for the thought that the mere mention of the alternative is not sufficient for making it relevant, but note that in our dialogues we do have more than the mere mention of the alternative. When A mentions the possibility of cleverly painted mules, there are various ways for B to resist allowing that alternative in as relevant. B could reject that alternative by, for example, saying “Painted mules, my eye! C’mon! That’s absurd! Get outta here with that crazy idea!” A possible contextualist view is that the mentioning of an alternative makes that alternative relevant if one gets away with making that alternative relevant, where one fails to get away with making it relevant if one’s interlocutor rejects the alternative’s relevance in some way like we’ve just imagined.(6) But in our dialogues, B does not resist allowing the alternative in as relevant. A does seem to get away with making the cleverly painted mule hypothesis relevant. So we have more than mere mentioning in our example; we have mentioning plus some form of non-rejection. (Indeed, beyond mere passive non-rejection, we seem to have some sort of positive acceptance.)

Still, on some possible contextualist views, even that may not be enough. Some contextualists may think that standards can only be raised if elevated standards are somehow appropriate to the context. So, for instance, many contextualists seem to think that higher-than-usual standards are appropriate when the stakes are unusually high, where it’s unusually important that the person in question be right. On one such possible contextualist view, mentioning a heretofore irrelevant alternative may make that alternative relevant only if the stakes are high enough to make it appropriate to allow that alternative in as relevant. Some such contextualists may deny that A’s mentioning of the painted mule hypothesis, even together with B’s apparent acceptance of that possibility, makes that alternative relevant. Such a contextualist can escape Yourgrau’s problem by claiming that A has failed to raise the standards for knowledge to such a level as to make B count as a non-knower, and thus B has no business saying she no longer knows.

But any contextualist will allow that under some circumstances, the standards for knowledge can be raised to unusually high levels. And regardless of what a particular contextualist thinks does manage to raise the standards, she will face a Yourgrau-like dialogue, ending with an absurd-sounding last line of the form “I did know then, but I no longer know now that…” where the line is completed by whatever it is that the contextualist in question thinks does manage to raise the standards. So I propose that contextualists not respond in the way we’ve been considering. So let’s just suppose (or pretend, if need be) that A has managed to raise the standards for knowledge in our dialogues to the point at which B and Henry no longer count as knowing that the animal is a zebra. It will turn out that, even so, there really isn’t much of an objection here.

3. Rejecting the Dialogues

The most important line of response to this objection is that, even where the standards for knowledge have been raised so that the subject in question no longer meets them, contextualism simply doesn’t predict that the speaker will say that the subject — whether the subject is the speaker himself (as in Dialogue 1), or some third party like Henry (Dialogue 2) — did know but no longer does. Putting in brackets one new word that I’m adding now, here’s how I responded a few years ago to the Yourgrau’s original dialogue (our Dialogue 1):

How shall the contextualist respond? The objection as I have put it forward, though it explains much of the initial resistance many feel toward contextualism, is based on a mistake. The contextualist believes that certain aspects of the context of an attribution or denial of knowledge affect its content. Knowledge claims, then, can be compared to other sentences containing context-sensitive words, like “here.” One hour ago, I was in my office. Now I am in the word processing room. How can I truly say where I was an hour ago? I cannot truly say, “I was here,” because I wasn’t here; I was there. The meaning of ‘here’ is fixed by the relevant contextual factors (in this case, my location) of the utterance, not by my location at the time being talked about.
Similarly, the contextualist may admit that the mentioning of the painted mules possibility affects the conditions under which one can truthfully say that one knows an animal to be a zebra: one now must be able to rule out that possibility, perhaps. But the contextualist need not, and should not, countenance the above dialogue. If in the context of the conversation the possibility of painted mules has been mentioned, and if the mere mention of this possibility has an effect on the conditions under which someone can be truly said to “know”, then any use of “know” (or its past tense) is so affected, even a use in which one describes one’s past condition. B cannot truly say, “I did know then that it was a zebra”; that would be like my saying “I was here.” B can [truthfully] say, “My previous knowledge claim was true,” just as I can say, “My previous location claim was true.” Or so I believe. But saying these things would have a point only if one were interested in the truth-value of the earlier claim, rather than in the question of whether in the present contextually determined sense one knew and knows, or didn’t and doesn’t.(7)

The objection seems to assume that, in describing someone’s past condition, the standards for knowledge are set by what the conversational context was at the time being talked about. My response was, and is, that it’s open to the contextualist to hold that when a speaker describes a past time, the standards that govern such talk are those set by the conversational context at the time of the speaker’s utterance. Indeed, this option is more than just open to the contextualist. Given what happens with context-sensitive terms in general, we should expect the epistemic standards to be set by the context at the time of utterance. Though it’s possible, I suppose, for there to be a context-sensitive term which operates in the way the objector to contextualism assumes (when describing a past situation, the term takes on the sense that would be determined by the context at the time being talked about), that’s not how context-sensitive terms usually seem to work. For instance, in describing my past location, the meaning of my use of “here” seems at least usually to be fixed by my present location, not by what my location was at the time being talked about. As I now say in New Haven, “When I was in Houston last year, David was here,” we all know that by “here”, I mean New Haven (my present location), not Houston (my location at the time I’m talking about).

4. Describing the Future and Describing Counterfactual Situations

It’s worth quickly noting that this lesson can and should be generalized to cover more than just talk about the past; talk about counterfactual situations and about the future should be given analogous treatment: When describing how things would have been or will be, as in describing what was, it’s the actual conversational context of the speaker at the time of the speaker’s utterance, rather than what would have been or will be the speaker’s context, that sets the epistemic standards. So don’t think that contextualism will have you say anything like the following last lines of Yourgrau-like dialogues:

If you hadn’t asked that, I [or Henry] would have known

After you ask that, I [or Henry] will no longer know.

5. Elusive Knowledge?

So contextualism won’t have you say anything like that. And in a closely related point, contextualism doesn’t imply that any knowledge appears or vanishes as conversational context changes.

It can seem to many that contextualism renders knowledge unstable or elusive in the sense that it would make our knowledge come and go — be gained or lost — as conversational context changes. And to many, this can seem a very problematic implication of contextualism, since it seems to them that knowledge in fact doesn’t appear and disappear due to changes in conversational context. I agree: Knowledge isn’t in that way elusive. But I insist that contextualism doesn’t make knowledge elusive in that way. Here I have to contend with friends as well as foes of contextualism. One of contextualism’s most prominent advocates, David Lewis, in his contextualist manifesto, “Elusive Knowledge,” makes many comments about the elusiveness of knowledge, including the following (emphasis added in all cases), which I’ve numbered for ease of reference. One bit of background: Lewis contends thatengaging in the practice of epistemology typically has the effect of raising epistemic standards. He writes:

1. Maybe epistemology is the culprit. Maybe the extraordinary pastime robs us of our knowledge. Maybe we do know a lot in daily life; but maybe when we look hard at our knowledge, it goes away. (p. 550)

2. In the strict context of epistemology we know nothing, yet in laxer contexts we know a lot. (p. 551)

3. Unless this investigation of ours was an altogether atypical sample of epistemology, it will be inevitable that epistemology must destroy knowledge. That is how knowledge is elusive. Examine it, and straightway it vanishes. (p. 560)

4. Imagine two epistemologists on a bushwalk. As they walk, they talk. They mention all manner of far-fetched possibilities of error. By attending to these normally ignored possibilities they destroy the knowledge they normally possess. (p. 565)

I certainly wouldn’t want to say any of the above. (Lewis, though he does write all the above, does seem to take it back in a way at the end of his essay, as we’ll see below in section 6. Still, he seems to me too willing to say such things.) If contextualism really implied that knowledge vanishes (3), is destroyed (3, 4), goes away (1), or is robbed from us (1) by means of conversational developments (often in conversations we’re no party to) that have no effect on how strongly we’re positioned with respect to the beliefs in question, then I’d probably reject contextualism. Thus, I can’t be surprised that others, who think contextualism has such implications, reject the view on those grounds. And I can’t be surprised that foes of contextualism think the view underwrites remarks such as those above, when one of contextualism’s strongest friends is the one making the remarks.

But let’s get clear, once and for all, about contextualism’s implications as the standards change so that a subject who used to meet them no longer does. To increase clarity, let’s consider a case involving third-person, rather than first-person, K-sentences. So: Henry’s over there, very confidently believing the true proposition P. Far away from him, Thelma and Louise are talking about whether he knows that P. Contextualism is true (we’ll suppose). To make things uncomplicated, let’s suppose that just two sets of standards come into play in Thelma and Louise’s conversation — “low” and “high” — and that throughout the story the strength of Henry’s epistemic position with respect to P remains constant at a level at which he meets the low, but not the high, standards. Initially, Thelma and Louise were in a context that selected the low standards, so they spoke the truth when they said, “Henry knows that P.” Now, however, their conversational context has changed; they’re now in a high context. So Thelma and Louise can no longer truthfully say that Henry knows, and in fact can now truthfully deny that Henry knows.

What has Henry lost? What has gone away, been destroyed, been robbed from him? “High” knowledge? No, he never had that. “Low” knowledge? No, that he still has.

Knowledge? Knowledge simpliciter? Well, if what’s meant by “knowledge simpliciter” is knowledge according to the one-and-only standards that ever govern attributions or denials of knowledge, then, of course, there’s no such thing on the assumption that contextualism is correct. (And if there were such a thing — if contextualism were wrong — then wherever those one-and-only standards are set, whether those uniquely correct standards are high or low, Henry has not lost this simple knowledge by means of Thelma and Louise’s conversational shenanigans: he either lacked it all along, or never had it.)

If you mean knowledge by our current standards, then, of course, it’s hard to say whether Henry has knowledge or lacks it. All I’ve specified so far about Henry is that he met (and meets) the “low” standards set by Thelma and Louise’s old context, that he fails to meet their “high” standards, and that the strength of Henry’s epistemic position didn’t change: he went from meeting to failing to meet due entirely to a change in their standards, and not at all because of a change in the strength of Henry’s epistemic position. No matter. We needn’t bother with more detail. Though we can’t yet tell whether or not Henry knows (whether or not he meets our current standards), we are already in a position to say whether or not Henry has lost knowledge of P. He hasn’t. Whatever our standards are right now, and however strongly or weakly Henry is positioned with respect to P, given only that the strength of his epistemic position hasn’t changed (and that we have been given), we can ascertain that Henry hasn’t lost knowledge. He of course hasn’t changed from meeting the standards we’re currently employing to failing to meet those same standards because of anything that transpired in Thelma and Louise’s conversation. He either knew and knows, or didn’t and doesn’t. Either way, he hasn’t lost his knowledge that P and no knowledge has gone away, been destroyed, or been robbed from him.

6. Lewis and “Semantic Ascent”

At the close of his paper, Lewis acknowledges that there’s something fishy about his saying that knowledge vanishes and that we first know and then fail to know:

But wait. Don’t you smell a rat? Haven’t I, by own lights, been saying what cannot be said? (Or whistled either.) If the story I told was true, how have I managed to tell it?…I said that when we do epistemology, and we attend to the proper ignoring of possibilities, we make knowledge vanish. First we do know, then we do not. But I had been doing epistemology when I said that. The uneliminated possibilities of error were not being ignored — not just then. So by what right did I say even that we used to know?

Lewis then admits that he “bent the rules” in making those comments — and I presume the same would go for the other comments we’ve been looking at. He then goes on to explain:

I could have had my say fair and square, bending no rules. It would have been tiresome, but it could have been done. The secret would have been to resort to ‘semantic ascent’. I could have taken great care to distinguish between (1) the language I use when I talk about knowledge, or whatever, and (2) the second language that I use to talk about the semantic and pragmatic workings of the first language. If you want to hear my story told that way, you probably know enough to the job for yourself. (p. 567)

And we do have to do the job for ourselves, for “Elusive Knowledge” comes to a close just one sentence later.

But it’s not hard to engage in a little semantic ascent and find substitutes for the object-level phrases I’ve been objecting to. Henry’s knowledge doesn’t vanish. He doesn’t first know and then fail to know. Rather, to be tiresome, we can say:

(A) First, Henry was such that the proposition expressed about him by the sentence “Henry knows” in Thelma and Louise’s conversation was true of him. But then (because their context changed so that “Henry knows” came to express a more demanding proposition) Henry was such that the (new) proposition that would have been expressed by “Henry knows” in Thelma and Louise’s context was not true of him.

But we don’t have to be that tiresome. There are devices for engaging semantic ascent that don’t involve explcitly putting K-sentences in quotation marks and explicitly saying things about the truth-values or truth-conditions of those sentences. With just a little set-up, one can, as I’ve been doing in this very paper, achieve that same effect by simply saying things along the lines of:

(B) First Henry met the epistemic standards set by Thelma and Louise’s context and then (because their standards went up) he failed to meet the standards set by their context.

Or, to be even less tiresome, we can say:

(C) First Henry counted as knowing, then he didn’t,

which really is only minimally more tiresome than is “First he knew, then he didn’t.” And, to explicitly deal with the numbered passage from Lewis’s “Elusive Knowledge” that I haven’t mentioned since displaying it, instead of Lewis’s

2. In the strict context of epistemology we know nothing, yet in laxer contexts we know a lot,

we can, and should, instead say the only slightly more tiresome,

(D) In the strict context of epistemology we count as knowing nothing, yet in laxer contexts we count as knowing a lot.

Lewis explains that he is engaging in an “expository shortcut, to be taken with a pinch of salt” (p. 566). But contextualists can utilize the likes of (B)-(D) in expository shortcuts to achieve the same effect as saying such truly tiresome things as (A). (B)-(D) build the tiresome “semantic ascent” of the likes of (A) into easy little phrases like “standards for knowledge”, and “counts as knowing”, which, when spelled out, amount to higher-level claims about the truth-conditions and truth-values of K-sentences. I’ve found that, in presenting contextualism, one or two (A)-ish statements (in which K-sentences are explicitly mentioned and something about their truth-conditions is explicitly stated) at the start of a paper or talk is all that’s needed for listeners or readers to be ready to understand the likes of (B)-(D). But for reasons we’ll look at in the next section, it seems a mistake to take rule-bending (breaking?) shortcuts of the likes embodied in phrases like Lewis’s 1-4.

7. The Fallacy of Semantic Descent

Don’t think that the likes of 1-4 express the contextualist’s initial position or consequences of it, and that the contextualist engages in “semantic ascent” — moves to likes of (A)-(D) — in order to get out of trouble. Rather, it’s only the likes of (A)-(D) that really express the contextualist’s commitments. The likes of 1-4 are only rule-bending and therefore potentially misleading shortcuts (obtained by some problematic semantic descent) to express (or mis-express) the contextualist’s commitments.

We might call the mistake of thinking that a theory or view has a certain implication, expressed in object-level language, when in fact its real implication is not what one thinks, but is rather some related higher-level statement, the “fallacy of semantic descent.” Theories according to which various types of sentences are held to be context-dependent in meaning are often subject to objections that depend on this fallacy. To get a feel for the fallacy in action, consider a case in which the higher level statement (the actual commitmentt) is uncontroversial. I assume we all believe that the phrase “this room” works in such a way that the following is true:

(E) It can happen that one person can truthfully say, of a physical object, A, “A is (wholly) in this room,” while, at the very same time, another person (who is in another room) can truthfully say of the very same physical object, “A is not (wholly) in this room.”

Now, suppose that someone objected to this view of yours by first claiming that, according to your view,

(F) One and the same physical object can at one and the same time both be and not be wholly in this room,

and then going on to suggest that (F) is absurd, counterintuitive, or implausible. The proper response to this objection, I take it, is to reject (F), to point out that it’s (E), and not (F), that you accept, and to point out that (F) does not follow from (E). To suppose that (E) commits you to (F) is to commit the fallacy of semantic descent.

It is not always easy to determine whether the fallacy of semantic descent is being committed. It is sometimes unclear whether the objector is really confusing his target’s commitments or whether she is rather, to use Lewis’s words, using the object level statement as an “expository shortcut, to be taken with a pinch of salt” to express her target’s commitments. In our above example, we might wonder whether our objector really thinks we believe or are committed to (F), or whether she’s just using (F) as a rule-bending device by which she hopes to express our real commitment, (E). After all, it seems we could — though it seems inadvisable — more or less explicitly agree to use (F) as a handy piece of rule-bending shorthand by which we’ll express (E). In that case, we should all remember that that’s what’s going on, and we should be careful not to come to doubt (E) because of how bad (F) sounds or how absurd (F) seems. After all, we had to break the rules to use (F) as a device for expressing (E). If (E) implied something that seems absurd, that would be grounds for doubting (E). But that an absurd-sounding statement can, through a rule-bending piece of shorthand, be used to express (E) casts no shadow on (E). At any rate, in the absence of any such explicit agreement, we might be left to wonder whether our objector is really committing the fallacy of semantic descent, or whether he rather just considers (E) itself to be crazy.

Consider just one actual, published example, whose target is not contextualism regarding K-sentences, but is rather a theory according to which a certain type of modal expression is held to be such that its “meaning is relative to the person who uses it.” In his essay “Certainty”, G.E. Moore (the target) considers statements of the form, ‘It is possible that p is true’. Engaging in some not-all-that-tiresome semantic ascent (note his talk about the meaning, truth-conditions, and truth-values of such statements), Moore writes the following:

The expression ‘It is possible that p is true’ is, though it looks as if it were impersonal, really an expression whose meaning is relative to the person who uses it. If I say it, that I should not know that p is false, is a necessary, though not a sufficient, condition for the truth of my assertion; and hence if two people say it at the same time about the same proposition it is perfectly possible that what the one asserts should be true, and what the other asserts false: since, if one of the two knows that p is false, his assertion will necessarily be false; whereas if the other does not know that p is false, his assertion may be, though it will not necessarily be, true.(8)

Now all of Moore’s main claims in the above passage are pretty clearly correct.(9) But Alan R. White objects to Moore’s account, writing, for instance, of “Moore’s mistake…of supposing that, since what is not known to me may be known to you, the same thing may be both possible and impossible.”(10) But Moore would not say or write that “the same thing may be both possible and impossible”. At least I’ve never caught such a statement in his writing. What he writes is:

(G) If two people say it [‘It is possible that p is true’] at the same time about the same proposition it is perfectly possible that what the one asserts should be true, and what the other asserts false.

And there are cases that establish that he’s absolutely right about that.(11) And (G) simply doesn’t imply any statement like

(H) The same thing may be both possible and impossible

I’m uncertain whether White is committing the fallacy of semantic descent here. I’m inclined to think that he is. (But read the first 52 pages of Modal Thinking and judge for yourself.) That is, I suspect that White thinks that (H) is something Moore is committed to. (Further, I think it’s because (H) strikes White as implausible that he rejects Moore’s account of the modal statements in question.) In fact, Moore is not committed to (H). (Indeed, more: Moore’s views about what “possible” and “impossible”, in the relevant sense, mean predict that what (H) expresses is, at the very least, highly problematic.) But maybe I’m wrong and White realizes that Moore is not committed to (H). Perhaps White is just “bending the rules” and using (H) as an “expository shortcut” by which he hopes to express Moore’s real commitment, (G). But then I don’t understand what White has against Moore’s account. Unless White considers (G) itself to be crazy. In fact, it’s because I really can’t see why anyone would think that (G) itself is clearly false that I suspect that White rejects Moore’s account at least in part because he mistakenly thinks that Moore is committed to (H).

There are places in the published literature where it at least seems that the fallacy of semantic descent is being committed against contextualist accounts of K-sentences. (We’ve seen one here: I would count Yourgrau’s use of his dialogue as an example of the fallacy. The contextualist is not committed to licensing B’s crazy last line. That line is just an object-level analogue of some meta-linguistic statements that, according to contextualism, B could appropriately and truthfully utter (were he philosophically sophisticated enough). Some other examples involve claims that contextualism implies statements of the types we’ve encountered in our Lewis passages 1-4.) But, in conversation, in my philosophical travels, I believe that, even among some very smart philosophers of language and of mind — I say “even” them because they’re the ones who, given their area, really should know better –, objections based on the fallacy of semantic descent form the single most common type of objection I’ve encountered. That is, the most common type of objection — at least as I divide types — is roughly of the form, “According to you, ….”, or “According to contextualism, …” where the complaint is completed by filling in some pretty implausible-sounding statement that I would in fact reject, that my views would lead me to reject, and which is such that the best explanation for why it’s being attributed to me is that it’s being confused with (or mistakenly taken to follow from) some higher level statement which I really do accept.

Let me quickly give one quite common example that’s a bit unlike the other examples of the fallacy that we’ve seen in this paper, in that it doesn’t involve Lewis-like statements about knowledge “vanishing” or “going away”, etc. I’ve frequently heard, occasionally from some very smart philosophers, roughly this: “According to you, whether or not Henry knows depends on our context. But how can Henry’s knowledge depend on our context?!!” Now, I would never say,

(I) Whether or not Henry knows depends on our context.

In fact, my views would warn me against saying any such thing. But what I would say is such things as,

(J) Whether we can truthfully assert various K-sentences about Henry depends (at least in part) on our context.

This because the content of those assertions are context-dependent, according to me. Now, maybe some of these objectors realize that I really reject (I) and they are just using (I) as what they hope is an obvious expository shortcut to express my real view, (J). This is unlikely, though, because when I do tell such objectors that I reject (I), they’re typically surprised by this. I have to suspect that such objectors are at least usually, and at least to some extent, committing the fallacy of semantic descent.

In light of this situation, I think it unwise to engage in any rule-breaking expository shortcuts to express the contextualist’s views and commitments. If folks find the contextualist’s actual commitments — like (J), above — to be strongly implausible, then, while I can’t myself understand why they think this, that’s fair enough. But given what seems to be a general tendency to be misled by the fallacy of semantic descent, so far from actually using the rule-bending shortcuts, like (I), to express the contextualist’s ideas, we should instead be explicitly warning folks to keep those misleading statements out of their minds as much as possible as they evaluate contextualism.


1. See Peter Unger, Philosophical Relativity (Minneapolis, Minn.: University of Minnesota Press, 1984).

2. “Contextualism and Knowledge Attributions,” Philosophy and Phenomenological Research 52 (1992): 913-929; see especially section III, pp. 924-928.

3. Australasian Journal of Philosophy 74 (1996): 549-567.

4. Palle Yourgrau, “Knowledge and Relevant Alternatives,” Synthese 55 (1983): 175-190; p. 183. Such a conversation, along with the worry that it creates problems for contextualism, was originally suggested to me by Rogers Albritton, who has been making such suggestions since well before Yourgrau’s article appeared.

5. For more on RA, and especially its relation to contextualism, see section II, pp. 918-929 of my “Contextualism and Knowledge Attributions.”

6. Here I’m loosely following a lead of David Lewis’s; in “Scorekeeping in a Language Game,” Journal of Philosophical Logic 8 (1979): 339-359, he writes: “If you say ‘Italy is boot-shaped’ and get away with it, low standards are required and the standards fall if need be” (p. 352, emphasis added).

7. “Contextualism and Knowledge Attributions,” p. 925.

8. G.E. Moore, Philosophical Papers (London: George Allen & Unwin LTD, 1959), p. 241.

9. For a vindication of Moore’s claims, see my UCLA dissertation, Knowledge, Epistemic Possibility, and Scepticism, pp. 58-72.

10. Alan R. White, Modal Thinking (Ithaca: Cornell UP, 1975), p. 52.

11. See, for instance, section 1 of my “Epistemic Possibilities,” Philosophical Review, 1991.

Skip to toolbar