1. DIFFERENTIATING CONTEXTUALISM AND THE AMBIGUITY THEORY OF ‘KNOWS’

In keeping with standard usage, I use the phrase epistemic contextualism, or more simply contextualism, to refer to the view that the truth conditions of knowledge ascriptions and denials (e.g. sentences of the form “S knows that P” and “S doesn't know that P”), in virtue of making a knowledge claim, shift in accordance with certain changes in the context of utterance. Or, to put it another way, contextualism is the view that the truth conditions of knowledge ascriptions and denials depend, in part, upon the attributor's conversational context because of the way in which ‘knows’ contributes to the claim.Footnote ⁵ Among the relevant aspects of context are both factors internal to the utterer (e.g. the utterer's purposes, interests, stakes, etc.) and factors external to the utterer (e.g. the conversation the utterer is contributing to, the community the utterer is a part of, etc.).Footnote ⁶

I also take the following clarification from Patrick Rysiew to express something essential about the view:

As Geoff Pynn has recently noted “[c]ontextualism is a broad tent,”Footnote ⁸ and the linguistic mechanisms by which context allows for the completion of a proposition differ among alternative types of contextualism. However, what remains essential in all cases is that context is required to recover the complete meaning and/or truth conditions of knowledge ascriptions and denials,Footnote ⁹ and that a shift in context can control a shift in truth conditions by changing the proposition expressed. This point will be useful later in clarifying the distinction between contextualism and the ambiguity theory of ‘knows’.Footnote ¹⁰

The ambiguity theory of ‘knows’ combines two theses. The first is that ‘knows’ has more than one sense – specifically more than one propositional sense, i.e. a sense that can properly be used in ‘knows that’ constructions.Footnote ¹¹ The second is that which sense of ‘knows’ is being employed plays a role in fixing the truth conditions of a knowledge ascription (in virtue of contributing to the meaning of the ascription). Thus, so long as the ambiguity theorist doesn't hold a view that requires that context be determined in order to fix meaning in general, the ambiguity theory allows that the meaning and truth conditions of a knowledge ascription can be determined even when the context is not. That is to say, the truth conditions of knowledge ascriptions can be known on the ambiguity theory even if the attributor's context is unknown or if the ascription is presented acontextually (if such a thing is possible).

It has often been noted that contextualism entails the following.

This upshot of contextualism plays an important role in the work contextualism is able to do. The ambiguity theory also entails SCWI. This is because on the ambiguity theory the first speaker can be using one sense, while the second speaker is using another sense, which results in the two sentences having different truth conditions that can pull apart as SCWI claims. As a result, the ambiguity theory also reaps the same benefits that contextualism gains from this upshot.

While until recently the ambiguity theory has not been much defended (or even discussed), it is a view that has been previously suggested and endorsed. Two such early ambiguity theorists are Norman MalcolmFootnote ¹² and Fred Feldman.Footnote ¹³ Malcolm's and Feldman's positions are very similar. Both posit two senses of ‘knowledge’ (what DeRose calls a “high” and a “low” senseFootnote ¹⁴) one of which is ‘knowledge’ in a rigorous, high-standards, “philosophical” sense and the other ‘knowledge’ in a practical, every-day sense.

While both Malcolm and Feldman adopt a “two senses” view of ‘knowledge, it is important to note that the ambiguity theory itself doesn't say anything about the number of senses ‘knows’ has or the conditions under which these different senses are satisfied.Footnote ¹⁵ This point has been noticed by Matthias Steup who has more recently advocated an ambiguity theory-like position which he dubs the “Multiple Concepts Theory” (MCT).Footnote ¹⁶ Steup writes that “[a]ccording to MCT, there are as many concepts of knowledge as there are different standards of knowledge.”Footnote ¹⁷ While Steup chooses to frame his discussion in terms of the multiple concepts of knowledge instead of the multiple senses of ‘knowledge,’ I take MCT to be completely compatible with the ambiguity theory.Footnote ¹⁸

Another recent advocate of a form of the ambiguity theory is Baron Reed. The metaphysical picture Reed advocates is that “knowledge in general is a determinable; the different knowledge relations that are grounded in particular degrees of justification are its determinates.”Footnote ¹⁹ He compares this to the nature of colors writing that, “[t]he same sort of metaphysical structure can be found in color: blue, for example, is a determinable with the various particular shades of blue as its determinates.”Footnote ²⁰ Reed combines this metaphysical account of blue with the following semantic account of ‘blue’: the “various usages of ‘blue’ are possible because it is ambiguous.”Footnote ²¹ And Reed takes the same semantic account to apply in the case of ‘knows.’Footnote ²²

The fact that the ambiguity theory itself doesn't say anything about the number of senses ‘knows’ has is a particularly important point because it shows that what distinguishes the ambiguity theory from contextualism is not the number of senses the theory posits. This point has sometimes gone unnoticed. For example, DeRose writes that,

By taking Malcolm's two senses view of ‘knowledge’ to be a “limiting case” of contextualism, DeRose implies that the only relevant difference between Malcolm's view and views in the family of positions championed by the contemporary defender of contextualism is the number of epistemic standards governing the truth conditions of knowledge attributions. This seems to be a mistake. After all, if there is nothing other than the number of senses posited that separates the ambiguity theory from contextualism, we lack an explanation as to what differentiates Reed's theory from the contextualist camp.Footnote ²⁴ Ambiguity theories of ‘know’ that are multivocal like Reed's and Steup's prompt one to note that there is a second, important difference between a view like Malcolm's and a view like DeRose's.

This second difference deals with what is required in order for a given utterance or sentence expression of a knowledge ascription or denial to pick out a proposition. On contextualism, what is required (in addition to the things we standardly think are required) is a context. On the ambiguity theory a context is not required, but a sense of ‘knows’ is. This in turn results in a difference between contextualism and the ambiguity theory as to what determines the truth conditions of a knowledge ascription or denial. For the proponent of contextualism, context is one such determinant, but for the proponent of the ambiguity theory context is not (at least not in any way beyond the way in which context may always be required). This difference between contextualism and the ambiguity theory is no mere difference in degree as is the difference between a view that advocates two epistemic standards versus a view that advocates more than two epistemic standards. This difference between contextualism and the ambiguity theory is a difference in kind.

In order to see clearly that this is a difference in kind and not just degree, one needs to clearly see the distinction, linguistically, between ambiguity and context-sensitivity. Adam Sennet makes the point nicely, writing that

Sennet's examples clearly show that a word can be ambiguous without being context-sensitive and that a word can be context-sensitive without being ambiguous. While both ambiguity and context-sensitivity allow for a change in content across context (as is evident in our uses of both ‘I’ and ‘bank’), with context-sensitive words the change in semantic content is due purely (to use Sennet's phrase) to the change in context, while with ambiguous words context plays a role relegated to the pragmatic where context merely makes it likely or sensible that a change in semantic content may occur via a change in the speaker's intention with certain changes in context.Footnote ²⁶ Even if one were to make the case that ‘bank’ is context-sensitive, the context-sensitivity of ‘bank’ wouldn't be because it was ambiguous and conversely with ‘I.’ This is, in part, because ambiguity is the non-context-bound property of a word having more than one meaning, any of which can be the meaning used in any context where the appropriate syntax is present,Footnote ²⁷ while context-sensitivity is the property of content variation resulting from a change in context. This distinction is in keeping with the general assumption in linguistics that context-sensitivity and ambiguity are distinct linguistic phenomena.Footnote ²⁸

One may acknowledge this difference in kind between context-sensitivity and ambiguity, but still want to resist that there is a difference in kind between contextualism and the ambiguity theory by suggesting that it is a specific element of context which fixes the sense of ‘knows,’ with that specific element of context being speaker's intention. There is a certain initial plausibility to this counterargument. After all, Cohen includes speaker's intentions among the elements that make up context,Footnote ²⁹ and in places DeRose gives a great amount of influence to speaker's intentions to determine context.Footnote ³⁰ It is true that a speaker's intention is part of the context of the utterance, and on the account of speaker meaning which I currently favor, it is the speaker's intention which fixes the sense in cases of ambiguity.

However, there are three important points to be made in response to this suggestion. First, in looking at the popular thought experiments used to ground contextualism – like DeRose's bank cases and Cohen's airport case – it seems clear that speaker's intention is only one of several constituents making up the speaker's context – including other constituents that are internal to the speaker (e.g. the utterer's purposes, interests, stakes, etc.) along with constituents that are external to the speaker (e.g. the conversation the utterer is contributing to, the community the utterer is a part of, etc.). The upshot of this is that, despite the fact that the exact boundaries of what counts as context for the proponent of context are often not made explicit, it is clear that much more than speaker's intention is typically taken to be part of context. Thus, even if we consider anything determined by speaker intention to be context-sensitive, the proponent of the ambiguity theory differs in kind from the traditional proponent of contextualism because the proponent of the ambiguity theory rejects the additional elements of context beyond speaker intention as having a role in determining the truth conditions of knowledge ascriptions and denials. (Or at least this will be the case for an ambiguity theorist who takes the sense of an ambiguous term to be determined by speaker's intention.)

Second, as Reed has noted, having determination of sense by speaker's intention count as a type of context-sensitivity makes ‘knows’ context-sensitive in only a very trivial sense that would certainly also make all ambiguous words context-sensitive along with just about every other word as well.Footnote ³¹ This seems to be a good reason to reject mere sensitivity to a speaker's intention as grounds for context-sensitivity.Footnote ³²

Third, this explanation seems to undermine the distinct ways in which we use context in relation to meaning. John Perry makes a useful distinction between the semantic and the presemantic which he describes in the following way,

I (along with PerryFootnote ³⁴) take this difference between ‘I’ and ‘bank’ to be a case of such a difference – the way in which context determines the meaning of ‘I’ is clearly semantic. That is, context remains relevant after the language, words, and meanings are all known. The task of determining which sense of ‘bank’ is being employed is a presemantic task. We use context to figure out what meaning a speaker has employed, but once we've successfully done that, there is nothing especially context-sensitive about ‘bank’ that remains. It seems clear to me that contextualists have traditionally seen context playing a robust, semantic role that cannot be relegated entirely to the presemantic. But part of what the ambiguity theory claims in stating that the variability in meaning is a result of an ambiguity in the term ‘knows’ is that the role context is playing is solely presemantic and as a result purely pragmatic. This seems to straightforwardly constitute a difference in kind.

That being said, my primary goal in this section has been to make clear the differences between the ambiguity theory and contextualism, as it is typically presented. I have done so by denying that the ambiguity theory is a type of contextualism, but at the end of the day my primary concern is not with taxonomy. If one wished to change or expand their characterization of contextualism such that the ambiguity theory counted as a type of contextualism, they are of course welcome to do so, but it is important that it be understood that there are key differences in kind that separate the ambiguity theory from traditional forms of contextualism and not merely a difference in the number of epistemic standards involved.

In what follows it will be useful to have a more specific form of the ambiguity theory in mind to compare to contextualism. Thus, for dialectical purposes only, I will endorse a “three senses” view of ‘knows,’ building upon the “two sense” positions adopted by Malcolm and Feldman.Footnote ³⁵ First, I'll employ a “low sense” of ‘knows’ such that it allows for most of our everyday knowledge ascriptions and denials to come out true. Second, I'll appeal to a “high sense” of ‘knows’ such that only very highly justified beliefs count as knowledge, but on which entitled epistemic certainty is not required. Third, I'll use a “certainty sense” of ‘knows’ such that the sort of justification that entitles one to epistemic certainty is required (i.e. such that the knowledge is infallible). These senses will be indicated at points with the labels knows_L (short for knows in the low sense), knows_H (short for knows in the high sense), and know_C (short for knows in the epistemic certainty sense).

2. A BRIEF INDEPENDENT CASE FOR THE AMBIGUITY THEORY

This paper is not meant to be a complete defense of the ambiguity theory. The independent linguistic grounding for the ambiguity theory, how the ambiguity theory fares compared to strict invariantism, and how we can explain the semantic blindness many of us have to the ambiguity of ‘knows’ (if it is in fact ambiguous) are all matters I've covered elsewhere.Footnote ³⁶ My fundamental concerns in this paper are showing that the ambiguity theory is distinct from contextualism and arguing that the ambiguity theory provides the same major benefits as contextualism.

But before making these arguments by comparison, it is worth briefly presenting some of the independent motivations for the ambiguity theory. After all, if contextualism already meets the goals of solving skeptical puzzles and preserving our everyday knowledge ascriptions, why look into another theory? The aim of this section is to provide initial philosophical and linguistic motivations for the ambiguity theory before arguing that the ambiguity theory shares the same major strengths as contextualism.

We'll start with a philosophical motivation. As noted in the previous section, both the ambiguity theory and contextualism entail the possibility of SCWI. However, note that only the ambiguity theory is compatible with the following.

Contextualism must reject DRWI because context-sensitivity (via whatever the semantic mechanism may be) allows sentences and/or utterances to pick out different propositions and to have varying truth-conditions across different contexts, but doesn't allow for a sentence to pick out varying propositions or to have varying truth-conditions within a particular context.Footnote ³⁸

In contrast, the ambiguity theory allows for DRWI because the semantic mechanism of ambiguity allows sentences and/or utterances to pick out different propositions and to have varying truth-conditions both across different contexts and within a particular context. SCWI requires what context-sensitivity and ambiguity have in common, namely, the ability for varying truth-conditions for a sentence or utterance across contexts. But DRWI requires what only ambiguity allows for: the ability for varying truth-conditions for a sentence or utterance within a particular context.Footnote ³⁹

The ambiguity theory's compatibility with DRWI is what allows the ambiguity theory to straightforwardly explain instances where DRWI is true. To see a plausible example of such a case, let's turn to the following scenario originally provided by Jessica Brown.

In this case the nurse's claim that the doctor knows which of the patient's kidneys is to be removed is meant to be taken intuitively as a true statement and appropriate response. This on its own doesn't favor DRWI or its rejection. However, consider Fantl and McGrath's claim that “the nurse in Brown's surgeon case could just as easily and just as properly have said, ‘Well, of course she's checking the chart; it's not enough to rely on her memory that it's the left kidney. She needs to know it is’.”Footnote ⁴¹ The implication of such a claim on the nurse's behalf amounts to an implicit denial that the doctor knows which of the kidneys is to be removed. Let us label the response that Fantl and McGrath claim the nurse could give (adding in an explicit knowledge denial) Nurse Response 2 (NR-2) and the original response that Brown claims the nurse could give Nurse Response 1 (NR-1). Our options are then as follows:

• NR-1: Of course, she knows which kidney it is. But, imagine what it would be like if she removed the wrong kidney. She should not operate before checking the patient's records.

• NR-2: Well, of course, she's checking the chart; it's not enough to rely on her memory that it's the left kidney. She needs to know it is, and without double-checking she doesn't know it.

Both NR-1 and NR-2 deal with the same subject (the surgeon), the same proposition (it is the left kidney of the patient that is to be removed) and are given in response to the same context.Footnote ⁴² NR-1 contains an explicit knowledge ascription while NR-2 contains an explicit knowledge denial concerning the same subject, proposition and context. For those with the intuition that both NR-1 and NR-2 represent plausibly true and appropriate responses, this poses a problem for contextualism, because the contexts prompting these responses are identical,Footnote ⁴³ yet both claims seem very plausibly true. Such intuitions pose no problem for the ambiguity theory. It seems to me that it frequently is the case that in one and the same context, a speaker is permitted to assert a knowledge ascription or denial, holding fixed the speaker and proposition. The ambiguity theory offers the cleanest explanation of such situations.Footnote ⁴⁴

Let's turn now to a linguistic motivation for the ambiguity theory. Ambiguous words come in two types: homonyms and polysemes.Footnote ⁴⁵ Homonymy is the phenomenon of one word having two or more meanings that are unrelated. Words like ‘bank’ and ‘bear’ are examples of the former and are called homonyms.Footnote ⁴⁶ Polysemy refers to the phenomenon of one word having two or more closely related meanings. Words like ‘crane’ and ‘arch’ are examples of this and are called polysemes. Any plausible ambiguity theory of ‘knows’ will claim that ‘knows’ is polysemous, not homonymous. A linguistic motivation for the polysemy of ‘knows’ can be found in noting the pervasiveness of polysemy among commonly used English verbs.

According to the Oxford English Dictionary ‘know’ is the eighth most commonly used English verb. The rest of the 25 most commonly used English verbs are ‘be,’ ‘have,’ ‘do,’ ‘say,’ ‘get,’ ‘make,’ ‘go,’ ‘take,’ ‘see,’ ‘come,’ ‘think,’ ‘look,’ ‘want,’ ‘give,’ ‘use,’ ‘find,’ ‘tell,’ ‘ask,’ ‘work,’ ‘seem,’ ‘feel,’ ‘try,’ ‘leave’ and ‘call.’Footnote ⁴⁷ These verbs do very well at passing common tests for ambiguity.Footnote ⁴⁸ For example, in the Conjunction Reduction test, one takes two sentences in which the possibly ambiguous word is supposedly being used in different senses and combines the two sentences into one while using the ambiguous term only once. One then looks to see if this new sentence is a zeugma (i.e. a sentence in which a single word applies to two different parts of the sentence in two different senses).Footnote ⁴⁹

Such zeugmas can easily be constructed using many of the verbs listed above. Take ‘is’ for example.

1. (1) She is a talented artist.

2. (2) She is running late.

3. (3) She is a talented artist and running late.

The zeugma turns on the dual usage of ‘is’ as indicating a persisting trait and a current state.

Zeugma can also be formed where the word standing in both relations to the ambiguous word is the same word. Take for example the following case using ‘made’.

1. (4) I made this [reed] basket.

2. (5) I made that [last second] basket [to win the game].

3. (6) I made this basket and that basket.

This can happen with constructions in subjects as well as with objects, as shown in this example with ‘tells’.

1. (7) He tells me that he has something to hide.

2. (8) His nervous demeanor tells me that he has something to hide.

3. (9) He and his nervous demeanor tell me that he has something to hide.

If one were to go through all 25 verbs listed above, one would find most lend themselves to forming zeugmas.

Similarly, it is relatively easy to craft sentences in which commonly used English verbs pass the Contradiction test (where two usages of a word are put in a sentence where the usage looks contradictory but isn't)Footnote ⁵⁰ as shown with the following examples.

1. (a) She made the basket but she didn't make the basket.

2. (b) It seems red but it doesn't seem red.

3. (c) Delisa called Skye, but she didn't call Skye.

4. (d) That's what I said, even though that's not what I said.

5. (e) I use my friends but I don't use my friends.

We can see why (a) is not contradictory if we think back to (4)–(6) and assign sinking the shot as the meaning of ‘made’ and crafting or creating to ‘make.’ In (b) we can avoid contradiction by taking ‘seems’ to mean looks visually and ‘seem’ to mean strike one as. For (c) we can avoid contradiction when ‘called’ means harkened after and ‘call’ means a certain type of contact via phone. With (d) contradiction is bypassed by taking the first ‘says’ in the loose sense to mean what is conveyed and the second ‘said’ to refer to the actual words uttered and their literal meaning.Footnote ⁵¹ And (e) can pass the contradiction test if the first instance of ‘use’ refers to utilizing in a non-abusive sort of manner and the second instance of ‘use’ refers to a sort of mere-means type of abuse of another, which Kant famously condemned. Note that to pass the contradiction test the sentences don't need to sound natural or fluid – they just need to avoid contradiction.

The initial plausibility of the view that generally commonly used English verbs are ambiguous, combined with the ability of so many of those verbs to pass classical ambiguity tests, provides us with good reason to think that most of English's most commonly used verbs are indeed ambiguous. Thus, without a reason to think that ‘know’ is unlike most other commonly used English verbs, this provides us with good reason to favor the view that ‘know’ is itself ambiguous. The force of this is increased when we consider the best explanation for why commonly used English verbs are ambiguous. The best explanation seems to be a functional claim that these commonly used English verbs need to be flexible and varied in order to do all the work that we use them for. The paucity of epistemic verbs available to English speakers and the varied and flexible functions ‘knows’ is called on to perform provides further reason to think that ‘knows’ is ambiguous like other commonly used English verbs.

The above argument is an indirect argument for the polysemy of ‘knows’ by arguing that it is part of a class of words that are almost always ambiguous. There are more direct arguments that can be made as well. One can argue that the different senses of ‘knows’ are semantically encoded by more than one word in another language – for example, the Latin cognitionem and scientia, particularly as they are used by Descartes.Footnote ⁵² While these are only initial independent motivations for the ambiguity theory, my hope is that they provide the context in which to motivate an interest in some comparative assessments between the ambiguity theory and contextualism.

3. THE STRENGTHS OF CONTEXTUALISM AND THE AMBIGUITY THEORY

As previously stated, the generally agreed upon strengths of epistemic contextualism are (1) that it can be used to solve persistent, skepticism-driven epistemological puzzles in a manner which explains why many people have competing intuitions that both skeptical and non-skeptical responses to such puzzles are plausible and (2) that it preserves the truth of our everyday knowledge ascriptions. In this section I argue that the ambiguity theory also has these strengths.

One type of puzzle which contextualism solves is exemplified in the following scenario from Stewart Cohen.Footnote ⁵³

It would appear that Smith has a standard for knowledge that differs from Mary's and John's.

The puzzle comes about by asking which (if any) party's knowledge ascription is true. Cohen thinks that to conclude that just Smith's, just Mary's and John's, or neither Smith's nor Mary's and John's ascription is true are all dissatisfactory options. After all, to conclude that Smith's ascription is true is to conclude that Mary and John could sensibly say “Smith knows the plane will stop in Chicago but we still need to verify that this is true.” But to conclude that either just Mary's and John's ascription is true or that neither Smith's nor Mary's and John's ascription is true is to conclude that many other everyday ascriptions of knowledge are incorrect and to move very strongly in the direction of skepticism.

Cohen concludes instead, by appeal to contextualism, that “[n]either standard is simply correct or simply incorrect.”Footnote ⁵⁵ Cohen writes that,

Many, like Cohen, consider each of the options he rejects dissatisfying. Thus, many consider it a virtue of a theory if it is able to supply an alternative. Contextualism does this. But so does the ambiguity theory.

On the ambiguity theory both Smith's knowledge attribution and Mary's and John's knowledge denial can be right in both their contexts, so long as Smith is using ‘know’ in the low sense and Mary and John are using ‘know’ in the high sense. Of course, Mary and John would be incorrect if they claimed Smith was wrong in claiming to have known that the flight would stop in Chicago. So long as Smith was using the word ‘know’ in the low sense (and it seems reasonable to think that he was), he is correct even in Mary and John's context.

To see why this is so, let P stand for the claim that the flight has a layover in Chicago. Let know_L stand for ‘know’ in the low sense. And let know_H stand for ‘know’ in the high sense. Smith's and Mary's and John's claims can now be expressed as follows:

Given that Smith saw on the printed itinerary that the flight had a layover in Chicago, Smith has sufficient evidence to make his claim that “I know_L that P” true. The differences between his low-standard context and Mary's and John's more demanding higher-standard context do not and could not make Smith's statement false. Were Smith to have overheard Mary and John say he didn't know, and were he to have said “No, I truly do know_L that P,” he would still be saying something true.

But while Smith's claim would remain true, it just wouldn't be very relevant, practically speaking, to Mary and John given their context.Footnote ⁵⁷ And given the relevance of the high sense of ‘know’ for Mary and John, it would be easy to see how Mary and John might carelessly conclude that Smith was wrong because they hadn't taken his intended sense into account.Footnote ⁵⁸ Thus, the ambiguity theory seems to be able to provide as satisfying a response as contextualism to Cohen's scenario and other scenarios of a similar type – namely, it shows how both Smith on the one hand and Mary and John on the other can be correct by showing that Smith is correct when using know_L and Mary and John are correct when using know_H.Footnote ⁵⁹

The other popular type of epistemological puzzle which the proponent of contextualism claims to be able to solve comes in the form of inconsistent triads of a particular structure. One such triad is the following:

1. (i) If I know I have hands, then I know I'm not a brain-in-a-vat (BIV).

2. (ii) I know I have hands.

3. (iii) I don't know I'm not a brain-in-a-vat (BIV).Footnote ⁶⁰

Cohen considers the best type of response to such a triad to be one in which not only is a solution provided as to which premises are true but in which an explanation of the independent plausibility of each of the premises is also provided.Footnote ⁶¹

The response which the proponent of contextualism is able to put forward is that this “skeptical paradox arises from inattention to shifts in context.”Footnote ⁶² Thus, the skeptic is correct in accepting (i) and (iii) and rejecting (ii) relative to the high standards context in which all three propositions are entertained. But this does not mean that we are not typically correct in accepting (ii) in most of the everyday contexts in which we believe that we have hands. What is important to note now is that the ambiguity theory is once again able to give an equally (if not more) satisfying response.

Just as the proponent of contextualism claims that the paradox arises from inattention to shifts in context, so the proponent of the ambiguity theory claims that the paradox arises from inattention to shifts in the sense of ‘know’ employed. If the low sense of ‘know’ is used consistently throughout, then (iii) is false while (i) and (ii) are true. If the certain sense of ‘know,’ know_C, is used consistently (or perhaps also if the high sense of ‘know,’ know_H), then (ii) is false while (i) and (iii) are true. The reason both (ii) and (iii) seem plausible is because for both there is a sense of ‘knows’ that makes the claim true.Footnote ⁶³ The skeptic and the Moorean go wrong only to the extent that they fail to recognize that there is a sense of ‘knows’ in which the respective opponent can sensibly affirm the proposition which they themselves reject.Footnote ⁶⁴ Thus, like contextualism, the ambiguity theory is able not only to provide a solution to this type of paradox but is also able to explain the initial plausibility of each of the premises.

This examination of the responses the proponent of the ambiguity theory can give to such epistemological puzzles also shows how the ambiguity theory preserves the truth of everyday knowledge ascriptions. This results from the fact that on the ambiguity theory, skeptical arguments are unsound when the low, everyday sense of knowledge is employed. (Or at the very least, if they are unsound on contextualism, they are also unsound on the ambiguity theory such that no theory has an edge over the other in this regard.) This shows that not only is the ambiguity theory able to solve the same epistemological puzzles that contextualism claims to solve, but that it also preserves the truth of everyday knowledge ascriptions just as well.

In this section I've aimed to show that the ambiguity theory has the same major strengths as contextualism. Contextualism explains how Smith and John and Mary can be correct in Cohen's airport case, but so does the ambiguity theory. Contextualism offers a solution to skeptical arguments and explains the appeal of each of the premises in such arguments, but so does the ambiguity theory. Contextualism preserves the truth of everyday knowledge ascriptions, but so does the ambiguity theory. Given these similarities, one might wonder what it is that explains contextualism's popularity and the ambiguity theory's relative obscurity. In the final section of the paper, I want to point out an explanation available to the ambiguity theorist about why it is that a proponent of contextualism could easily misidentify context as the key to solving these epistemological problems as opposed to the different senses of ‘knows.’

4. AN AMBIGUITY-THEORY-FRIENDLY EXPLANATION OF THE APPEAL OF CONTEXTUALISM

What is at the center of explaining why the proponent of contextualism could easily misidentify context as opposed to ambiguity as the key to solving a number of epistemological problems is the semantics/pragmatics distinction. The explanation the ambiguity theorist has available is that the role of context is purely pragmatic instead of semantic as the contextualist claims. In other words, on the ambiguity theory context affects only which knowledge ascriptions and denials it would be appropriate to endorse while it is the sense of ‘knows’ which affects the truth conditions of those knowledge ascriptions and denials. Appealing to these distinctions is not a novel suggestion for those arguing against contextualism. Jessica Brown and Patrick Rysiew have both developed detailed accounts of the pragmatic role of context in their respective defenses of a moderate, subject-insensitive form of invariantism,Footnote ⁶⁵ and more recently Baron Reed has made appeal to this distinction in defending his preferred form of the ambiguity theory.Footnote ⁶⁶ The appeal of this explanatory option is noted by proponents of contextualism as well. For example, Keith DeRose writes that “the chief bugaboo of contextualism has been the concern that the contextualist is mistaking variability in the conditions of warranted assertability of knowledge attributions for a variability in their truth conditions”Footnote ⁶⁷ and acknowledges that this idea is a frequent first reaction to contextualism.Footnote ⁶⁸

Brown, Rysiew and Reed all appeal to Grice's Cooperative Principle, which is the maxim to “make conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged.”Footnote ⁶⁹ Thus, a shift in our intuitions about whether an ascription or denial of knowledge is correct is to be explained by a context-derived shift in what's appropriate to endorse. For both the advocate of the ambiguity theory and the advocate of invariantism, this pragmatic explanation of context provides a tidy solution to the type of data that contextualists typically offer in favor of their position. However, the fact that the ambiguity theory allows for differing truth conditions for knowledge ascriptions and denials depending on which sense of ‘knows’ is used gives the pragmatic explanation of context additional power on the ambiguity theory that it lacks when combined with the moderate subject-insensitive invariantism of Brown and Rysiew. This feature of the ambiguity theory is what allows it to circumvent the most pressing problems DeRose raises for the pragmatic account of warranted assertability for the classical invariantist.

To see why the ambiguity theory has this explanatory advantage, consider that on a form of classical invariantism like Brown's or Rysiew's, meaning and therefore truth conditions must remain fixed even when context changes what is appropriate to say. Thus, those defending invariantism via a pragmatic explanation of context will need to account for instances in which it is appropriate to utter or otherwise endorse literally false assertions and most likely will also need to explain why utterances that intuitively seem true are literally false. (For example, an invariantist who holds a moderately non-skeptical classical view of knowledge will need to explain why Mary's and John's knowledge denial of Smith is literally false despite appearing true.) There is disagreement over how much of a cost this is for a theory,Footnote ⁷⁰ and I myself do not see this challenge as insurmountable. However, the ambiguity theory appears to be able to skirt this difficulty entirely.

Unlike proponents of classical invariantism, ambiguity theorists are able to avoid dealing with instances of literally false but either appropriate or seemingly-true utterances. They can claim that on the ambiguity theory the pragmatic role that context plays is merely to indicate to or guide the listener in understanding which sense of ‘knows’ is likely being used by the speaker via the speaker's context. This is because on the ambiguity theory, unlike on either invariantism or contextualism, in a given context there is more than one proposition that can be picked out by any given knowledge ascription or knowledge denial.

This phenomenon of context pragmatically conveying which sense of an ambiguous word a speaker is using is perfectly clear in dealing with homonyms whose homonymy is a mere morphological accident like ‘bank’ (the financial institution) and ‘bank’ (the side of a river). Imagine that we are having a picnic by a river and that there is a duck down by the side of the river. If I tell you that “There's a duck down by the bank” it is clear to you that by ‘bank’ I mean the side of a river. And, given that there is a duck down by the river, I've said something true. Whether or not there is a duck down by any financial institution isn't relevant to whether or not I've said something true, and you most likely will not take it to be relevant because the context has pragmatically indicated to you that by ‘bank’ I mean the side of the river. And if it is false that there is a duck down by any financial institution, this does not cause my utterance of “There's a duck down by the bank” to be a case in which I've said something strictly speaking false but only appropriate or seemingly true. In short, the ambiguity theory has the advantage over invariantism of being able to utilize a pragmatic explanation of context on which context serves to pragmatically convey which proposition a knowledge utterance picks out as opposed to pragmatically conveying something potentially false given the fixed proposition that the knowledge utterance picks out.

The fact that context plays an important role in what is being pragmatically conveyed in many knowledge ascriptions and denials provides the ambiguity theorist with an even stronger explanation for why one might mistakenly take context to be doing the semantic work that the ambiguity theorist claims the sense of ‘knows’ is doing. The source of this additional explanatory power is rooted in the fact that it seems very reasonable to think that there is a strong correlation between usages of certain senses of ‘knows’ and certain types of contexts.

For example, it seems plausible to surmise that in most everyday situations the “low” sense of ‘knows’ is the sense that speakers use in casual conversation. Similarly, those situations in which someone is likely to use the “high” sense of ‘knows’ are almost always situations in which there is some other mark which indicates that the context has shifted into a high standards situation (e.g. “Are you sure you know that? After all couldn't it be the case that you're a brain-in-a-vat?”) and in which there is a corresponding reason for why the high sense of ‘knows’ is (at least taken to be) relevant. Given this high correlation between certain types of contexts and the usage of certain senses of ‘knows,’ it would be easy to mistake the high correlation that results between certain types of contexts and certain truth conditions for knowledge ascriptions and denials as a case in which the change in context was responsible or a sufficient reason for the change in the truth conditions.

In this paper I've shown how the ambiguity theory is distinct from contexualism, provided some initial motivations for the ambiguity theory, argued in favor of the ambiguity theory by showing how the ambiguity theory shares the major strengths contextualism does, and offered an ambiguity-theory-friendly explanation of why one might first gravitate to contextualism over the ambiguity theory even if the ambiguity theory is the correct theory. This paper focuses primarily on the relationship between the ambiguity theory to contextualism, as part of a larger general case for the ambiguity theory. I suggest that the considerations here provide good reason for those drawn to contextualism to give serious consideration to the ambiguity theory as well.Footnote ⁷¹