You are on page 1of 28

Relativism about Knowledge

Abstract

It is, in one sense, a truism that all knowledge is relative. At least, it is not wildly controversial to hold that whether or not a subject knows a given proposition is relative to: some or all of her beliefs (whether she believes the proposition in question); her environment (the presence or absence of fake barns, say); her faith in her own discriminatory abilities; what other people have said to her regarding this proposition. To hold any of these factors to be relevant to determining, for example, whether a proposition concerning morality is true might in some circles be enough for you to be branded a moral relativist; however, taking them to be relevant to whether a proposition is known would not have the same e ect. us to characterize relativism about knowledge we need to be more precise than we might be for other kinds of relativism; we cannot label people as relativists simply because they say that knowledge is relative to the beliefs and circumstances of the knower, since everyone would agree with that. Suppose there are such things as truths about epistemic justi cation that statements about epistemic justi cation express complete, truth-evaluable propositions rather than incomplete or underspeci ed propositions, imperatives, or simply the approval or disapproval of the utterer. One important kind of truth about justi cation will concern the epistemic status of particular beliefs for an enquirer, i.e. whether a speci c belief is justi ed by a particular totality of evidence in particular circumstances; another will consist of general truths about justi cation, for example truths about how much evidence of a certain kind counts for or against certain kinds of beliefs. We could then state epistemic relativism roughly as the proposal that the obtaining of these truths about justi cation depends not only on the total informational state of the subject, but also on which particular epistemic standard or system is relevant in each case. For the purposes of this paper, I take it that the principal interest of such a theory of justi cation lies in its capacity to motivate a view widespread in the humanities, which I shall call epistemic pluralism: this is the view that di erent societies or communities can have radically di erent knowledges, all of which are deserving of equal respect, because none can be assessed independently of a
I have no intention to endorse any speci c internalist or externalist proposal about the nature of justi cation; those with strong views about justi cation are invited to construe subsequent talk of justi cation depending on evidence in whatever way best ts their preferred theory of justi cation.

given epistemic system.

us the interesting forms of epistemic relativism will be those

which see justi cation as relative to the epistemic system current across a community; it is not intended that the standards for justi cation might vary between di erent inhabitants of the same community. An epistemic relativist in the current sense is not primarily motivated by the problem of making sense of variations in knowledge-ascription across high-stakes and low-stakes scenarios within a community; thus the current debate is distinct from the competition between contextualism (DeRose Stanley ), subject-sensitive invariantism (Hawthorne ; ) and relativism (MacFarlane a) within mainstream epistemology, where

the standards to which knowledge and justi cation are relative may be said to vary even between speakers in the same conversation. Further, epistemic pluralism in the current sense is not intended to force the recognition of a plurality of equally valid concepts of knowledge, so that which concept is picked out by the word knows may vary from one community to the next. Such a suggestion does not adequately capture the view under consideration, for what is intended is that, even employing our concept of knowledge alone, we should recognize radical divergence between communities with regard to what is known. A paradigmatic example is found in Boghossian ( a: - ): the Lakota, a Native American tribe, should be recognized as knowing that their ancestors rst entered the Americas from a subterranean spirit world, while we may equally truthfully present ourselves as knowing that the Lakota entered the Americas across the Bering Strait. e reason why many archeologists are persuaded by such an apparently paradoxical view is that they believe that ( rst-world) science is just one of many ways of knowing the world (Boghossian a: ). e epistemic pluralist view at issue, then, is in the rst place a thesis about the possibility of radical diversity with regard to what is known, possibly backed up with a pluralism about possible ways of arriving at knowledge. It does not help us make sense of such a suggestion to imagine that the Lakota have a concept of knowledge radically alien to our own; what is at issue is whether we can make sense of the seemingly paradoxical knowledge-claims made on the Lakotas behalf by members of our own community, who share our own concept of knowledge (whatever it is). Recent discussion of epistemic relativism in the current sense has addressed the question of how, and whether, the theory can be formulated so as to avoid what I shall call the Acceptance Problem: this is the problem that accepting epistemic relativism would leave us incapable of recognizing the normative force of our own epistemic judgements that we cannot accept both epistemic relativism and our own epistemic beliefs. On one side of the debate, Boghossian ( a) argues that, when properly formulated, epistemic relativism

has the consequence that all our pre-theoretic judgements about justi cation are strictly, literally false; on the other side it is alleged that Boghossians argument relies on a mistaken account of what the propositional content of statements about justi cation ought to be, according to the epistemic relativist (Neta ; Kalderon ). Here I shall argue that an Acceptance Problem remains, even a er the controversy about propositional content has evaporated a view also defended by Wright ( ); however I shall suggest that e problem for the problem can be avoided by adopting a less nave view of the connection between our acceptance of an epistemic system and the obtaining of such a system. the epistemic relativist is then to answer the question, what is it for an epistemic system to obtain within a community, if it is not simply for the system to be accepted by that community? Insofar as I suggest a way to avoid the Acceptance Problem, my presentation is sympathetic to epistemic relativism. However, in the closing sections I return to the connection between epistemic relativism and the wider aim of establishing epistemic pluralism. Here I suggest that a proper understanding of the kind of epistemic relativism needed to support epistemic pluralism reveals an incoherence in the latter doctrine. Even if there is no decisive argument against epistemic relativism, there is a strong case against epistemic pluralism.

Boghossians Challenge and its Critics


Boghossian ( a) poses a version of the Acceptance Problem for epistemic relativism:

he claims that the epistemic relativist cannot go on accepting her pre-theoretic epistemic system because she is now committed to treating the epistemic principles that constitute this system as strictly, literally false. Boghossian recommends an error-theoretic approach to ordinary assertions about justi cation: faced with a statement of a particular epistemic judgement such as (C) Copernicanism is justi ed by Galileos observations we must reform our talk so that we no longer speak simply about what is justi ed by the evidence, but only about what is justi ed by the particular epistemic system that we happen to accept.
Boghossian Boghossian , p. , p.

is recommended reformation is especially puritanical: we do not abandon our usual discourse about justi cation because it is misleading (although true), but rather the thought is that particular epistemic judgements are uniformly false, and so must be replaced by judgements about what is entailed by the epistemic systems that we happen to accept. us we can make sense of what epistemic relativism amounts to, according to Boghossian: we might state it as the view that only explicitly relativistic epistemic judgements should be endorsed, because only explicitly relativistic judgements are capable of expressing truth. So too, mutatis mutandis, for sentences: we should only utter explicitly relativistic sentences about justi cation, because no others can express truth. A sentence will count as explicitly relativistic i it includes a clause of the form according to current epistemic system ES.... Sentences about justi cation that fail to include such a clause will be false, because they will attribute absolute justi cation where justi cation is only ever relative to an epistemic system. e problem for the epistemic relativist now comes into focus: since accepting epistemic relativism (in Boghossians formulation) requires us to accept that only explicitly relativistic judgements can be true, and since our pre-theoretic epistemic judgements are not explicitly relativistic, accepting epistemic relativism commits us to counting all our pretheoretic particular epistemic judgements as false. But particular epistemic judgements are instances of general epistemic principles, and plausibly any general principle with many false instances is itself false. us epistemic relativism, in Boghossians presentation, commits us to counting all our epistemic principles as false, and consequently we cannot but reject the epistemic system which is constituted by these principles. But then there is a direct line from acceptance of epistemic relativism to the rejection of the epistemic system that we do in fact endorse. Epistemic relativism was introduced to widen the range of acceptable epistemic judgements by allowing for divergence across communities with regard to what counts as justi ed by a given totality of evidence. But now it seems that accepting the theory leaves us incapable of endorsing even our own epistemic practices. Critics of this argument have emphasized and Boghossian himself ( b) has been

quick to point out that other formulations of epistemic relativism are possible, which do not incorporate an error-theory about ordinary epistemic discourse, and consequently do not commit us to the falsity of our own epistemic system. We should notice two in
Boghossian a, p.

particular: one, which we might call Content-Relativism, suggests that ordinary sentences and beliefs about justi cation have relativistic content; the other, which might be described as Truth-Condition Relativism, suggests that ordinary epistemic sentences and beliefs do not have relativistic content, but instead have relativistic truth-conditions. According to Content-Relativism, our pre-theoretic claim (C) that Copernicanism is justi ed by Galileos observations is true, because the content of the proposition would be more perspicuously expressed using the sentence (C*) According to the current epistemic system ES, Copernicanism is justi ed by Galileos observations. Although the original tokening of the proposition, using (C), suggested that justi cation is a two-place relation between belief and evidence, the full statement of the propositional content given by (C*) reveals justi cation to be a three-place relation between belief, evidence, and the current epistemic system. Since (presumably) our current epistemic system applauds Galileos reliance on observation rather than scripture to form beliefs about the heavens, what we said (and what we believe) turns out to be true. An alternative account is given by Truth-Condition Relativism, which nds a way to count our pre-theoretic epistemic judgements as true without altering the propositional content expressed. Here, the suggestion is that there is nothing misleading about our original statement in fact, the content of the proposition Copernicanism is justi ed by Galileos observations is, as it seems to be, one involving a two-place relation between beliefs and evidence. However, the truth-condition for this proposition is relativized: we no longer speak of the proposition being true or false simpliciter, but rather true for an epistemic system. is has the consequence that two commentators on Galileos beliefs could if they were subject to di erent epistemic systems exhibit blameless disagreement about one and the same proposition even if their total evidential state was the same: the sentence Copernicanism is justi ed by Galileos observations could express the same content for both of us, while being true for my epistemic system, yet false for your epistemic system. Boghossian ( b) considers and explicitly rejects both of these alternative formulations:

Content-Relativism is ruled unacceptable because it is implausible to suppose that users of the reinterpreted sentences intend their remarks to be elliptical for some relational sentences ( b: ), while Truth-Condition Relativism is rejected because it requires us to say that people
ose who believe that blameless disagreement of this kind is a distinguishing feature of relativism will deny that my Content-Relativism is properly described as relativism. So be it: at this point I am not interested in legislating over terminology.

didnt know what the truth-conditions of their own thoughts were, that when they stated those truth-conditions simply by disquoting, they said something false. Neither of these arguments should convince us. First, it is implausible to contend that the logical structure attributed to a propositional content must match the intentions and intuitions of ordinary speakers. For one thing, such a requirement is too strong: for example, it would prevent us from endorsing Russells theory of descriptions (because most speakers do not intended to express an existentially quanti ed content when they use a de nite description). Moreover, it may seem strange to credit ordinary speakers with any intentions about the relativity of their judgements at all: while it may be true that users of sentences about justi cation do not intend to express relativized propositions, it may be equally true that users of such sentences have nothing that could count as an intention to express propositions that are not relativized. In most cases it may be that the intentions of users of such sentences do not give us any guidance as to whether statements about justi cation have relativistic content. is is especially so given that the conceptual apparatus necessary to form an intention to express a relativized or unrelativized propositional content may be beyond the reach of the majority of those stating epistemic judgements. Finally, if the purpose of allowing the intuitions and intentions of ordinary speakers to guide our choice of logical structure was to eliminate the phenomenon of semantic blindness, where language-users are disconcertingly ignorant of the content of their own utterances, the attempt does not succeed: some language users will continue to exhibit semantic blindness even under the current proposal, namely those theorists who are convinced that the sentences in question do express a relativized content. e argument against Truth-Condition Relativism is even less persuasive. Boghossian alleges that ascribing relativistic truth-conditions to a proposition p requires us to deny the ordinary disquotational truth-condition, (D) p is true i p, in favour of a relativized truth-condition, (D*) p is true i p relative to F. Although it seems clear that the Truth-Condition Relativist should endorse (D*), it is unlikely that she must also reject (D) as Boghossian supposes. When we learn from (D*)
Boghossian See DeRose b, p. for this last argument applied to debates between contextualism and invariantism.

that the truth-conditions of p are the same as those of p relative to F, this is a lesson which applies to both occurences of p in the original disquotation (D). Since both sides of the biconditional (D) have the same (relativistic) truth-conditions, (D) is true, and we are not in a position to impute massive error to anyone who believes (D). Since TruthConditional Relativism does not suggest that our ordinary beliefs about truth-conditions are mistaken, it need not be rejected on that count. So the epistemic relativist has two viable alternatives to Boghossians error-theoretic construal, each of which allows her to count ordinary pre-theoretic epistemic judgements as true. Nevertheless, I shall suggest that a version of the Acceptance Problem threatens even these alternatives which is to say that endorsing either the Truth-Conditional or Content versions of epistemic relativism leaves us incapable of accepting our own epistemic system. First, a few points about the prima facie connection between general epistemic principles, epistemic systems, and speci c epistemic judgements. It is natural to think of an epistemic system is as a collection of general epistemic principles principles about what kinds of evidence justify what kinds of belief. Further, it is natural to explain a believers propensity to form certain kinds of belief on the basis of certain kinds of evidence by appealing to the believers acceptance of a general principle licensing her beliefformation. For example, we might explain a creationists disregard for evolutionary theory by adverting to his acceptance of an epistemic principle according to which scriptural evidence trumps observation of the fossil record. us it is also natural, when relativizing judgements about justi cation, to treat them as relative to the epistemic system that we ourselves accept. Finally, since epistemic systems are collections of general epistemic principles, it is natural to suppose that a particular judgement about justi cation is true relative to (or according to) a system in the sense that that speci c judgement is entailed by the epistemic principles contained within the system plus the epistemic circumstances that obtain. If we think of epistemic principles as encoding a system for deciding whether a given belief is justi ed, then a speci c judgement about the epistemic status of a given belief will be true according to the principles because it is a consequence of the principles that, in the circumstances, the given belief is justi ed. If this is a plausible reconstruction of the epistemic relativists views, then what should she say about the truth-conditions of the general principles which make up her own epistemic system? As I have argued, there are two viable ways to relativize truths about justi cation; these can be applied to general epistemic principles as follows:
See Boghossian a, p. - for a defence of these natural assumptions about the connection between epistemic systems, principles, and judgements.

Content-Relativism:

e statement of epistemic principle p expresses the

content p relative to the epistemic system ES which I accept. Truth-Condition Relativism: e statement of epistemic principle p ex-

presses the content p, but this content is not true or false simpliciter; instead it is true relative to the epistemic system ES which I accept. Choice between these options will largely be determined by our approach to disagreement: on Content-Relativism adherents to di erent epistemic systems who apparently disagree in their epistemic principles or judgements will not genuinely disagree, for the relevant epistemic system is written into the content of any proposition they endorse: when a member of community A accepts principle p they are really accepting `p relative to community As epistemic systeme, while a member of community B who rejects principle p really rejects
`p relative to community Bs epistemic systeme.

Conversely, Truth-Condition Relativism

preserves disagreement: when we dispute principle p we dispute the same propositional content regardless of which epistemic system is current. Some would argue that preserving genuine disagreement is a desideratum of any acceptable relativization (MacFarlane ); however for many the attraction of relativism about a certain discourse consists in its promise to make good on the feeling that debate within the discourse is somehow pointless or insubstantial, by revealing that there is no genuine disagreement. For that reason I shall not attempt to arbitrate between Content-Relativist and Truth-Condition Relativist approaches to epistemic relativism. A second reason for withdrawing from the debate between these two formulations is that discussion can proceed without prejudicing decision between the two: it makes sense to ask how relativization to an epistemic system a ects the conditions under which a statement of epistemic principle counts as true, whether we believe that the relativity is part of the propositional content of the principle, or should be captured by adding an extra parameter to the evaluation of a non-relativistic content. In both cases the suggestion is that a statement of principle p will be true i p is true relative to an accepted epistemic system. But what is it for an epistemic principle to be true relative to an epistemic system? On the natural view outlined above, any epistemic principle contained within a system will be licensed by true relative to or true according to that system simply in virtue of being part of that system, by trivial self-entailment. If p is part of epistemic system ES then system ES trivially entails p and consequently p is true according to ES. But on the current account, to be part of an individuals epistemic system is simply to be a general epistemic principle accepted by that individual; if our epistemic principles are true because

part of our epistemic system, and they are part of our system because accepted by us, then our epistemic principles are true simply in virtue of our acceptance of them. at is enough to make trouble for the epistemic relativist who has reasoned this far. She is unconcerned by Boghossians original objection, that all her pre-theoretic epistemic judgements turned out to be false, because there are two ways of construing those judgements so as to count them as true. However, when we apply those suggestions Content-Relativism and Truth-Condition Relativism to the epistemic principles to which the assessor appeals to ground her particular epistemic judgements, it turns out that the truth of those principles is determined only by her decision to accept them. at is to say, my epistemic system gets to be true (for me) simply because I accept it. But to understand this is immediately to realize that, had I not accepted the system, the principles it contains would exert no authority over me. is suggests that the principles in question are in some sense optional and it is hard to see how a principle or rule can exert authority over us if our acceptance is optional in that sense. Crispin Wright, with a characteristically nice turn of phrase, neatly pinpoints that the issue of rationally unconstrained acceptance ( : ) of epistemic principles is the central problem for a relativist account of justi cation: if our acceptance of a principle is up to us, we can no longer credit that principle with the kind of normative authority consistent with epistemic judgement. However, Wright supposes that this kind of free acceptance of principle is an inevitable corollary of any view according to which epistemic systems might di er between communities: such a view requires that there are general propositions about epistemic ... justi cation, whose basic place in ones epistemic ... system goes with their acceptance being e ectively rationally unconstrained. As I shall argue later, we should not be too hasty to assume that there are no constraints on what epistemic system obtains, above and beyond our free choice to accept that system: it cannot be assumed that an Acceptance Problem a icts every epistemic relativism worthy of that name, simply because it a icts epistemic relativism as currently formulated. Can this version of the Acceptance Problem be avoided by understanding acceptance as a community-wide phenomenon rather than a matter of individual choice? is made, for independent reasons, by Kalderon: [In Boghossians discussion] Epistemic judgements are relativized to episWright , p.

is suggestion

temic systems that an individual accepts. In the context of social constructivism, wouldnt the more relevant formulation be in terms of epistemic systems that a community agrees upon? It is clear that treating acceptance of an epistemic system as a community-wide phenomenon is an improvement, for the current suggestion is that epistemic standards vary between communities rather than within them; we need to foreclose on the possibility that standards may vary between individuals in one and the same community. (Indeed, one plausible answer to the vexed question of what constitutes a community for the purposes of epistemic relativism is that shared membership of a community is determined by shared epistemic system.) Nevertheless, this suggestion does not enable us to avoid the Acceptance Problem, as is clear when we see that even community-wide acceptance is determined by the fact that individuals within that community accept the system for themselves. Perhaps not everyones judgement counts as much as anyone elses (Kalderon : ). Still it will be the case that if we the community had chosen di erently, a di erent epistemic system would have obtained. Even if I (as a single member of the community) do not have absolute authority over the system that gets accepted, my opinion counts for something; moreover, we the community have to take collective responsibility for the system we have: there is nothing else to blame other than our free choice. Compare an idealized democracy: no-one has absolute authority to pick a government, yet which government is elected depends only on the unconstrained choice of the people. Treating acceptance of an epistemic system as community-wide agreement still leaves us with the problem that the choice of system is up to us or optional, and again it is hard to see how the principles contained within that system can exert authority over us if our (communal) acceptance of them is optional.

Rejecting Epistemic Systems;

ree Kinds of Relativity

Two other putative solutions to the Acceptance Problem present themselves; in this section I outline them and argue that they fail. One is to reject the very idea of an epistemic system to deny that there is such a thing as a set of epistemic principles which all the members of a community share. Notice that we need to do more than simply claim that the shared epistemic system is, like the frictionless plane, an idealization not found in nature, but useful for the purposes of theory: as long as the notion of a shared epistemic system plays a central role in a relativist theory of justi cation of the kind under discussion, the problem
Kalderon , p.

will remain that anyone who accepts that theory of justi cation thereby undercuts his allegiance to whatever epistemic principles he has. Rather, to avoid the current problem it is necessary to remove the notion of a shared epistemic system from our theory altogether. If there are no such things as epistemic systems, then it no longer makes sense to ask whether, and how, we should go on accepting our own epistemic system in the face of the epistemic relativists revelation that acceptance of that system is up to us. It is indeed plausible that there are no such things as epistemic systems. But even if this is true, it is of little help to the epistemic relativist. To motivate relativism about justi cation, we need to be able to make sense of the idea that the background against which particular epistemic judgements are made might vary between communities; and as I have already suggested to generate an epistemic relativism consistent with the epistemic pluralist agenda at issue, we need to make sense of the idea that the background against which particular epistemic judgements are made remains constant across members of the same community. How else is this to be done, if not by introducing the notion of a set of epistemic principles which are shared across a community and which constitute an epistemic system? To talk of a community sharing an epistemic background which it fails to share with another community is to represent a community as sharing a way of thinking about justi cation which might not be shared by others; and there does not seem to be a way of capturing how a way of thinking could be the same within a group and di erent from that of other groups, save by introducing the notion of a set of accepted general principles about justi cation in short, an epistemic system. So rejecting the notion of an epistemic system solves the Acceptance Problem, at the cost of making the current epistemic relativist proposal unintelligible. An alternative suggestion takes a more nuanced approach to the question of whose epistemic system counts when making judgements about justi cation. Suppose we have a speci c epistemic judgement concerning an individual, for example, (GO) Galileos astronomical beliefs were justi ed by his observations. e epistemic relativist suggests that the truth or falsity of a proposition such as (GO) depends in part on the epistemic system at issue. But whose epistemic system? A natural suggestion, in line with much current debate on analytic relativism, is that what counts in assessment of the truth or falsity of (GO) is the epistemic system of the assessor. (GO) is true when assessed by us, because we share Galileos scienti c epistemic system, although
See MacFarlane : for a principled attack on Boghossians notion of an epistemic system. See especially MacFarlane b

it is false when assessed by someone like Cardinal Bellarmine whose thought operates within an epistemic system according to which scripture trumps astronomical observation when it comes to determining cosmology. Call this view the assessor-sensitivity of justi cation. It is likely that it is such a view that is at issue in Boghossian ( a); at least that is a plausible reconstruction based on his insistence that any statement of an epistemic judgement should by reformulated to include explicit reference to the person making that judgement, by means of the indexical I. However, a view that takes justi cation to be assessor-sensitive is not the only way to construe epistemic relativism; nor is it obviously what the defenders of epistemic relativism have in mind; nally (as I shall argue in the closing sections of this paper) it is not the right way for the epistemic pluralist or social constructivist to deal with the sort of relativity needed if knowledge is to turn out to be socially located. We should take note of two alternatives in particular. One, which we might call contextualism about justi cation, is that the epistemic system relevant to assessing any claim about justi cation is the epistemic system selected by the context of use: the system current for the person who tokens that claim, e.g. by uttering a sentence expressing it. is position di ers somewhat from the contextualism about knowledge familiar from the literature: contextualism is more usually invoked to make sense of divergent knowledge-ascriptions between di erent members of the same community. Nevertheless, the central idea is easily grasped: whereas assessorsensitive accounts see the relevant epistemic system to be that to which the assessor belongs, contextualist accounts will see the relevant epistemic system to be the one current in the community of the tokener of the claim. While these accounts will deliver the same verdict in the case where tokener and assessor are the same (e.g. where I discuss the truth or falsity of my own epistemic statements and beliefs), they will come apart in cases where we assess a statement about Galileos epistemic status which is tokened by some third party: here the contextualist will assess according to the third partys epistemic system, while the defender of assessor-sensitivity will assess according to her own epistemic system. It does not seem that endorsing contextualism about justi cation will get us very far with the current problem. Since contextualist and assessor-sensitive accounts will agree that our own statements of epistemic principle are to be assessed in light of our own epistemic system, we still have the worrying result that the principles that comprise our epistemic system get to be true simply by being part of the epistemic system that we accept. e Acceptance Problem is not avoided by adopting contextualism, because contextualism retains the problematic insistence that our own epistemic system is true because we accept it.

Moreover, it is possible to make a strong case that contextualism about justi cation does not capture what the epistemic pluralist has in mind. Consider the situation when three epistemic systems are in play: Galileos, our own, and that of some third community call it Community C. When a member of Community C expresses an epistemic judgement about Galileo, for example by tokening (GO), and saying that Galileos astronomical beliefs were justi ed by his observations, the contextualist should say that the truth or falsity of that utterance depends on whether Galileos beliefs were justi ed according to the standards current in community C. But plausibly, this is not what the epistemic pluralist intends: when evaluating claims about justi cation of this kind, it makes sense to evaluate according to our own epistemic system, and it makes sense to evaluate according to Galileos epistemic system; but why should a claim about Galileo be evaluated (by us) according to an epistemic system which neither we nor Galileo share? e situation here is markedly di erent from that encountered in standard debates about contextualism: whereas it is plausible that the truth of an epistemic claim is covariant with the situation of the speaker when such variation amounts merely to the di erence between high stakes and low stakes scenarios within a community, it is much less appealing to say that epistemic claims are to be evaluated according to the situation of their tokener where this is a matter of selecting one out of a range of possibly salient complete epistemic systems. A di erent, third view sees justi cation to depend, not on the epistemic system of the assessor, but rather on that of the person forming the belief whose justi cation is under debate. According to this way of thinking which I shall describe as the subject-sensitivity of justi cation what counts when we assess Galileos beliefs qua justi cation is not our epistemic system but that of Galileo himself, because Galileo is the owner of the belief whose justi cation is at stake. Such a view of justi cation is o en assumed among defenders of epistemic relativism. I o er two illustrative examples. ... a kind of relativism about justi cation, saying that whether someone is justi ed in believing p in light of evidence E depends crucially on their background beliefs and credences. the reason that if it visually seems to Galileo that there are mountains on the moon, then Galileo is prima facie justi ed in believing that there are mountains on the moon is that, in Galileos community of inquiry, Observation is an agreed-upon epistemic principle. In each case, the suggestion is that epistemic relativism is best played out as a view that
MacFarlane Kalderon , p. . My italics. , p. . My italics.

connects the justi cation of an enquirers beliefs with the background epistemic system that the enquirer himself accepts; there is no prospect of someone from a di erent community overriding the enquirers claim to justi cation by imposing an alien epistemic system on the believer. us the suggestion is parallel to subject-sensitive invariantism in standard high-stakes/low-stakes cases in epistemology; both theories share the view that what counts is the situation of the enquirer, not that of whoever happens to be tokening or assessing a proposition about the epistemic status of that enquirer. One problem for this subject-sensitive version of epistemic relativism is that it does, a er all, deliver absolute truths about justi cation: there is no prospect of securing blameless disagreement about Galileos epistemic status, because there is only one correct answer to the question was Galileos belief justi ed by his observations? Once we have factored in the epistemic system within which Galileo was working, we nd that Galileos belief either was or was not justi ed by the evidence according to that system, and as it happens, this epistemic system is the only relevant one so there is only one right answer to the question. A second problem for subject-sensitive epistemic relativism is that it might not prompt any revision of classical epistemology a er all. Suppose that we adopt the nave account according to which an epistemic system is a set of principles accepted or believed by the enquirer. en any di erence in epistemic system between two enquirers will entail a di erence in the background of beliefs or accepted propositions against which they must assess their evidence. And it is not news, from the point of view of conventional epistemology, that di erences in background beliefs can lead to di erences in which beliefs are justi ed by what evidence: Marcias justi cation in believing that it is pm when she sees the clock can be a ected by the presence or absence of the belief that the clock is broken, and more generally on the constitution of her total belief-set. Even if these consequences are relatively easy for the subject-sensitive epistemic relativist to live with, there is a much more serious problem for the position, namely that it still does not completely avoid the Acceptance Problem. Our judgements about the epistemic status of other peoples beliefs plausibly do not face an Acceptance Problem, for in assessing the beliefs of someone from a di erent community we assess according to the epistemic system current within that community, and since we can use that system for the purposes of our assessment without having to accept the system for ourselves, there is no problem about how we might accept that system, even given its dubious status as optional for the members of that community. However, two issues remain: rst, how can we make sense of forming a normative epistemic judgement about someone within a di erent community,
See Hawthorne ( ) and Stanley ( )

using an epistemic system that we do not share? To class a belief as unjusti ed, say, is to judge that the subject ought not to have formed it on the basis of the evidence at hand; but why should an epistemic system which we do not share have any consequences for our judgements about which beliefs someone ought or ought not to have formed? Second, and worse, the Acceptance Problem remains for judgements about our own evidence-based beliefs, and those of other members of the community. If I judge that I was justi ed by my evidence in believing that p, I judge myself using the epistemic system that I accept. But how can I go on accepting that system once I absorb the epistemic relativist revelation that the authority of this system consists only in the fact that my community accepts it? Although I do not use my own epistemic system in order to make judgements about the epistemic status of beliefs formed by member of other communities, I still need to accept my epistemic system, otherwise there is no yardstick by which the epistemic status of my own beliefs can be measured. Replacing assessment-sensitive epistemic relativism with its subject-sensitive cousin does not enable us to avoid the Acceptance Problem.

A Repair to Epistemic Relativism


Nevertheless, epistemic relativism can be repaired to the point where it is at least coherent, if not ultimately plausible. What is needed is to reject the idea that an epistemic system derives its authority over a community simply from the fact that the community accepts that system; the problem that remains is how the choice between epistemic systems is determined if not by the acclamation of the populace. Such a response to the current problem is suggested by the wide gulf between the claim that epistemic standards are local to communities, and contingent on the circumstances within those communities, on the one hand, and the idea that epistemic standards apply within a society because accepted by that society, on the other. e former idea is a much weaker thesis than the latter; yet both are run together by the opponents of epistemic relativism. I shall suggest that a thesis of the former kind is defensible where the latter is not. Consider this relativistic picture: there are many di erent communities, and for each of them there is a distinct epistemic system. Depending on which avour of epistemic relativism you prefer, these systems will have normative authority either (i) over the epistemic status of beliefs formed by members of the community (subject-sensitive epistemic relativism) or (ii) over the truth-values of statements about justi cation assessed by members of the community (assessor-sensitive epistemic relativism) or (iii) over the truth-value of statements about justi cation tokened by members of the community (contextualist epistemic relativism). Crucially, however, the fact that an epistemic system obtains within

a given community does not depend on facts about whether or not the community accepts that system. is is of course not to say that there will be no members of the community who accept the system: merely to say that the communitys acceptance of the system is the result of the systems obtaining for other reasons. People will accept epistemic principles because they ascertain that these principles obtain, rather than the principles obtaining simply because people decide that they should. No Acceptance Problem threatens, because the normative force of the epistemic system that obtains within the community does not depend on the communitys acceptance of the system. is position is clearly a form of epistemic relativism, for it relativizes truth about justi cation to an epistemic system such that, had the relevant epistemic system been a di erent one, the truth about justi cation would have been di erent. However, it lacks the extraneous commitment to the view that the obtaining of an epistemic system depends only on the communitys acceptance of that system, and thereby avoids the most serious problem for formulations of epistemic relativism current in the literature. Yet a problem remains: what, if not acceptance, determines that my system obtains in my community, and your system obtains in your community? is is not a fatal objection, for answers are possible. One would take a pragmatist form: the choice of epistemic system is determined by what works from the point of view of e ective belief formation; since what works will di er between societies, so will the epistemic system that obtains. Another answer is familiar from mainstream debates about contextualism, and is also suggested by some remarks of Neta ( : ): an epistemic system will be more stringent where it is more important to arrive at the correct answer. A community where resources are scarce and danger is rife will be one where the epistemic standards are higher and the epistemic system more exacting. It is, then, possible to make sense of epistemic relativism. However, we can say this much in opposition: any proponent of such a view about justi cation owes us a decision between subject-sensitive, assessor-sensitive and contextualist forms of the proposal; even more importantly, she must be willing to explain what it is for an epistemic system to obtain within a community, now that it is clear that this cannot simply be a matter of the communitys accepting that system. Any determinate answer to the latter question will involve endorsing a theory (e.g. pragmatism about justi cation) that brings problems of its own; however, until an answer is given epistemic relativism remains merely a promissory note rather than a complete proposal about how we should understand the possibility of divergent truths about justi cation.

Epistemic Relativism and Epistemic Pluralism


My discussion so far has suggested that epistemic relativism, narrowly conceived as a thesis about the truth-conditions or content of propositions about justi cation, is not an obviously self-defeating position. Where there is a cogent argument against the position, it will be one that poses an explanatory challenge to the epistemic relativist: what determines that any particular epistemic system has normative authority over our epistemic judgements, if not simply the fact that we happen to accept that system? Nevertheless, I think the preceding clari cation of options for the epistemic relativist provides material for a sound argument against epistemic pluralism, the view that di erent communities can have radically di erent knowledges all of which are deserving of equal respect. Some form of epistemic pluralism is necessary, if not su cient, for holding that knowledge is socially located or socially constructed in any interesting or controversial sense: how can knowledge depend on the community of the knower if there is no substantial di erence from one community to the next with regard to what is known? My contention will be that the epistemic pluralist must be willing to say both that justi cation is subject-sensitive that truths about justi cation are relative to the subject whom the proposition is about while also saying that truth in general is assessor-sensitive, i.e. that all propositional truth is relative to the person who is assessing the proposition for truth or falsity. Since (I claim) propositions about justi cation cannot be both subject-sensitive and at the same time assessor-sensitive, epistemic pluralism is incoherent. My argument depends on the premise that justi cation and truth are necessary for knowledge (even if they are not su cient for it); I shall make no attempt to defend that assumption, and consequently my argument will have no relevance for pluralists who deny that knowledge requires truth, or are willing to use knowledge as a term for any rationally-held belief. I shall also simplify my presentation in two relatively inconsequential ways. First, I shall talk of communities as the knowing subjects, although communities only know things in a derivative sense: our community knows about special relativity in the sense that scientists within our community know about special relativity. Second, when talking about justi cation I shall omit explicit reference to evidence: for a is justi ed in believing p please understand a is justi ed by her evidence in believing p. I hope nothing important hinges on these shortcuts. First, why must justi cation be subject-sensitive? I have already o ered illustrative quotes to suggest that the subject-sensitivity of justi cation is what defenders of epistemic relativism have in mind; further, there is good reason to suppose that such a view is required if epistemic pluralism is to get o the ground. Notice that if we are to accept that

(AK) Community A knows proposition p we must also accept (AJ) Community A is justi ed in believing proposition p. In previous discussion, I suggested three ways of determining the epistemic system to which (AJ) is relative. One is assessor-sensitivity, which is to say that the truth of (AJ) depends on the epistemic system current for the person deliberating over whether to accept or reject it. Another is context-sensitivity, which I glossed as the view that the truth of (AJ) depends on the epistemic system current for the person uttering or otherwise tokening (AJ). It is clear that neither of these two options will not do here. If (AJ) is assessor-sensitive, then we should have to accept or reject it on the basis of our epistemic system. Since, by hypothesis, community A is di erent from our own, and may have a radically di erent epistemic system, treating (AJ) as assessor-sensitive does not guarantee that we will accept (AJ); indeed in many cases where the epistemic pluralist may want to ascribe knowledge, the requirement of assessing claims to justi cation according to ones own epistemic system will prevent us from counting a community as having their own special knowledge because we shall not be able to assess their belief as justi ed: the belief in question is not justi ed with regard to our epistemic system, and that is the one that the assessor-sensitivity of justi cation would require us to employ in assessing such a claim. Context-sensitivity should be rejected for similar reasons: in cases where we are both tokeners and assessors of (AJ), the contextualist will select our own epistemic system as the relevant one, in which case there is no way to recognize a communitys beliefs as justi ed although they are not licensed by our own epistemic system. Instead, then, if we want to take seriously the claim that di erent communities have their own distinctive knowledges, it will be necessary to treat claims such as (AJ) as relative to the epistemic system current in the believers own community. at is to say, we accept (AJ) because we realize that belief in p is justi ed according to the epistemic system current in community A, and accept that what counts from the point of view of assessing justi cation is not the epistemic system of the assessor, but rather that of the subject who forms the belief that is to be assessed as justi ed or unjusti ed. e second part of my argument is the claim that the epistemic pluralist must endorse a general claim about the assessor-sensitivity of propositional truth. To establish this it is necessary to look more closely at the scope and aims of epistemic pluralism. Epistemic pluralism may be seen as incorporating two commitments: one is that di erent communities may know di erent sets of propositions; another is that there is no sense in which

any of these sets of known propositions is better than any other. Yet the conjunction of these claims may be true as a matter of empirical fact, without forcing the revision of classical epistemology: it is a truism that di erent communities know di erent things, just as we know things which our ancestors did not, and vice versa (for example, I do not know what shape a mammoth footprint is). Moreover, it is easy to imagine circumstances in which two communities know di erent propositions, yet neither knowledge-set is better than the other. We merely have to imagine a situation in which, say, one community knows a lot about arable farming, and the other knows a lot about animal husbandry. So the conjunction of Variance and Equal Validity is not enough to produce a distinctive epistemic pluralist view. Instead, I suggest that epistemic pluralism should be understood in terms of the hypothesis of radical di erences between communities with regard to what is known. One way of making room for the possibility of radical di erences between di erent knowledges is via the subject-sensitive epistemic relativism just outlined, for then proposition p may be known in one community because justi ed by the standards current in that community, yet fail to be known in another community because not justi ed by the standards current there even though both communities have the same total evidence. Yet this by itself is not enough to deliver a truly radical di erence between communities knowledges, nor is it obviously what epistemic pluralists have in mind. First, notice that (so long as we retain the normal picture of knowledge as factive) what can be known will still be limited by what is true; the only divergence in knowledges between communities will be where some truths are not known in a community because that community is not in a position to form beliefs about those truths which live up to the communitys standards for justi cation. e only sense in which knowledges can di er between communities is that di erent communities will know di erent subsets of truths, depending on their circumstances; but this does not do much to distinguish epistemic pluralism from classical theories of knowledge, since it is uncontroversial that di erent communities will know di erent amounts of the truths that there are. A second reason for dissatisfaction with the current account of epistemic pluralism is that has no room for the kind of examples that epistemic pluralists endorse. To return to our paradigmatic example from Boghossian, the epistemic pluralist will say that the Lakota know that their ancestors came from inside the Earth, while we know that the Lakotas ancestors came from Asia across the Bering Strait ( a: - ). Here it is clear that the sense in which knowledge sets may be radically di erent across communities is that di erent
See Boghossian a, p. for these commitments as de ning features of constructivism in the humanities.

knowledge-sets may be incompatible with each other that my community might know p while at the same time your community knows not-p or some other proposition q that is incompatible with p. What is intended by epistemic pluralists in the humanities is not merely knowledge-sets which di er between communities, but rather the possibility of con icting propositions being known within the knowledge-sets of di erent communities. us it seems that a de ning feature of epistemic pluralism should be the acceptance of incompatible knowledge-sets. e need for the epistemic pluralist to make room for incompatible knowledge-sets enables us to see why epistemic pluralism is committed to the assessor-sensitivity of truth. Notice that if we want to accept (AK) Community A knows proposition p we must also accept (AT) Proposition p is true. And if we mean to endorse the thesis of knowledges which are radically di erent in the sense that they are incompatible with each other, we also should be prepared to accept the pair (BK) Community B knows proposition not-p and (BT) Proposition not-p is true. It is tempting to say that there is no sense in which we could ever accept the conjunction of (AT) and (BT), on the grounds that, by disquotation, accepting the conjunction (AT) Proposition p is true & (BT) Proposition not-p is true. is the same as accept the contradiction p & not-p. If that is so, the possibility of radically di erent knowledges cannot so much as get o the ground: if there is no sense in which we can endorse the conjunction (AT) & (BT), then we

cannot accept the claim that community (A) and (B) have incompatible knowledge-sets. For many, this may be enough to show that the idea of incompatible knowledge-sets is untenable, and must be abandoned along with epistemic pluralism itself. For others, it will suggest that epistemic pluralists must be using the word knowledge to indicate some non-factive concept, in which case it is plausible that classical epistemologist and epistemic pluralist are simply talking past one another, labouring under the misapprehension that their opponent is using the word knowledge to talk about the same thing as they are. However, I shall suggest that there is an alternative option available to the epistemic pluralist, which is to say that the truth required for knowledge the truth which is at issue in (AT) and (BT) is relative, rather than absolute truth. If the conjunction (AT) & (BT) can be understood as saying that p is true for community A while not-p is true for community B, then there is a sense in which we can endorse that conjunction, and hence endorse the claim that community A knows p while community B knows not-p, without giving up the idea that knowledge requires truth. endorse (AK) Community A knows proposition p & (BK) Community B knows proposition not-p because we are able to accept something like the following: (K) p is true (for Community A) and justi ed (according to Community As standards), while not-p is true (for Community B) and justi ed (according to Community Bs standards). It seems that (K) enables us to make sense of the possibility of incompatible knowledgesets while maintaining the attractive idea that justi cation and truth are necessary for knowledge. I should note that on some relativist schemes, it is not possible even to accept (K), because there is no room for a semantic mechanism according to which we could endorse p is true for community A unless we were also prepared to endorse p is true. e suggestion, then, is this: we can

On such a system, there is no prospect of talking about how things are for people who do not share our own circumstances. Fortunately, there is good reason not to impose such a restriction: if we disallow the locution true for then there does not seem to be an easy way to state the relativists own position, that what is true for us may di er from what is true for you. Indeed, it is possible to make a persuasive case that, for the relativist, true for is the fundamental notion, and uses of true in the object language are to be explained in terms of the more primitive true for (Cappelen and Hawthorne : - ).

But is the epistemic pluralist proposal ultimately coherent? I shall suggest that it is not, because it attempts to combine two incompatible views about what propositions are true relative to. Consider the claim that p is true (for Community A). What kind of relativity is this? If the relativity in question is to extend to all propositions which might be known within the community, then it cannot be subject-sensitivity: for one thing, not every proposition has a logical subject (It is raining, something is in the cellar); for another, making truth relative to the circumstances of the subject of a proposition would mean that truth values were xed, once and for all, by facts about whoever the proposition is about for example, claims about the origin of the Lakotas ancestors would be xed once and for all by the circumstances of the Lakotas ancestors; but what is intended is that the truth about the Lakotas ancestors varies between investigating communities. It might be thought that some kind of global contextualism could capture the kind of relativity at issue; the suggestion would be that p is true relative to Community A because truth is relative to the context of use of p, and Community A in some sense count as the users of p. However, there are notable problems with this approach, deriving from the fact that, in assessing community As claims to knowledge, we are ascribing truth and falsity to their beliefs rather than their utterances, and consequently ascribing truth and falsity to propositional contents rather than the sentences used to assess them. First, it is not obvious that we can assign a context of use to a belief, as it is hard to isolate any one circumstance in which the believed content might be said to be used. Second, what is intended here is that communities A and B have genuinely incompatible beliefs; but standard contextualism destroys the possibility of genuine disagreement, since it relativizes sentential truth-values to contexts of use by allowing the context to determine which proposition is expressed; within such a framework there is little room for the truth-values of a given propositional content to vary according to context of use, since the necessary variation is already delivered by variation in the content expressed by a use of a sentence on an occasion. What is needed to preserve the epistemic pluralist theory of radical disagreement between communities is a form of non-indexical contextualism (MacFarlane ), according to which truth-

conditions, but not propositional content, vary according to the context of use. Yet even here we are prevented from counting the propositional content of Community As beliefs as true: since the notion of truth employed in such a contextualist account is sentential rather than propositional truth, the only way in which we can ascribe truth to the content of a belief is derivatively, by assigning truth to the sentence, (PT) What community A believe in believing that p is true But since we are uttering (PT), the context of use is our context; and p need not be true according to our context so there is no guarantee that (PT) will be true either. If the only way to endorse the content of Community As beliefs is by uttering a sentence claiming that sentence to be true, and the truth-conditions of sentences are relative to the context of use, then there is no way to endorse the content of the beliefs of those in a di erent context. Contextualism leaves us incapable of counting Community As beliefs as true unless those beliefs also happen to be true for us. One option remains, which avoids the problems consequent on contextualist approaches. at is to make truth relative to the context of assessment. Such an assessment-relative conception of truth can apply to propositions just as well as it applies to sentential truth: one proposition can have many di erent assessors as easily as can one sentence. Moreover, this option delivers the correct verdict with regard to the truth-value of Community As beliefs. Although p may be false according to our context of assessment, and so is false for us, the members of Community A, since they believe that p, count as assessors of p just as we do; consequently there is a sense in which we can agree that p is true for Community A, by saying that p is true as assessed by Community A. So it seems that, in order to accept the epistemic pluralist claim that community A might know p while another community knows an incompatible proposition such as not-p, we need to accept both that what counts for determining the truth of a proposition about epistemic justi cation is the situation of the subject of the proposition (subject-sensitivity), and that what counts for determining propositional truth in general is the situation of the assessor of the proposition (assessor-sensitivity). But since the latter claim is a general one, it should apply no less to propositions about justi cation; thus it seems the epistemic pluralist is committed to saying both that what counts, for propositions about justi cation, is the situation of the subject, and that what counts, for propositions about justi cation, is the situation of the assessor. is, I claim, is incoherent. In the following paragraphs I explain and respond to the most signi cant objection to this charge of incoherence.
See MacFarlane , p. and Cappelen and Hawthorne , p. for similar points.

One objection to the argument I have just sketched will be especially prominent: doesnt the whole thing just rest on some unsatisfactory fudging around the issue of disquotation? I have argued thus: to accept that community A knows that p, we must accept that community A is justi ed in believing that p, and this is something we cannot accept by our standards of justi cation; therefore a claim like (AJ) Community A is justi ed in believing proposition p. must be subject-sensitive rather than assessor-sensitive. But surely we can nd an easier way to accept (AJ). Suppose we simply accept the assessor-sensitivity of truth for all propositions; then anyone in community A who assesses proposition (AJ) will do so with regard to his own epistemic system, and count (AJ) as true. In that case, we can say that (AJ*) (AJ) is true (for Community A). from which it is a simple matter to infer (by disquotation) the original proposition (AJ) Community A is justi ed in believing proposition p, which we are now licensed to accept. is is an interesting suggestion, but I do not think it works, for a reason already hinted at: within a relativistic framework, we cannot apply a rule of disquotation of the kind that takes us from (AJ*) to (AJ), simply because what is true for someone or other need not be true for me. Indeed, allowing the move from p is true (for someone or other) to p would result in us endorsing contradictions, for then we could move from accepting (AT) Proposition p is true (for community A) & (BT) Proposition not-p is true (for community B). to accepting p & not-p. In order to count community A as knowing that p, we have to accept (AJ), not (AJ*), and the only way we can accept (AJ) is if we hold that the relevant epistemic system for claims about justi cation is not that of the assessor our own but rather than of the subject whose justi cation is in question.

A second objection is that there is no incoherence in claiming that propositions about justi cation are both subject-sensitive and assessor-sensitive, because there is no limit to the number of extra parameters we may introduce to the evaluation of such a proposition. Indeed, if relativists are happy to accept that context-sensitivity and assessor-sensitivity may coexist (MacFarlane b: ), why not accept that subject-sensitivity and assessorsensitivity may coexist in the evaluation of propositions about justi cation? Here the response is that subject-sensitivity and assessor-sensitivity cannot coexist because they are each competing to do one and the same job namely, to select the community whose standards, experience and values are relevant to the evaluation of the proposition in question. us subject-sensitivity suggests that the community whose standards are relevant to evaluation are those of the community to which the subject of the epistemic proposition belongs; the speci c sense in which these standards are relevant is that they incorporate an epistemic system according to which the subject is to be judged qua justi cation. Yet assessor-sensitivity says that the community relevant to evaluation is that of the assessor; any community-speci c factors which make a di erence to the evaluation of a proposition (including, but not limited to, the epistemic system at play in the community) should be those belonging to the assessors community. ese two claims subject-sensitivity and assessor-sensitivity cannot both be upheld, because then we would in many cases have to select two communities as determining the truth of a proposition about judgement, and these communities are likely to disagree in the evaluation they license. A third objection the last I shall consider here is that I may be setting up a straw man in attributing to the relativist a view that predicts the assessment-sensitivity of all propositions. Perhaps the intention of the epistemic pluralist might be captured more perspicuously by the suggestion that the truths which form the content of the supposedly divergent knowledges held by di erent communities for example, truths about human origins, science, and the world are assessor-sensitive, while truths about the epistemic status of those beliefs are not assessor-sensitive, but rather subject-sensitive. is is a coherent position, but not one that it makes much sense to endorse. As I noted earlier, to accept that truths about justi cation are subject-sensitive is to assert that there is one correct answer to the question, Was Galileos belief justi ed by his evidence?, since subject-sensitivity holds that there is only one epistemic system (that of the putative knower) that is relevant to the evaluation of any claim about justi cation. So the proposed emendation of epistemic pluralism leaves us committed to the view that there is, in a sense, absolute truth about the justi cation of peoples beliefs, yet there is no absolute truth in areas such as science, biology, and everyday descriptions of material objects. is seems to get things the wrong

way round! If there is no absolute truth about the world, why should there be how could there be absolute truth about whether or not my beliefs were licensed by my evidence? Although it is possible to challenge the prima facie appearance of incoherence in the epistemic pluralists commitment to both the subject-sensitivity of justi cation and the assessment-sensitivity of propositional truth, it now seems that the incoherence in genuine. e epistemic pluralist now has two options: one is to adopt the modest proposal that justi cation is subject-sensitive, although truth in general is not assessment-sensitive. In that case, what can be known within a community will be limited by what is (absolutely) true, in which case we no longer have the possibility of radical disagreement, and the theory no longer captures what the epistemic pluralist intends: it cannot be that the Lakota know that they came from the spirit world, and yet we know that they came across the Bering Strait, for when properly construed these claims are incompatible, and only one can be known. Alternatively, it is possible to drop the subject-sensitivity of justi cation, in which case propositions about justi cation will be assessor-sensitive like everything else. But then, as I have argued, there is no way of counting adherents to other epistemic systems as justi ed, since our evaluation of any claim about their epistemic status must be governed by the assessors (our) epistemic system. In that case, although there may be a pluralism of truth, we cannot recognize a pluralism of knowledge, for all of our assessments of claims about justi cation (ours and everyone elses) must be governed by our own parochial epistemic system. e Lakota do not count as knowing because their epistemic practices do not measure up to our own epistemic standards.

A erword: Reason vs. Relativism


It is a point well made that the source of many peoples relativistic beliefs about what would otherwise be called matters of fact is not, as it happens, any cogent philosophical argument, but rather the result of a backlash against colonial arrogance in imposing civilized belief-systems on the weaker party, and of the widespread belief that the authority of reason, and the attendant rhetoric of objectivity, is a mask for the interests of power.
See Boghossian a, p. for the same complaint levelled at theories which hold that the only absolute , p. facts there are, are facts about our beliefs. Kalderon , p. . e same point is found already in Boghossian

If that is so, it is possible to make a case that the whole project of responding to relativism by means of rational argument is mistaken, for it does not address the actual causes of relativistic belief. us Kalderon suggests that

e source of relativistic conviction is relevant to the rhetorical e ectiveness

of undermining the arguments advanced in its favour. If the source of relativistic conviction does not lie with the cogency of these arguments, then undermining them would leave relativistic conviction untouched. It is strange that Kalderon makes his point in this way, as an allegation about the rhetorical e ectiveness of undermining arguments for relativism, when his context is a discussion of the cogency of Boghossians argument against epistemic relativism. Plausibly, a cogent argument against a position can retain its e ectiveness even in situations where we recognize that successful demolition of philosophical arguments for that position will be dismissed as an irrelevance by our opponent. Nevertheless, the accusation that relativistic conviction is impermeable to rational argument is a worrying one, and I shall nish by o ering two remarks in defence of the current project of approaching relativism and epistemic pluralism through rational argument. First, philosophical argument is not merely evangelical: it is not the case that engaging in rational debate is to be done only for the purpose of changing the minds of those with whom we disagree. A primary aim of philosophical dispute (apart from the obvious one, of getting at the truth) is normative: to establish what we ought to believe. us it makes little or no di erence to the value of philosophical argument if there exist people who (for whatever reason) are so stubborn that no argument can unseat their prejudices. Clearly there are such people; but that does not make rational debate any less worthwhile. Second, it is nave to adopt a picture of the relationship between relativistic conviction and philosophical argument in favour of relativism, such that the latter can make a di erence to the former only if the latter was the cause of the former. Suppose that I have a prejudice in favour of the view that we have Free Will (I do). at prejudice is not caused by any philosophical argument it is, if anything, the product of my upbringing and of the kind of society I inhabit, plus a healthy dose of what I nd comfortable to believe. Nevertheless, the cogency of arguments in favour of Free Will can make a di erence to what I believe: if I examine all the rational arguments in favour of Free Will and nd them wanting, I may become aware that my prejudice is merely a prejudice, and to that extent nd myself less inclined to hang on to it once I become aware of the strength
Kalderon , p.

of the arguments pushing in the other direction. Just so, I suggest, even those whose relativistic conviction was originally produced by factors other than reasoned argument may be a ected by a successful demonstration that the arguments in favour of relativism do not succeed. is will be especially so if as may actually be the case widespread acceptance of relativism is combined with the widespread misconception that relativistic theses are trivially coherent, philosophically unproblematic or even rationally mandated by current theory. seldom heeded. us, I suggest, there is work to be done in making sense of epistemic relativism and epistemic pluralism, even in a world where the advice of philosophers is

References
Boghossian a. Fear of Knowledge Against Relativism and Constructivism. Oxford: b. What is Relativism? In P. Greenough and M. Lynch (eds.), Truth and Realism. Oxford: Oxford University Press. Cappelen, H., and Hawthorne, J. University Press. DeRose, K. DeRose, K. : i)a( . Contextualism and Knowledge Attributions. In Philosophy and : . . Bamboozled by Our Own Words: Semantic Blindness and Some . . Knowledge and Lotteries. Oxford: Oxford University Press. . Epistemic Relativism. In Philosophical Review a. i)a( . : . e Assessment Sensitivity of Knowledge Attributions. In Oxford Phenomenological Research . Relativism and Monadic Truth. Oxford: Oxford Clarendon Press.

Arguments Against Contextualism. In Philosophy and Phenomenological Research Hawthorne, J. Kalderon, M. E. MacFarlane, J.

Studies in Epistemology : Society : - .

b. Making Sense of Relative Truth. In Proceedings of the Aristotelian . Relativism and Disagreement. In Philosophical Studies . Nonindexical Contextualism. In Synthese Neta, R. Stanley, J. Wright, C. : i)a( - . . : . i)a( . : - .

. Boghossian, Bellarmine, and Bayes. In Philosophical Studies . In Defense of Epistemic Relativism. In Episteme : . Fear of Relativism? In Philosophical Studies :

. Knowledge and Practical Interests. Oxford: Oxford University Press.

You might also like