Author Information: Paul Faulkner, University of Sheffield, paul.faulkner@sheffield.ac.uk

Faulkner, Paul. 2013. “Two-Stage Reliabilism, Virtue Reliabilism, Dualism and the Problem of Sufficiency.” Social Epistemology Review and Reply Collective 2 (8): 121-138.

The PDF of the article gives specific page numbers. Shortlink: http://wp.me/p1Bfg0-Vf

Please refer to:

• Fuller, Steve 2012. “Social Epistemology: A Quarter Century Itinerary.” Social Epistemology 26 (3-4): 267-283.
• Special Issue 2: On the Future Direction of Social Epistemology.

Abstract

Social epistemology should be truth-centred, argues Goldman. Social epistemology should capture the ‘logic of everyday practices’ and describe socially ‘situated’ reasoning, says Fuller. Starting from Goldman’s vision of epistemology, this paper aims to argue for Fuller’s contention. Social epistemology cannot focus solely on the truth because the truth can be got in lucky ways. The same too could be said for reliability. Adding a second layer of epistemic evaluation helps only insofar as the reasons thus specified are appropriately connected to reliability. These claims are first made in abstract, and then developed with regard to our practice of trusting testimony, where an epistemological investigation into the grounds of reliability must inevitably detail the ‘logic of everyday practices’.

Section 1

“Mainstream epistemologists are lovers of truth; at least they are comfortable doing epistemology with the truth concept in hand” (Goldman 2009, 3). I agree, and I think for good reason: despite the widespread disagreement over what is the nature of knowledge, there is near universal agreement that knowledge is factive; if one knows that p, it is true that p. It follows that whether one knows that p or not will be determined by the connection one’s thinking that p has with the fact that p. Alvin Goldman’s various epistemologies — his causal theory of knowledge, reliabilism, the two-stage reliabilism that is the topic of this paper, and his more recent veritism — then represent different ways of working through this thought.[1] Now this conception of epistemology is largely one that I share. However, this paper articulates a criticism that develops one made by Steve Fuller who has a quite different view. When reviewing Goldman’s Knowledge in a Social World, Fuller (2000, 575) had the following to say:

Goldman assumes a pre-sociological sense of ‘the social’ as the aggregation of individuals. Accordingly, social life occurs only in observable interactions between individuals. Upbringing, training … do not figure in Goldman’s theory.

This, he thinks, is misguided: philosophy should borrow from sociology and recognise “the logic of everyday practices that constitute the social order” (574). The social “pre-figures” observable individual interactions because these interactions are framed by reasoning that is “local” and socially “situated” (574). This criticism is levelled against Goldman’s veritism, but it could be equally made against any of his epistemologies. The ambition of this paper is then to articulate this criticism as a problem for Goldman’s two-stage reliabilism, and so for theories with this same form.

Section 2

According to reliabilism, a belief is justified to the degree that the process producing it is reliable. As with any influential theory, numerous criticisms followed. Reliability, it was argued, is not necessary for justification since the brain-in-a-vat, in being able to articulate justifications of its beliefs, is justified in its beliefs. But few of the brain’s belief forming processes are reliable (Lehrer and Cohen 1983; Cohen 1984). Nor does it seem that reliability is sufficient for justification because it is possible to imagine processes that are reliable but whose deliverances one would be unjustified in believing. Clairvoyance seems to be one such process (BonJour 1980). “Two-stage reliabilism” was, at one point, Goldman’s response to these criticisms.[2] According to this theory, epistemic evaluation involves two stages.

The first stage features the acquisition by an evaluator of some set of intellectual virtues and vices. This is where reliability enters the picture. In the second stage, the evaluator applies his list of virtues and vices to decide the epistemic status of targeted beliefs. At this stage, there is no direct consideration of reliability (Goldman 1994, 300).

Again:

When a process or method is judged to have a high proportion of true outputs, it is viewed as a warrant-conferring process or method. … I shall call this first stage of the epistemological story the standard-selection stage … The second stage … is the standard deployment-stage. In this stage, members of the community apply the chosen standards by judging whether individual beliefs (either actual or hypothetical) are warranted as a function of whether they are arrived at (or sustained) by approved processes or methods (Goldman 1999b, 11).

Goldman then clarifies: “the criterion appealed to is reliability, not judged reliability or believed-reliability. Actual reliability is the criterion that the community tries to apply” (Goldman 1999b, 11). The theory of justification proposed might be thus formalised:

S’s belief that p is justified if and only if
        (1) S believes that p on the basis of reliable process X; and
        (2) S believes
            (a) that p is delivered by X, and
            (b) that X is reliable
(Otherwise put: that p is delivered by X and forming belief on the basis of the deliverances of X is a virtue and is believed to be a virtue by S.)

This theory of justification then gives straightforward solutions to the problem that reliability alone is neither necessary nor sufficient for justification. With respect to the brain-in-a-vat, Goldman observes “that an epistemic evaluator will match the victim’s vision-based processes to one (or more) of the items on his list of intellectual virtues, and therefore judge the victim’s beliefs to be justified” (1994, 295). And with respect to the clairvoyant, Goldman notes that this lies in “a class of putative faculties, including mental telepathy, ESP, telekinesis, and so forth that are scientifically disreputable. It is plausible that evaluators view any process of basing beliefs on the supposed deliverances of such faculties as vices” (296). Otherwise put, the clairvoyant is not justified because condition (2) in (2SR) is not satisfied. And although the brain-in-a-vat’s beliefs are not justified, because condition (1) in (2SR) is not satisfied, (2SR) can still capture a sense in which the brain-in-a-vat belief’s are in good epistemic standing: for each, condition (2), which is a necessary condition on being justified, is satisfied.

Now two-stage reliabilism represents a relatively brief period in Goldman’s thinking (and I’ll have more to say on this shortly).[3] But its epistemic significance is much greater than this because two-stage reliabilism formulates a virtue epistemological position. Thus, and for instance, the account of justification it proposes is essentially similar to that presently advanced by Ernst Sosa. In Sosa’s hands the two stages of epistemic evaluation become two kinds of knowledge: animal and reflective. Here is an early statement of this distinction.

One has animal knowledge about one’s environment, one’s past, and one’s own experience if one’s judgements and beliefs about these are direct responses to their impact — e.g., through perception or memory — with little or no benefit of reflection or understanding.

One has reflective knowledge if one’s judgment or belief manifests not only such a direct response to the fact known but also understanding of its place in a wider whole that includes one’s belief and knowledge of it and how these come about (Sosa 1991b, 240).

A more recent statement makes it clearer that animal knowledge is to be understood in terms of the necessary condition (1) in (2SR). Animal knowledge is possessed, Sosa claims, when this condition is satisfied:

A belief amounts to knowledge only if it is true and its correctness derives from its manifesting certain cognitive virtues of the subject, where nothing is a cognitive virtue unless it is a truth conducive disposition (2009, 135).

Reflective knowledge then comes with the believer having “an epistemic perspective on his belief, a perspective from which he endorses the source of that belief, from which he can see that source as reliably truth conducive” (135). That is, with the satisfaction of condition (2) in (2SR). In short, knowledge and justification derive from the operation of a reliable process, or virtue, combined with an endorsing belief or, more broadly, an endorsing epistemic perspective.

It is the contention of this paper that two-stage reliabilism does not provide the straightforward response to the problem of sufficiency that Goldman hoped, and which was briefly described above. Once it is acknowledged, as two-stage reliabilism does acknowledge, that reliability alone is not sufficient for justification adding an endorsing belief, or an epistemic perspective, does not provide a simple solution to the problem. The contention, more precisely, is that this strategy can work but it can do so only if further conditions are met. For the case of testimony, which will be central to this paper, what is also needed is an account of how our epistemic perspective involves reasoning, which as Fuller would say, is ‘local’ and ‘socially situated’.

Before moving to introduce this objection, I hazard this observation: Goldman abandoned two-stage reliabilism precisely because he realized that the problem of sufficiency was not so easily resolved. Or at least this is suggested, I think, by his recent criticism of Sosa’s virtue theory. Adding a coherent set of beliefs that provide an epistemic perspective on a reliable process of belief formation cannot be enough for justification, Goldman observes, because this coherence could derive from no more than guesswork. Further conditions need to be met, and he suggests that

What is necessary for added justifiedness, then, is that the subject must arrive at a belief in his system’s coherence by some sort of reliable process of coherence detection (Goldman 2004, 88).

But then, Goldman asks, why isn’t reliability enough? That is to say, why go down the route of adding a second stage of epistemic evaluation at all? [4] I think that this worry is exactly right but I start from the assumption that this route is compelled because of the problem of sufficiency. What needs to be worked out is precisely what is needed for coherence itself to be a reliable process.

Section 4 will outline the problem of sufficiency or, what might be called the new problem of sufficiency, and suggest a solution. Section 5 will consider this solution in the context of giving a two-stage reliabilist account of testimonial knowledge. It is here, I suggest, that there is a truth to Fuller’s worries. In the next section, section 3, I consider first the recent claims by Jennifer Lackey that testimonial knowledge poses a general problem for virtue epistemological theories to the extent that they take a certain form.

Section 3

One element that is prominent in recent virtue epistemological theories is that virtues are conceived as abilities. Thus, Sosa also defines reflective knowledge as “defensibly apt belief; i.e. apt belief that the subject aptly believes to be apt, and whose aptness the subject can therefore defend” (Sosa 2007, 24). And

Aptness requires the manifestation of a competence, and a competence is a disposition, one with a basis resident in the competent agent, one that would in appropriately normal conditions ensure (or make highly likely) the success of any relevant performance issued by it (Sosa 2007, 24).

More explicitly, John Greco proposes this analysis of knowledge:

(KSA) S knows p if and only if S’s believes the truth (with respect to p) because S’s belief that p is produced by intellectual ability (Greco 2009, 18).

This interpretation of virtue epistemology, Jennifer Lackey argues, is roundly refuted by the case of testimonial knowledge. She gives this case:

CHICAGO VISITOR: Having just arrived at the train station in Chicago, Morris wishes to obtain directions to the Sears Tower. He looks around, approaches the first adult passerby that he sees, and asks how to get to his desired destination. The passerby, who happens to be a lifelong resident of Chicago and knows the city extraordinarily well, provides Morris with impeccable directions to the Sears Tower by telling him that it is located two blocks east of the train station. Morris unhesitatingly forms the corresponding true belief (Lackey 2009, 29).

What this case straightforwardly shows is that “while Morris clearly knows on the basis of testimony that the Sears Tower is two blocks east of the train station, he does not deserve the requisite kind of credit for truly believing this proposition” (Lackey 2009, 29). I think that this is correct: in acquiring testimonial knowledge, one acquires knowledge through relying on others. It is then the speaker’s competence, or in some cases that of someone else in the testimonial chain, that matters, not one’s own competence.[5]

The case of testimonial knowledge suggests that knowledge cannot be analysed in virtue terms; that is, if virtues are conceived as intellectual abilities. However, this is not how Goldman conceived of them in formulating two-stage reliabilism. In “Epistemic Folkways and Scientific Epistemology”, virtues name positive answers to the question ‘How does X know?’

In answer to this question it is common to reply, ‘He saw it’, ‘He heard it’, ‘He remembers it’, ‘He infers it from such-and-such evidence’, and so forth. Thus basing belief on seeing, hearing, memory, and (good) inference are in the collection of what folk regard as intellectual virtues (Goldman 1994, 300).

Virtues, thus conceived, are no more than broadly characterized ways of forming belief. Such a characterisation of virtue can equally be found in Sosa; for instance at one point Sosa defines “aptness” thus:

The ‘aptness’ of a belief B relative to an environment E requires that B derives from what relative to E is an intellectual virtue, i.e., a way of arriving at belief that yields an appropriate preponderance of truth over error (Sosa 1991a, 144, my italics).

This formulation, like Sosa’s distinction between animal and reflective knowledge quoted above, suggests two-stage reliabilism: reflective knowledge requires aptness (condition (1) of (2SR)) and justification or aptness that can be defended (condition (2)). [6]

Virtue epistemological theories, I then propose, encompass both what Lackey calls the credit view of knowledge, and what Axtell calls virtue reliabilism.[7] For both theories virtues are reliable processes. What makes these processes virtues for virtue reliabilism is then the believer’s cognition of them as such. So virtue reliabilism understands knowledge as reliably produced true belief that one has a reflective perspective on (defined as (2SR) above). Whereas what makes a reliable process a virtue on the credit view is its being the ability of an individual. So the credit view understands knowledge as the product of one’s own intellectual abilities (defined as (KSA) above). This view, I agree with Lackey, is refuted by the case of testimonial knowledge. However, virtue reliabilism, as a form of two-stage reliabilism, is not so refuted.

Moreover, Lackey’s own dualist theory of testimony is two-stage reliabilist in form. According to Lackey,

an adequate view of testimonial justification or warrant needs to recognize that the justification or warrant of a hearer’s belief has dual sources, being grounded in both the reliability of the speaker and the rationality of the hearer’s reasons for belief (Lackey 2008, 177)

This adequate theory is dualism.

Dualism. For every speaker A and hearer B, B justifiably believes that p on the basis of A’s testimony that p only if: (1) B believes that p on the basis of the content of A’s testimony that p, (2) A’s testimony that p is reliable or otherwise truth conducive, and (3) B has appropriate positive reasons for accepting A’s testimony that p.[8]

That is, acquiring a justified belief from testimony requires that the source of the testimonial deliverance be reliable, which is requirement (1) in (2SR). And it requires the believer have beliefs that endorse the acceptance of this deliverance, which is essentially the satisfaction of (2) in (2SR).[9]

Two-stage reliabilism, then, is an influential theory receiving expression in virtue reliabilist theories and Lackey’s dualism. This branch of virtue epistemology is not straightforwardly refuted by the case of testimonial knowledge. However, in the next section, I would like to argue that adding a second layer of epistemic evaluation does not straightforwardly resolve the problem of sufficiency.[10] In the section after I return to the case of testimonial knowledge.

Section 4

Knowledge is factive: if one knows that p, it is true that p. If it is then assumed that belief is a proper component of knowledge, it follows that knowledge implies true belief. However, so the standard argument goes, knowledge cannot be analysed simply as true belief because one might believe the truth merely by luck or accident.[11] For example, suppose a gambler has a policy of always placing a bet on the horse with the longest odds. This bet never pays, but overtime the gambler succumbs to the fallacy of thinking that next time it will pay. After all, he thinks, one day he must win with this bet. One day the bet does pay off; the gambler believed that the horse with the longest odds would win, and it did. Did he know this? Of course not, the argument runs, it was a crazy accident that all the other horses fell at the first hurdle and complete luck that his betting policy, and method of forming belief, led to a win and true belief. What is required for knowledge, reliabilism proposes, it that there be a non-accidental connection between a subject’s belief and the truth of this belief. What is required is that a method of forming belief reliably yields the truth. Since the gamblers policy reliably loses him money, this requirement is not satisfied in this case.

The problem of sufficiency for reliabilism is that this requirement of reliability is not enough to eliminate the element of epistemic luck. The objective goodness of a process or method of forming belief at yielding truths is not enough. Originally, this point was made by imaging clairvoyant powers (BonJour 1980). And the case of the gambler can be modified in this way. Suppose that the gambler doesn’t always place a bet on the horse with the longest odds, he only does so when he has a ‘hunch’, as he would say, that luck is with him. On these occasions, he believes this horse will win, and it always does so because his hunch is actually the manifestation of a clairvoyant power. This power operates to reliably produce such a feeling whenever the horse with the longest odds is going to win. Now further suppose that the gambler’s belief is not sensitive to his track record of success: he is responding to the fact that he feels lucky in each case and nothing else. In this case, while it is not accident that his clairvoyance leads him to form a true belief, it is still a matter of luck relative to his basis for forming these beliefs, which is just that ‘he feels lucky’, that the beliefs thus formed are true.[12]

Two-stage reliabilism is a development of reliabilism that enables a response to this problem. The reliability of the process or method of belief formation, it may be acknowledged, is not sufficient for justification. What is also needed is the belief that the process or method is reliable. As it stands the strategy of betting on the horse with the longest odds is recognized by the clairvoyant gambler as a bad one. He knows that it is crazy, and will only lead to him losing money. For this reason he only adopts this strategy when he is feeling lucky. And the policy of doing this is not something that he has any beliefs about; it is just something that he does. He thereby fails to have an epistemic perspective on this policy. If he did have an epistemic perspective, and in particular if he recognized that the policy had a good track record, then he could be justified in belief. But lacking these beliefs about his way of forming belief, the beliefs that his adoption of this policy results in lack justification. Otherwise put, the clairvoyant gambler might satisfy condition (1) in (2SR), but he is not justified because he does not satisfy condition (2).

However, now consider a third case. Suppose that the clairvoyant gambler is in addition quite superstitious. He is not merely sensitive to his hunches of good luck, in addition he is sensitive to the fact that he has adopted the policy of responding to these hunches. And being superstitious he thinks that this is a good policy to adopt. That is, he thinks it is a good policy to place a bet on the horse with the longest odds on those occasions when he has a hunch that his luck is in. This is to say that the method of forming belief that this policy identifies figures on his list of virtues. Now consider the occasion where this superstitious clairvoyant gambler has a hunch that his luck is in, places a bet, believes this betting policy is a good one and believes that the horse he has betted on will win. In this case, the superstitious clairvoyant gambler has satisfied the conditions two-stage reliabilism requires for justified belief. The gambler employs a reliable method of belief formation and he believes that the method he employs is a good one — it is on his list of virtues – and believes the method to be operative in this case. So when his belief turns out to be true, as it inevitably does, when all the other horses fall at the first hurdle, two-stage reliabilism delivers the verdict that the gambler knew his horse would win. This is the wrong verdict. It is still true that the gambler is insensitive to matters of track record; he just believes that the policy of acting on his lucky hunches is a good one because he is superstitious. Superstition is not the kind of ground for belief that can convert an unjustified belief (in the case of the clairvoyant gambler) into a justified belief (this case of the superstitious clairvoyant gambler). Otherwise put, conditions (1) and (2) in (2SR) might be satisfied, but the way in which (2) is satisfied is wrong.

In developing two-stage reliabilism Goldman says two things that might offer some response to this problem. First, there is some constraint as to what can figure on a subject’s list of virtues and vices. Epistemic evaluators, Goldman says, “inherit their lists of virtues and vices from other speakers in the linguistic community” (1994, 297). And while one can hypothesize that the superstitious clairvoyant gambler had ‘being responsive to hunches of luck’ on his list of virtues, it is less plausible to suppose that the gambler’s policy would receive a general social endorsement. What this suggests, in effect, is that a further sub-condition be added to (2SR) to give:

(2SR*) S’s belief that p is justified if and only if
        (1) S believes that p on the basis of reliable process X; and
        (2) S believes
            (a) that p is delivered by X, and
            (b) that X is reliable
        (3) S believes (2a) and (2b) because the wider social community believes it.

The problem with this response is that it is good only insofar as the beliefs of the wider social community are in good epistemic standing. But it seems possible that wider social endorsement could be as poorly grounded as that of the superstitious clairvoyant gambler. Morever, Goldman recognises that there can be “different sub-cultures in the linguistic community” (1994, 297). And it does not stretch plausibility to suppose that there might be one that is superstitious in the way that superstitious gambler is.

Thus, second, Goldman recognises that it can always be asked, “when are beliefs really justified, as opposed to being held justified by this or that community? A natural response is: a belief is ‘really’ justified if and only if it results from processes (or methods) that really are reliable, and not merely judged reliable by our epistemic community” (Goldman 1999b, 11). The project of scientific or naturalised epistemology is then to work out what processes of belief formation are actually employed, and what processes are actually reliable.[13] Of course, in the case imagined the error does not lie in what is believed — the superstitious clairvoyant gambler does employ a reliable method and is correct in believing that he does — but in the grounds of this belief. However, this error would be removed if these grounds were scientific rather superstition. Thus, if this suggestion is other than a return to straightforward reliabilism, it is the suggestion of a replacement condition (2) in (2SR).

(2SR**) S’s belief that p is justified if and only if
        (1) S believes that p on the basis of reliable process X; and
        (2) The scientific community believes
            (a) that p is delivered by X, and
            (b) that X is reliable

The problem with this response is that it undercuts two-stage reliabilism’s response to clairvoyance. Consider again the second case of the clairvoyant gambler, but suppose now as a variant on this case that the scientific community has discovered the existence of clairvoyance as a power and that the clairvoyant gambler has this power as described. In this variant case the conditions of (2SR**) are satisfied. But all the reasons for thinking that the clairvoyant gambler does not know and is not justified remain (assuming that the clairvoyant gambler’s background of belief is unchanged, and, in particular, that he is ignorant of this scientific endorsement). An alternative way of reading Goldman’s suggestion is then to regard it as proposing, not a replacement condition (2), but a further condition in the manner of (2SR*). This gives:

(2SR***) S’s belief that p is justified if and only if
        (1) S believes that p on the basis of reliable process X; and
        (2) S believes
            (a) that p is delivered by X, and
            (b) that X is reliable
        (3) The scientific community believes that X is reliable.

The problem with this response is that it does not in fact change anything with respect to the superstitious clairvoyant gambler. If background of belief of this gambler, and the superstitious linguistic community of which he is a member, remains unchanged, then the reasons for thinking that this gambler does not know and is not justified remain unchanged. It is still true that the gambler merely believes that the policy of acting on his lucky hunches is a good one because he is superstitious. And this is not the kind of ground that can convert an unjustified belief (in the case of the clairvoyant gambler) into a justified belief (this case of the superstitious clairvoyant gambler).

Two-stage reliabilism proposed to solve the problem of sufficiency that confronted reliabilism as a theory of justification by adding a second layer of epistemic evaluation. It is not enough to yield justified outputs that a process be reliable, there is in addition the requirement that the subject have a couple of epistemic beliefs, namely that he believe that the process is reliable and that the output is delivered by the process. The basic problem with this solution, I have argued, is that the subject’s grounds for these epistemic beliefs might be poor. Goldman suggests two ways of responding to this problem, which are two ways of safeguarding against this possibility: first, by adding the requirement that the epistemic beliefs be held because they are collectively held; and, second, by adding the requirement that these beliefs are scientifically endorsed. The problem with the first requirement is wider social endorsement might be equally ill grounded; and the problem with the second requirement is that it need not speak to the subject’s grounds for these epistemic beliefs.

What is needed to resolve this problem is simply the requirement that the subject’s grounds for the epistemic beliefs specified in condition (2) of (2SR) be good. More precisely, what is needed is that these grounds either be, or be determined by, those facts that determine the truth of condition (1). Justification requires that the facts that determine the reliability of the process of belief formation also provide the subject’s grounds for believing that this process is reliable. The problem arises when the facts that determine (1) are different to the facts that determine (2); the problem is a case where these two conditions are satisfied independently of each other and so are disconnected from one another. In order to resolve this new problem of sufficiency, two-stage reliabilism should be formulated thus:

(2SR-final) S’s belief that p is justified if and only if
        (1) S believes that p on the basis of reliable process X; and
        (2) S believes
            (a) that p is delivered by X, and
            (b) that X is reliable
        (3) the facts that determine (1), determine (2).

Condition (2) is not satisfied in the case of the clairvoyant gambler. Condition (3) is not satisfied in the case of superstitious gambler clairvoyant. But if this gambler were scientific, rather than superstitious, all three conditions would be satisfied.

The task of giving a two-stage reliabilist theory for any given ‘virtue’ or way of forming belief must then detail how condition (3) is satisfied; that is, how it is that these reliability determining facts also determine the believer’s beliefs about reliability or the epistemic goodness of the given process of belief formation. In the next section, I return to the case of testimony and consider how this desideratum is to be satisfied in this case.

Section 5

Consider again Lackey’s case CHICAGO VISITOR.

Having just arrived at the train station in Chicago, Morris wishes to obtain directions to the Sears Tower. He looks around, approaches the first adult passerby that he sees, and asks how to get to his desired destination. The passerby, who happens to be a lifelong resident of Chicago and knows the city extraordinarily well, provides Morris with impeccable directions to the Sears Tower by telling him that it is located two blocks east of the train station. Morris unhesitatingly forms the corresponding true belief (Lackey 2009, 29).

With respect to this case, Lackey says that “Morris clearly knows on the basis of testimony that the Sears Tower is two blocks east of the train station” (29). And I think that this is right. Now the passerby, as described, is reliable in her testimony. So according to (2SR) and dualism, as quoted above, Morris knows that the Sears Tower is two blocks east of the train station because he has “appropriate positive reasons” for believing this passerby’s bit of testimony (Lackey 2006, 170). According to Lackey, this requirement of reasons should be understood very minimally. The reasons merely need be such “that they render it, at the very least, not irrational for her [in this case Morris] to accept the testimony in question” (Lackey 2008, 181). Thus, these reasons need not justify Morris’s testimonial belief, they just need ensure that his acquiring this belief is not irrational, which is to say that they need to render his acquisition of belief explicable.[14] Only so much is required, according to (2SR), because this is all that is needed for an epistemic perspective on this instance of belief acquisition. Ultimately, justification comes from the reliability of the passerby’s utterance.

So what reasons does Morris have? Given the weakness of this requirement, one would expect that the intuition that Morris testimonially knows to be robust across a wide range of answers to this question. And I think that this is the case. To illustrate, consider a variation, which might be called SUPERSTITIOUS CHICAGO VISITOR.

Having just arrived at the train station in Chicago, Maurice wishes to obtain directions to the Sears Tower. He looks around and approaches an adult passerby wearing a black felt hat as he thinks that this is a sign of honesty, and asks how to get to his desired destination. The passerby, who happens to be a lifelong resident of Chicago and knows the city extraordinarily well, provides Maurice with impeccable directions to the Sears Tower by telling him that it is located two blocks east of the train station. Maurice unhesitatingly forms the corresponding true belief.

Given that the passerby “knows the city extraordinarily well”, doesn’t it remain true that Maurice gets to know that the Sears Tower is two blocks east of the train station? My feeling is that Maurice does get to know this, even though he has a rather odd reason for approaching this passerby. As in the case CHICAGO VISITOR, Maurice is informed of the right directions by a speaker with “impeccable” credentials and thereby knows where to go.

However, if this is the case, there is a sharp contrast between SUPERSTITIOUS CHICAGO VISITOR and the case of the superstitious clairvoyant gambler. How is it that the reason provided by superstition is enough when added to one reliable process (the testimonial case) but not enough when added to another reliable process (the gambling case)? The obvious answer is that in the case of testimony, reliability is sufficient: Maurice does not need any reasons to be justified because it is enough that the passerby’s bit of testimony is reliable. This obvious answer I would reject for general reasons: for the reasons described in the last section, reliability cannot alone be sufficient for justification. And for particular reasons: there are reasons particular to testimony for thinking that testimonial acceptance must be supported by reasons, or so I have argued at length elsewhere.[15] The explanation I would rather propose is that under the most plausible description of these cases both Maurice and Morris approach the passer-by with an attitude of trust. Since we think that testimonial knowledge can be got on trust, we recognise that even Maurice is put in a position to know the whereabouts of the Sears Tower by the passerby’s testimony. [16] Trust, I now want to suggest, is sufficient to enable this, when many other weakly grounded reasons might not be so, because it satisfies condition (3) in (2SRfinal).

To elaborate on this let me first, briefly, say what I take the attitude of trust to be and say how it is that this attitude is reason providing (See Faulkner 2011, §6.1-6.3). Trusting is something we do: it is the act of putting oneself in a position of depending on something happening or someone doing something. And trust is an attitude we can have, or fail to have, in these situations of dependence. Properly speaking, trust is two similar but distinct attitudes. I can trust my alarm to go off or my car to start in that I depend on these things happening and expect that they will. But the expectation that accompanies our depending on someone doing something is often normative. When we depend on someone doing something, we can expect them to do that thing in the sense that we expect this of them and will be prone to various reactive attitudes if they do not. While both senses of trust can be found in our testimonial interactions, it is the latter thicker sense that is relevant here.

To illustrate this, suppose I am waiting for you in a café. I depend on you showing up, maybe in no stronger way than that I will be deprived of your company if you do not, or maybe in some more substantial fashion: you are to hand over the monies that will allow me to take to the high seas. Either way, I depend on you turning up, and I expect you to do so. This expectation is normative: I will feel let down by you if you fail to show up. And it has a specific sense: I don’t just expect you to come to the café for whatever reason, but for the reason that I am here waiting for you. To say that I trust you will be here shortly is just to say that I depend on you in this respect and have this expectation of you. In the case of testimony, we trust others, in this sense, for the truth. And this is to say we depend on others for information and expect them to tell us what they know because we need this information. Again the expectation here is specific: we expect others to tell us what they know because we need this information too. This I think will have been the attitude with which Morris, or indeed Maurice, approached the passerby. How then is this attitude reason giving? And how does this reason satisfy condition (3) in (2SR — final)?

Take the case where I am waiting in the café for you. I expect something of you: that you see my waiting here as a reason to show up, and show up for that reason. In holding this expectation, I presume that you will be sensitive to this reason and so presume, other things being equal, that you will be moved by it. But that is to say that, other things being equal, I presume you will show up. And if I didn’t presume this, my attitude in waiting for you would not be one of trust. So trusting involves a background set of presumptions. Other things being equal, these presumptions rationalize the act of trust. Thus the short answer to the question, “Why am I still waiting for you?” is “Because I trust you will show up”. And the long answer to this question, which states a piece of theory, is that I am still waiting because of the background set of presumptions that come with the attitude of trust and which render my trusting rational, other things being equal. Similarly, trust can rationalize our uptake of a piece of testimony. Thus the short answer to the question, “Why believe what the speaker says?” is “Because I trust her for the truth”. And the long answer, which again states a piece of theory is that I believe what the speaker says because of the background set of presumptions that come with the attitude of trust. In the case CHICAGO VISITOR these presumptions are to the effect that the passerby will see Morris’s need to know the directions to the Sears Tower as a reason for telling him how to get there, and will so be moved to tell him.

Now recall the final formulation of two-stage reliabilism.

(2SR-final) S’s belief that p is justified if and only if
        (1) S believes that p on the basis of reliable process X; and
        (2) S believes
            (a) that p is delivered by X, and
            (b) that X is reliable
        (3) the facts that determine (1), determine (2).

And consider the case CHICAGO VISITOR. Condition (1) is satisfied because the passerby “knows the city extraordinarily well”. Condition (2) is satisfied or satisfied in a fashion, I suggest, because Morris’s attitude is one of trust and thereby has the reason that trust gives. That is, in presuming that the explanation of the passerby’s telling what she does is his need to know the whereabouts of the Sears Tower, Morris presumes that the passerby knows this, and so is reliable in what she says. This presumption falls short of the belief that the passerby is reliable in what she says, and this reason is not justificatory, but it still renders Morris’s acquisition of belief explicable. And it does so through giving Morris something like an epistemic perspective on this process of belief formation. This is then enough for Morris to be justified or know how to get to the Sears Tower because, I suggest, and only because, condition (3) is satisfied.

The reason that rationalizes Morris’s testimonial uptake is no more than Morris’s attitude of trust (maybe combined with some sensitivity to when trust should be withheld). Condition (3) is nevertheless satisfied in this case because there is a common cause of Morris’s attitude and the reliability of the passerby’s testimony (or at least the testimony side of this reliability). The availability of trust — our ability to think in terms of the thick concept of trust characterized — stems from our being in a community with certain social norms of trust.[17] These norms structure our evaluative practices and how we think about such trust situations as that described by CHICAGO VISITOR. On Morris’s side, these norms allow Morris to presume that the passerby will respond to his request for directions by telling him where the Sears Tower is, if she knows its location. So they allow Morris to approach the passerby with an attitude of trust; they determine the availability of trust as an attitude that Morris can adopt. On the passerby’s side, these norms make it likely that she will regard Morris’s request as a making manifest a reason, namely that given by Morris’s need, for telling Morris how to get to the Sears Tower. Insofar as the passerby acts on this reason, the norms thereby determine that her testimony is reliable. Thus the facts that determine reliability determine a reason for expecting reliability. So while trust-based reasons do not offer justificatory grounds for belief, and merely render Morris “not irrational” in believing what the passerby says, the reason trust provides is sufficient to ensure that Morris gets to know what the passerby tells him because with it reason and reliability are appropriately connected. This is not true in the case of the superstitious clairvoyant gambler. But it remains true in SUPERSTITIOUS CHICAGO VISITOR if Maurice’s attitude is, as it plausibly is, one of trust.

Section 6

As it stands, two-stage reliabilism fails to resolve the problem that reliability is insufficient for justification. Adding a second level of epistemic evaluation, a belief about — or epistemic perspective on — reliability does not resolve this problem all the while it remains possible for this belief to be ill-grounded. What justification, or knowledge, requires is that there be an appropriate connection between the reliability of a belief forming process or method and the belief that this process or method is reliable. What is required is that the belief be grounded in awareness of the reliability or the facts that determine it. This requirement, I suggested, can then be added as a further (third linking) condition in the formulation of a two-stage reliabilist theory of justification. The application of this theory to any given way of forming belief then requires, in addition to consideration of whether or not that way of forming belief is reliable, how any belief in this reliability might be grounded and so consideration of how this linking condition might be satisfied. When it comes to giving an epistemological theory of testimony, the fact that we allow knowledge to be acquired in cases like CHICAGO VISITOR necessitates an account of how the kind of reason for belief that the audience possesses in this case might be appropriately connected to the reliability of the testimonial belief that is formed.

The challenge of providing such an account must be met by Lackey’s Dualism, given its two-stage form, if it is provide a theory of testimonial justification and knowledge (and is to be more than a statement of necessary conditions). Such an account is then delivered, I suggested, by the proposal that in this case belief is formed on trust such that it is trust that provides the audience’s reason for belief. However, filling in the details of this account then involves the description of trust as a thick evaluative concept that “locally situates” reasoning. And this is to agree with the letter, if not the detail, of Fuller’s criticism of Goldman’s social epistemology: this social epistemology is epistemologically adequate only insofar as it has the resources to provide these kinds of ‘sociological’ descriptions.

References

Axtell, Guy, ed. 2000. Knowledge, belief, and character: Readings in virtue epistemology. Lanham, MD: Rowman and Littlefield.

BonJour, Laurence. 1980. “Externalist theories of empirical knowledge.” Midwest Studies in Philosophy 5 (1): 53-74.

Cohen, Stewart. 1984. “Justification and truth.” Philosophical Studies 46: 279-96.

Faulkner, Paul. 2010. “Norms of trust.” In Social epistemology, edited by Adrian Haddock, Alan Millar and Duncan Pritchard, 129-147. Oxford: Oxford University Press.

Faulkner, Paul. 2011. Knowledge on trust. Oxford: Oxford University Press.

Fuller, Steve. 2000. “Social epistemology: A philosophy for sociology or a sociology of philosophy?” Sociology 34 (3): 573-578.

Goldman, Alvin. 1967. “A causal theory of knowing.” Journal of Philosophy 64 (12): 357-372.

Goldman, Alvin. 1979. “What is justified belief?” In Justification and knowledge, edited by G. S. Pappas, 1-23. Dordrecht: D. Reidel.

Goldman, Alvin. 1994. “Epistemic folkways and scientific epistemology.” In Naturalizing Epistemology, 2^nd ed., edited by Hilary Kornblith, 291-316. Cambridge, MA: The MIT Press.

Goldman, Alvin. 1999a. Knowledge in a social world. Oxford: Clarendon Press.

Goldman, Alvin. 1999b. “A priori warrant and naturalistic epistemology.” In Philosophical perspectives, epistemology (Philosophical Perspectives Annual Volume) (Volume 13), edited by James E. Tomberlin, 1-28. Malden MA: Wiley-Blackwell.

Goldman, Alvin. 2004. “Sosa on reflective knowledge and virtue perspectivism.” In Ernest Sosa and his critics, edited by John Greco, 86-95. Oxford: Blackwell.

Goldman, Alvin. 2009. “Social epistemology: Theory and applications.” Royal Institute of Philosophy Supplement 64: 1-18.

Greco, John. 1993. “Virtues and vices of virtue epistemology.” Canadian Journal of Philosophy 23 (3): 413-432.

Greco, John. 2003. “Knowledge as credit for true belief.” In Intellectual virtue: Perspectives from ethics and epistemology, edited Michael DePaul and Linda Zagzebski, 111-134. Oxford: Oxford University Press.

Greco, John. 2009. “Knowledge and success from ability.” Philosophical Studies 142 (1): 17-26.

Kornblith, Hilary. 2004. “Sosa on human and animal knowledge.” In Ernest Sosa and his critics, edited by John Greco, 126-34. Oxford: Blackwell.

Lackey, Jennifer. 2006. “It takes two to tango: Beyond reductionism and nonreductionism in the epistemology of testimony.” In The epistemology of testimony, edited by Jennifer Lackey and Ernest Sosa, 160-89. Oxford: Oxford University Press.

Lackey, Jennifer. 2008. Learning from words: Testimony as a source of knowledge. Oxford: Oxford University Press.

Lackey, Jennifer. 2009. “Knowledge and Credit.” Philosophical Studies 142 (1): 27-42.

Lehrer, Keith and Stuart Cohen. 1983. “Justification, truth and knowledge.” Synthese 55 (2): 191-207.

Moran, Richard. 2001. Authority and estrangement: An essay on self-knowledge. Princeton: Princeton University Press.

Pritchard, Duncan. 2005. “Virtue epistemology and the acquisition of knowledge.” Philosophical Explorations 8 (3): 229-243.

Riggs, Wayne D. 2002. “Reliability and the value of knowledge.” Philosophy and Phenomenological Research 64 (1): 79-96.

Sosa, Ernest. 1988. “Intellectual virtue in perspective.” In Knowledge in perspective: Selected essays in epistemology, edited by E. Sosa, 270-94. Cambridge: Cambridge University Press.

Sosa, Ernest. 1991a. “Reliabilism and intellectual virtue.” In Knowledge in perspective: Selected essays in epistemology, edited by Ernest Sosa, 131-47. Cambridge: Cambridge University Press.

Sosa, Ernest. 1991b. “Knowledge and intellectual virtue.” In Knowledge in perspective: Selected essays in epistemology, edited by Ernest Sosa, 225-44. Cambridge: Cambridge University Press.

Sosa, Ernest, ed. 1995. Knowledge in perspective: Selected essays in epistemology. Cambridge: Cambridge University Press.

Sosa, Ernest. 2007. A virtue epistemology: Apt belief and reflective knowledge, volume I. Oxford: Clarendon Press.

Sosa, Ernest. 2009. “Human knowledge, animal and reflective.” In Reflective knowledge: Apt belief and reflective knowledge, volume II, edited by E. Sosa, 135-53. Oxford: Clarendon Press.

Categories: Articles

Tags: Alvin Goldman, dualism, Paul Faulkner, social epistemology, Steve Fuller, two-stage reliabilism, virtue reliabilism