2. THE INITIAL VIEW

We are not simply retainers of propositional knowledge. We also know how to acquire it. You, for example, know how to figure out whether your bathroom faucet is currently leaking, you know how to figure out whether your favorite sports team won more games this season than last season, you know how to figure out the sum of 294 and 3342, and so on. In normal circumstances when you exercise such know-how you typically gain (and retain) propositional knowledge. If you know how to figure out whether the faucet is leaking and you use that know-how, the typical result will be knowledge that the faucet is leaking (if it is leaking) or knowledge that the faucet is not leaking (if it is not leaking). My thesis is that justification is crucially connected to this kind of know-how.

Here is one way of capturing the idea:

The above account purports to explain the justification of beliefs in terms of propositional knowledge: a justified belief is what you get when you exercise your knowledge how to know that.Footnote ¹

A few aspects of (T1) need clarification. Following Hawley (Reference Hawley2003: 20–2), know-how is here relativized to “circumstances” by which I mean a thinker's potential environment and other relevant conditions that could obtain in a world. Take Greco's example of Jeter.Footnote ² He knows how to hit baseballs, but not in just any environment. War zones, hurricanes, and sand storms are all environments relative to which Jeter does not know how to hit baseballs. Similarly, although Jeter knows how to hit baseballs, he does not know how to hit baseballs under just any conditions. Jeter does not know how to hit baseballs with bats that weigh eight hundred pounds, or baseballs traveling at the speed of sound, or in the dark. The intuitive distinction between environments and conditions isn't crisp and permits overlap. Perhaps being in a place where intoxicating drugs are in the air counts as a condition, perhaps as an environment. Sorting out the relevant difference is unnecessary, and for present purposes I package them into know-how's circumstances. As Hawley points out, this relativization of know-how to circumstances is typically left implicit in our ordinary ascriptions of know-how.Footnote ³

Second, concerning condition (i), I take it that S knows how to know that P in C only if the following conditions are met:

These conditions are not intended to identify what it is in virtue of that one possess knowledge how to know that P in C. Rather, they are meant to highlight certain constitutive features involved in the target sort of know-how. Each of (i.a)–(i.c) involves reference to a way of φ-ing. Like most discussions of know-how, I offer no informative analysis of a way of φ-ing. This is unproblematic for it's easy to think of so many perfectly commonplace examples that we can be confident that the notion is well-grounded.Footnote ⁶

The core idea behind (i.a) is that one cannot know how to φ in C unless it's possible to φ in some way in C (Hawley Reference Hawley2003; Bengson and Moffett Reference Bengson, Moffett, Bengson and Moffett2011: 192). The core idea behind (i.b) is that knowing how to know that P in C is a matter of having a way of knowing that P which one knows how to exercise. When it comes to acquiring propositional knowledge apt examples will include our usual ways of responding to our perceptual, memorial, and introspective experiences to arrive at beliefs based on those experiences.Footnote ⁷ The idea behind (i.c) is that exercising one's know-how in general must lead to success when one is in the right kind of circumstances. Without this kind of success condition it would just be counterintuitive to regard one as possessing the target sort of know-how. The importance of conditions (i.a)–(i.c) will become apparent below.

Third, condition (ii) captures the thought that if S's belief is to be justified, S's believing P has to result from S's exercise of her knowledge how to know that P in C. It's not enough that one knows how to know that P in C and that one believe P in her actual circumstances. Rather, one's believing P must be brought about (and sustained) by one's knowledge how to know that P.

Lastly, the relativization of know-how to circumstances cannot be overemphasized. Otherwise the present view would imply the problematic thesis that: a way of knowing P simpliciter can be a way of falsely believing P simpliciter. The view I'm advocating has no such implication for there is no such thing as a way of knowing P simpliciter. All ways of knowing propositions are circumstance-relative and only demand the “counterfactual success” indicated in (i.c). So what the present view is committed to is this: a way of knowing P in certain circumstances can be a way of falsely believing P in other circumstances. For this reason merely exercising one's knowledge how to know that P in C doesn't entail that one knows P or that one is in C. In this sense, exercising know-how is non-factive. Moreover, it is crucial to emphasize that this relativization does not give way to the idea that someone can exercise her knowledge how to know that P in C by doing anything they want in any possible circumstance they find themselves in. There must be restrictions. A pilot can exercise her knowledge how to fly planes in normal circumstances by operating an advanced flight simulator, whereas she cannot exercise her knowledge how to fly planes in normal circumstances by dancing the mambo in a drunken stupor.Footnote ⁸ Although scope of this paper does not permit us to consider the various ways one might attempt to specify the intuitive restrictions that exist here, it should be clear from these sorts of examples that such restrictions do exist and our intuitive grip on them is enough for us to effectively theorize with.

3. REFINEMENTS

(T1) cannot be true. For, although exercising know-how is not factive, knowing how to know that P in C is quasi-factive. According to (i.a)–(i.c) one can have such know-how only if there is a way of coming by propositional knowledge of P in C, and one would succeed in knowing that P were one to exercise that know-how in C.Footnote ⁹ Accordingly, one can possess knowledge how to know that P in C only if P is true and knowledgeable when one is in C. Because of this, (T1) rules out the possibility of justified false beliefs in C as well as having justified true beliefs in C that cannot be known due to epistemic luck (as in Gettier cases). But that is implausible … or so says epistemological orthodoxy.

We can make space for such beliefs to be justified by changing (T1)'s circumstantial parameter in the following way. Let ‘C_@’ refer to the type of a thinker's actual circumstances and ‘the C _P s’ refer to the collection of the types of circumstances in which it is possible for S to know how to know that P (because conditions (i.a)–(i.c) are satisfiable in each of the C _P s). We can then adjust (T1) to permit justified false beliefs and justified Gettier beliefs as follows:

So long as (T2) allows C_@ to be distinct from each of the C _P s it creates the possibility of having justified false beliefs as well as justified Gettier beliefs. For even if such beliefs cannot be known in one's actual circumstances, according to (T2) one can yet have a justified belief in them so long as there are other circumstances where they are knowledgeable. Thus (T2) easily avoids (T1)'s problem (see section 4.1 for discussion of examples).

However, it is clear that S cannot have justification to believe P in any circumstances she find herself in just because it's possible for her to know how to know that P in some circumstances.Footnote ¹¹ So we need to specify a suitable relation that holds between S, C_@, and the C _P s. This is the business of the I-relation. The most plausible specification of the I-relation will be one that appreciates what is common between Gettier cases and cases where one is misled so as to have a justified false belief: they are cases where the thinker in question is, in some way or other, unaware of the unfortunate nature of her circumstances. Put differently, in such circumstances S is in a position where her actual circumstances are in some way “indistinguishable” from circumstances where she would not have been misled and would have had propositional knowledge.

Appreciating these points naturally directs us to the following specifications of the I-relation:

When inserted into (T2), this amounts to saying that justifiably believing P requires that S doesn't know that she's not in a circumstance where it's possible for her to know that P (Williamson Reference Williamson2000). Due to the ease with which one can fail to have knowledge – e.g. defeaters, Gettier factors, etc. – requiring an absence of knowledge is too weak a condition to be plausible. If so, here's an alternative specification of the I-relation:

When inserted into (T2), this amounts to saying that justifiably believing P requires that S lack adequate reason to think she's not in a circumstance where it's possible for her to know that P. The basic idea behind Dogmatist ways of restricting (T2) is that they permit one to have justification in, say, classical skeptical cases (e.g. BIV scenarios) because they are cases where, although one cannot know how to know that P in those circumstances, skeptical circumstances are in a certain sense “indistinguishable” from circumstances in which one does know how to know that P and, for all one knows, one is in such a case. The same holds in more mundane cases of misleading evidence and gettier cases, for what is characteristic of those cases is one's inability to discern that one's evidence is misleading or that they've been subject to epistemic luck.Footnote ¹³

Of course one might think Dogmatist specifications of the I-relation are too weak, maintaining that one must also have some sort of positive reason to think one is in the kind of circumstances where obtaining propositional knowledge is possible. Hence:

Such specifications are more demanding than their Dogmatist counterparts, requiring one to have adequate reason to think/know that one is in epistemically favorable circumstances with respect to the belief in question.Footnote ¹⁴ Perhaps there are less demanding permutations that fall between those outlined above. But in what follows I will defend no particular specification of the I-relation. This is a choice-point for advocates of the knowledge-first approach to justification I'm defending.

So (T2) avoids the problems generated by (T1)'s tight circumstantial parameter. But there is a lingering problem that (T2) faces. For no matter how loose one's circumstantial parameter is, some propositions are beyond knowledge, e.g. the necessarily false. But this conflicts with the common thought that we can have justified beliefs in necessarily false propositions. Frege, arguably, justifiably believed the Naive Comprehension Axiom.

The solution to this problem is to focus on the fact that one may believe P as a result of exercising a way of knowing in C despite the fact that P cannot be known. For even if one cannot know how to know that P – because P is necessarily false – one can still believe P as the result of an exercise of a way of knowing other propositions. For example, one way of coming by propositional knowledge is believing on the basis of testimony you have strong reason not to doubt. Suppose Frege asserted that the fifth proposition on the board is a logical truth, though it is actually a necessary falsehood. Frege's student might give the assertion some consideration, and see no problem with that particular assertion, and believe the assertion on the basis of Frege's testimony. Although the student has not exercised a way of knowing the fifth proposition (since it cannot be known), she has exercised her knowledge how to acquire propositional knowledge of other propositions on the basis of testimony by forming her belief in the particular way that she did. The same is also true of Frege. Frege knew how to acquire propositional knowledge of a certain range of logical and mathematical propositions. The way in which Frege acquired such propositional knowledge was perhaps complex, involving no simple movement from, say, the intuition that P to belief in P. It likely involved further checks and balances we needn't pause over. The key observation here is that in arriving at his belief in the Naive Comprehension Axiom in, say, the intuition-driven way that he did, Frege exercised his knowledge how to acquire propositional knowledge of other propositions in an intuition-driven way.Footnote ¹⁵

Here's a permutation of (T2) that homes in on these facts. Letting ‘the C _W s’ refer to the collection of the type of circumstances in which W is a way of gaining propositional knowledge of certain propositions:

Condition (i”) tells us that S must have a way of acquiring propositional knowledge to have justification. As discussed above, believing on the basis of testimony one has reason to trust (as Frege's student did) and believing on the basis of intuitions (in the way that Frege did) are ways of acquiring propositional knowledge. Other ways will involve forming beliefs on the basis of perceptual experiences, memorial experiences, reasoning under certain conditions, and so forth. Condition (ii’) tells us that S must believe P as an exercise of some subset of those ways if she's to have justification. (J = KH) is, save one minor adjustment below, the general account of justification I will be defending on the grounds that it can account for a greater range of intuitive facts about justification than any other competing knowledge-first account.

First, (J = KH) grounds justification in an uncontroversial fact: we are knowers who have ways of acquiring propositional knowledge. Moreover, this is a fact that is closely linked to justification whatever one's view of justification happens to be. For whenever we successfully exercise our knowledge how to acquire propositional knowledge of P by employing some such way it follows that we not only know that P but that we also have a justified belief in P. (J = KH) explains why this is. And even when we unsuccessfully exercise that know-how, and hence fail to acquire propositional knowledge, the result is typically a justified belief. (J = KH) says this is not merely an accidental correlation between know-how and justification. Rather, this kind of know-how plays a key role in explaining justification.

Second, (J = KH) vindicates the now-common thought that justification is crucially connected to intellectual virtue (Zagzebski Reference Zagzebski1996; Sosa Reference Sosa2007; Turri Reference Turri2010; Greco Reference Greco2010, Reference Greco2012). For knowing how to gain propositional knowledge is such a virtue. Accordingly, (J = KH) is in keeping with broadly virtue-theoretic approaches to epistemic justification. It differs in that standard virtue theories are developed in the context of abilities rather than know-how. I have here preferred a know-how based account for two reasons. First, Hawley (Reference Hawley2003) has done a great bit of work elucidating the general structure of know-how. Second, it's intuitive to think that, provided know-how and abilities are distinct, know-how is a more likely place to look to ground propositional knowledge and justification (more on this in the Kelp discussion below).

Third, (J = KH) can explain the way in which the dogmatist/conservative debate of almost the last two decades hasn't been just a pointless debate in the theory of justification (Pryor Reference Pryor2000; Cohen Reference Cohen2002; Wright Reference Wright2004; White Reference White2006; Silins Reference Silins, Gendler and Hawthorne2008; Silva Reference Silva2013). For there is a substantive question as to just which specification of the I-relation is correct. And many of the considerations raised by dogmatists and conservatives of a non-knowledge-first approach to justification can be brought to bear on this question with relative ease.

Finally, although (J = KH) is not clearly a form of internalism,Footnote ¹⁶ it is internalist-friendly insofar as it respects certain core internalist intuitions. For suppose S exercises her knowledge how to acquire propositional knowledge by believing P on the basis of her perceptual experience. Notice that being in the sort of conditions needed to acquire perceptual knowledge is not an essential part of S's exercise of that know-how. For just as a pilot can exercise her knowledge how to fly a plane when unwittingly placed in a cleverly constructed flight simulator, a thinker can exercise her knowledge how to acquire propositional knowledge when she's the unwitting victim of an envattment procedure. Ways of acquiring propositional knowledge are constituted by ways of transitioning to belief states. So every exercise of a way of acquiring propositional knowledge in C is an exercise of a way of transitioning to a belief state. This exercise is constituted by triggering certain belief-forming dispositions and capacities which – because it's a way of knowing in C – would yield knowledge if one were in C. But one needn't be in C for those dispositions and capacities to be triggered. So conditions (i”) and (ii’) don't appear to carry any strongly externalist conditions in their wake. Thus, if S* knows how to do the same things as S does, and S* exercises that same know-how as S does in believing P on the basis of her perceptual experience, it will follow from (J = KH) that S* has justification iff S does – so long as condition (iii’) is satisfied for both S and S*.Footnote ¹⁷ Thus, provided one's preferred specification of condition (iii’) doesn't generate any problems, advocates of (J = KH) can endorse:

It is beyond the scope of this paper to defend (NEG). But as Neta and Pritchard (Reference Neta and Pritchard2007) point out, (NEG) is among the most widely influential internalist intuitions in recent epistemology. Accordingly, (J = KH)'s compatibility with (NEG) is a significant mark in its favor.Footnote ¹⁹

In certain ways, (J = KH) resembles two recent knowledge-first views of justification. First, according to Lisa Miracchi, a justified belief is, roughly, a belief that is the result of the exercise of some competence to know, where such competences are reliable abilities to know (Miracchi Reference Miracchi2015). Accordingly, hers is a thoroughly reliabilist account of justification, making the accommodation of internalist intuitions like (NEG) impossible. And the only apparent way to get her view to accommodate (NEG) is to drop the reliabilist constraint on abilities and to introduce something like the circumstance-relative machinery developed above in the progression from (T1) to (J = KH). Thus, insofar as internalist intuitions are pervasive and well-grounded, Miracchi's account faces an objection stemming from (NEG), one that (J = KH) nimbly dodges. This is not to say that Miracchi is unable to explain away the intuition in terms of some other positive epistemic property distinct from justification. Like any other externalist she can appeal to blamelessness or excusability or some other non-justificatory property.Footnote ²⁰ But advocates of (J = KH) needn't bother.

The second recent knowledge knowledge-first view of justification is Christoph Kelp's view. According to Kelp: one has a justified belief that P if and only if one competently believes that P. Here's his account of competent belief:

Like Miracchi's preferred account, Kelp prefers an account of justification that is cast in terms of competences to know that P rather than knowing how to know that P. But unlike Miracchi's view his view is potentially able to accommodate (NEG) due to its relativization of the relevant competences to circumstances.Footnote ²²

So why prefer (J = KH) to (J = CB)? The first reason to prefer (J = KH) to (J = CB) is that (J = CB) lacks anything like an I-relation. (J = KH)'s inclusion of an I-relation not only locates a legitimate place for the dogmatist/conservative debate in epistemology, but it also protects (J = KH) against counterexamples that (J = CB) faces. These counterexamples to (J = CB) crop up in connection with the fact that one can exercise an ability to (or knowledge how to) know propositions in a certain range of favorable circumstances C, while being in circumstances, C*, that one has good reason to think are unfavorable for acquiring propositional knowledge. For example, suppose you knew there were cadres of crazed scientists roaming the land, envatting brains at every turn. Suppose further that you had vivid memories of boffins armed with envattment equipment storming your home and subduing you. Furthermore, suppose that the beginning of the envattment procedure created various visual blips and other discontinuities in your visual field, blips which are persisting. In such circumstances you could know that you were the subject of an envattment procedure and that your perceptual beliefs no longer reliably track reality. Yet despite your new unfavorable circumstances, you could still exercise your competence to acquire propositional knowledgeFootnote ²³ in favorable circumstances by forming perceptual beliefs in response to your visual experiences in the way you normally would. (J = CB) implies that such post-envattment perceptual beliefs would be justified because they result from an exercise of your ability to acquire propositional knowledge in favorable circumstances. But surely such beliefs would be unjustified. (J = KH) is protected against any such problematic implication given its inclusion of an I-relation. For such beliefs are formed in the face of one's knowledge of one's own envattment, and hence one's knowledge that one's circumstances are no longer fit for forming perceptual beliefs about one's immediate environment.Footnote ²⁴

A defender of (J = CB) might object that any way of forming beliefs that doesn't halt the perceptual belief-forming process in response to such envattment information is not really a competence to know. But that is incorrect. For there are two kinds of competences (and kinds of know-how) to be distinguished. First there are the simple, unreflective competences that take no note of factors that indicate that one might be in unfavorable circumstances for the employment of the simple competence. Simple competences are what young children possess, children who have not yet come to realize (nor are they in a position to realize) the many ways in which one can be misled by one's environment. Yet, as with all competences according to Kelp, these simple competences are circumstance relative. So a simple competence to know that P in favorable circumstances remains a competence that one can possess and exercise even if one is not actually in favorable circumstances.

But there is a second, more epistemically advanced competence that we develop over time. It's a competence to selectively employ the simple competence; a competence that is (at least partially) responsive to factors that indicate that one's circumstances are not fit for the employment of the simple competence. The advanced competence does not make the simple competence any less of a competence in favorable circumstances. It just helps us appropriately respond to facts pertaining to whether or not we're actually in such favorable circumstances, i.e. the circumstances relative to which the simple competence is indeed a competence. Without something broadly like an I-relation our ability to monitor our circumstances and appropriately respond to information indicating that we are in circumstances unfit for employing a given competence becomes justificatorily irrelevant. This simply isn't plausible. A way out of this quandary lies with an I-relation.

(J = CB) can obviously be augmented so as to include an I-relation, and I would advise advocates of (J = CB) to do so. Is there any reason to prefer (J = KH) to an augmented version of (J = CB)? To fix ideas, take a view of justification as competent belief that includes an I-relation and employs more of the ideology of (J = KH) than Kelp's thesis:

One way of evaluating (J = KH) vis-à-vis (J = CB+) is to see if there are any properties of justification that we would expect a theory of justification to explain, and to see if both (J = KH) and (J = CB+) prove equal to that task. For example, justified belief seems to be a cognitive achievement in at least the sense that a thinker whose belief is justified is creditworthy for believing as she does (more on this in section 4.3).Footnote ²⁵ Do exercises of know-how as well as exercises of competences constitute a cognitive achievement? Bengson and Moffett (Reference Bengson, Moffett, Bengson and Moffett2011) have argued on behalf of intellectualism that behavioral-dispositional states (and hence competences) are unable to explain the fact that there is a constitutive connection between know-how and cognitive achievement.Footnote ²⁶ If they are right, then any competence-based approach to justification will be unable to explain the fact that justified belief is a cognitive achievement. This would afford us a strong reason to prefer (J = KH) to (J = CB+) as an explanation of what justifies belief.

But let's assume anti-intellectualists can rally an explanation for know-how's connection to cognitive achievement, and for simplicity let's assume that the version of anti-intellectualism at issue reduces knowing how to know that P in C to a competence to know that P in C.Footnote ²⁷ This may appear to be a reason to favor (J = CB+) over (J = KH). But it's not. For if knowing how to know that P in C is reducible to a competence to know that P in C, then (J = KH) will be reducible to (J = CB+). In this case, (J = KH) turns out to be true and able to explain what justifies belief just because (J = KH) is (J = CB+) and because (J = CB+) is true and able to explain what justifies belief. Given such a reduction, (J = KH) and (J = CB+) are theses that neither conflict nor compete as explanations of justified belief.

So while (J = KH) remains a viable explanation of justification for intellectualists who think competences cannot constitute cognitive achievements, (J = CB+) does not. But those anti-intellectualists who reduce the relevant know-how to competences and advocate (J = CB+) will not have any reason to reject the explanation of justification that (J = KH) offers. They'll just take (J = CB+) to be a more metaphysically illuminating explanation of justification because it clues us in to the nature of know-how. There are obviously many rabbit trails here for there are many different ways of being an (anti-)intellectualist. If the number of (anti-)intellectualist analyses of know-how is n, then there will be as many ways of endorsing (J = KH). What I hope to have shown is both that (J = CB+) and (J = KH) are compatible, and that (J = KH) is something intellectualists and anti-intellectualists needn't disagree about despite their differing opinions on the nature of know-how.Footnote ^{28 ,} Footnote ²⁹

4. COLLATERAL BENEFITS OF (J = KH)

Various knowledge-first views of justification have been proposed in the last decade. (J = KH) joins their ranks. But what is there to be said in favor of (J = KH) vis-à-vis alternative knowledge-first accounts of justification? Much and more. My strategy in what follows is to highlight intuitively plausible properties of justification that (J = KH) is able to accommodate without introducing any kind of error theory.