Interim Review
© 2017 Mark McBride, CC BY 4.0 https://doi.org/10.11647/OBP.0104.08
As we now transition from Part One to Part Two of the book, let’s take stock of what has been achieved in Part One and what remains to be achieved in Part Two. First, what has been achieved? Essentially, immediate justification (such as perceptual justification as understood by the dogmatist) allows for basic knowledge, and basic knowledge seems to give rise to at least two problems. (A) It would allow that provides an adequate response to the sceptic. (B) It would allow a ‘too easy’ inferential route from basic knowledge to other knowledge. This was my pattern of objection to dogmatism, immediate justification, and basic knowledge. In Chapter One, I argued that Wright’s position (which does not accept that there is immediate perceptual justification to believe the Moorean premise ‘I have hands’) still faces the (MOORE)-transmit problem. In Chapter Three, I argued that, even if (MOORE)-reasoning and (EK)-reasoning do transmit warrant and do not involve any epistemic circularity, they still exhibit an epistemological limitation. At that point we could have concluded that the prospects of allowing for immediate justification and basic knowledge would be rather good.
In Chapter Four I argued that a concept central to the work of philosophers debating over immediate justification and basic knowledge — viz. transmission failure — is secure from a particular challenge. Specifically, I argued that transmission failure is possible on all or most plausible accounts of the sources of warrant-or-evidence for future reasoning. Of course, the notion of transmission failure discussed in that chapter is connected with epistemic circularity; and a core upshot of dogmatism is that there is no epistemic circularity in (MOORE)-reasoning. The challenge faced by the dogmatist is to respond to problems (A) and (B) without appeal to epistemic circularity. Thus, the importance of Chapter Four is more theoretical than dialectical.
The dialectical importance of the final chapter in Part One is, by contrast, extreme. There, I argued that epistemological dogmatism is vulnerable to a strong objection. Here, I consider whether there are any good responses to that objection.
1. Reacting to Chapter Five’s Objection to Dogmatism
I closed Chapter Five by suggesting that a defender of dogmatism about justification or dogmatism about knowledge may well look to offer reasons to reject (J-Closure) or (K-Closure), respectively. In the coming chapters — Chapter Six and, in particular, Chapter Seven — I oppose, specifically, John Hawthorne’s recent defense of (K-Closure). Those chapters join Hawthorne in, arguendo, assuming that something in the region of conclusive reasons or sensitivity is a condition on knowledge, before going on to render those conditions in as plausible a form as possible.1
In section 5 of the Introduction I mentioned a way — involving taking conclusive reasons or sensitivity to characterise justification — of extending any rejection of closure from (K-Closure) to (J-Closure).2 I noted, though, that even with conclusive reasons or sensitivity conceived as a necessary condition for justification, that would not guarantee an exception to (J-Closure). This is because: although if the conclusion did not meet the conclusive reasons or sensitivity condition it would not be justified, the premise might also not be justified, despite meeting the conclusive reasons or sensitivity condition. However, I noted that in any actual case there might be no reason to suppose that the premise, which meets the conclusive reasons or sensitivity condition, nevertheless fails to meet some other necessary condition for justification. Consider the dialectical context where someone is trying to pin a problem on dogmatism by using (J-Closure) to proceed from justification to believe (1) to justification to believe (3). The person posing this problem is content to assume that the conditions for justification to believe (1) are met, and we can add that (1) meets the conclusive reasons or sensitivity condition. So it is hard to see that the problem-poser has any good reason to reject the continuing assumption that there is justification to believe (1) or justified belief in (1), even with conclusive reasons or sensitivity now added as a necessary condition on justification. But (3), though putatively justified, fails the conclusive reasons or sensitivity test. So there is plausibly an exception to (J-Closure).
In sum, while I believe such an extension is worthy of serious consideration, it must be conceded that I have offered no decisive argument against (J-Closure).3 So, even if I can make out a successful (fully general) case for rejecting (K-Closure) — and so for rejecting the argument against dogmatism about knowledge — I cannot presently claim that that success carries over to vindicate rejection of (J-Closure) — and so rejection of the argument against dogmatism about justification.
However, I am not resigned to accepting that dogmatism about justification is false, and one might reject the arguments against dogmatism by other means than a rejection of closure principles. Here, I briefly suggest an alternative way of responding to the arguments presented in Chapter Five against dogmatism. This alternative — not pursued in Chapter Five — involves a form of departure from the standard Bayesian picture. To some degree, which of these two routes a defender of dogmatism pursues — and I do not claim them to be exhaustive — will depend on the costs associated with, respectively, rejection of closure principles and a departure from standard Bayesian assumptions.
Davies’s (2009) account, on which I build in Chapter Three, “involve[s] some departure from the standard Bayesian requirement that a probability distribution must assign some probability, high or low, to each proposition, including [MOORE(3) An external world exists]” (356).4 We can take it that this also includes: The table is not white with red lights shining on it.
These propositions — the conclusions of (MOORE)-reasoning and (EK)-reasoning respectively — may be outside the subject’s conceptual repertoire when the subject first comes to believe the premise, ‘I have hands’ or ‘The table is red’, on the basis of a perceptual experience as of having hands or as of a red table. Once the conclusion proposition is grasped, and the entailment is recognised, the subject is committed to assigning a posterior credence to the conclusion that is no lower than the posterior credence assigned to the premise. However, we might depart from the requirement that a coherent assignment of prior credences should encompass propositions of which the subject has, at that stage, no understanding. While much work would need to be done, it might be that this kind of departure from the standard Bayesian picture could offer another way to respond to Chapter Five’s arguments against dogmatism.
We have briefly reviewed two possible methods of resisting Chapter Five’s arguments against dogmatism. I now want to review a third possible method. Nico Silins (2007) can be taken to claim that White’s arguments — which I sought to capture in Chapter Five — only establish (I) that it is impossible for anyone ever to acquire evidence that justifies ordinary empirical propositions unless they already had, prior to acquiring that evidence, propositional justification for rejecting sceptical hypotheses, and so do not establish (R) that our justification for ordinary empirical propositions essentially rests on or is supported by our justification for rejecting sceptical hypotheses.5
This is a promising strategy. Does Silins make good on his claim? (To the extent that he does, of course, the dogmatist will have another possible escape route from Chapter Five’s objection.) Consider the following example: (Prime) If it is a necessary truth that every thinker always has (propositional) justification for certain a priori knowable propositions, then it is impossible ever to acquire evidence that justifies me in believing that I have hands unless I already had, prior to acquiring that evidence, propositional justification for believing the a priori proposition that there is no largest prime number; however it would not be that my justification for believing that I have hands essentially rests on or is supported by my justification for believing that there is no largest prime number. A supporter of Silins might suggest there is an analogy between Prime and the case(s) involving sceptical hypotheses.6 This analogy would suggest that, even if, for example, it is impossible to be justified in believing that I have hands on the basis of a perceptual experience as of hands unless I have antecedent warrant to believe the anti-sceptical proposition that there is an external world, still the warrant to believe that I have hands might not depend on the antecedent warrant to reject the sceptical hypothesis.
Here is a dogmatist-friendly view of what the Prime example establishes:
(iv**) If one has justification to believe ‘I have hands’, after having perceptual experience as of having hands then one must have had justification to believe there is no largest prime number antecedently to having perceptual experience as of having hands.
This is of the same form as (iv) in the argument against j-dogmatism in Chapter Five. The point of the Prime example is that it obviously does not follow that j-dogmatism is false. It does, let us accept, follow from (iv**) that:
(v**) Perceptual experience as of having hands’ ability to provide justification to believe ‘I have hands’ is not modally independent of whether one has antecedent justification to believe that there is no largest prime. (One cannot have one justification without having the other.)
However, there is clearly a gap between that (allowing that it does follow) and ‘j-dogmatism is false’, which does not follow. The point is that from the modal dependence in (v**) it clearly does not follow that justification to believe ‘I have hands’ depends constitutively on justification to believe that there is no largest prime — even though one cannot have one justification without having the other. Having a justification to believe that there is no largest prime does not partly constitute having a justification to believe ‘I have hands’; it is not part of what it is for one to have a justification to believe ‘I have hands’. And I take it that epistemic dependence is a kind of constitutive, and not merely modal, dependence.
I think there is much merit in Silins’s strategy. I also think that we might not need even to grant that the argument of Chapter Five establishes the relation of modal dependence between epistemic warrants. We could allow that the argument does establish:
(iv—) If one has justification to believe ‘I have hands’, after having perceptual experience as of having hands then it must be that a coherent assignment of credences would have assigned a high credence to an anti-sceptical hypothesis antecedently to one’s having perceptual experience as of having hands.
And then we could ask, with Pryor (2012: 282): “What’s the relation between probabilistically confirming and supplying warrant?”.
Still, it is not entirely clear to me, at this point, how best to respond to Chapter Five’s objection to dogmatism. While I have reviewed several options for the dogmatist — options which I take to be very much live and, indeed, promising — much remains to be done before any utterly convincing response to the objection could be presented.
I certainly do not at all assume that the objection to dogmatism is unanswerable. Indeed, my provisional view is that dogmatism remains a tenable position. Nevertheless, it is reasonable to consider the worst-case possibility that, ultimately, dogmatism cannot be defended against the objection. It would be natural, at that point, to consider taking Roger White’s (2006) line, whose focus, recall, is specifically on dogmatism about justification.
White (2006: 552–53), not interested in questioning (J-Closure), considers retreating from dogmatism about justification to a distinct thesis which he calls JUSTIFICATION BY DEFAULT (cf. Pryor 2000: 519).7 At points in the book up until now, we’ve had cause to highlight elements of White’s position as they were relevant. At this point, we do well to state it in full:
Suppose that we abandon dogmatism, and insist that in order to gain perceptual justification for believing that P, we must have independent justification for believing that we are not victims of a visual illusion that P. We could nevertheless insist that we have a kind of default justification for assuming the general reliability of our perceptual faculties. We are entitled to believe that our faculties tend to deliver the truth unless we have some positive reason to doubt this.8 Our faculties are generally reliable only if skeptical alternatives rarely obtain. So if I’m justified in taking my faculties to be reliable, then I need give very little credence to skeptical hypotheses (unless of course I have reason to suspect that one obtains). On the view that I’m sketching I do not need to explicitly believe in the reliability of my faculties or the falsity of sceptical alternatives in order to gain justification from perceptual experience, but justification for this reliability is available to me nevertheless […]9 [I]t is denied that the justification for ruling out skeptical alternatives requires some prior empirical ground. This justification is available a priori by default. I am justified in my perceptually based beliefs, according to this view, in just the circumstances that the dogmatist claims that I am. Such a view seems to have all of the advantages of dogmatism but avoids all of my objections. No doubt the view requires closer examination and development. But if we are attracted to something in the ball park of dogmatism, it seems to be the right place to look.
What follows? Suppose we retreat from dogmatism, allow that justification to believe quotidian propositions epistemically depends on justification to believe an anti-sceptical hypothesis or reliability hypothesis, and accept — with White — that we have such JUSTIFICATION BY DEFAULT. A corollary of such a retreat would appear to be that we could then say that the solution to problems (A) and (B) — outlined at the outset of this Interim Review — is that the relevant reasoning involves epistemic circularity in each case.10 However, the situation is not so straightforward. I showed in Chapter One (in effect) that there can still be transmission of some warrant even where there is a kind of epistemic circularity. More on this later, in section 2.
Now if we have justification by default for a reliability hypothesis, we might not need to believe it, still less to know it, in order to know quotidian propositions: there could be antecedent justification to believe, but no antecedent belief and so no antecedent knowledge. Still, any such knowledge of quotidian propositions wouldn’t be (the logically stronger) Pryorian basic knowledge, because it depends epistemically (constitutively) on one’s having justification to believe the reliability proposition. Could it, though, be (the logically weaker) Cohenian basic knowledge? Interestingly, this is precisely the case about which it was said in the Introduction (p. 10) that it is at best unclear that it is a case of Cohenian basic knowledge.
Though White’s focus is on justification, the argument of Chapter Five did apply to knowledge, and not just justification. But the most that the argument in Chapter Five would establish is that if one knows, with source K, some quotidian proposition, then one was antecedently in a position to know that K is reliable. That is not the same as saying that one would antecedently know that K is reliable. So far, so good for Cohen’s basic knowledge. However, if the argument in Chapter Five establishes that the justification to believe the quotidian proposition depends epistemically (constitutively) on the antecedent justification to believe that K is reliable, then I have said that it is at best unclear whether this is Cohenian basic knowledge. On the other hand, if the argument does not establish the constitutive epistemic dependence, then we should say that knowledge of the quotidian proposition is Pryorian (and thus Cohenian) basic knowledge — and that the challenge to dogmatism was unsuccessful.
2. Looking Back to Part One and Ahead to Part Two
Now what follows with respect to the work done in Part One and the work to be done in Part Two? First, what follows with regard to Part One? In Chapter One we saw that Wright, contra the standard view, faced the difficult (MOORE)-transmit problem — a problem standardly only taken to be faced by dogmatists. This problem — the particular problem of giving a plausible account of why something seems wrong with (MOORE)-reasoning given it can transmit warrant — cannot be answered by Wright by pointing out that there’s non-transmission of (disjunctive) warrant by the lights of his (non-)transmission theses. Wright cannot so answer because there’s transmission of evidential warrant by the lights of two other plausible (non-)transmission theses.11 Crucially, in each case — that is, in both the case involving Wright’s theses and the case involving the two other plausible theses — the warrant under consideration is provided by the same experience, a visual experience as of having hands. (For why invocation of the two theses that license transmission of evidential warrant is not question-begging against Wright, see my introduction, and subsequent discussion, of theses (A)-(D) in Chapter One, section 2.4ff.) One natural idea would be for Wright to borrow the kind of story that I tell in Chapter Three (on behalf of the dogmatist) about the problem of easy knowledge. But Chapter One has shown that Wright cannot happily borrow that kind of story, especially given his (1985) position on the non-factuality of the proposition that there is an external world.
Insofar as dogmatism turns out not to be a live option, Wright would no longer be at a dialectical disadvantage vis-à-vis dogmatism. Nevertheless, insofar as Wright faces the (MOORE)-transmit problem, Chapter One can be taken to pose the following question to Wright: how do you answer the (MOORE)-transmit problem? Note also that in Chapter One (at 1.1) I wrote: “Assume, then — with the dogmatist and Wright — that something (at least) seems wrong with (MOORE)-reasoning.” Meanwhile, in Chapter Three I claimed that reasoning fundamentally the same as (MOORE)-reasoning — namely, (EK*)-reasoning — is less unsatisfactory than (EK)-reasoning. Is there any conflict between these claims from Chapter One and Chapter Three? No. All that the main argument of Chapter Three requires is that (EK)-reasoning and (EK*)-reasoning are different, in that (EK*)-reasoning seems less unsatisfactory than (EK)-reasoning. Importantly, it does not require that either (EK*)-reasoning or (MOORE)-reasoning is wholly satisfactory. Indeed it diagnoses an epistemic limitation in each form of reasoning, namely that neither argument can be used in the epistemic project of settling the question — albeit that limitation doesn’t preclude knowledge.
Moreover, we can note that, if Wright ends up with a first time perceptual warrant for the proposition that there is an external world, then his position is, to that extent, similar to the dogmatist’s position, and Wright needs to explain why (MOORE)-reasoning seems not to be a satisfying response to the sceptic. That is quite consistent with the idea that (EK)-reasoning seems to be somehow epistemically worse than (EK*)-reasoning. I take Chapter Three to have established that none of the principles that have been proposed can account for this difference.
Chapters Two and Three then looked at the problem of easy knowledge. These chapters explore a legitimate question: is there a solution to the problem of easy knowledge that is available to the dogmatist? My answer is: yes, there is. So if there is something wrong with dogmatism, it is not that it cannot respond to the problem of easy knowledge. Of course, we must now ask how things stand if the argument of Chapter Five shows dogmatism to be untenable. Then, the fact that justification for (EK1) depends on justification for (EK3) means that we have a kind of epistemic (justificatory) circularity. So that is at least part of the solution to the problem of easy knowledge (unlike the situation in Chapter Three). There might still be an additional problem if an argument like that made in Chapter One shows that there is an (EK)-transmit problem (despite the circularity). However, that raises issues quite different from those in Chapter Three.
Finally, what follows with regard to the ensuing Part Two — our exploration of certain conditions that some philosophers have claimed are necessary for knowledge, viz. conclusive reasons, sensitivity and safety? It bears emphasising that we need not assume that we have basic knowledge to make exploration of conclusive reasons, sensitivity,12 and safety worthwhile. Moreover, now that we have seen that a defender of dogmatism about knowledge may well want to offer reasons to reject (K-Closure), the outcome of our enquiry into (K-Closure) in Chapters Six and Seven (and an additional closely related knowledge-closure principle in Chapter Six)13 becomes of increased interest.14
1 It might seem that a defense of dogmatism about knowledge (a position which, concededly, hasn’t clearly been endorsed by any philosopher thus far, but which has been articulated and which is independently interesting) ideally ought not to rest on such an assumption. This is not because these conditions are peculiarly anti-dogmatist in spirit; it is rather just that a defense of dogmatism should rely on as few assumptions as possible, in order to avoid creating hostages to fortune. But note: if such a defense works, there couldn’t be knowledge by inference of the conclusion of (EK) (or (MOORE)).
2 If this extension works, there couldn’t be justification by inference of the conclusion of (EK) (or (MOORE)). So (EK)-reasoning would be an example of transmission failure. So, apparently, there could be transmission failure that was not related to epistemic circularity.
3 Recall, (J-Closure) is defined in terms of justification to believe. While there any surely reasons to reject a closure principle defined instead in terms of justified belief, such a principle is not presently germane.
4 A departure of this kind, for Davies (2009: 356) results from “[s]upposing […] that there were a theory of evidential support with the following two features. First, Moore’s experience as of hands would provide no support for MOORE(1) [I have hands] given the negation of MOORE(3) as background assumption but, second, Moore’s experience could provide support for MOORE(1) even without the positive adoption of MOORE(3) as a background assumption.” Indeed, more generally Davies’s distinction between two epistemic projects (and associated resultant distinctions) may be difficult to capture in the standard Bayesian framework. Cf. Pryor (2012: 282–83) for a related “propos[al] to set […] probabilistic considerations aside for […] discussion [of transmission-failure]” on account of “many complications at play”. More generally on this issue — that is, resolution of potential conflict between formal considerations and a family of views of which dogmatism is a member — see Pryor (2013).
5 Let us not probe further into some of the notions introduced in (R) — e.g. essentially rests on, is supported by — other than to note that they are not Bayesian notions as such (likewise for the notions of immediate versus mediate evidential support). White (2006: 555, n. 14) gestures towards this Silins point, appealing to the “in virtue of” relation. Put somewhat vernacularly, the point might be that there is some slippage here between different notions of independence in the arguments as formulated in Chapter Five.
6 Note, though, while justification for believing that there is no largest prime number, ex hypothesi in Prime, is necessary a priori, it’s plausible that default justification for rejecting sceptical hypotheses, if it exists, is contingent a priori (cf. Hawthorne 2002 and Cohen 2002: 320–22).
7 As White notes, JUSTIFICATION BY DEFAULT bears obviously similarities to Wright’s appeal to the notion of unearned warrant; Pryor rejects these stances.
8 Cf. Burge’s (1993, 2003) view, on which we have such an entitlement to rely instead on our faculties.
9 Thus, White has in mind propositional, and not doxastic, justification.
10 Insofar as Cohen doesn’t appeal to epistemic circularity to solve problems (1) and (2), we can be sure — confirming Chapter Three’s account of his position — that his position is not simply described in terms of such a retreat.
11 It might be objected that Moore’s argument fails completely to reveal why our acceptance of its conclusion has the status of having any warrant whatsoever. One way of understanding this objection is that the evidential warrant that is transmitted from premise (1) of (MOORE) is — according to Wright’s view — dependent on the unearned warrant to accept premise (3) of (MOORE). Wright (2004) indicates a philosophical theoretical account of the nature of that unearned warrant but, of course, this piece of philosophical theory will not be within the grasp of ordinary non-philosophical thinkers. It will therefore be correct that Moore’s argument does not itself reveal why (3) has any warrant whatever — just as it does not reveal why (1) has any warrant whatever. On this understanding of this objection, (3) is in no worse shape than (1).
12 It is noteworthy that Kripke (2011: 189–90) says: “Even if [Nozick’s sensitivity] account is restricted to perceptual (‘noninferential,’ or ‘basic’) knowledge — and I think [it is] much more plausible if so restricted — the red barn problem can still arise.” We encounter the ‘red barn’ problem, and related issues, in Chapter Seven.
13 This is the principle on which Dretske was originally focused which, unlike (Closure), does not involve the performance of a competent deduction.
14 We’ve seen several times that one might consider an effort to ‘carry over’ arguments against (K-Closure) to (J-Closure) by, for example, taking the conclusive reasons or sensitivity condition to characterise justification.