Open Book Publishers logo Open Access logo
  • button
  • button
  • button
GO TO...
Contents
Copyright
book cover
BUY THE BOOK

Introduction*

© 2017 Mark McBride, CC BY 4.0 https://doi.org/10.11647/OBP.0104.01

This book explores two sets of issues in contemporary epistemology. The first part explores issues surrounding the category of basic knowledge (or justification) — that is, at a first pass, knowledge (or justification) which is immediate, in the sense that one’s justification for the known proposition doesn’t rest on any justification for believing other propositions.1 The second part investigates issues surrounding knowledge-closure and various conditions, namely conclusive reasons, sensitivity and safety, which some philosophers have claimed are necessary for knowledge. Each part of the book is substantial (there are five chapters in the first part and four in the second), and the two sets of issues — while evidently of independent interest — are interrelated in several ways (on which more below). The Conclusion, which includes a prospectus for further work, ties the safety condition on knowledge (Chapters Eight and Nine) back to the notion of failure of transmission of epistemic warrant (an absolutely central notion in Part One).

Before getting to the substance of the book, some terminological remarks are in order. Knowledge is taken to be a propositional attitude. Thus our focus is on so-called ‘knowledge-that’, as contrasted with other potential forms of knowledge — most saliently, perhaps, so-called ‘knowledge-how’.2 ‘Justification’ and ‘warrant’, unless otherwise stated, are used interchangeably in this book to refer to that which makes it epistemically appropriate to believe a proposition, and this is commonly referred to as propositional justification or warrant. Thus — using the ‘warrant’ nomenclature — if S has propositional warrant for a proposition, p, we can talk of S having warrant to believe p (or warrant for (believing) p). Also, S has doxastic warrant for p iff S has propositional warrant for p and S believes p on the basis of his propositional warrant for p. If this is the case, one can talk of S having a warranted belief in p (or, perhaps, being warranted in believing p). My choice of which term to use on any particular occasion is determined by the term most commonly used in the debate into which I’m entering. Relatedly, I assume, consistently with there being degrees of belief or credences, that there is a workable notion of binary belief. Unless otherwise stated, when I refer to ‘belief(s)’ in this book, I am making reference to this binary notion.

1. Starting Points

The starting points for our enquiry are G. E. Moore’s (1939) ‘Proof’ of an external world and Crispin Wright’s (1985) discussion of sceptical arguments and, especially, his introduction of the notion of transmission failure. The following argument can be extracted from Moore’s paper:3

(WARRANT FOR 1) I am having a visual experience as of having hands.

(MOORE)

(1) I have hands.

(2) If I have hands an external world exists.

(3) An external world exists.

My interest is not primarily in this argument and its logical properties but in reasoning with the argument. This reasoning involves inference, a mental activity, in contrast with the argument itself, which is an ordered set of propositions. On the distinctions between argument and inference, the logical and the psychological, and for strong views on the disconnect between the distinctions’ relata, see Harman (1986, 2010), who “reject[s] the idea that deductive rules like modus ponens are in any way rules of inference” (2010: 152).

As it is formulated here, (WARRANT FOR 1) is a proposition about Moore’s experience. According to one interpretation of the argument, this proposition is a premise and the transition from this premise to (1) involves inductive inference. According to another interpretation, the visual experience described in (WARRANT FOR 1) makes it epistemically appropriate to believe (1) — and the transition from experience to belief is not one of inference. Whichever way the epistemic support for (1) is understood, the critical question is whether this support is transmitted across the simple modus ponens inference to the anti-sceptical conclusion (3). Wright suggests that the (MOORE) argument does not provide a satisfying response to the sceptic (1985: 437):

Once the hypothesis is seriously entertained that it is as likely as not, for all I know, that there is no material world as ordinarily conceived, my experience will lose all tendency to corroborate the particular propositions about the material world which I normally take to be certain.

Thus, Wright continues (437): “Only if Moore already has grounds for [(3)] does [(WARRANT FOR 1)] tend to support [(1)].”4

Wright’s purpose is not, of course, to promote scepticism; he notes that scepticism could be avoided “if it could be reasonable to accept a group III proposition [e.g. (3)] without reason; that is, without evidence” (1985: 459). In later work (e.g. 2004) he refers to the envisaged unearned warrant as ‘entitlement’.5 If we have an unearned and non-evidential warrant for (3), then (WARRANT FOR 1) does support (1). But, Wright says, it would be a mistake to think that this evidential support is transmitted across the inference from (1) to (3) (1985: 436–37):

It simply is not true that whenever evidence supports a hypothesis, it will also support each proposition which follows from it. The important class of exceptions illustrated are cases where the support offered to the hypothesis is conditional upon its being independently reasonable to accept one in particular of its consequences.

Thus, on Wright’s account, there is failure of transmission of epistemic warrant from premise to conclusion in cases with the structure of the (MOORE) argument. Warrant is not transmitted from the premise (1) to the conclusion (3) because the support for (1), provided by the visual experience as of having hands, is conditional or dependent on its being antecedently and independently reasonable to accept (3).6

Authors writing about warrant transmission and transmission failure often present principles concerning the (non-)transmission of warrant, and such principles will be mentioned at many points in this book. I do not definitively commit to any particular such principles. Throughout, I simply explore these principles, taking them — in particular, the results they mandate — to be answerable to a more general notion of epistemic circularity. Plausibly, it is because the (MOORE) argument exhibits a kind of epistemic circularity that it does not provide a satisfying response to the sceptic. There would be no evident circularity, and no correspondingly obvious explanation for the unsatisfying character of the (MOORE) argument, if our warrant to believe (1) (‘I have hands’) did not rest or depend on a warrant — whether earned or unearned — to believe or assume (3) (‘An external world exists’).

2. Immediate Justification and Basic Knowledge

We shall say that a subject has immediate or basic justification for believing p iff that subject is justified in believing p, and that justification doesn’t rest or depend on any justification for believing other supporting propositions. This account of immediate justification is related to the notion of epistemic antecedence.7 I prefer to leave the notion intuitive, and not to commit to a substantive analysis thereof. However, should I need to commit to a substantive analysis of epistemic antecedence, let me tentatively endorse James Pryor’s (2000: 525): “Your justification [i.e. warrant] for believing p1 is antecedent to your justification [i.e. warrant] for believing p2 just in case your reasons for believing p1 do not presuppose or rest on your reasons for believing p2. Your reasons for believing p1 can not beg the question whether p2.” It should be noted that the analysans contains several interrelating notions — presuppose; rest; reasons — which may themselves call for further analysis.8 I shall make use of a notion of epistemic dependence, related to Pryor’s notion of epistemic antecedence as follows: your warrant to believe p is antecedent to your warrant to believe q just in case your warrant to believe p does not depend on your warrant to believe q. A notion of epistemic antecedence more demanding than Pryor’s would require, in addition, that warrant to believe q does depend on warrant to believe p.

I explore a contrasting notion of temporal antecedence in Chapter Two — a notion which some philosophers (e.g. Zalabardo 2005) seem, upon analysis of their work, to take to be the important notion of antecedence in these debates. Here I offer several quick notes about the relationship between temporal antecedence and epistemic dependence (and the theses I come to put on display will be implicitly utilised in Chapter Two). First:

(TE1) If warrant for p is temporally antecedent to warrant for q, then it’s not the case that warrant for p is epistemically dependent on warrant for q.

If one can have warrant to believe p without (yet) having warrant to believe q, then the warrant to believe p cannot be epistemically dependent on warrant to believe q. One could, though — it would seem — have a warranted belief p without (yet) having a warranted belief q, even though warrant to believe p is epistemically dependent on warrant to believe q. Pryor (2012: 271) describes a position (not his own), according to which warrant to believe p could be epistemically dependent on warrant to believe q, even though warranted belief in p would not need to be based on warranted belief in q (nor even on warrant to believe q).

Second:

(TE2) It is not the case that: If warrant for p is temporally antecedent to warrant for q, then warrant for q is epistemically dependent on warrant for p.

One could have warrant to believe p without (yet) having warrant to believe q, even though warrant to believe q is not epistemically dependent on warrant to believe p. After all, warrant to believe q might be quite unrelated to warrant to believe p. Part of my aim in Chapter Two’s engagement with Jose Zalabardo — when I next return to consider notions of temporal antecedence — is to argue that epistemic antecedence and dependence are the important notion in these debates.

So, returning squarely to the epistemic domain, let us suppose — as an example of immediate or basic justification — that the mere having of a visual experience as of having hands gives one defeasible perceptual justification to believe that one has hands. Pryor introduces the notion of defeasible justification in the following way (2000: 517): “Our perceptual justification for beliefs about our surroundings is always defeasible — there are always possible improvements in our epistemic state which would no longer support those beliefs”. Similarly, defeasible justification is “justification that does not guarantee that our beliefs are correct” (2000: 518).9 If a subject has knowledge on account of having immediate or basic justification then we shall say that the subject has immediate or basic knowledge.

Importantly, focusing on the preceding sentence, it is ‘if’ and not ‘iff’: I want to leave open, and later in the book explore, a different form of knowledge, explored by Stewart Cohen (2002, 2005), aptly called ‘immediate’ or ‘basic’, which, importantly, is not — better: need not be — arrived at on account of having immediate or basic justification. (It may be that the ‘basic’ terminology sits easier with this form of knowledge than the ‘immediate’ terminology.) Cohen (2002: 310) calls knowledge delivered “prior to one’s knowing that the source is reliable” basic knowledge.

How best to understand this Cohenian basic knowledge? Consider (Cohen 2002: 309):

KR: a potential knowledge source K can yield knowledge for S only if S knows that K is reliable.

Allowing for basic knowledge involves rejection of KR. Here is my understanding of such a rejection: it allows for knowledge of a kind that can be yielded for S by a potential knowledge source K even if S does not know K is reliable. It is not saying that only knowledge that is actually yielded for S prior to S’s knowing that its source is reliable can be basic knowledge. Instead, even when S does know that K is reliable, S can still continue to acquire basic knowledge from source K. S can still acquire knowledge that could have been yielded for S by K even if S did not know K is reliable. (Of course, it is also the case that, if S does know that K is reliable, then S can acquire knowledge based on S’s knowledge that K is reliable.) In sum, I understand Cohen’s basic knowledge as a modal notion: a case in which K yields for S knowledge that p and S happens to know that K is reliable may still be a case of basic knowledge in Cohen’s sense if S’s knowledge that p is of a kind that could be yielded for S by K even if S did not know K is reliable.

Assuming I’m right on all this, the pressing question would become: what would it be in virtue of that we could have Cohen’s basic knowledge? A natural thought at this point would be, given Cohen’s focus on knowledge, to appeal to a notion of epistemic antecedence or dependence for knowledge: we could have Cohen’s basic knowledge in virtue of knowing p not being epistemically dependent on knowing K is reliable. Now Cohen doesn’t provide us with such a notion, but we might — at this point — draw on Pryor’s (2000: 525) “exten[sion of] th[e] notion of epistemic priority to knowledge”:

[Y]ou count as knowing p1 antecedently to knowing p2 just in case you know p1 and p2, and the justification on which you base your belief in p1 is antecedent to the justification on which you base your belief in p2.

Now Pryor (2000: 521) himself notes that “connections between justification and knowledge are complicated”, so we must be cautious here. I suggest that a subject S who knows p and knows that the source of that knowledge is reliable (q) can have Cohenian basic knowledge if:

the justification on which S bases his belief in p does not epistemically depend on the justification on which he bases his belief in q.

I also suggest that a subject S who knows p and does not know that the source of that knowledge is reliable (q) can have Cohenian basic knowledge if:

the justification on which S bases his belief in p does not epistemically depend on any justification to believe q.

It seems implausible that a subject S who knows p and knows that the source of that knowledge is reliable (q) can have Cohenian basic knowledge if:

the justification on which S bases his belief in p epistemically depends on the justification on which he bases his belief in q.

And it seems at best unclear that a subject S who knows p and does not know that the source of that knowledge is reliable (q) can have Cohenian basic knowledge if:

the justification on which S bases his belief in p epistemically depends on a justification to believe q.

Now this is only a first pass, but I hope to have gone some way towards connecting epistemic antecedence and dependence for knowledge (Cohen’s focus) with epistemic antecedence and dependence for justification/warrant (my chief focus, and Pryor’s). Moreover, though I have not settled whether or not we can in fact have Cohenian basic knowledge in any particular case, I have given a view of what it is in virtue of which we might have it.

It remains to depict the relationship between this Cohenian basic knowledge and Pryorian basic knowledge. Suppose, then, that some knowledge is basic knowledge in Cohen’s sense. Does it follow that the corresponding justification is an immediate justification in a Pryorian sense? The answer is ‘no’. Cohen (2002: n. 4) asserts: “One could consistently hold that basic knowledge in my sense must be based on other beliefs, provided it need not be based on the belief that the belief source is reliable.” (And we may take it if the knowledge is based on knowledge of, or justification to believe, other propositions, it would still be basic in Cohen’s sense.) Straightforwardly then, this knowledge wouldn’t be Pryorian basic knowledge, which involves justification which isn’t based on any other beliefs (or on knowledge of, or justification to believe, any other propositions). Thus, Cohenian basic knowledge need not be Pryorian basic knowledge.

What about the converse entailment? Is it the case that Pryorian basic knowledge must be Cohenian basic knowledge? This time, the answer is ‘yes’. Cohen (2002: n. 4) asserts: “[A]ny basic knowledge in the traditional foundationalist sense will be basic in my sense as well.” Now while it’s not entirely clear what Cohen means by ‘traditional’,10 this assertion is an important clue as to Cohen’s intentions. Pryorian basic knowledge is arrived at on account of immediate justification — justification not based on any other beliefs. A fortiori, then, it isn’t based on “the belief that the belief source is reliable” (Cohen 2002: n. 4), which is the key to securing Cohenian basic knowledge. Thus, Pryorian basic knowledge must be Cohenian basic knowledge.

Thus, summing up, we have two forms of knowledge which can aptly be referred to as basic knowledge. As a conceptual matter, there is no bar to a piece of knowledge being both Pryorian and Cohenian basic knowledge. But Pryorian basic knowledge and Cohenian basic knowledge are independent of one another: Cohenian basic knowledge need not be Pryorian basic knowledge. They are not, however, both-ways independent of one another: Pryorian basic knowledge must be Cohenian basic knowledge. Pryorian and Cohenian basic knowledge are, then, logically distinct — in the sense of not being extensionally equivalent. These two forms of basic knowledge are at stake as we move through Part One of the book, and are contrasted further in the Interim Review. (I hereinafter take care, where there is a potential for confusion, to explicitly distinguish between these two forms of basic knowledge.)

3. A Pattern of Objection

Having introduced immediate justification and basic knowledge, we can say that there would be no evident epistemic circularity in the (MOORE) argument if our perceptually based knowledge were to be Pryorian basic knowledge; that is, if the justification to believe (1) provided by the visual experience described in (WARRANT FOR 1) were to be immediate — not dependent on a justification to believe any other proposition, such as (3). According to the epistemological position known as dogmatism (Pryor, 2000, 2004), merely having a perceptual experience that represents the world as being a certain way provides immediate, though defeasible, justification to believe that the world is that way.11 So the epistemological dogmatist cannot appeal to epistemic circularity to explain why the (MOORE) argument does not provide a satisfying response to the sceptic.

On the face of it, this is a problem for dogmatism; and the problem generalises. Stewart Cohen (2002, 2005) considers arguments such as the following:

(WARRANT FOR 1) The table looks red.

(TABLE)

(1) The table is red.

(2) If the table is red, then it is not white with red lights shining on it.

(3) The table is not white with red lights shining on it.

Here it might be said, by analogy with what Wright (1985) says about the (MOORE) argument, that it is only if one has antecedent warrant (perhaps unearned warrant) for (3) that the visual experience as of the table being red, described in (WARRANT FOR 1), supports (1). If this is, indeed, the structure of dependence of justification in the (TABLE) argument, then the argument involves a kind of epistemic circularity and provides another example of transmission failure. This seems to fit well with the intuitive view that simply looking at the table and following through the (TABLE) argument is not an adequate way of assuring oneself that one is not being misled by tricky red lights. (Wright, 2003: 60–63, discusses this same example — but with a red wall instead of a red table.)

However, according to dogmatism, merely having a perceptual experience that represents the table as being red provides immediate, though defeasible, justification to believe (1) (‘The table is red’). Specifically, this justification does not depend on an antecedent justification to believe (3) (‘The table is not white with red lights shining on it’). So there is no epistemic circularity and no failure of transmission of warrant.12 This is a problem for dogmatism — the problem of easy knowledge (Cohen, 2002: 313): “It seems very implausible to say that I could in this way come to know that I’m not seeing a white table illuminated by red lights.”13

Cohen (2002, 2005) raises a second problem for views that allow immediate justification and basic knowledge: what I shall, for contrast purposes, refer to as the problem of easy evidence, or bootstrapping. Here, the question is whether a reasoner is in a position to know, by racking up enough pairs of the form <I am having a visual experience as of P, P> (e.g. <I am having a visual experience as of the table being red, The table is red>), that my visual perception is reliable. In this book, I bracket consideration of the problem of easy evidence or bootstrapping. Principally this is because I am presently inclined to follow Wright’s (2007: 43–44) proposal:

A pool of evidence should be regarded as providing inductive confirmation of a hypothesis only if it is reasonable to consider it as drawing upon a representative sample. And that in turn requires that a significant prior probability for the thesis that counterexamples would have shown up in the sample if there are any [sic.]. But the body of ‘confirming’ data compiled by chalking up pairs of the form, <It appears visually to me that P, P>, in the way described has no chance of containing any counterexamples to the contention of the reliability of my visual appearances. So it provides no inductive support for that contention in any case, for purely general methodological reasons quite independent of Dogmatism.14

I would simply add: if, by contrast, the body of confirming data does have a chance of showing up counterexamples to the contention of the reliability of my visual appearances, it can provide inductive support for that contention. Now I do not, for one moment, pretend that this can be the end of the matter (for one thing, there has been a wealth of interesting, recent literature on this issue); it does, though, at least explain my motivation for not devoting extended treatment to the bootstrapping problem in the book.15

The problem of explaining why the (MOORE) argument does not provide a satisfying response to the sceptic about the external world, and the problem of explaining why the (TABLE) argument does not intuitively provide assurance that one is not the victim of a trick of the light, instantiate a pattern of objection to dogmatism, immediate justification, and basic knowledge. And, at a very general level, it is this pattern of objection which forms the core of Part One of the book.

4. The Backdrop of Foundationalism

Before turning to the second part of the book, a matter pertaining to the backdrop against which the book is written — in particular, the first part of the book — must be addressed. There are two reasons for delving into this. First, at a general level it is important to be clear about the scope of the issues addressed head-on in the book. Second, and relatedly, it is important to note a particular epistemological approach falling outside the scope of the book, as the adoption of this approach would lend a very different view of issues central to the first part of the book.

In a nutshell, Part One of the book takes place against the backdrop of a form of foundationalism.16 And it should be noted that on a (much more) coherentist epistemology, many of the issues opened up in this book — e.g. the problem of easy knowledge — may well take on a very different form, or even disappear. (Relatedly, on such an epistemology, it is not clear how there could be instances of epistemic antecedence or dependence between two propositions, of the kind which might underwrite instances of transmission failure.) This is all a little gnomic. To make it less gnomic I need to do three things: first, say a little more about foundationalism and coherentism; second, explain what I mean by ‘backdrop’; and finally, explicate why it is an upshot of a (particular) coherentist approach that issues such as the problem of easy knowledge alter form, or even dissolve. Pursuing these issues — in particular, addressing the upshots of coherentism for issues in this book — will involve analysis of Ernest Sosa’s (2009) position on the problem of easy knowledge.

First, then, what are foundationalism and coherentism? Foundationalism first. Pryor (2001: 100–01) succinctly states the matter thus:

According to the foundationalist, the justification for all our beliefs ultimately traces back to a set of ‘basic beliefs’, which we have immediate justification for using.

(It is worth emphasising that Pryor is using ‘immediate justification’ in the sense delineated earlier in this introductory chapter.) We can take this to be core foundationalism — a thesis to which all foundationalists must be committed. Within the genus foundationalism we can follow Pryor in prising apart two species — traditional (or robust) and modest.

There appear to be two characteristics or hallmarks of traditional foundationalism; one is a claim pertaining to the nature of the basic beliefs, and one is a claim pertaining to the security with which such beliefs are held by the believer. The traditional foundationalist approach to the nature and security of basic beliefs can be summarised thus:

Traditionally, foundationalism thought that these basic beliefs could only concern the nature of one’s current thoughts and experiences. (And perhaps some a priori matters as well.) All our other empirical beliefs, including beliefs about our perceptual environment, had to rest inferentially on this austere foundation of beliefs about our current mental states. These basic beliefs were thought to be exceptionally secure. They were often claimed to be infallible and indubitable. According to many foundationalists, the reason why these beliefs were so secure was that we had a non-propositional ‘direct apprehension’ of the mental states the beliefs were about. (Pryor 2001: 101)

Traditional foundationalism came in for persuasive criticism on all counts; and philosophers of a foundationalist bent have, typically, responded not by rejecting core foundationalism, but by adopting a more modest version of foundationalism than that found in its traditional precursor. This modesty is evident in a weakening of the foregoing ‘security claim’:

Newer forms of foundationalism are quite ‘modest’, in that they allow basic beliefs to be fallible, revisable, less than maximally justified, and so on. They only require that these basic beliefs be immediately justified […] (Pryor 2001: 101)

Additionally, there has been a sea-change in the ‘nature claim’ as held by modest foundationalists: “Some of these new foundationalists allow beliefs about our perceptual environment to qualify as basic” (Pryor 2001: 101). Finally — crucial for our coming contrast with coherentism — to be a modest foundationalist is not to rule out “important epistemic roles for facts about coherence, e.g. they may allow facts about coherence to defeat or to strengthen one’s justification of a belief, even a basic belief.” (Pryor 2001: 101) What, then, makes modest foundationalism a genuine form of foundationalism? It is its commitment to the above-mentioned core foundationalism — its commitment to a set of basic beliefs for which we have immediate justification. It is this modest foundationalism which serves — in a sense to be explained — as the backdrop for Part One of the book (and hereinafter I use the term ‘foundationalism’ to refer to this modest form thereof).

Now to coherentism. It should be evident from the foregoing that our resulting taxonomy is going to be more complex than a simple foundationalism/coherentism divide according to which foundationalism allows no epistemic role for coherence: foundationalism, we have seen, can allow some epistemic role for coherence. Before getting to our ultimate taxonomy, let us, in broad-brush form, set out a working thesis of coherentism:

According to the coherence theory of justification, also known as coherentism, a belief or set of beliefs is justified, or justifiably held, just in case the belief coheres with a set of beliefs, the set forms a coherent system or some variation on these themes […] By a traditional account of coherence we will mean one which construes coherence as a relation of mutual support, consistency or agreement among given data (propositions, beliefs, memories, testimonies etc.) (My emphasis) (Olsson 2003/2012)17

Let us refer to an account of justification of this form as pure or robust coherentism. Its purity or robustness is arrived at by dint of the italicised ‘just in case’ in the above quotation. Finally, let us refer to an account of justification allowing a significant role for coherence — in the sense outlined in the above quotation — but not adopting the ‘just in case’ claim distinctive of pure/robust coherentism as impure or modest coherentism.18 (It is regrettably not, I think, as simple a matter as replacing ‘just in case’ with ‘if’/‘only if’ in the above quotation to arrive at impure or modest coherentism.) And just as modest foundationalism can be supplemented by a role for coherentism, let us, plausibly, assume that impure/modest coherentism can be supplemented by a role for foundationalism.

We are now in a position to state our tripartite taxonomy: (modest) foundationalism; impure/modest coherentism; and pure/robust coherentism. One thing should be readily apparent. Foundationalists can allow and impure/modest coherentists do allow a role for coherence in the epistemic justification of a belief or set of beliefs. The natural next question then becomes: what determines whether an account of epistemic justification that allows a role for coherence in epistemic justification gets to be foundationalist or coherentist? As one might anticipate, there is no clean answer to this question. My working assumption is that the significance of the role ascribed to coherence determines this: the less significant, the more likely we are to have a foundationalist theory; the more significant, the more likely we are to have a coherentist theory.19

Now, with this taxonomy in hand, I can explain more fully what I mean by the ‘backdrop’ against which this book operates. I’ve said the book takes place against the backdrop of a form of foundationalism. We’re now in a position to see two related things. First, given the form of foundationalism in question is modest (and not traditional/robust), this backdrop is not stipulating away any possible role for coherence in a theory of epistemic justification. Second, and relatedly, this backdrop is not stipulating away impure/modest coherentism. As regards ‘backdrop’ itself, what precisely do I mean in this context? I absolutely do not mean that (modest) foundationalism’s truth is presupposed: on the contrary, much of the first part of the book is an exploration — a testing — of its viability. (Indeed, one of the chief interlocutors in Part One of the book is Wright — a philosopher who would not ascribe to foundationalism in the sense delineated here.) Instead I mean something a little more nebulous: the first part of the book takes seriously the viability of (modest) foundationalism (compatibly with which, it also, we have seen, takes seriously a role for coherence in epistemic justification). Limitations of space regrettably mean that something has to give, and a corollary of erecting this foundationalist backdrop to the book is that extended treatment is not given to pure/robust coherentism.20

Beyond the independent importance of making this backdrop clear, why spend so much time on this ostensibly merely terminological matter? Because, as I want to show now, this backdrop has important upshots for the broad shape of problems considered in the first part of the book. In particular, I want to show that if — and, importantly, only if — pure/robust coherentism is adopted, problems such as that of easy knowledge alter their form entirely.21 Put differently: problems such as that of easy knowledge — and putative solutions thereto — take the form as presented in this book against the backdrop of (modest) foundationalism, a backdrop which takes seriously both modest foundationalism and impure/modest coherentism (plausible forms of which assign a role for foundationalism).

One good way into this is by introducing Sosa’s (2009) recently articulated views. More specifically, I will do this by focusing on Sosa’s stance on the problem of easy knowledge — a problem which receives extended scrutiny in Chapters Two and Three of the book. Cohen (2002: 309) introduces this problem by presenting the following proposition, with which we are already familiar:

KR: a potential knowledge source K can yield knowledge for S only if S knows that K is reliable.

Sosa (2009: 211) nicely charts consideration of proposition KR as presenting us with a dilemma: to affirm KR is to “face the problem of vicious circularity”; to deny KR is to face the problem of easy evidence, or bootstrapping (and also the problem of easy knowledge as delineated above). Here, then, is the beginning of Sosa’s (2009: 239–40) response to this dilemma:

The right model for understanding reflective22 justification is not the linear model whereby justification is a sort of liquid that flows through some pipe or channel of reasoning, from premises to conclusion. (Such flow is linear, unidirectional; the pipe or channel “transmits” the justification — or warrant, or epistemic status.) A better model is rather that of the web of belief, whereby the web is properly attached to the environment, whilst its nodes can also gain status through mutual support. Any given node is thus in place through its connections with other nodes, but each of them is itself in place through its connections with the other nodes, including the original given node.

Now there is much to be said about this suggestive passage. First, though — and most importantly for present purposes — by looking at the passage in context we can be fairly sure that, though Sosa is acknowledging an epistemic role for coherence, he is not elaborating a pure coherentist picture here. As Wright (2011: 38) puts it: “So Sosa’s view, although his recourse to the metaphor of the web emphatically commits him to coherence as a source of justification, cannot be such a pure coherentism.”23 It will, at most, be impure/modest coherentism. We can drive this point home by noting Sosa’s remark that the web be “properly attached to the environment”.24 Now, while there is nothing, so far, to commit Sosa’s impure/modest coherentism to supplementation by foundationalism, when we come to piece together Sosa’s position on KR such supplementation is hinted at by Sosa.

Before moving on to Sosa and KR — a topic with which we shall conclude our introduction of Part One of the book — let’s pause to consider a pure coherentist picture with respect to the easy knowledge case and the above-noted dilemma arising from consideration of KR. Wright (2011: 38) eloquently, but suggestively, puts things thus:

Notably, there is no easy knowledge problem for pure coherentism […] The Dilemma is precisely a problem about how to understand the justificational architecture of […] rational beliefs independently of considerations of their systematic integration into a larger system. The ground is cut from under the Dilemma by the pure coherentist’s willingness to disavow that there is any such well-conceived species of […] rationality.25

Now this is all just to vindicate my claim that if we adopt pure coherentism, then problems such as the problem of easy knowledge — and answers thereto — are not going to take the form they take in this book.

It remains for me to vindicate — or, better, take a first step towards vindicating — the corresponding only if claim. Inevitably, I must be more cautious — less confident — about this claim. My strategy will be to show that Sosa’s impure/modest coherentism addresses the problem of easy knowledge in much the form in which it is presented in this book. I will then perform an inference to the best explanation: all (plausible) impure/modest coherentist positions address the problem of easy knowledge in much the form it is presented in in this book.26

Wright (2011: 39) notes that:

Sosa’s tendency seems to be to try to address the Dilemma by, as it were, distributing the distinction in the two kinds of knowledge [animal and reflective] across the horns. So: there has to be some knowledge (the animal) for which KR fails if knowledge is to be possible at all; but there also has to be some knowledge for which KR holds if knowledge is to allow rational scrutiny and organisation under a fully responsible epistemic perspective.27

Yet Wright (39) is correct to point out that, at this stage, we can simply “present the Dilemma as one for reflective knowledge”.28 Thus, so far, we have nothing like the radically different form — dissolution — of the problem of easy knowledge as mandated by adoption of pure coherentism. How, then, does Sosa explain the accomplishment of a transition from mere animal knowledge that, say, the table is red, to reflective knowledge of the same proposition?29

Sosa suggests (2009: 239):

Consider […] one’s justification for a given commitment (or its status as epistemically appropriate): say a commitment that lies behind one’s belief that one sees a red wall [e.g. that in the present instance the appearance of the wall displays the actual colour of the wall] — might one’s reflective rational justification for that commitment gain a boost through one’s now basing it in part (perhaps in some very small part) on the belief that one does see a red wall? How are we to understand such boost in reflective rational justification?

Now this is all somewhat suggestive in nature. Here is one way —pursued by Wright (2011) — of connecting these suggestive comments with Sosa’s impure coherentist account of epistemic justification. It is noteworthy that the foregoing suggestive comments of Sosa’s occur immediately before the above-quoted ‘web-as-contrasted-with-pipeline’ analogy is drawn by Sosa. It is thus natural to take Sosa’s answer to the question of how to understand such ‘boosting’ to be provided (in part) by the ‘web model’. As Wright (39–40) puts it:

[T]he idea seems to be that once a suitable web of beliefs is in place, able to receive and integrate new animal knowledge and take it up into reflective awareness, a modest degree of transmission of warrant—not enough, presumably, to invite the charge of “easy warrant”—from immediately believed (animally known) premises to those of their deductive consequences that articulate presuppositions for the acquisition of that knowledge, becomes possible.

Let us assume this is an adequate interpretation of Sosa. The important point for present purposes is that this impure coherentist account addresses the problem of easy knowledge much as it is presented in this book.30 More specifically, even though this impure coherentist account, on this view, assigns the foregoing ‘boosting’ role to coherence, the KR problem of understanding the ‘justificational architecture’ of beliefs independently of their coherence is still a live one. The impurity of this coherentist account ensures that this is a live question; in contrast with the previously considered pure coherentist account on which it is not live. It remains for me to perform an inference to the best explanation, namely: this is so for all impure coherentist approaches.

I take myself to have done the three things I set out to do in this section, namely: to explicate further foundationalism and coherentism; to clarify what I mean by ‘backdrop’; and finally, to explain why it is an upshot of a (particular) coherentist approach that issues such as the problem of easy knowledge alter form, or even dissolve. (In this final task I have also sketched a view of a sophisticated impure coherentist response to the problem of easy knowledge: Sosa’s).

5. Conditions on Knowledge31

Part Two of the book proceeds to examine three putative conditions on knowledge which have received much scrutiny from epistemologists since their proposals: conclusive reasons (Dretske, 1971: 1):

R is a conclusive reason for P if and only if, given R, ~◊~P (or, alternatively, ~◊ (R.~P));

sensitivity (Nozick, 1981: 172):

[S knows that p only if] if p weren’t true, S wouldn’t believe that p;32

and safety (Sosa, 1999: 378):

Call a belief by S that p “safe” if: S would not believe that p without it being so that p.33

Exploration of these conditions on knowledge is evidently of independent interest (and I say more about this in the Overview of Part Two of the book). For now, I want to introduce two issues of particular interest when considering links between the two parts of the book.

The first issue concerns the conclusive reasons and sensitivity conditions on knowledge — conditions I attempt to render in as plausible a form as possible in Chapters Six and Seven, respectively. These two closely related conditions, while not identical, each license a rejection of knowledge-closure. (The matter is slightly more complicated than this: Essentially, to show a necessary condition on knowledge is not closed, isn’t quite yet to show knowledge itself is not closed.34 But this complication can be bracketed for present purposes.) For example, while one has a conclusive reason for/sensitively believes the table is red, one does not have a conclusive reason for/sensitively believe what is known to be entailed by this proposition, viz. the table is not: white and cleverly lit by red lights. This (controversial) rejection of knowledge-closure — licensed by adoption of conclusive reasons/sensitivity as a necessary condition on knowledge — leads to a straightforward solution to the problem of easy knowledge (assuming rejection of KR):35 we can have basic knowledge that the table is red, without it following that we can have (easy) knowledge that sceptical hypotheses don’t obtain (by simple deduction or inference from said basic knowledge).

The problem of easy knowledge is one problem for dogmatism; a second is presented in Chapter Five. Rejection of knowledge-closure seems to promise a possible solution to that second problem as well — at least, for dogmatism about knowledge. But the second problem also arises for dogmatism about justification, and here it is instead rejection of a corresponding justification-closure principle which promises a possible solution.36

Do, then, my tentative arguments against knowledge-closure in Chapters Six and Seven carry over to constitute arguments against justification-closure? (Interestingly, Wright (2011: 31–32), for example, reads Dretske to be rejecting knowledge- and justification-closure.) To establish that they do carry over I would need to claim that these conditions (conclusive reasons and sensitivity) characterise justification. (‘Characterise’ is a hedging term.) Suppose one claimed that conclusive reasons or sensitivity are necessary conditions on justification. But to show that a necessary condition on justification is not closed, isn’t quite yet to show justification itself is not closed — for the premise may fail to meet other necessary conditions for justification. Still, in any actual case, there might be no reason to suppose that the premise, which meets conclusive reasons or sensitivity, fails to meet any other necessary condition for justification. (And in the Interim Review I consider whether the dialectical context of a defense of dogmatism is one such case.) Absent any such failure, one can reasonably suppose that the premise is justified and that one has an exception to justification-closure. The exception to justification-closure would be explained by the failure of closure for the necessary condition (conclusive reasons or sensitivity), even though failure of closure for the necessary condition does not (for the reason given) logically guarantee failure of justification-closure. (Such is typically the way with explanation.)

Now even amongst defenders of these conditions, it is difficult to discern unqualified endorsement of these ‘carry over’ claims. (Interpretation of Dretske and Nozick on this issue is difficult.) But, while it would be inadvisable for me to push any such claim too far,37 I do believe it is worthy of serious consideration.38

I pick up on, and discuss, these issues further in the Interim Review (and beyond). And this linkage between the two parts of the book should be kept in mind as the conclusive reasons and sensitivity conditions on knowledge are scrutinised in detail in Chapters Six and Seven.39

The second linkage issue connects the safety condition on knowledge — scrutinised in detail in Chapters Eight and Nine — with an absolutely central issue in the first part of the book: failure of transmission of (epistemic) warrant. Martin Smith (2009) has proposed an intriguing account of failure of transmission of knowledge40 in terms of the safety condition. (In particular, Smith (2009: 172) claims “that a failure to transmit safety is a sufficient [but not necessary] condition for a failure to transmit knowledge.”)41 I take it if there is failure of transmission of warrant, there is failure of transmission of knowledge (insofar as warrant is a necessary condition on knowledge);42 but the converse entailment does not hold (insofar as warrant may transmit, but some other necessary condition on knowledge may fail to do so). Thus, there can be failure of transmission of knowledge without failure of transmission of warrant. (If such a possibility obtains, that would mean that warrant would transmit, but not knowledge.) One might then ask: if Smith explains failure of knowledge transmission in terms of safety, how will that help us understand failure of warrant transmission, if the two — failure of knowledge and failure of warrant transmission — can come apart? The answer is that, while there is no entailment from the former to the latter, the former may well be playing an important explanatory role in many instances of the latter. I discuss Smith’s proposal further in the Conclusion. For now it suffices to note that this second linkage between the two parts of the book should be kept in mind while the safety condition on knowledge is scrutinised in detail in Chapters Eight and Nine.

I have aimed to produce a straightforward and uncluttered text, and to present the main lines of argument with maximum clarity. As a consequence, a not insubstantial portion of the words in this book are to be found in footnotes. I hope that an initial reading of the main text of each individual chapter will provide a clear view of its claims and strategy. The footnotes are not, however, dispensable; they provide background and connections, elucidation, caveats, and concessions. Ultimately, my investigation of the viability of allowing for basic knowledge, and into various putative conditions on knowledge, is to be judged on the basis of the full text, and not just the main text, of this book.


* All footnotes are the author’s own unless otherwise stated.

1 Our chief focus in this book is on empirical knowledge (and extensions of it through inference/deduction), though inevitably, at times, consideration of non-empirical knowledge becomes salient.

2 For a good overview of knowledge-how, and its contested relationship with knowledge-that, see Fantl (2012).

3 In this book I do not enter into the exegetical question of how best to interpret Moore (1939).

4 For more on this, see Davies (2004: 215, 217).

5 Such entitlement would seem to be a form of a priori default warrant. For Wright’s latest word on entitlement, see his (2014).

6 For more on this, see Davies (2004: 217–21) and McGlynn (2014). See Moretti and Piazza (2013) for a good survey of transmission issues.

7 The very idea that there could be genuine instances of this to-be-explained notion of epistemic antecedence is seemingly called into question by some philosophers (cf. Chapter Four).

8 A cautionary note about terminology: Pryor (2004) uses the term ‘liberal’ — to contrast with ‘conservative’ — to refer to stances endorsing immediate justification of this kind. (Such stances are liberal in the sense of not requiring — as a conservative stance would — justification for believing the other supporting propositions. And Pryor’s dogmatism is one such liberal stance.) Huemer (2001: ch.5), meanwhile, coined the term ‘phenomenal conservatism’ to refer to (and endorse: see, inter alia, 2006, 2007) a stance of a kind which Pryor would refer to as ‘liberal’. (For a recent collection of essays on dogmatism and phenomenal conservativism, see Tucker (2013a).) I avoid these labels, and thus this potential confusion, in this book.

9 Cf. Williamson (2000: 265–66) on the putative both-ways independence of defeasibility and non-factiveness.

10 Indeed, his introduction of ‘evidentialist foundationalism’ — a Pryorian thesis — muddies the water somewhat.

11 See Wedgwood (2011: n. 4) for a “claim that the process of taking one’s sensory experiences at face value is primitively rational [, which] is similar in spirit to the position that James Pryor (2000) has called ‘dogmatism’.”

12 Pryor (2004, 2012) suggests instead a dialectical failing. (For more exploration of this dialectical route, see Burge (2003) and Markie (2005).) Davies (2009), meanwhile — on whom we focus in Chapter Three — explores both epistemic and dialectical realms in consideration of these arguments. (For discussion of Davies, see Coliva (2010) and Pérez Otero (2013).) For a useful survey article, see Carter (2012).

13 For more on this, see Davies (2004: 236–37).

14 Wright (2011: 36) has since qualified this claim somewhat (in response to conversation with Cohen), and it is an interesting question whether the qualification is necessary.

15 I note that William Alston (1989) called bootstrapping ‘epistemically circular’ reasoning — and Alston’s usage has become popular. While I don’t dissent from this terminology, I use ‘epistemically circular’ in a different sense in this book. Finally, see Cohen (2010) and Wedgwood (2013) for interesting recent remarks on bootstrapping.

16 See Pryor (2001: sec.2) for a good overview of, and extensive list of references for, foundationalism’s development in recent epistemology. I shall be drawing on material from Pryor (2001) in what follows in the main text. See also Fumerton (2000/2010) for a good survey of foundationalism.

17 Olsson (2003/2012) goes on to provide an illuminating analysis of coherentism in much more detail.

18 Wright (2011) also adopts the ‘pure’/‘impure’ coherentism distinction, though he puts a somewhat different gloss on it. (The gloss is not so different as to render impermissible, in this Introduction, quoting Wright when using the distinction.) We will shortly come to see an instance of impure or modest coherentism which Wright (2011) labels ‘frictional coherentism’.

19 Three points: (1) Regarding significance: it is notable that the two examples, reproduced in the main text, which Pryor gives of a role that a modest foundationalist can ascribe to coherence are defeating or strengthening (presumably pre-existent) justification. (2) I assume that the impure/modest coherentist can allow a role for factors which are neither coherentist nor foundationalist (that is, I assume that it is not the case that factors that are not coherentist are ipso facto foundationalist). Nothing crucial hangs on this, however. (3) With all this in hand, we can plausibly describe our taxonomy as exclusive; it will almost certainly not be exhaustive.

20 And so the antonym of ‘takes seriously’ is not ‘takes unseriously’.

21 It is worth pointing out that, just as traditional/robust foundationalism has fallen out of fashion on account of the extremeness of its stance, so too has pure/robust coherentism fallen out of fashion.

22 I assume broad familiarity with Sosa’s (2007, 2009) difficult animal/reflective distinction, though I delve further into this distinction later.

23 Wright (2011) adds, here, in a footnote (n. 14): “And is much the better for that, many would hold. It is pure coherentism that is open to the McDowellian complaint about ‘frictionless spinning in the void’.”

24 This indicates some impurity in the coherentism only on the plausible assumption — which I hereby make — that a pure coherentist is bound to deny that the web of belief (as a whole) needs to be properly attached to the environment (as a whole).

25 I have omitted here Wright’s distinction between access- and management-rationality.

26 Two points: (1) This inference is made more plausible by the sophisticated nature of Sosa’s account of epistemic justification. (2) At n. 19, I acknowledged that our tripartite taxonomy will likely not be exhaustive. In particular, neither Wright nor the sceptic would seem to fall neatly into any of our three categories, but the claims just made about Sosa hold for the sceptic and Wright too. Thus to establish my ‘only if’ claim, I likewise perform a second inference to the best explanation: all (plausible) approaches to epistemic justification falling outside our tripartite taxonomy address the problem of easy knowledge in much the form it is presented in in this book.

27 This is similar to Cohen’s (2002) approach to KR (detailed in Chapter Three) — an approach Cohen acknowledges to be drawing on Sosa’s work.

28 In a rhetorical vein, Wright (2011: 39) continues: “How are we to advance from the animal knowledge that the wall is red to reflective knowledge that it is so? If reflective knowledge requires reflective knowledge of the presuppositions of its acquisition, how is the latter to be accomplished?”

29 Wright (39) pauses to consider an impure coherentist answer, which he labels frictional coherentism, which is — perhaps surprisingly — not pursued by Sosa. In broad outline it is a “pure coherentism restricted to reflective knowledge and warrant”. In essence, on this answer, we would have no epistemic role for coherence at the level of animal knowledge, and then an exclusive epistemic role for coherence in transforming animal knowledge into reflective knowledge. The key point for present purposes is that, insofar as frictional coherentism is not unrestricted pure coherentism, the KR problem of understanding the ‘justificational architecture’ of beliefs independently of their coherence is still a live one.

30 It should be emphasised that Sosa is interested, in the chapter under consideration, in both what we have called the problem of easy knowledge and the problem of easy evidence, or bootstrapping.

31 For a good survey article of recent debate over these conditions, see Comesaña (2007).

32 Recent discussion of the (conclusive reasons and) sensitivity condition as a live option as a necessary condition on knowledge is not vast, but cf. DeRose (1995, 2011), Adams and Clarke (2005), Black and Murphy (2007), Cross (2010). For signs of change, see Becker and Black (2012), though it is to be noted that only a small number of the essays therein end up defending sensitivity. Roush’s recent work (2005, 2010, 2012) on the sensitivity condition bears a special mention here. However, as her account (2012: 244) is one which “uses probability rather than counterfactuals, and in which the sensitivity condition is neither a necessary nor a sufficient condition”, it does not receive extended treatment in this book.

33 Safety, meanwhile, has proven more popular than sensitivity as a necessary condition on knowledge. Indeed, it is its very popularity which demands the inclusion of a chapter in this book focusing on it: Juan Comesaña’s (2005) putative counterexample to safety is the chief focus of Chapter Eight. For some other recent putative counterexamples to safety as a necessary condition on knowledge, see Vogel (1987: 212, 1999: 165), Hawthorne (2004: 4–5), and Hudson (2007). And for an interesting recent discussion of “a safety theoretic conception of knowledge”, see Hawthorne (2007). Finally, Comesaña (2007: n. 21) interestingly notes that “under the assumption that p is truly believed, if p is sensitive then it is safe (under the revised semantics of subjunctive conditionals [gestured at in Chapter Eight, n. 2 of this book]). Therefore, counterexamples to safety as a necessary condition on knowledge are also (further) counterexamples to sensitivity.” But caution is in order: Sosa, on whose account of safety we focus in Chapter Eight, does not always formulate safety in terms of subjunctive conditionals.

34 See Chapter Seven, n. 6

35 The problem of easy knowledge only arises for theorists (like the dogmatist) who reject KR.

36 And I take it that if dogmatism about justification is false, then dogmatism about knowledge is also false. Importantly, the Interim Review makes clear that knowledge-/justification-closure rejection is not the only possible solution for the dogmatist.

37 One reason is that I do not have a particularly substantive account of justification. This book simply takes justification to be that which makes it epistemically appropriate to believe a proposition.

38 Cf. Chapter Six, section 2.6, where I am prepared to talk of “conclusive reasons in some sense justify[ing] belief in propositions for subjects”.

39 For a good survey article on closure principles, including an extensive list of references, see Kvanvig (2006). It is worth noting also that many ‘closure’ principles which we encounter in the book, insofar as they involve deduction or inference, are closely related to ‘transmission’ principles. For subtle distinctions between different versions of closure, and their relationships to problems in this area, see Blome-Tillmann (2006). And for a closure-based proposed solution to the problem of easy knowledge, see Black (2008).

40 Smith takes it to be something of an historical accident that ‘transmission questions’ are formulated in terms of warrant or justification, and not knowledge.

41 It is again possible — though not common amongst participants in these debates — to conceive of the safety condition as characterising warrant or justification. (Smith, for example, does not do so, but cf. section 2.3 of Chapter Eight, where I am prepared to talk of safety playing some form of justificatory role.)

42 And so, if there is not failure of transmission of knowledge, there is not failure of transmission of warrant.