Knowledge First?

Aidan McGlynn

• Publisher: Palgrave Macmillan
• Year: 2014
• ISBN:
  9781137026460
• Topic: Philosophy
• Subtopic: Epistemology in Philosophy

According to a long tradition, questions about the nature of knowledge are to be answered by analyzing it as a species of true belief. In light of the apparent failure of this approach, knowledge first philosophy takes knowledge as the starting point in epistemology. Knowledge First? offers the first overview of this approach.

1

Introduction: Lessons from Gettier

1.1 The analysis of knowledge

Let’s begin with the standard narrative of the history of epistemology. In the Theaetetus and the Meno, Plato discusses the nature of knowledge, and an account seems to emerge and be tentatively endorsed (at least in the latter dialogue). According to that account, knowledge isn’t just true belief, since one might make a lucky guess, but all that we need to add is a justification condition to rule such cases out; to know that P is to have a justified true belief that P. This is commonly known as the justified true belief (JTB) account or the tripartite account of knowledge. Philosophers rested content with the JTB account for almost the next 2500 years, until two-and-a-half pages published in Analysis in 1963 changed everything. There, Edmund Gettier presented two counterexamples to the JTB account which demonstrated that its three conditions are too weak; while they may be individually necessary for knowing, they are not jointly sufficient. This triggered 50 years of ingenious, sophisticated, and often ridiculously complicated attempts to identify the mystery factor X that would result in knowledge when added to justified true belief (or alternatively, when replacing justification). However, these attempts proved to be plagued by further counterexamples, with some eventually arguing that epistemologists were buying into assumptions that made such counterexamples unavoidable (Zagzebski 1994). While a few promising candidate accounts of knowledge have emerged recently, none looks remotely problem-free and none has enjoyed widespread endorsement, with the result that we entered the 21st century with little consensus about how to solve what’s become known as ‘The Gettier Problem’. Attempts to offer analyses of other philosophically interesting concepts such as meaning, goodness, and causation have met with similar fates. Small wonder, then, that some have begun to suspect that the entire project of ‘analysis’ was a wrong turn, and have tried to offer alternative visions of the central concerns and ambitions of philosophy.

This narrative is no doubt crude and misleading in many respects (even if one ignores the implication that epistemologists care about nothing but analysing knowledge). For one thing, it’s far from clear that Socrates or anyone else depicted in the Meno really does endorse the JTB account of knowledge (see, for example, Shope 1983: 12–17). Some have doubted whether the account has really ever enjoyed the orthodox status that the standard narrative accords it. For example, Plantinga writes (1993: 6–7):

Of course there is an interesting historical irony here: it isn’t easy to find many really explicit statements of a JTB analysis of knowledge prior to Gettier. It is almost as if a distinguished critic created a tradition in the very act of destroying it.

Moreover, ‘Gettier’ cases, as they are commonly known, appear to occur much earlier. In Western philosophy, examples with the same essential structure have been attributed to Russell (1912: chapter 13 and 1948: 170–1), while similar examples are said to have been formulated considerably earlier still within the Indian tradition (as reported by Matilal 1986: 136–7). However, it’s not clear that these examples were put forward with the same intent as Gettier’s, even if the examples themselves are very similar.

However, perhaps what’s most misleading about the standard narrative is the idea that there is a unified philosophical activity or project, philosophical or conceptual analysis, which we find in both Plato’s dialogues and throughout the post-Gettier attempts to refine or replace the JTB account. It’s true that much of this tradition involves attempts to state necessary and sufficient conditions for a subject to know that P, and that typically it’s taken to be a constraint on such attempts that the conditions should be in some sense prior to knowledge itself. Beyond this, however, lies mostly fragmentation and unclarity. As the familiar label ‘conceptual analysis’ suggests, some philosophers see the project as that of analysing the concept of knowledge in terms of simpler, antecedently understood concepts. On one particularly demanding version of this conception of analysis, associated with G. E. Moore, the complex, conjunctive concept should have the same meaning as the target concept. It’s this conception of analysis which, in the presence of the assumption that the meaning of concepts that one grasps must be transparent to one, leads to Moore’s paradox of analysis; it seems like any correct analysis of a concept one grasps must be as uninformative as the claim that all bachelors are eligible but unmarried men. More relaxed conceptions of analysis are possible. For example, one might hold that knowledge is justified true belief that meets some further condition, but deny that the concept of knowledge is identical with, or has the same meaning as, the corresponding complex concept; metaphysical complexity need not engender conceptual complexity (see Chapter 8). The problem with the standard narrative is not merely that it fails to specify what analysis is, but rather that it fails to acknowledge that it isn’t even clear that there is any one set of aims that correctly characterizes what knowledge’s would-be analysers took themselves to be trying to accomplish (cf. Ichikawa and Steup 2012).

Still, crude and misleading as it no doubt is, the standard narrative will suffice for our purposes here. This introductory chapter will accomplish three main tasks. First, in Section 1.1, I’ll offer a brief overview of the debate sparked by Gettier’s counterexamples to the JTB account, though I will only touch on a few choice moments from 50 years of vigorous discussion. This will serve to give a flavour of this discussion without getting bogged down in the details, but more importantly it will also introduce theoretical machinery and thought experiments that we will draw on throughout the rest of the book. In a similar spirit, Section 1.2 focuses on so-called lottery propositions as well as examining the safety condition on knowledge that lottery propositions are often thought to motivate. Finally, Section 1.3 lays out the knowledge first approach to epistemology and related topics in the philosophies of mind and language, as it arises as a response to the ‘Gettier Problem’. We will consider a number of respects in which knowledge might be regarded as ‘first’, and we will single out two as being of particular significance. According to the first, knowledge is the ‘unexplained explainer’ (Williamson 2000: 10). Knowledge cannot be analysed or elucidated in terms of ingredients that are traditionally regarded as more basic, such as truth, belief, and justification, but we can use our grip of the distinction between knowledge and ignorance to shed light on other philosophically interesting phenomena (including, perhaps most significantly, justification). The second key respect in which knowledge might be thought primary is that it is to displace belief as the central and most important mental state with mind-to-world fit. Not coincidentally, the chapters that follow this introduction are organized into two principal parts, each part corresponding to one of these two ways in which knowledge might be regarded as first.

1.2 Gettier cases

According to the JTB account, one knows that P if and only if one has a justified true belief that P. The justification in question here is doxastic justification, as contrasted with propositional justification. For a proposition P to be propositionally justified for one is for one to have justification to believe P, and that’s a status P can enjoy whether or not one actually exploits that justification in forming a belief that P. For a belief to be doxastically justified is for it to be properly based on one’s justification for P; what’s required for proper basing is controversial, and we’ll leave it as a placeholder here. The important point is that the JTB account says that to know that P it’s necessary and sufficient to have a true, doxastically justified belief that P.

That the JTB account offers plausible necessary conditions for knowledge is not entirely uncontroversial. The thesis that only truths can be known – the factivity of knowledge – is implicated in several epistemic paradoxes, and some have proposed rejecting it as an escape (see Cook 2013: chapter 6 for an overview). Moreover, non-factive uses of ‘knows’ and its cognates seem to be quite common even amongst fluent speakers of English, and one might suggest that this is evidence that knowledge isn’t factive (or at least, that we should be wary of linguistic arguments in favour of the factivity of knowledge).¹ Radford (1966) offers a well-known attempted counterexample to the entailment from knowledge to belief, and this remains influential in some quarters (for example, Myers-Schulz and Schwitzgebel 2013). Finally, a number of epistemologists have suggested dropping the justification requirement on knowledge.² However, I’m simply going to take the JTB account to identify plausible necessary conditions for knowledge. I won’t attempt to defend that here, since this isn’t going to be at issue in what follows. Let us now turn to Gettier’s argument that these conditions aren’t sufficient.

Gettier cases have the following structure. A subject S has a justified belief that P, and some other proposition Q clearly follows logically from P. S infers Q from P, thereby coming to have a justified belief. As it happens, S’s evidence for P is misleading, and P is in fact false. In contrast, S has no independent evidence for Q, but it is in fact true. S’s belief that Q seems to meet all three conditions imposed by the JTB account; S’s belief is doxastically justified and true. Yet, it doesn’t seem that S knows Q. Gettier offers two examples that fill in this template, but the following modification due to Lehrer (1965: 169–70) is particularly clear and compelling:

Imagine the following. I see two men enter my office whom I know to be Mr. Nogot and Mr. Havit. I have just seen Mr. Nogot depart from a Ford, and he tells me that he has just purchased the car. Indeed, he shows me a certificate that states he owns the Ford. Moreover, Mr. Nogot is a friend of mine whom I know to be honest and reliable. On the basis of this evidence, I would be completely justified in believing

P1: Mr. Nogot, who is in my office, owns a Ford.

I might deduce from this

H: Someone in my office owns a Ford

I would then be completely justified in believing H. However, imagine that, contrary to the evidence, Mr. Nogot has deceived me and that he does not own a Ford. Moreover, imagine that Mr. Havit, the only other man I see in my room, does own a Ford, though I have no evidence that he (or I) owns a Ford.

Here it seems that one has a justified and true belief that someone in one’s office owns a Ford, and yet it also seems that that one doesn’t know this. If that’s right, then it looks like the three conditions offered by the JTB account are not jointly sufficient for knowledge, even if they are individually necessary.³

Gettier doesn’t assume any particular substantive theory of what justification is, though, as he points out (1963: 121), his argument does turn on some assumptions about justification. First, he assumes that justification obeys a closure principle, which we’ll formulate as follows:

If one has a justified belief that P, one knows that Q follows from P, and one competently infers Q from P, then one has a justified belief that Q.

It is a principle like this that allows us to conclude that one’s belief that someone in one’s office owns a Ford is justified when one has competently inferred this conclusion from one’s justified belief that Mr. Nogot, who is in one’s office, owns a Ford. So stated, the principle almost certainly stands in need of refinement, but we won’t concern ourselves with those details here; let’s just assume for now that some suitably refined version of the principle for justification is true and will underwrite Gettier’s argument.⁴ The second assumption Gettier makes is that one can have a justified but false belief that P: that justified belief, unlike knowledge, is not factive. On the face of it, this assumption is very plausible. In Lehrer’s Gettier case, for example, one seems to have extremely good evidence for believing that Mr. Nogot owns a Ford, and one forms the belief on the basis of that evidence. What more could be required for one’s belief to be justified?

One early response to the problem raised by Gettier suggests that something else is indeed needed for one’s belief to be justified, in the sense required for knowledge. According to some epistemologists, to have a justified belief that P, in the relevant sense, is to believe P on the basis of evidence that entails P. From the perspective adopted by such infallibilists, as we’ll call them, the Gettier problem is a symptom of the error involved in adopting a fallibilist account of justification – an account that allows that one might meet the justification condition on knowledge despite basing one’s belief that P on evidence that doesn’t entail P.⁵ This response has not enjoyed much popularity, both because it strikes many as being almost as counterintuitive as the claim that the subject in a Gettier case knows the proposition in question and because it might seem to grant too much to the sceptic who wishes to make a case that we have no justified beliefs or knowledge (for example, Unger 1975). However, there are contemporary descendants of this approach that may fare better, including the thesis, endorsed by several proponents of knowledge first philosophy, that one has a justified belief that P just in case one knows that P. We’ll discuss this account of justified belief in Chapter 3.⁶

Another response starts with the observation that Gettier’s cases (and close variants like Lehrer’s) involve a subject inferring a conclusion from a falsehood. In light of this, one of the earliest responses involved trying to exclude Gettier cases by adding a fourth condition to the three provided by the JTB account, requiring that one’s belief not be inferred from a falsehood (for example, Clark 1963). However, while this condition does well enough when it comes to ruling out cases that share the same structure as Gettier’s original cases, there are a number of other apparent counterexamples to the JTB account which it seems powerless to disarm. Let’s consider two well-known examples to illustrate this point.

In the first, adapted from Russell (1948: 170–1), Ashley looks at her hitherto impeccably reliable clock and sees that it reads two o’clock. She forms the belief that it is two o’clock on this basis. As it happens, Ashley’s clock stopped exactly twelve hours prior, and she just happened to look at it during a minute when trusting it would give her a true belief. Now it’s plausible that Ashley believes, at least implicitly, that her clock is functioning properly, and this belief is false. However, there’s little plausibility in the suggestion that she infers her belief that it’s two o’clock in part from this false belief, and so it looks like the proposed fourth condition is met. Yet, this seems like just as effective a counterexample to the JTB account as Gettier’s own examples. A more radical example also creates trouble for this proposed fourth condition. Suppose that Henry is driving through the countryside, and he stops in front of a barn and comes to believe that the structure in front of him is a barn on the basis of its visual appearance. He’s right; the structure is a barn. However, he’s in barn-façade county, and the barn he happened to stop by is the only genuine barn in a region heavily populated with very convincing façades. Again, Henry seems to have a justified belief that the structure is a barn, and that belief is true, but many are reluctant to credit him with knowledge.⁷

Moreover, a number of epistemologists have recently made a plausible case that one can know P even if one infers P from a falsehood (e.g., Hawthorne 2004: 57fn19; Warfield 2005; and Klein 2008). To give one of Warfield’s examples, suppose that one is about to give a lecture, and one counts the number of people in the audience, reaching a total of 63. From this, one infers the conclusion that the 100 handouts one has printed off will suffice. However, there are only 62 people at the lect...

Table of contents

• Cover
• Title
• 1  Introduction: Lessons from Gettier
• Part I  Knowledge as the Unexplained Explainer
• Part II  Knowledge as a Mental State
• Notes
• Bibliography
• Index