In 2019, I developed an entirely new version of Introduction to Philosophy. (I retired in 2023, in the wake of the Covid pandemic and the death of my father.) Profound changes in the student population over the preceding years, as well as a shocking increase in the number of students plagiarizing exams and papers (aided by online sources and tools), led me to create a course centered entirely around a series of carefully crafted lectures that I would publish weekly on Blackboard (an online bulletin board) and which students who were enrolled could download. The course focused on a number of core philosophical subjects – Reasoning; Personhood and Identity; Mind and Body; Freedom; Knowledge; Reality; God; and Value and Obligation – and the lectures covered the central questions that arise in these areas, as well as some of the more influential positions on them. And while I made some reference to historical figures, at its heart the course was topical in nature.
The unit on knowledge was divided into three lectures, covering the following topics.
—Knowledge how and knowledge that
—The a priori/a posteriori distinction
—Whether or not knowledge is true, justified belief
—The question of what constitutes justification or warrant
First, we discussed the distinction between “knowledge-how” and “knowledge-that” that Gilbert Ryle makes in his paper, “Knowing How and Knowing That.” Since the rest of the unit would be concerned with the tripartite theory of knowledge and with questions regarding epistemic warrant, I thought it important that students understand that while we would be focusing on knowledge-that, it is neither the only sense in which we know nor the most fundamental. And as students are under the impression that competent performance – knowledge-how – is a function of the possession of some body of propositional knowledge – knowledge-that – it is important to help them see that knowledge-how is in fact prior to knowledge-that: that acquiring propositional knowledge is itself a kind of performance that can be done competently or incompetently, and that good habits of scholarship must be developed before good scholarship can be produced.
Having made it clear to students that for the rest of the unit, we would be concerning ourselves with questions regarding knowledge-that, I moved on to a discussion of the a priori. The idea of empirical knowledge is relatively straightforward, in the sense that there is nothing mysterious per se about the notion that one can come to know things on the basis of what one perceives. Certainly, there are reasons why we might doubt the accuracy of what we see, hear, etc., but unless one is considering the issue from the most abstract of perspectives, these remain in the background and have little effect on our day to day practices. There may be skeptical reasons for doubting the existence of planets and moons, but those tasked with studying our solar system or deeper space are unconcerned with them.
A priori knowledge, on the other hand, is inherently puzzling. On what grounds could one know something, other than some kind of experience? Before we can address this question, however, students have to understand that there is such a thing as a priori knowledge in the first place (at least prima facie), and the way I do it is through a simple arithmetic example. While we may have come to learn that two plus two equals four by counting pebbles or marbles or the like, no amount of counting could possibly confirm or disconfirm this mathematical truth. After all, if you gathered the pebbles and counted five, rather than four, you wouldn’t conclude that you’d discovered a counterexample to the proposition that two plus two equals four, but rather that you’d counted wrong.
The question remains, then, on what grounds we know things like this, and I suggested that we might find answers if we consider some other things that we know a priori, such as statements that describe synonymies, like ‘bachelors are unmarried men’, and logical truths, like ‘bachelors are bachelors’. In these cases, the relevant epistemic grounds lie in knowing the meanings of the words, and I explained to students that some philosophers once thought that mathematical statements might be grounded in the same way; that one might reduce them to statements of pure logic and thereby confirm and disconfirm them in the manner that we confirm and disconfirm logical truths.
I told the students that virtually no one believes this today, but I also observed that even if they did, there would remain a further problem with regard to moral statements, for which it also seems that there can be no empirical justification (I pointed out that it does not follow from any statement regarding what is the case that anything ought – or ought not – to be the case), and which clearly cannot be confirmed merely by examining the meanings of the words out of which they are constructed. So we would still need some account of the grounds on which some a priori statements are justified, even if the logicist program had been a success.
The next thing we discussed was the tripartite theory of knowledge – the idea that knowledge is true, justified belief – and I spent a few moments describing each of the elements individually. I wondered aloud why we might not simply conceive of knowledge as true belief, which led to a short discussion of the sense in which the ascription of knowledge to a person may have an honorific dimension; that we sometimes are reluctant to say people know things that they have arrived at entirely by accident, which foreshadowed a discussion that we would have further along in the unit regarding justification and epistemic virtue/obligation.
In communicating the most famous critique of the tripartite theory of knowledge, I provided students with a simple version of a Gettier-style case that I have developed from an example given by Jonathan Dancy in his Introduction to Contemporary Epistemology (1985):
Imagine that it is 1981 and you have just finished watching John McEnroe defeat Bjorn Borg in the finals at Wimbledon. You call me and tell me that McEnroe beat Borg, and if asked, I would say that you know it, insofar as you (a) believe it; (b) are justified (i.e. have a good reason for) believing it; and (c) it is true.
Suppose, however, that it turns out that the Wimbledon video feed was unavailable this year and that what you had watched was last year’s match. Assume also, for the sake of argument, that McEnroe had beaten Borg last year as well.
What Gettier had done is come up with counterexamples to the tripartite theory of knowledge: cases in which a person might satisfy all three conditions of the theory and yet, still not be appropriately described as knowing the thing in question.
I discussed two strategies for confronting the Gettier cases. The first involves adding an additional “defeasibility” criterion to the tripartite theory – as advanced by Keith Lehrer and Thomas Paxson in their 1969 paper, “Knowledge: Undefeated Justified True Belief” – while the second involves a causal criterion, as proposed by Alvin Goldman in his 1967 paper, “A Causal Theory of Knowing.”
I began by pointing out to the students that in all of the Gettier and Gettier-style cases, there is always something that, had one known it, would have defeated the justification for one’s belief. In the Wimbledon case, if one had known that one was watching last year’s match and one did not otherwise know the outcome of this year’s match, one would not have believed that McEnroe beat Borg this year. The idea, then, is to add a “no defeaters” criterion to the tripartite theory of knowledge, turning it into a quadripartite one:
S knows P if and only if: (1) S believes P; (2) S is justified in believing P; (3) There are no other things, N, that would defeat one’s justification for believing P; (4) P is true.
This solution is ultimately unviable in my estimation, for reasons that Lehrer and Paxson consider in the paper. One may think one has been confronted with a defeater that later turns out not to be one, and one can never really know whether every possible defeater has been considered. I spoke for a few minutes about Lehrer and Paxson’s “Tom Grabit” case which describes a scenario in which one believes that one has seen Tom Grabit steal a book from the local library. The justification for this belief is later defeated, however, upon speaking with Grabit’s mother, who claims that Tom is out of town, but that his identical twin brother was in the library. Later still, one finds out that Tom’s mother is a pathological liar and that Tom has no twin brother, so perhaps one’s belief that Tom stole the book is justified after all. You get the drift.
We then turned to the second strategy for dealing with Gettier cases, in which we conceive of justification causally. One knows P, when one’s belief that P is causally connected to P in the relevant way, and P is true. By way of illustration, I discussed two knowledge-granting modalities that would seem to require such a causal connection: perception and memory. I can only know that I saw my favorite cup on the table, if the cup in question was the actual cause of my perception. If I found out later that in fact, what I had seen was a meticulously created hologram, I would conclude that I hadn’t seen my favorite cup on the table after all. Similarly, I can only remember having taught my class this morning if teaching my class was the actual cause of the memory. If it turns out I’d overslept and dreamt that I’d taught my class, I could no longer correctly say that I remembered it. And If one considers the Gettier cases again, one quickly appreciates that in each case, the relevant causal connection is missing. My belief that McEnroe won Wimbledon this year was not caused by watching this year’s match, but last year’s. The point overall is that according to the causal theory of knowledge, one only knows something if the actual truth-maker is the cause of one’s belief.
I characterized this view as “construing justification causally,” but as I explained to my students, one might argue that it really eliminates justification altogether. The causal theory of knowledge is known as an “externalist” theory, because whether my belief is warranted is a matter of whether it is properly related to the facts and not whether I have a good reason for believing it, and we can imagine cases in which one might be causally connected to the facts in the right way but also be unaware of it. Imagine that one is able to predict the future with 100% accuracy and that as a matter of fact, there is a genuine, scientifically respectable way that your brain does it, of which you are unaware. According to the traditional conception of justification, where being justified means that one has a good reason for believing something, a reason that one knows and can rehearse, one’s belief would not be justified under such circumstances, but if the causal theory of knowledge is correct, then it would seem that one would be.
A consequence of being an externalist with respect to justification is that it would seem to undermine the normative dimension of epistemic warrant, which in turn makes it difficult to make sense of the ideas of epistemic obligation and virtue. I explain to students that this need not be thought of as a deal-breaker, but that they always should consider things carefully, when a view comes along that while attractive in some number of ways, renders it difficult or impossible to continue to make sense of things that one has already accepted and which otherwise seem useful. We rightfully think that people should have good or better grounds for the things they believe, and we rightly praise those who do, and on an externalist picture of justification like this one, it is hard to see how we legitimately do that.
Our final topic was the structure of knowledge, in which I offered sketches of both classical Foundationalism and Coherentism. The foundationalist believes that justification is essentially inferential; that to say P is justified by F is to say that in some way it can be correctly inferred from F. As I explained to the students, this requires one to commit to the idea that there are some number of “basic” beliefs that require no justification themselves and by which all of the rest of the beliefs in our system of knowledge are justified. Otherwise, we wind up with an infinite regress of justifications, which ultimately means that none of our beliefs are justified.
If a belief requires no justification, then it must be self-evident, necessarily true, or some variation thereupon. As I pointed out to the students, we have already encountered statements that have these qualities, namely those expressing synonymy relations and logical truths. Perhaps, epistemically basic beliefs are beliefs like these? The trouble is that basic beliefs are supposed to provide epistemic foundations for everything else that we believe, which would suggest that they must be substantial in terms of their content, and not empty, in the way that synonymies and logical truths are. And this would seem to be a feature, not a bug; that is, it would seem that semantic substance and logical necessity are inversely related which, if one meditates upon it, makes perfect sense.
I described Coherentism as replacing Foundationalism’s linear model of justification with a holistic one. Beliefs are not justified by inference from each other, one to one, but instead, a belief is justified if it “coheres” with all of the other things that we are already justified in believing. By ‘cohere’, we mean that the belief in question is consistent with the others and in some manner, “mutually supporting.”
Coherentism’s main advantage is that it does not face an infinite regress of justifications and thus, need not find a set of beliefs that are both substantial and necessarily true. But it is also at a significant disadvantage, insofar as it severs the ties between justification and truth: complete fictions, after all, can be coherent. Consistency is a purely internal property of a set of statements, and if we were to make enough of the idea of the beliefs in a coherent system being “mutually supporting” so as to reconnect a coherent set of statements with reality, then it will have to be by construing “mutually supporting” in terms of some notion of valid inference, which renders us essentially foundationalists again, saddled with Foundationalism’s problems.
