So we saw in part one that knowledge requires true belief, that is, it requires getting it right. But we also saw that knowledge actually requires a lot more than mere true belief. And the question then became, well what do we need to add to true belief in order to know? Well there's been one very influential account of what it takes to turn true belief into knowledge which has been until quite recently the dominant view in the theory of knowledge. And this is known as the classical account of knowledge. The classical account of knowledge has been defended right back into antiquity. There's a version of this position in Plato for example. Until quite recently, it was just the standard view about the nature of knowledge. Here is what it says. It says knowledge is true belief, that is, knowledge requires truth, it requires belief. And then it also requires a third condition, which is usually called a justification condition. So what is the difference on this view between knowledge and mere true belief, merely getting it right? While the difference is that when one knows, one has a justification for what one believes, where that means one has good reasons that one can offer in support of what one believes. So go back to our example from part one of the juror forming their beliefs through prejudice. As opposed to the juror forming their belief through carefully attending to the evidence that's offered and thinking their way through that evidence and the directions they've been given. In the one case, the case of the person believing through prejudice, although they get it right, or they end up with a true belief, they can't offer good reasons in favor of what they believe. I mean, the best they can say is that they believe what they do because of prejudice. But of course that's a terrible reason for believing a proposition in this case. Because, I mean aside from anything else, prejudice isn't a good way of forming true beliefs about whether or not someone's guilty. In contrast, the juror who carefully attends to the evidence and thinks things through, when asked why they believe what they do, they can offer a justification. They can offer good reasons in support of their belief. So it seems then the difference between knowing as opposed to merely getting it right is the possession of a justification. It's the possession of good reasons in support of why you believe what you do. And think again about those two intuitions that we encountered at the end of part one, the anti-luck intuition and the ability intuition. Well, take the anti-luck intuition first. That says that if you know, then your true belief is not simply a matter of luck. Well, if you've got good reasons in support of you belief, well then it isn't a matter of luck, is it? It's obviously, if you've got good reasons, that means that you got to the truth because you formed your belief in the right kind of way. And similarly, the ability intuition, the claim that if you know, then your true belief is down to your cognitive abilities in some important way. Well, if you can offer good reasons, well, where do those reasons come from, if they weren't derived from you using your cognitive abilities? Just like a juror who knows, uses their cognitive abilities to attend to the evidence and think their way through the evidence in order to form a correct judgement. So the classical account looks very plausible. The difference between knowing as opposed to merely getting it right is that one has a justification. So, the classical account was very dominant for a long time and that's because it just looks so credibly plausible, knowledge is justified true belief. The view is often called a tripartite analysis of knowledge because it has three parts to it. It's very simple, it's truth, belief, justification. That's all that's required to know. Unfortunately, the classical account of knowledge doesn't work. This came to light in the early 60s on account of a very famous paper, a journal article by a man called Edmund Gettier. Now, this paper is the stuff of legend in philosophy. It's just two and a half pages long. But it completely demolishes the classical account of knowledge. What Gettier offered were cases where you have agents who have got true, justified beliefs, so that is they meet the conditions imposed by the classical account of knowledge, and yet they don't know. In particular, what's distinctive of Gettier cases is that the agent concerned doesn't know because it's just a matter of luck that their beliefs are true. So these are what's called Gettier counterexamples. And we're going to look at some of these counterexamples, these cases, right now. Now notice the examples that we're going to look at, they're not the original examples that Gettier offered. His examples are quite complicated. But we don't need to use Gettier's complicated examples to make the point that he wanted to make. There are much simpler examples that can make this point. The first one we're going to look at is the famous Stopped Clock Case. So this was first offered by Bertrand Russell, though at the time he didn't realize it was a Gettier case. He used this example to illustrate a different point. But imagine that someone came downstairs one morning, and they form their belief about what the time is by looking at the clock in their hall. And let's suppose that this clock has been very reliable up until now, and there's no reason to think there's anything wrong with it today. The time it's telling looks plausible, it's roughly correct, and so on. So you know it's roughly what time it is in the morning, and what the clock says pretty much corresponds to that. So you've formed a belief about what the time is by looking at the clock. And now let's stipulate that the belief that you've formed is true, so you've formed a true belief by looking at this clock. Now the belief is clearly justified because you've got good reasons for believing what you do. The clock is a reliable clock. You've got independent reasons for thinking that what the clock says is broadly correct. You've no reason to doubt that the clock is giving you a wrong answer, and so on. So you've gotta justify a true belief. Here's the twist in the tale though. Suppose the clock has stopped. Suppose it stopped 12 hours ago, or 24 hours ago if you're using a 24 hour clock. So what you're looking at right now is a stopped clock. However you just so happen to be looking at the stopped clock in the one time in the day, or two times in the day depending on whether it's a 12 or 24 hour clock, when it's showing you the correct time. Here's the crux of the matter. You can't come to know what the time is by looking at a stopped clock, even if you happen to get a true belief about what the time is, as has happened in this case. And part of the reason for that is that getting a true belief about the time by looking at the stopped clock is just too lucky. If you've got a true belief, it's just a matter of luck that your belief is true. And so, we're offending here against the anti-luck intuition on knowledge, which we saw in part one. So it seems what we've got is justified true belief. So we got a belief that satisfies the condition of the classical account of knowledge. And yet we haven't got knowledge. And in particular we haven't got knowledge because we've got a true belief where it's just a matter of luck that the belief is true. Here's a second Gettier star case, this time due to Roderick Chisholm, who was an American philosopher. Imagine a farmer looking into a field and seeing in clear daylight, and so forth, what looks very much like a sheep in the field. And so on this basis they form the belief that there's a sheep in the field. Now imagine this belief is true. There really is a sheep in the field. So they've got a true belief. And indeed, on any plausible conception of what justification involves, they've got a justified true belief there's a sheep in the field. I mean after all, normally that you can see something in clear daylight which looks like something which you're very familiar with, in this case the sheep, if you're a farmer. That's a good reason that one can offer, a good justification for why one has the belief that one does. So the farmer, it seems, has a justified true belief. Here's the crux though. Imagine we set the case up so that what the farmer's looking at is not a real sheep. What they're looking at is some sheep shaped object. Let's suppose it's a big hairy dog or a sheep shaped rock or something like that, so something that looks like a sheep, but which isn't in fact a sheep. But now imagine there really is a sheep in the field. It's just it's hidden from view behind the thing that the farmer's looking at. So what the farmer's got is a true belief and a justified true belief. But intuitively, he doesn't know there's a sheep in the field because what he's looking at isn't a sheep. And again, the moral of this seems to be that he doesn't know because his belief is just lucky. He's got a true belief that could have very easily been false. And the reason why it could very easily have been false is because what he's looking at isn't a sheep, but just a sheep shaped object. If there hadn't by chance been a sheep hidden from view behind the sheep shaped object he's looking at, then his belief wouldn't have been true. So again, we've got a Gettier case, justified true belief, which doesn't amount to knowledge. And it doesn't amount to knowledge, because even though the belief is true or justified, it's just a matter of luck that that belief is true. Now one way of responding to Gettier cases is to try and criticize the cases themselves. So, for example, in the sheep case you might say, well, is the farmer really believing that there's a sheep in the field as opposed to believing that that thing over there that they're looking at is a sheep? I mean if that were true, then this would make a big difference. Because of course the belief that thing over there is a sheep, well that's a false belief, as opposed to the belief there's a sheep in the field. And in general with Gettier cases, there's usually a bit of fancy footwork that one can do to try and get out of the problem by fiddling about with the case. But it ought to be clear that although this might work for particular kinds of cases, it's not in general going to be a good way in dealing with Gettier cases and with the problem that they pose. The point is that there's a general formula for creating Gettier cases. So insofar as one can find a problem with a particular kind of case, then we just get back to the formula and create a new one. And so if we want to deal with the Gettier problem, we've got to engage with these cases en masse, and not simply deal with them piecemeal, one by one. So what's the formula for creating Gettier cases? Well it basically has two steps. The first step is to have an agent forming beliefs in such a way where their belief would normally be false. So, so think about the farmer forming a belief by looking at a sheep shaped object rather than a sheep. So they're forming a belief whether there was a sheep in the field by looking at something which isn't a sheep. Or think about the agent looking at a stopped clock to find out what the time is. So in both cases you've got them forming their belief in such a way they would normally end up with a false belief. But now we just adapt the case so that as it happens, they end up with a true belief. So in the stopped clock case we just stipulate that they happened to look at the clock at the one time of the day when it's showing the right time. In the sheep case we stipulate that they just happened to be looking at a sheep shaped object which has a genuine sheep hidden from view behind. So with that formula in play we can construct Gettier cases at will. And that means that the Gettier problem is quite a serious problem. It's not a problem that can be dealt with simply by focusing on the details of each particular case. Rather one needs to find a very general way of excluding Gettier cases in one's theory of knowledge. So in the immediate aftermath of Gettier's article, people thought that maybe there'd be a simple solution to this. Maybe, for example, all we need to do is to take the classical account, the tripartite account, three-part account, and just add a fourth part onto it. Add some fourth condition which gets around the Gettier problem. But it soon became clear that it's not as straightforward as it looks. And in fact, it's I think pretty much widely accepted now that there is no straightforward of just simply adding an extra condition onto the classical account to fix the problem caused by Gettier cases. Now, I can't go through all the different proposals that people have made in this regard. So what I want to do is just pick out one proposal which might on the face of it look very plausible and try and give you a sense of some of the problems that this proposal faces. So here's a view that someone called Keith Lehrer has put forward, and various other people have tried to defend versions of this. The thought is that we need to add to the classical account a fourth condition which says that your belief is not based on any false assumptions, or false lemmas. So a lemma is just a, for our purposes we can just think of as an assumption. So this is called the no false lemmas view. So knowledge become justified, true belief, where the true belief is not based on any false lemmas. So no false lemmas account looks plausible on the face of it. You just add this extra condition to knowledge, and thereby you exclude the Gettier cases. The devil, however, as so often in philosophy, is in the detail. In particular we need to be given a principled way of understanding what an assumption is or a lemma is, in this context. And that's actually quite a tricky thing to do. In particular, we don't want to be offered a way of thinking about assumptions, about lemmas, which is so broad that it excludes even genuine cases of knowledge as being knowledge. But equally we don't want it so narrow that it fails to exclude Gettier cases as cases of knowledge. So consider, for example, a very narrow way of thinking about assumptions. We might think that an assumption is something that one actually thinks about as an assumption in forming one's beliefs. So in the stopped clock case, for example, we might imagine that in forming the belief about the time by consulting the clock, our agent actually thinks to himself, I'm assuming here that the clock is working, or something like that. The problem with that is that it seems psychologically implausible. When we form beliefs about the time, we seem, quite properly, to just simply directly form the belief. We don't think to ourselves, start to list in our minds what assumptions we're making in forming the beliefs that we do. And remember, the subject has no reason to think that the clock isn't working. So, they might go through this process if they've got a reason for thinking the clock isn't working. But insofar as that what they're faced with is a working clock, it seems the right thing to do is just forming a belief about the time. So this conception is a narrow conception of what an assumption is. It doesn't seem to do the work that it's meant to do. It doesn't exclude Gettier cases. But now imagine a broad conception of assumption. Suppose we think an assumption is just some false belief that one has, which is germane in some way to the target belief that you're forming in the Gettier case, and which is false. So here we get the right result in the Gettier cases. So it probably is true in the case of the stopped clock, for example, that our subject believes, falsely as it turns out, that the clock is working. So they're not consciously thinking about this at the time, but they do, on some level, believe it. Here's the tricky thing though, if you think of assumptions in this very broad way, then the danger is that lots of genuine case of knowledge are now going to be excluded. After all, of all the many things that we believe at any one time, some of those things that we believe will be false. So who is to say these false things that you believe, which may be very peripheral to the kinds of beliefs that you're forming right now. Who is to say that they in virtue of being false shouldn't deprive you of knowledge of the things you're believing right now? So what is it that says that they're not somehow assumptions in play in some implicit way in your forming your beliefs? So that's the tricky thing. We have to be given some principal way of understanding what an assumption is, such that is generates the right kind of result. And that's actually notoriously difficult to do. So the Gettier problem is much more difficult to resolve than it may at first appear. In particular, it's not amenable to a quick fix. Now there's two sorts of very general questions, issues that are raised by the Gettier problem. The first is whether or not justification is even necessary for knowledge. So the Gettier problem demonstrates that justification isn't sufficient with true belief in knowledge, that you need at the very least something else. But of course one might at this point wonder whether maybe justification is the problem here. Maybe rather than trying to find some extra condition to add to justification, maybe we should rethink what we require of knowledge over and above mere true belief. Maybe we don't need justification at all. Maybe some other kind of condition is what we should add to knowledge. And this is one possibility that epistemologists have explored in the aftermath of Gettier's famous article. And the second problem which might be related to the first is the issue of, well, if the justification condition doesn't eliminate knowledge-undermining luck, right. If the justification by itself can't respond to our intuition about what we call the part one, the anti-luck intuition, it can't explain how when we know we've got true belief that it isn't down to luck. Well then what kind of condition would do that? So it seems we've got a basic demand that we lay down on a theory of knowledge, that we want an account of how it is when we know that our true belief isn't down to luck. Gettier cases show that one can have justified true beliefs that are down to luck. And so the question then becomes, well, what kind of condition must we add to knowledge in order to be confident that we've got cognitive success, true belief that isn't down to luck? And that's quite a puzzle. We thought justification would do that, it doesn't do that. So what kind of condition would do that? Okay, here are conclusions from part two. First thing we looked at was the classical account of knowledge. This is the idea that when one knows one has a true belief, one gets it right. And in addition one has a justification, that is, one can offer good reasons in support of what one believes. We saw this has been a very influential theory of knowledge. But we also saw that it's not tenable on closer inspection. And the reason why its not tenable is down to Gettier cases, or the Gettier problem in general. So Gettier cases are cases of justified true belief where you don't have knowledge. And the reason why you don't have knowledge is there's just a matter of luck that your belief is true in these cases. And then finally we saw that the Gettier problem isn't amenable to an obvious solution. To illustrate this, we looked briefly at the no false lemmas account of knowledge, which tries to simply add a fourth condition onto the theory of knowledge, onto the classical account in order to solve Gettier problems. And we saw that the devil lay in the details there. It was very difficult to get a formulation of this view such that it did the work that it was meant to do. So we're left then with a fundamental problem. And the problem is, what is knowledge? If it's not justified true belief, then, and it seems it isn't, that's the moral of the Gettier cases, well then what is knowledge?