There is a saying that I like: "even a stopped clock is right twice a day". Now this is usually used as a bit of a mean joke when someone gets something right when you might not expect them to. However, we can use this as an example of a 'Gettier Case' (I think it is probably my favourite one). As I alluded to at the end of the last post, Edmund Gettier was a philosopher who demolished the classical definition of knowledge (Justified true belief) with a concise paper consisting of a few examples where there is a belief that is true and justified but is quite clearly not knowledge. Why is this important? Well his observation, demonstrated through his examples, made it quite clear that we don’t really have a great way of describing why one thing would count as knowledge, and another thing would just count as fluke, or a lucky chain of events. This is an issue that can impact significantly on how confident we can be in some of the things that we ‘know’. Let’s delve a bit deeper.
The Problem
Now the publication that started this all off has to have one of the greatest word count to impact ratios of any academic paper. In 1963 Gettier published ‘Is justified true belief knowledge?’, with a concise two and a half pages that quite clearly answered his title question in the negative. He gave two examples that demonstrated the problem with the justified true belief definition, and the epistemology world has been discussing it for the following 50 plus years. Whilst his cases do work, they aren’t my favourite examples. As such, I will use the clock case that I have just hinted at. This has actually been attributed to Bertrand Russell, albeit in a different context, but I think it nicely highlights the problem.
You have just come downstairs in the morning. You look at the clock and note it is 8.50 (we’ll use the example of it being an analogue clock). This is indeed the time. However, the clock actually stopped working at 8.50pm last night and it just so happens that you are looking at it exactly 12 hours later. So did you know the correct time when you looked at the clock? The instinctive response is that of course you didn’t. You are looking at a broken clock, therefore you didn’t actually know the time, even though by chance it was correct. However, the JTB definition would have said that you did. The time was correct (true), you believe it to be true, and looking at a clock is a very reasonable (justified) way to know what time it is. Therefore, you have a justified true belief. I think from this you can get a good idea of what the Gettier cases show. In essence, they show that, with a bit of luck, you can actually end up with numerous examples of justified true belief, that we intuitively feel are clearly not knowledge. Indeed, there is even a formula that we can create to describe what would constitute a Gettier case:
You have just come downstairs in the morning. You look at the clock and note it is 8.50 (we’ll use the example of it being an analogue clock). This is indeed the time. However, the clock actually stopped working at 8.50pm last night and it just so happens that you are looking at it exactly 12 hours later. So did you know the correct time when you looked at the clock? The instinctive response is that of course you didn’t. You are looking at a broken clock, therefore you didn’t actually know the time, even though by chance it was correct. However, the JTB definition would have said that you did. The time was correct (true), you believe it to be true, and looking at a clock is a very reasonable (justified) way to know what time it is. Therefore, you have a justified true belief. I think from this you can get a good idea of what the Gettier cases show. In essence, they show that, with a bit of luck, you can actually end up with numerous examples of justified true belief, that we intuitively feel are clearly not knowledge. Indeed, there is even a formula that we can create to describe what would constitute a Gettier case:
- Take a belief that is justified but would normally be false
- Modify the details of the case so that, by luck, it is actually true
So What?
The non-epistemologists amongst you may be asking the above question. Fair enough. However, I think there is a very important issue that is highlighted within this. Indeed, it comes back to the very title of this series: how do we know? If we can, through luck, end up with a scenario that ends up looking just the same as knowledge, what actually is knowledge? If having a reasonable justification in your true belief isn’t enough, then what is enough? This creates two big opposing problems in my mind.
The first is the problem of skepticism. By working through some of the problems that Gettier Cases pose, we may actually end up finding that we don’t know very much at all. One answer to the Gettier problem is to make our true belief infallible. This is also problematic. I am not planning on going deep down the rabbit hole of skepticism in this series, but infallibility is quite a high bar to clear for something as important to our daily lives as knowledge. Indeed, some forms of radical skepticism (Descartes’ thought experiment is what springs to my mind) would suggest that we can ‘know’ almost nothing. The presence of hallucinations and illusions, the fallibility of our senses, and the innate urge to find patterns where they may not exist, mean we may not actually be well connected to the reality that we think we are. The Matrix might actually be true! If this is hard to believe (which I find that it is), what is harder is where you draw the line between what is possible (we may be in the Matrix if we really think it all through), and what is probable and useful. But that means we have to find some nondescript threshold, somewhere away from infallibility, that defines when we truly know something.
The second point, following on from this first conclusion, is that of luck. If luck can play such a role in how things play out, where does this leave our system of knowledge. Isn’t this just an excellent breeding ground for the retrospective attribution of justification that we see in the different pseudosciences? If my horoscope uses broad enough language, it can be true enough times to make me stop and think. And if justification doesn’t actually help us separate what we know, then the pseudoscientific stories that get retrofitted are as useful to defining knowledge as any others (I am almost certainly overstating the point here though). So to summarise, it is actually a bit of an issue when we want to start to look at the different methods of getting to the truth as there almost seems to be some need to set a threshold for the concept of knowledge to make sense.
The first is the problem of skepticism. By working through some of the problems that Gettier Cases pose, we may actually end up finding that we don’t know very much at all. One answer to the Gettier problem is to make our true belief infallible. This is also problematic. I am not planning on going deep down the rabbit hole of skepticism in this series, but infallibility is quite a high bar to clear for something as important to our daily lives as knowledge. Indeed, some forms of radical skepticism (Descartes’ thought experiment is what springs to my mind) would suggest that we can ‘know’ almost nothing. The presence of hallucinations and illusions, the fallibility of our senses, and the innate urge to find patterns where they may not exist, mean we may not actually be well connected to the reality that we think we are. The Matrix might actually be true! If this is hard to believe (which I find that it is), what is harder is where you draw the line between what is possible (we may be in the Matrix if we really think it all through), and what is probable and useful. But that means we have to find some nondescript threshold, somewhere away from infallibility, that defines when we truly know something.
The second point, following on from this first conclusion, is that of luck. If luck can play such a role in how things play out, where does this leave our system of knowledge. Isn’t this just an excellent breeding ground for the retrospective attribution of justification that we see in the different pseudosciences? If my horoscope uses broad enough language, it can be true enough times to make me stop and think. And if justification doesn’t actually help us separate what we know, then the pseudoscientific stories that get retrofitted are as useful to defining knowledge as any others (I am almost certainly overstating the point here though). So to summarise, it is actually a bit of an issue when we want to start to look at the different methods of getting to the truth as there almost seems to be some need to set a threshold for the concept of knowledge to make sense.
Attempted Solutions
As you may be able to tell, these issues have made epistemologists uneasy. It feels like there should be an answer to this problem, given that we seem to be able to know what we mean when we say know. This has led to a number of efforts to both repair the JTB definition, often by adding some extra component, and efforts to create an entirely new definition. Whilst I won’t go through many of these, there are a number of themes that have been put forward.
A couple of examples include the no-false lemmas and no-defeaters arguments. These appear to be centred around the confidence of the justification process. A lemma is a term for a false assumption, and a defeater is a piece of information that exists (although not known to the believer) that would ‘defeat’ their justification. The proponents of these responses have suggested mitigating against these factors by adding them as a fourth component of JTB, thereby patching it. These seem to perform fairly well in some of the Gettier cases that have been described. However, greater exploration can lead us back to the spiral of problems that we see in skepticism when we have to decide where we want to set our thresholds. What counts as a false assumption? How wide do we have to cast our net to make sure that there are no defeaters? If we think about this broad enough we end up back with the problem that we may not actually know anything with adequate confidence.
The reliabilist theory comes up against similar problems. This theory replaces the concept of justification with the component that the true belief must be formed from a reliable process. This seems like another reasonable definition, but comes apart again with some simple examples-the lottery example probably being the best. Using a reliabilist approach, we would be able to say that we know that we haven’t won the lottery every time that we buy a ticket, because the odds would say that this is a reliable route to a belief (>99.999%). However, it runs against our intuitions so say that we know this even before the numbers have been drawn. This leads us to think that there is something different from pure probability that goes into determining whether we know something.
A couple of examples include the no-false lemmas and no-defeaters arguments. These appear to be centred around the confidence of the justification process. A lemma is a term for a false assumption, and a defeater is a piece of information that exists (although not known to the believer) that would ‘defeat’ their justification. The proponents of these responses have suggested mitigating against these factors by adding them as a fourth component of JTB, thereby patching it. These seem to perform fairly well in some of the Gettier cases that have been described. However, greater exploration can lead us back to the spiral of problems that we see in skepticism when we have to decide where we want to set our thresholds. What counts as a false assumption? How wide do we have to cast our net to make sure that there are no defeaters? If we think about this broad enough we end up back with the problem that we may not actually know anything with adequate confidence.
The reliabilist theory comes up against similar problems. This theory replaces the concept of justification with the component that the true belief must be formed from a reliable process. This seems like another reasonable definition, but comes apart again with some simple examples-the lottery example probably being the best. Using a reliabilist approach, we would be able to say that we know that we haven’t won the lottery every time that we buy a ticket, because the odds would say that this is a reliable route to a belief (>99.999%). However, it runs against our intuitions so say that we know this even before the numbers have been drawn. This leads us to think that there is something different from pure probability that goes into determining whether we know something.
Summary
To summarise, I think the Gettier problem brings into focus the problem of balancing skepticism against luck. If we are too strict on our requirements for certainty then I think Descartes may have been correct when analysing that we would appear to know almost nothing. However, if we are not stringent enough, it appears that luck can begin to play a much more prominent role and hence the existence of Gettier cases (as well as cases of mistaken belief). My intuition from all of this is that, like quite a few other things, knowledge is a spectrum rather than categorical. You can know something to a greater or lesser degree, and indeed this may be reflected in some of the other words that we supplement the verb ‘to know’ with: certain, probable, possible, etc. As I imagine we will explore as we delve into the philosophy of science, much of our scientific method recognises this. We use an iterative and probabilistic approach to much of our knowledge (I am particularly thinking of the famous p value here) that increments in its confidence over time. As such, I think the vagueness of the definitions that we noted above are actually appropriate. We know something if we have a justified true belief with a strong degree of confidence against luck. And this is why I think the scientific method and mindset is so important; we must go looking for this luck to avoid being deceived. Indeed, I am beginning to believe that many of the answers to our epistemological challenges may be found in these anti-deception strategies. However, I guess I can’t yet say that I know this...
Links & References
- Lacewing, M. Philosophy for AS and A level: Epistemology and Moral Philosophy. Routledge. 2017.
- Epistemology. The Basics of Philosophy. https://www.philosophybasics.com/branch_epistemology.html
- Stanford Encyclopedia of Philosophy. Epistemology. 2020. https://plato.stanford.edu/entries/epistemology/
- Wireless Philosophy. Philosophy - Epistemology. Youtube. 2016. https://www.youtube.com/watch?v=r_Y3utIeTPg&list=PLtKNX4SfKpzUxuye9OdaRfL5fbpGa3bH5&index=1
- University of Edinburgh. Introduction to Philosophy. Coursera.
- Hetherington, S. Gettier problems. Internet Encyclopaedia of Philosophy. https://iep.utm.edu/gettier/
RSS Feed
