The Gettier Problem

[thegrons]

Is anyone familiar with the Gettier Problem? In short Gettier claims to have found a counter-example to the tripartite theory of knowlege viz. JTB knowledge is a justified true belief.

Here is a Gettier example: A man looks out the window into his pasture and sees a cow about 100 yards out; hence he forms the belief “there is a cow in the pasture”. Unknown to him the cow he was looking at was a stuffed animal, but there really is a cow in the pasture though it was hidden behind the barn. So The man has a true belief, but was it justified?

Another Gettier example would be: A man digs in his pocket and feels some change and forms a belief “I have a dollar of change in my pants”. But when he pulls out the change he only has .90 cents, BUT he had a dime in his back pocket so again he has a true belief BUT was it justified?

These Gettier examples have seemed to perplex philosophers for years because of the issue of justification and the only seemingly possible answer to these issues were somewhat ad hoc. They were ad hoc in the sense the to say that the man was never really justified seems to run against our intuitions.

I think I have the answer that shows that these Gettier problems are not really problems at all and furthermore they turn on a misrepresentation of what propositions one would really form. For example in the cow in the pasture example, would someone really form the proposition “There is a cow in the pasture” or would they really believe “There is a cow 100 yards out in the pasture in my line of site, (occupying the space of the toy animal)”? And the man with the change in the pocket would not form the proposition “I have a dollar in my pants” but rather the object of his belief would be in the change in the front pocket.

So in these two and all Gettier type examples, the believer never had a true belief to begin with. Gettier misrepresents what the prudent man would believe and his attack on the tripartite view of knowledge is really slight of hand, or at best a fallacy of hasty generalization. I believe that Gettier cannot show that the JTB analysis is flawed and to say that the issue of a false belief is not ad hoc.

Any comments?

 

[MarkThompson]

Yes, it is well known that Gettier destroys the standard presentation of the JTB theory of knowledge. You complain that “Gettier misrepresents what the prudent man would believe,” but JTB purports to be a definition of knowledge, to which there must be no counterexamples, not a rough, average-case description. It therefore does not matter that Gettier’s examples are unusual or that most people would not form the specific propositional beliefs at issue.

What is, to me, notable about Gettier-type examples is the frequency with which you find the phrase “by sheer coincidence” in them. I have never done the work to flesh out this hypothesis, but I have long suspected that the Gettier problem could be largely solved by requiring an appropriate causal relationship between the subject matter of the belief and the justification by which it is held.

 

[thegrons]

Thanks Mark but must not a counter example be cogent and applicable? I maintain that Gettier builds a strawman and then attacks it. Certainly if I see a mechanical sheep in my pasture, the object of my faith or belief is the object that occupies the points in space and time of the mechanical sheep and not some sheep in the pasture out of sight; I still maintain that the Gettier examples are straw fallacies and not an appropriate attack on the tripartite view of knowledge.

 

[Taestron]

Gettier wasn’t saying that these beliefs wouldn’t be justified. If we believe something with sufficient evidence, then that belief is justified. For example, we would be justified in thinking that “there is a cow in the pasture” because there is no sufficient reason for us to believe that the stuffed animal was not a live cow. This is important; it means we can have justified but false beliefs.

I’m not sure making explicit the exact belief solves the problem. You said that the real belief in question was “there is a cow 100 yards in front of me.” While this belief is clearly false, I question the possibility of having this belief without also having the belief “there is a cow in that field.” Are these one and the same belief? If not what is there exact connection? Surely you might be able to distinguish them cognitively, but regardless, you have the belief “there is a cow in that field.”

The issues arising out of Gettier problems would be solved if truth was not a criterion for knowledge. Take the cow example. The issue at hand is whether or not I am justified in my belief that there is a cow in the field. Take the real cow out of the equation so that the decoy cow is left. If I have no reason to believe that the cow is a decoy, then I can be justified in my belief “there is a cow in the field.” Now here is the real issue. Whether or not that cow is real or fake does not affect my belief; I will act as though the belief is true even if it is false because I am justified in believing it is true. Since there is no cognitive distinction between a true justified belief and a false justified belief, I don’t think truth should be a criterion for knowledge.

What Gettier problems demonstrate is that there are cases where we have justified beliefs that are only coincidentally true. This weakens the appeal of the traditional JTB formation of knowledge. Personally, I think this is a good thing for epistemology. When discussing theories of knowledge, philosophers overwhelmingly focus on “justification.” You do see discussions of “belief,” but rare are the discussions of “true” (unless they are discussing Gettier). And so, removing the truth criterion will only formalize what epidemiologists already do.

 

[MarkThompson]

Thanks Mark but must not a counter example be cogent and applicable? I maintain that Gettier builds a strawman and then attacks it. Certainly if I see a mechanical sheep in my pasture, the object of my faith or belief is the object that occupies the points in space and time of the mechanical sheep and not some sheep in the pasture out of sight; I still maintain that the Gettier examples are straw fallacies and not an appropriate attack on the tripartite view of knowledge.

The concept of a “straw man” is inapplicable in this kind of philosophy; it is irrelevant that the examples are farfetched and indeed might never have occurred in history. Gettier’s article, as he says in its opening sentence, is a response to the “[v]arious attempts [that] have been made in recent years to state necessary and sufficient conditions for someone’s knowing a given proposition” (emphasis added). A true set of necessary and sufficient conditions, be definition, could not admit of any counterexamples, even “farfetched” ones. If the proposed counterexamples were actually impossible, then they would fail, but these clearly are not.

Similarly, your notion that “the object of my faith or belief is the object that …” misses the mark. The JTB theory is a theory about propositional knowledge (i.e., S knows that P, where P is some proposition), not “objective” (“object-oriented”?) knowledge. You could try to build up an alternative epistemology in which beliefs have physical “objects” in the way you indicate, but that is not what Gettier and his interlocutors (midcentury analytical philosophers) are talking about.

 

[MarkThompson]

Since there is no cognitive distinction between a true justified belief and a false justified belief, I don’t think truth should be a criterion for knowledge.

The problem with that idea is that this whole endeavor has always been an attempt to formalize a definition of “knowledge” that largely dovetails with our ordinary, common-language conception of it. Since people would almost uniformly reject sentences of the sort “He knew P to be true, yet it was actually false” as a contradiction in terms, we have always taken truth as a fundamental criterion in this regard. Sure, we could develop a “philosopher’s version” of the word “knowledge” as a term of art bearing little in common with the ordinary English word, but that is pointless. We might as well just call it “justified belief”; what does it add to say, “and philosophers term this ‘knowledge’ within the context of academic epistemology”?

 

[Perplexity]

I’m not sure every JTB formulation is vulnerable to Gettier, for instance:

S is justified in believing that p iff S believes that p and were S to encounter a defeater to p which she took herself to be unable to adequately fend off, her belief in p would extinguish.

S knows that p if she believes that p, is justified in believing p, and p is true.

 

[MarkThompson]

I’m not sure every JTB formulation is vulnerable to Gettier, for instance:

S is justified in believing that p iff S believes that p and were S to encounter a defeater to p which she took herself to be unable to adequately fend off, her belief in p would extinguish.

S knows that p if she believes that p, is justified in believing p, and p is true.

Well, that is a JTB+ formulation. Once you add a fourth condition, you have moved beyond the realm under consideration in Gettier’s article. In your case, you will want to say that it is the justification that is indefeasible, not p itself, since, as an ex hypothesi truth, p cannot be defeated.

Of course, it is nettlesome to introduce a mandatory, unlimited counterfactual into the equation. The real problem with the no-defeaters condition is that it is too difficult to define and limit the term. For instance, perhaps a completely one-sided harangue of discouraging – but true – remarks about the fallibility of the senses, the shakiness of memory, and the deviousness of con artists would be enough to break S’s confidence in the justification of p. Is that a “defeater”, such that, if so, S really doesn’t know p? Or must we say that the defeater only counts if it is appropriately “fair”, that it must present all the relevant information? If that’s the case, we eventually wind up having to posit a borderline-omniscient S in our counterfactual.

 

[Perplexity]

Well, that is a JTB+ formulation. Once you add a fourth condition, you have moved beyond the realm under consideration in Gettier’s article. In your case, you will want to say that it is the justification that is indefeasible, not p itself, since, as an ex hypothesi truth, p cannot be defeated.

Of course, it is nettlesome to introduce a mandatory, unlimited counterfactual into the equation. The real problem with the no-defeaters condition is that it is too difficult to define and limit the term. For instance, perhaps a completely one-sided harangue of discouraging – but true – remarks about the fallibility of the senses, the shakiness of memory, and the deviousness of con artists would be enough to break S’s confidence in the justification of p. Is that a “defeater”, such that, if so, S really doesn’t know p? Or must we say that the defeater only counts if it is appropriately “fair”, that it must present all the relevant information? If that’s the case, we eventually wind up having to posit a borderline-omniscient S in our counterfactual.

I tried to anticipate objections like these The defeater needn’t be appropriately ‘fair’, for example, unless its being so were relevant to S taking herself to be unable to adequately fend it off. This should ward off fears of quasi-omniscients. What’s important is whether S takes herself to be unable to fend the defeater off.

But, I’m not too invested in this to be honest. (I find things like Bayesian solutions, or Harman’s no false lemma principle more weighty).

Oh, I thought I’d ask about your aversion to counter-factuals given your indicated inclination towards causal solutions?

 

[Qoeleth]

The Gettier problem rests on the ‘justified’ component. Now, the only way justification of belief can be assessed is if it leads to true belief. If it leads to true belief in any particular instance, it is, by definition, justified.

The man who sees the model cow, and then concludes that there is a cow in the field, actually has JTB. His means of knowing, since it lead to a true belief, is, by this, justified.

On the other hand, if there wasn’t a cow, his belief would then be false, and therefore his means of knowing would be unjustified, in that instance.

For this reason, the ‘justified’ aspect seems redundant. Why not go with simply TB?

 

[Taestron]

The problem with that idea is that this whole endeavor has always been an attempt to formalize a definition of “knowledge” that largely dovetails with our ordinary, common-language conception of it. Since people would almost uniformly reject sentences of the sort “He knew P to be true, yet it was actually false” as a contradiction in terms, we have always taken truth as a fundamental criterion in this regard. Sure, we could develop a “philosopher’s version” of the word “knowledge” as a term of art bearing little in common with the ordinary English word, but that is pointless. We might as well just call it “justified belief”; what does it add to say, “and philosophers term this ‘knowledge’ within the context of academic epistemology”?

It is precisely because I want to build a model of practical knowledge that I want to remove the truth criterion. IMO, it is the truth criterion that opens up Gettier problems that makes epistemology more academic than it needs to be. I repeat there is no cognitive difference between a true justified belief and a false justified belief. Honestly, the first time I considered that knowledge needed to be true was when I studied JTB theory. I acted as if my beliefs were true regardless of whether or not they were true; it is impossible to believe something I think is false. I think the popular conception of knowledge is on my side these days. Is there any real epistemological difference between “She knew p to be true, but p was false” and “She knew p to be true, but p was only true for her”? In postmodernism’s hayday (in the humanities), it might be time to focus less on objective versions of knowledge and it explore its more subjective features.

I’m not sure every JTB formulation is vulnerable to Gettier, for instance:

S is justified in believing that p iff S believes that p and were S to encounter a defeater to p which she took herself to be unable to adequately fend off, her belief in p would extinguish.

S knows that p if she believes that p, is justified in believing p, and p is true.

Stop me if I am wrong, but this does not seem to solve Gettier problems. In the original formation of these problems, S was invincibly justified in her beliefs (i.e. she did not encounter a defeater nor had she reason to expect a defeater was present), but her belief was only coincidentally true. Here lies the problem. This formulation is meant to be an objective definition of knowledge, but if there is not link between justification and truth, the objective element starts to crumble. Plus the statement I bolded seems unnecessary in how we know things. It does not matter whether my computer is actually on or not; I believe it is, I am justified that it is, and so I act as if it is true.

The Gettier problem rests on the ‘justified’ component. Now, the only way justification of belief can be assessed is if it leads to true belief. If it leads to true belief in any particular instance, it is, by definition, justified.

The man who sees the model cow, and then concludes that there is a cow in the field, actually has JTB. His means of knowing, since it lead to a true belief, is, by this, justified.

On the other hand, if there wasn’t a cow, his belief would then be false, and therefore his means of knowing would be unjustified, in that instance.

For this reason, the ‘justified’ aspect seems redundant. Why not go with simply TB?

Because I can have a belief in the flying spaghetti monster without any justification for it. If it turns out that the flying spaghetti monster is true without any change in my evidence, can you really say that I have knowledge of the flying spaghetti monster? Or to hit closer to home for most on this forum, you can have all the evidence in the universe for the existence of a god, but it turns out the athiests are correct without any change in the existing evidence. Are you not justified in your belief? I would like to think you are.

I think your conception of justification is contrary to the standard conceptualization. There are accepted(ish) criterion to justification (although some details are fuzzy): Cause and effect, Correlation, Induction, Deduction, etc. If these processes provide evidence for something which outweighs the evidence for the contrary, then we can have justification for a belief. To say someone’s means of knowing is unjustified because the belief is false when there is no other difference seems to destroy the understanding of justified.

 

[MarkThompson]

For this reason, the ‘justified’ aspect seems redundant. Why not go with simply TB?

Because most people are unwilling to apply the label “knowledge” to beliefs acquired in unreasonable ways, such as astrology. Suppose, for instance, that Smith is an average person who is no more likely to develop cancer than anybody else. At a fair, a tarot card reader predicts that he will get cancer. Smith, who is into mysticism, believes her, and coincidentally a year later he does in fact get cancer. Shall we then say, “Yes, Smith had foreknowledge of his cancer”? Most people would not consider the “lucky guess” to qualify as knowledge.

On the other hand, if there wasn’t a cow, his belief would then be false, and therefore his means of knowing would be unjustified, in that instance.

As Taestron remarks, this is an impoverished version of “justification” which renders the criterion essentially meaningless. It would also result in having to say that a person is “justified in believing his horoscope” on those occasions when, randomly, the predictions turn out to be true – something which most of us who do not believe in astrology would never say. At best, you have simply succeeded in divorcing the word “justification” from anything like its ordinary meaning.

 

[MarkThompson]

It is precisely because I want to build a model of practical knowledge that I want to remove the truth criterion. IMO, it is the truth criterion that opens up Gettier problems that makes epistemology more academic than it needs to be. I repeat there is no cognitive difference between a true justified belief and a false justified belief. Honestly, the first time I considered that knowledge needed to be true was when I studied JTB theory. I acted as if my beliefs were true regardless of whether or not they were true; it is impossible to believe something I think is false. I think the popular conception of knowledge is on my side these days. Is there any real epistemological difference between “She knew p to be true, but p was false” and “She knew p to be true, but p was only true for her”? In postmodernism’s hayday (in the humanities), it might be time to focus less on objective versions of knowledge and it explore its more subjective features.

Yes, I suppose that from a sort of solipsistic cognitive point of view there is no difference between a justified belief and a justified true belief. Then again, I don’t see that there is any difference between a justified belief and a (bare) belief, either. In any event, epistemology is classically interested with how the individual relates to the outside world, so it is customary to posit the existence of objective truths and then ask how the mind can have access to them. One can do philosophy that ignores the outside world, but any definition of “knowledge” that is established within that framework will not necessarily bear much resemblance to the commonplace concept of knowledge. If you accept that, then fine; but the results are likely to be of limited applicability.

 

[MarkThompson]

I tried to anticipate objections like these The defeater needn’t be appropriately ‘fair’, for example, unless its being so were relevant to S taking herself to be unable to adequately fend it off. This should ward off fears of quasi-omniscients. What’s important is whether S takes herself to be unable to fend the defeater off.

I confess I’m not sure what that means, then. Could you give an example that illustrates the difference?

Oh, I thought I’d ask about your aversion to counter-factuals given your indicated inclination towards causal solutions?

As I said, I’ve never really put in the effort to think it through, so perhaps you’re right that there is some sort of tension there. It’s just that I’ve noticed that Gettier cases always seem to operate on a “… and just by coincidence” basis, so it strikes me that introducing a criterion that eliminates “justified beliefs that are only coincidentally true” may go a long way to resolving the problem.

 

[hicetnunc]

Any comments?

Very interesting post…thank you!

I think there is different take on all of this that sometimes gets overlooked. Simply put, the problem Gettier tackles is a problem of analytical philosophy, and the solution is likewise a problem of analytical philosophy.

You do not find this problem in Thomism, because, quite frankly, Thomism does not define knowledge as being ‘justified, true belief’ (in fact, neither does Aristotle, and although it is attributed to Plato in the Theaetetus, it is a position that Plato ultimately rejects and many have argued that modern philosophers continue to project their own “problems of epistemology” onto him). It is true that you will find some books treating of the JTB theory in reference to Aquinas (for example, Chisolm and Inwagen), even these are typically efforts to reconcile analytical philosophy with Thomism, but I’m not sure it quite works.

In fact, not only is this a problem of analytical philosophy, it is a problem of epistemology. There was no such thing as “epistemology” in Scholastic and Christian philosophy (the term itself is coined by Ferrier, a 19th century Scottish analytical philosopher) who drove a wedge between “knowing” and “being”.

In Thomism, knowing and being are not distinct in the way they are in analytical philosophy and Gettier’s problems. In Thomism, as in Aristotle, knowledge is a mode of being; when one knows something, one takes on the form, so to speak, of the thing known. Knowledge is thus an ontological relationship (and not just an “epistemological” one) between knower and known. There is, of course, a lot more that could be said about this, but the bottom line is that I think your intuition about Gettier is correct, and that there is no real challenge in his thought-experiments to the possibility of knowledge or of defining it adequately enough to enable us to reason the Truth.

 

[thegrons]

So many responses to comment on??? I think I agree that Gettier attempts to treat epistemology in a deductive sense, of which it is not. The issue of justification is subjective, the issue of “true” or truth begs the question in that we intuitively hold to a correspondence theory of truth which begs the question. In fact all issues of truth and knowledge either beg the question or are subject to an infinite regress and the Problem of the Criteria. Not even the physicist can know “brute facts” but must rely upon her presuppositions regarding time, space, and matter. And this takes her on an infinite regress or circular reasoning.

Gettier takes advantage of a “truncated” proposition that one is likely to form such as “there is a cow in the pasture” instead of “there is a cow 100 yards in front of me in the pasture”. Most of us would truncate this proposition for sake of ease and time. Yet through the doctrine of charity we can correct the truncation and restate the proposition that we all know needs correcting. We all intuitively know that the corrected proposition is what the believer had as the object of her faith.

The tripartite view of knowledge, as well as any other coherence, or foundational views all have their issues; yet my only comment is that Gettier is playing fast and loose with his examples. How many on this post would truly believe that in the Gettier examples; the believer would not have held the proposition if it weren’t for the .90 change or the toy cow? Again the doctrine of charity allows us to repair propositions and even add them when needed. For example the proposition " The Pope knows there are nine planets; therefore the Pope believes there are nine planets". This proposition is not valid; however, the doctrine of charity would allow us to repair the argument and add the premise “If the Pope knows there are nine planets, then the Pope believes there are nine planets”.

I would apply the same repairs to the Gettier examples and thus remove the ostensible dilemma of there being a JTB, when in fact the believer only had a JB. This I believe shows the defect in Gettier.

 

[Qoeleth]

Because most people are unwilling to apply the label “knowledge” to beliefs acquired in unreasonable ways, such as astrology. Suppose, for instance, that Smith is an average person who is no more likely to develop cancer than anybody else. At a fair, a tarot card reader predicts that he will get cancer. Smith, who is into mysticism, believes her, and coincidentally a year later he does in fact get cancer. Shall we then say, “Yes, Smith had foreknowledge of his cancer”? Most people would not consider the “lucky guess” to qualify as knowledge.

As Taestron remarks, this is an impoverished version of “justification” which renders the criterion essentially meaningless. It would also result in having to say that a person is “justified in believing his horoscope” on those occasions when, randomly, the predictions turn out to be true – something which most of us who do not believe in astrology would never say. At best, you have simply succeeded in divorcing the word “justification” from anything like its ordinary meaning.

Imagine the question “is knowledge gained from astrology justified?”. We would normally say no, because it tends to be wrong, or vague. Now, if it happened that astrology produced accurate information 99% of the time, we would then conclude that knowledge gained from astrology was, in fact, justified. Is this not right?

So, how else is the means of gaining information ‘justified’ except on the basis of whether it produces truth?

 

[Qoeleth]

I think your conception of justification is contrary to the standard conceptualization. There are accepted(ish) criterion to justification (although some details are fuzzy): Cause and effect, Correlation, Induction, Deduction, etc. If these processes provide evidence for something which outweighs the evidence for the contrary, then we can have justification for a belief. To say someone’s means of knowing is unjustified because the belief is false when there is no other difference seems to destroy the understanding of justified.

Imagine this. Prof. Smith tells me two things- one about maths, and one about history. Now, what he tells me about maths turns about to be true. So, I would be justified if I believed Prof. Smith about that proposition. But what he tells me about history turns out to be false. So, if I had believed him, I would have been unjustified. If Prof. Smith tells me many propositions, some true and some false, I would be justified in believing the true one. The problems is, unless I know what his area of expertise is, I have no way of telling before hand which I would be justified in believing (i.e. which are true).

However, I concede that an apparently true belief, might actually be false, if the process of whereby it is accounted for are false (assuming the causality is implicit or explicit in the proposition). For example, if a horoscope says, “Because Venus is in line with Jupiter, you will get cancer.” Even if I do get cancer, the proposition (as a whole) is still false, and therefore one would not be justified in believing it.

 

[MarkThompson]

Imagine the question “is knowledge gained from astrology justified?”. We would normally say no, because it tends to be wrong, or vague. Now, if it happened that astrology produced accurate information 99% of the time, we would then conclude that knowledge gained from astrology was, in fact, justified. Is this not right?

This is a completely different conception of justification than the one you mentioned before. In your previous post, you wrote: “If it leads to true belief in any particular instance, it is, by definition, justified.” Sure, if astrology worked 99% of the time we would probably say that beliefs derived from it were justified. (That is, of course, a different hypothetical.) But let’s suppose we have a crystal-ball psychic who is a fraud and whose predictions are accurate only 10% of the time. According to your previous post, beliefs deriving from her predictions are justified on those 1-in-10 occasions when, randomly, they happen to be true. In reality, however, most people would say that consulting a fraudulent psychic never justifies a belief, even when she happens to get lucky.

To give another example, suppose that at a traveling fair they have a “cancer diagnosis” machine, where you grip two electrodes and the machine tells you whether you have cancer or not. In fact, however, the electrodes are nonfunctional and the machine is set up to give the same diagnosis to everybody: “You do not have cancer.” If it were correct that anything that “leads to true belief in any particular instance” were “by definition, justified,” then we would have to conclude that such a machine actually provides justification for people who, as it turns out, don’t have cancer, to believe that they are cancer-free.

So, how else is the means of gaining information ‘justified’ except on the basis of whether it produces truth?

There is a difference between correlating justification with the general reliability of a process in producing true beliefs, and correlating it with the truth of the specific belief in question. The cancer-diagnosis machine, for instance, is not generally reliable, so it cannot justify beliefs even in those “particular instance” when it happens to “produce truth.”

 

[thegrons]

Imagine this. Prof. Smith tells me two things- one about maths, and one about history. Now, what he tells me about maths turns about to be true. So, I would be justified if I believed Prof. Smith about that proposition. But what he tells me about history turns out to be false. So, if I had believed him, I would have been unjustified. If Prof. Smith tells me many propositions, some true and some false, I would be justified in believing the true one. The problems is, unless I know what his area of expertise is, I have no way of telling before hand which I would be justified in believing (i.e. which are true).

However, I concede that an apparently true belief, might actually be false, if the process of whereby it is accounted for are false (assuming the causality is implicit or explicit in the proposition). For example, if a horoscope says, “Because Venus is in line with Jupiter, you will get cancer.” Even if I do get cancer, the proposition (as a whole) is still false, and therefore one would not be justified in believing it.

The fallibility thesis does allow for a little wiggle room; for example, it may have been considered knowledge that the world was flat until it was disproved. Especially in empirical matters, we often have to grant knowledge about certain issues with the understanding that there may be a time in the future that what we once treated as knowledge is no longer knowledge. Ultimately any knowledge can be challenged, even a priori knowledge. I could very easily argue that I don’t accept the “convention” that there are no square circles. I can argue that there is a possible world where A is taller than B and B is taller than A; after all if we really press the issue of grounding our knowledge, every thing ultimately comes back to our worldviews and presuppositions. None of us can know the world intrinsically nor can we see things objectively. I studied epistemology for 3 years only to find out that there are no incorrigible truths, even to us who ground our reason in an infinite God. Logic, science, epistemology, all of it really gets down to rhetoric or the “art of persuasion”.