(●)The Gettier problem asks why a justified true belief can still fail to be knowledge.
+A farmer stands at the edge of his field at dawn and sees, in the middle distance, what looks exactly like one of his sheep. He believes there is a sheep in the field. He is right: there is. But what he is looking at is a large white dog, and the real sheep is lying out of sight behind a rise in the ground. His belief is true, and he had every reason to hold it. Did he know there was a sheep in the field?
Almost everyone says no. He was right, but he was right by luck. Puzzles like his are called Gettier cases, and they matter because of a very old idea about what knowledge is.
- Epistemology, from the Greek for knowledge, is the part of philosophy that studies knowledge and justification.
- An analysis breaks an idea into its conditions, the things that must all be true for it to apply, the way "bachelor" breaks into "unmarried" and "man."
- On the classic analysis, knowledge is justified true belief. You know something when you believe it, it is true, and you are justified in believing it, meaning you have good reason to.
- The idea is old. In the Theaetetus Plato considered whether knowledge is true belief with an account of why it is true, though the dialogue ends without accepting any definition. In the Meno he compared true beliefs without such an account to statues that wander off unless they are tied down. Many twentieth-century philosophers held versions of the analysis, among them A. J. Ayer and Roderick Chisholm, and the farmer's story is adapted from Chisholm's 1966 case.
In 1963 Edmund Gettier, a young philosopher at Wayne State University, published a three-page paper, "Is Justified True Belief Knowledge?", with two cases.
- The ten coins. Smith and Jones have applied for the same job. Smith has strong evidence that Jones will get it (the company president told him so) and that Jones has ten coins in his pocket (he counted them). So Smith believes that the man who will get the job has ten coins in his pocket. In fact Smith gets the job, and, unknown to him, he also has ten coins in his pocket. His belief is true and justified, and it is not knowledge.
- Barcelona. Smith has strong evidence that Jones owns a Ford. From that he reasons to "Either Jones owns a Ford, or Brown is in Barcelona," picking Brown's location at random. Jones does not own a Ford. But Brown, by pure chance, is in Barcelona, and an either-or statement is true when either half of it is. The belief is true and justified, and again it is not knowledge.
Cases like these are counterexamples: a single case that meets every condition of a definition and still is not the thing defined, the way one black swan is enough to sink "all swans are white."
Gettier was not the first to notice the pattern. Bertrand Russell (1948) described a man who looks at a stopped clock he thinks is running, at the very moment it happens to show the right time. In the eighth century, some twelve centuries before Gettier, the Buddhist philosopher Dharmottara told of a traveler who sees a mirage of water and finds real water hidden under a rock where the mirage appeared. But it was Gettier's paper that set off the search for the missing ingredient, and that search has lasted ever since.
(●)Philosophers have proposed many fourth conditions, and externalists have taken the problem as evidence for their view.
+Philosophers responded in the obvious way. If three conditions are not enough, add a fourth condition: one more requirement a true, justified belief must meet before it counts as knowledge. The list of candidates grew long.
- No false premises. A premise is a claim you reason from, the "because" behind a conclusion. Michael Clark (1963) noticed that in both of Gettier's cases Smith reasoned from a false one ("Jones will get the job," "Jones owns a Ford"). Perhaps knowledge is justified true belief reached without relying on anything false.
- No defeaters. Keith Lehrer and Thomas Paxson (1969) proposed that knowledge is justified true belief that no hidden truth would defeat. That is, there is no truth the believer is missing which, if added to what they know, would take away their good reason.
- The right cause. Alvin Goldman (1967) proposed that the fact must cause the belief in the right way. On this proposal, the ten-coins belief fails because what produced it was facts about Jones, not the coins in Smith's own pocket.
- Tracking. Robert Nozick (1981) proposed that the belief must track the truth, so that you would not believe it if it were false.
- Safety. Ernest Sosa (1999) and Duncan Pritchard (2005) proposed that the belief must be safe, so that it could not easily have been false.
- Skill. Sosa and John Greco proposed that the belief must be true because of the believer's skill, as an archer's arrow hits because of her aim and not because of the wind.
- Give up. Timothy Williamson (2000) concluded that knowledge cannot be broken into parts at all. It should be taken as basic: a starting point used to explain other ideas, rather than something explained by them.
Each proposal met new counterexamples of its own. The most famous is fake barn country. Henry drives through a district full of barn facades, flat fronts with nothing behind them, built to look like barns from the road. He happens to look at a real barn and believes, "That is a barn." His belief is true, and his eyes are working well. But he would have believed the same of any of the fakes, and most philosophers say he does not know. Alvin Goldman published the case in 1976 and credited it to Carl Ginet.
Externalism is the view that a belief's standing (whether it is justified, whether it counts as knowledge) can depend on facts outside the believer's mind, such as whether the belief was caused in the right way or produced by a reliable process. Many externalists took the Gettier problem as a reason to adopt their view. Their thought was this: if the missing ingredient is an outside link between belief and fact, then knowledge is not a matter of the believer's reasons after all.
(●)The Gettier problem collapses once two senses of justification are kept apart.
+Timothy and Lydia McGrew answer the externalist reading of Gettier in a 1998 paper whose title states their thesis: "Internalism and the Collapse of the Gettier Problem." It was later reprinted as the first chapter of their 2007 book. They defend internalism, the view that justification depends only on what lies within the believer's mind: experiences, memories, other beliefs, and the reasoning that connects them.
Their whole answer turns on noticing that "justified" has two senses.
- In the first sense, which they call legitimation, a belief is justified when it is reasonable given what the believer has to go on. Legitimation does not require that the believer's premises be true. Smith's belief in the ten-coins case is legitimate, since anyone in his place would have believed the same.
- In the second sense, which they call justification proper (strict justification, for short), a belief reached by reasoning is justified only if the premises it rests on are true. Smith's belief is not justified in this sense, because it rests on "Jones will get the job," and that is false.
- Internalism, on this picture, is a claim about the first sense. Whether a believer is reasonable depends only on what lies within her mind.
- The second sense reaches outside the mind for one thing only, the truth of the premises. That is the same kind of fact the classic analysis already required of the belief itself, which had to be true.
Their thesis is that every Gettier case is a case of legitimation without justification proper. Somewhere, openly or silently, the believer always relies on a false premise. They build on what they call the Russellian solution, after Bertrand Russell, who had argued as early as 1912 that a true belief drawn from a false premise is not knowledge. On this view the classic analysis survives, provided "justified" is read in the stricter sense. And the move to externalism, they conclude, is neither useful nor necessary.
The argument for that view is deductive. Its premises work like a locked chain: if each link holds, the conclusion cannot slip out. That does not mean each premise must be certain beyond all doubt. It only has to be more plausible than its denial.
(P1)Every Gettier case relies on a false premise.
+Look closely at any Gettier case and you find the believer leaning on something false. That falsehood is what makes the truth a matter of luck.
(1) A belief relies on any premise its support passes through.
- To say a belief relies on a premise is to say that the believer's evidence supports the belief only by way of that premise.
- She need not ever put the premise into words. Most of what any of us leans on, we never say.
- Think of a woman at a bus stop who expects the 8:10 because the timetable says so. She never tells herself "this timetable is current," but tell her it is last year's and her confidence goes at once.
- That gives a simple test. A belief relies on a premise when learning that the premise is false would take the support away.
(2) The two original Gettier cases rely on a false premise openly.
- In the ten-coins case Smith reasons from "Jones will get the job." In the Barcelona case he reasons from "Jones owns a Ford." Both are false.
- Take the false premise away, and the luck goes with it. Suppose Smith had the same kind of evidence about himself instead: the president tells him he will get the job, and he has counted the coins in his own pocket. Then his belief that the man who will get the job has ten coins would be well grounded, and there would be nothing lucky about it.
(3) The other classic cases rely on one silently.
- The farmer believes there is a sheep in the field because "that animal is a sheep." The animal is a dog.
- The man at the stopped clock believes the time because "this clock is working." It has stopped.
- The traveler at the mirage believes there is water because "what I see over there is water." It is a mirage.
- None of them says the premise aloud. Each one relies on it all the same, and in every case it is false.
(4) Luck enters exactly where the falsehood does.
- What makes these beliefs lucky is a mismatch. The believer's reasons point to one fact, and the belief comes out true because of a different one. The ten-coins belief is true because of Smith's pocket, but every reason he had was about Jones's.
- Think of a letter addressed to the wrong house that still reaches the woman it was meant for, because by chance she is visiting that house the day it arrives. It found her, but nobody would say the address was right.
(5) The harder cases hide the false premise instead of removing it.
- Philosophers have constructed cases meant to contain no false premise at all: fake barn country (D1), and a case about a Ford in which the reasoner never believes the false claim (D2).
- Each one still relies on a false assumption about the believer's situation, such as "the barn-like things around here are barns." Henry never says it, any more than you say "this chair will hold me" before you sit down, but his belief leans on it all the same. Tell him the county is full of fakes, and his confidence goes at once.
- Whether every such case yields to this reply is the most contested part of the view (D1, D2, D11).
Therefore, every Gettier case relies on a false premise, openly or silently, and the false premise is the source of the luck.
(P2)A belief that relies on a false premise is not justified in the sense knowledge requires.
+The word "justified" has two senses. Knowledge needs the stricter one.
(1) We already use the word in two senses.
- Say a jury convicts a man on strong evidence, and the evidence turns out to have been planted. We say the jurors were reasonable. They did nothing wrong. We also say the conviction was not justified, since it rested on falsehoods.
- The first verdict is about the jurors' reasoning, which is what legitimation means. The second is about whether the case was actually made, which is justification proper. Both ways of speaking come naturally, and they come apart exactly when the evidence is false.
(2) Knowledge needs the case actually made.
- To know something is to have the truth in hand for a reason that connects to it. A chain of reasoning with a false link does not connect, however carefully it was forged.
- Think of a rope bridge with one rotten plank in the middle. A careful hiker, stepping lightly, may get across safely by luck. That does not make the bridge sound.
- Only a premise the support actually runs through can do this damage. A false belief sitting idle beside the reasoning leaves the support intact, like a miscount by one when any count near fifty would do (D7).
(3) The stricter sense adds no outside condition except truth.
- Justification proper asks one thing more than legitimation does: are the premises the belief rests on true? Truth is a fact about the world, of course. But it is the same outside fact the classic analysis already required of the belief itself.
- What it does not ask about is how the belief was caused, or whether some process in the believer's brain is reliable. Those are the outside links externalists add (D4).
- Think of an accountant checking a ledger. She asks whether each figure is correct and whether the totals follow from them. She does not ask which clerk typed the entries or how reliable the office printer is.
(4) The looser sense keeps everything internalists care about.
- Smith was reasonable, and so was the farmer looking at the dog. Each believed what his evidence supported, and internalism praises them for it.
- What they lacked was not good reasoning but true starting points. That is why they fell short of knowledge.
Therefore, a belief that relies on a false premise may be reasonable, but it is not justified in the strict sense that knowledge requires.
(C)Therefore, Gettier cases are not justified true beliefs in the sense knowledge requires.
+Gettier's cases are not counterexamples to the analysis of knowledge as justified true belief, once "justified" is read in the stricter sense. Every Gettier case relies on a false premise (P1). A belief that relies on a false premise lacks the stricter justification that knowledge needs (P2). So no Gettier case meets the justification condition in the sense that matters.
That pattern of reasoning is a syllogism, an argument with two premises and a conclusion. This one has a valid form, which means that if the premises are true, the conclusion must be true. Its skeleton is: all A are B; all B are C; so all A are C. Filled in, it reads like this. Every Gettier case is a case of relying on a false premise; every case of relying on a false premise lacks strict justification; so every Gettier case lacks strict justification.
(1) The classic analysis survives Gettier's cases.
- Gettier's cases count against justified true belief only when "justified" means legitimation. Read it as justification proper, and none of them meets the justification condition, so none is a counterexample.
- You might wonder where the demand for true reasons stops, since every reason seems to rest on reasons of its own. Followed all the way down, the true reasons end in a foundation: a belief the believer does not reason to at all, because she has it directly, such as her awareness of a white, woolly shape in her field of vision right now.
- Could some other case meet all three conditions in the strict sense and still fall short of knowledge? That is a further question, and the strongest attempt is a recipe for building one (D3).
- The classic analysis needed sharpening, not replacing.
(2) Externalism is not needed to solve the problem.
- The Gettier cases have been taken as evidence that knowledge needs an outside link between belief and fact. On this account it needs only true reasons, and truth is the one outside fact every account of knowledge already requires.
- That matters, because the externalist links come with costs of their own (D4).
(3) The answer leaves the believer able to check.
- Suppose a person wonders whether her belief has been "Gettiered," caught in a Gettier case without her realizing it. She does not need to inspect hidden causes or the reliability of her brain.
- She needs to ask whether her premises are likely to be true, and that is a question she can investigate (D8). The farmer, had he wondered, could have walked over to the animal and looked.
See also:
Philosophy / Epistemology: Justified True Belief (JTB)Philosophy / Epistemology: InternalismPhilosophy / Epistemology: Externalism
On the Gettier problem and the analysis of knowledge:
Ayer, A. J. The Problem of Knowledge. London: Macmillan, 1956.
Chisholm, Roderick M. Perceiving: A Philosophical Study. Ithaca, NY: Cornell University Press, 1957.
Chisholm, Roderick M. Theory of Knowledge. Englewood Cliffs, NJ: Prentice-Hall, 1966.
Clark, Michael. "Knowledge and Grounds: A Comment on Mr. Gettier's Paper." Analysis 24 (1963): 46-48.
Conee, Earl, and Richard Feldman. "The Generality Problem for Reliabilism." Philosophical Studies 89 (1998): 1-29.
Gettier, Edmund L. "Is Justified True Belief Knowledge?" Analysis 23 (1963): 121-123.
Klein, Peter D. "Useful False Beliefs." In Epistemology: New Essays, edited by Quentin Smith. Oxford: Oxford University Press, 2008.
Lehrer, Keith, and Thomas Paxson Jr. "Knowledge: Undefeated Justified True Belief." Journal of Philosophy 66 (1969): 225-237.
McGrew, Timothy, and Lydia McGrew. "Internalism and the Collapse of the Gettier Problem." Journal of Philosophical Research 23 (1998): 239-256.
McGrew, Timothy, and Lydia McGrew. Internalism and Epistemology: The Architecture of Reason. London: Routledge, 2007.
Merricks, Trenton. "Warrant Entails Truth." Philosophy and Phenomenological Research 55 (1995): 841-855.
Nagel, Jennifer, Valerie San Juan, and Raymond A. Mar. "Lay Denial of Knowledge for Justified True Beliefs." Cognition 129 (2013): 652-661.
Russell, Bertrand. The Problems of Philosophy. London: Williams and Norgate, 1912.
Russell, Bertrand. Human Knowledge: Its Scope and Limits. London: George Allen & Unwin, 1948.
Shope, Robert K. The Analysis of Knowing: A Decade of Research. Princeton, NJ: Princeton University Press, 1983.
Critics and rival views engaged:
Feldman, Richard. "An Alleged Defect in Gettier Counter-Examples." Australasian Journal of Philosophy 52 (1974): 68-69.
Goldman, Alvin I. "A Causal Theory of Knowing." Journal of Philosophy 64 (1967): 357-372.
Goldman, Alvin I. "Discrimination and Perceptual Knowledge." Journal of Philosophy 73 (1976): 771-791.
Greco, John. Achieving Knowledge: A Virtue-Theoretic Account of Epistemic Normativity. Cambridge: Cambridge University Press, 2010.
Lehrer, Keith. "Knowledge, Truth and Evidence." Analysis 25 (1965): 168-175.
Machery, Edouard, et al. "Gettier Across Cultures." Noûs 51 (2017): 645-664.
Nozick, Robert. Philosophical Explanations. Cambridge, MA: Harvard University Press, 1981.
Pritchard, Duncan. Epistemic Luck. Oxford: Clarendon Press, 2005.
Skyrms, Brian. "The Explication of 'X Knows That p'." Journal of Philosophy 64 (1967): 373-389.
Sosa, Ernest. "How to Defeat Opposition to Moore." Philosophical Perspectives 13 (1999): 141-153.
Sosa, Ernest. A Virtue Epistemology: Apt Belief and Reflective Knowledge, Volume I. Oxford: Clarendon Press, 2007.
Starmans, Christina, and Ori Friedman. "The Folk Conception of Knowledge." Cognition 124 (2012): 272-283.
Warfield, Ted A. "Knowledge from Falsehood." Philosophical Perspectives 19 (2005): 405-416.
Weinberg, Jonathan M., Shaun Nichols, and Stephen Stich. "Normativity and Epistemic Intuitions." Philosophical Topics 29 (2001): 429-460.
Williamson, Timothy. Knowledge and Its Limits. Oxford: Oxford University Press, 2000.
Zagzebski, Linda. "The Inescapability of Gettier Problems." Philosophical Quarterly 44 (1994): 65-73.
Zagzebski, Linda Trinkaus. Virtues of the Mind: An Inquiry into the Nature of Virtue and the Ethical Foundations of Knowledge. Cambridge: Cambridge University Press, 1996.
Historical sources:
Plato. Meno. 97d-98a.
Plato. Theaetetus. 201c-210b.
Biblical citations:
Luke 1:2; Luke 1:4. Scripture quotations are from the English Standard Version.
+Fake barn country has no false premise. Henry simply sees a real barn with working eyes, yet he does not know, so the Russellian solution fails.
1. Fake barn country is the hardest case for the false-premise view.
The case was built to remove the false premise. Henry does not reason from anything; he looks, and he sees a barn. If he has no false premise and still does not know, then something other than a false premise is missing. The most natural candidate is an outside fact: the fakes nearby.
2. Henry's belief rests on a silent assumption about where he is.
A look at a barn front supports "that is a barn" only because, in the world as Henry takes it to be, things that look like barns from the road are barns. That assumption is part of what makes the look good evidence. In fake barn country, it is false.
- Think of a woman who reaches into a bowl of what she takes to be ordinary candy. A prankster has filled most of the bowl with soap shaped like candy, and she happens to pick a real piece. Her belief "this is candy" rested on "the things in this bowl are candy," and that was false.
- The assumption also explains something else. Henry and a driver in ordinary farm country can have the same experience and the same good reasons, so that both are equally reasonable, and yet only the second one knows. Their legitimation is identical. What differs is that the assumption they both rely on is true for one of them and false for the other.
3. A belief about the world is inferred, not simply seen.Classical foundationalism is the view that knowledge rests on foundations that are certain, such as your awareness of your own present experience. On that picture, what Henry has with certainty is his experience of a barn-like shape. "That is a barn" is something he reaches from that experience, as the best explanation of it, and that reasoning relies on what he takes his surroundings to be like. When his picture of his surroundings is badly wrong, the reasoning has a false link.
- Think of hearing a knock and concluding that someone is at the door. The sound is what you have directly. "A visitor" is the best explanation of it, given what you assume about your house. But if a loose shutter has been banging all afternoon, the assumption is false.
4. Not everyone accepts the reply.
Critics answer in two ways. Some say the silent assumption is invented after the fact, since Henry never believed anything about the district (D11). Others deny that seeing a barn involves any inference at all.
The test in (P1) answers both. A belief relies on an assumption when learning it was false would take the support away. Tell Henry the county is full of fakes, and his confidence that he is looking at a barn collapses at once. It was leaning on the assumption all along, the way anyone who stands up is leaning on "the floor will hold" without ever thinking it.
See also:
Philosophy / Epistemology: Classical Foundationalism
+Some Gettier cases avoid false premises entirely. A careful reasoner can infer straight from her evidence to "someone in the office owns a Ford" without ever believing the false claim, and still be Gettiered.
1. The case is a real challenge.
Keith Lehrer (1965) devised the original. A colleague named Nogot gives Smith excellent evidence that he owns a Ford, and Smith infers that someone in the office owns a Ford. That is true, but only because another colleague, Havit, owns one.
Richard Feldman (1974) pressed a sharper version. In it, Smith never believes "Nogot owns a Ford" at all. He reasons straight from true statements about his evidence (say, Nogot's papers and Nogot's word) to "someone in the office owns a Ford." Every premise he actually uses is true.
2. The evidence supports the conclusion only through the false claim.
Nogot's papers and Nogot's word support "someone in the office owns a Ford" only because they support "Nogot owns a Ford." That is the only route by which they bear on the conclusion at all. Whether or not Smith says it to himself, the support runs through a falsehood.
- Think of a train that passes through a station without stopping. It still went through that station. And if the station was closed by a flood, the train did not get through.
3. Withholding belief does not change what the evidence supports.
Strict justification asks whether the premises a belief's support depends on are true, whether or not the believer has put them into words or formed a belief about them (P1). Smith's evidence is all about Nogot, so it makes "someone in the office owns a Ford" probable only to the extent that it makes "Nogot owns a Ford" probable.
Believing the first while declining to believe the second does not open a new route to the truth. All Smith has done is decline to name the route he is using. On that reading the careful reasoner still relies on a false premise, and the case is legitimation without justification proper.
+Linda Zagzebski's recipe: any account on which justification does not guarantee truth can be Gettiered. Take a justified false belief, add a stroke of luck that makes it true, and the account fails.
1. The recipe is powerful.
Zagzebski argued in 1994 that Gettier problems are inescapable for any analysis on which the condition doing justification's job leaves open the possibility of a false belief. Her recipe has two steps:
- Start with a belief that meets the condition but is false.
- Then change the story so that the belief comes out true by an accident unconnected to the reasons.
The condition is still met, and the belief is now true, but it is not knowledge. The only escape, she concluded, is a condition that guarantees truth, and then justification and truth are no longer separate ingredients.
2. Justification proper closes the gap in deductive reasoning.
The recipe needs a belief whose reasons point one way and whose truth comes from somewhere else. Justification proper requires the reasons themselves to be true. Now suppose they are true and the reasoning is deductive, each step forced by the last as in arithmetic. Then the conclusion cannot be false, and the recipe has no false belief to start from.
- Think of two navigators who both reach port. One had a correct chart and sailed well. The other had a wrong chart and caught a lucky current. Zagzebski's recipe can always produce the second navigator. What it cannot do is turn the first into the second without making the chart wrong.
3. Probable reasoning is the hardest test.
Some reasoning, called inductive, moves from true premises to a conclusion they make only probable: every Sure-Fire match a man has struck has lit, so this one will. Brian Skyrms (1967) built the classic hard case from exactly that kind of reasoning. A man strikes a Sure-Fire match. Impurities in it keep the friction from lighting it, but at that very moment a rare burst of radiation ignites it anyway. His belief that it would light is true, his premises about past matches are true, and he does not know.
- The answer is that inductive reasoning relies on an unspoken premise: that this case is like the others in what makes the conclusion true. Here that premise is false, because this match, with its impurities, is not like the others. The premise passes the test in (P1, D11), too. It is about the match, and a man told about the impurities would stop expecting a flame.
- Where that premise is true, the conclusion comes out true for the same reason the earlier ones did, so the luck the recipe needs has no place to enter. The believer can be only probably sure of the premise. That limits how confident she should be, not whether her reasons are true.
- The McGrews' paper takes up an induction case among the counterexamples it answers. Whether this reply covers every such case is debated.
4. A knowledge-making condition that guarantees truth is the right result.
Zagzebski's argument ends with a price: escape the recipe, and justification is no longer separate from truth. Once the two senses of "justified" are kept apart, the price turns out to be small.
- Legitimation, the part the believer supplies, stays fully separate from truth. Smith had it, and so did the man with the match, while each relied on something false. What guarantees truth is the whole knowledge-making condition, true reasons included. And that is just what one should expect of knowledge.
- Zagzebski chose the same way out herself. In her later account, knowledge arises from acts of intellectual virtue that reach the truth because of the virtue, so the condition that makes a belief knowledge guarantees that it is true. Others have argued directly that whatever turns true belief into knowledge must guarantee truth.
- Taking that escape costs the believer nothing. What she must supply is what she always had to supply: good evidence. The truth of her belief, and of what it rests on, is supplied by the world, and no account of knowledge ever asked her to supply that herself.
+The Gettier problem shows what externalists said all along: knowledge needs an outside link between belief and fact. Even the false-premise answer concedes it, since whether a premise is true is a fact outside the mind.
1. Justification proper does reach outside the mind.
The strongest form of this objection turns the McGrews' answer against them. Whether Smith's premise is true is a fact about Jones, not about Smith's mind. So justification proper is partly external, and that looks like the externalists' point granted.
The first half is right. Truth is a fact about the world, and justification proper asks for it.
2. Truth is not the link externalists add.
Every account of knowledge since Plato has required that the belief be true. Justification proper takes that same requirement, extends it to the premises the belief rests on, and adds nothing else from outside.
The externalist links are a different kind of thing: a causal chain, a reliable process, a belief that could not easily have been false. They are facts about how the belief came about, or about how it would have fared in other circumstances, and the believer may have no way even in principle to look at them.
- Think of two inspectors at a footbridge. One asks whether every plank in it is sound. The other asks whether the mill that cut the planks usually cuts sound ones. The first is asking about the bridge in front of him. The second is asking about something else, and a good mill can still ship a rotten plank.
- Whether the believer is reasonable, which is what internalism is about, still depends only on what lies within her mind.
3. The externalist links have Gettier problems of their own.
Each outside link has been tried, and each has run into trouble:
- Causal accounts. Henry's belief in fake barn country is caused by a real barn in the ordinary way, and he still does not know. Goldman, who had proposed the causal account in 1967, offered the barn case in 1976 as a reason to move beyond it.
- Reliabilism, the view that a belief is justified when a reliable process produced it. Reading a clock is a reliable process, and it is how Russell's man got his lucky belief from a stopped one. Henry's eyesight is reliable too. Goldman's answer was a further condition: the believer must be able to tell the real thing from relevant look-alikes. But which look-alikes count as relevant is then a question the view has to settle case by case.
- Safety, the requirement that the belief could not easily have been false. The trouble shows up with a necessary truth, one that could not have been otherwise, such as a sum in arithmetic. A belief in a necessary truth could not have been false at all, so a simple safety condition counts even a lucky guess at one as safe. Think of a schoolboy who blurts out that 17 times 23 is 391 without working it out. He happens to be right, and on the simple condition his guess passes. Safety theorists have had to revise the condition, for example by asking whether the same way of forming beliefs could easily have gone wrong on similar questions.
None of these links guarantees that the belief is true, so Zagzebski's recipe works against every one of them, and she applied it to externalist accounts as well as internalist ones (D3).
4. Externalism is neither useful nor necessary here.
That is the McGrews' verdict. If true reasons close the gap, the outside link is not needed. And since every externalist link has needed patches of its own, it does not close the gap better.
See also:
Philosophy / Epistemology: ExternalismPhilosophy / Epistemology: Internalism
+After sixty years of failed analyses, the lesson is that knowledge cannot be analyzed at all. Timothy Williamson is right that knowing is a basic state of mind, not belief plus truth plus something else.
1. The record of failure is real.
Robert Shope's 1983 survey of the debate filled a book, and the counterexamples kept coming. Williamson's case is more than a tally of failures, though.
Some states of mind can be had only toward truths. You can see that it is raining only if it is raining, and you can remember that the door was locked only if it was. Williamson argues that knowing is a state of mind of this kind, a state in its own right and the most general of them. If that is so, the long run of failed analyses is just what one would expect.
2. Failure to agree is not proof of impossibility.
Many hard problems in philosophy and mathematics went unsolved for long stretches before a solution was found. Fermat's Last Theorem waited more than 350 years for its proof. Sixty years of debate shows that the problem is hard. It does not show that no analysis is correct. And the false-premise analysis, after all, was proposed in the first year and is still defended.
- Think of a stubborn lock and a ring of keys. If most of the keys have failed, that does not show none will fit. It means you should try the most promising one carefully.
3. Taking knowledge as basic does not remove the work.
Smith still does not know, and someone still has to say why. Williamson's own explanation is that knowledge requires safety: one who knows could not easily have been wrong in a similar case. He denies that this amounts to an analysis. It is a safety condition all the same, though, and it brings safety's troubles with it (D4).
4. Justification does not go away either way.
Even if knowledge could not be analyzed, people would still need to know whether their beliefs are reasonable and whether their reasons are true. Those are the questions internalism answers.
+"Justification proper" is a convenient redefinition. Calling a belief unjustified because its premises are false simply defines the Gettier problem away.
1. The worry is fair.
A theory that saves itself by changing the meaning of its key word deserves suspicion. If "justified" meant "reasonable" in 1962 and means "reasonable with true premises" now, then the problem may have been dodged rather than solved.
2. The requirement was there before Gettier.
Ordinary speech already tells a reasonable verdict apart from a case actually made (P2), and a proof apart from a convincing argument with a flaw in it. Russell, for his part, argued in 1912 that a true belief drawn from a false premise is not knowledge.
- His example is a man who thinks the late prime minister was Balfour, and so believes the late prime minister's surname began with B. He is right, because it was Campbell-Bannerman, but he does not know it.
- And the false-premise proposal came back within months of Gettier's paper. The distinction is old. What the McGrews added was names for its two halves.
3. The test is whether the distinction does real work.
A good distinction explains things we already believe. This one explains why Gettier subjects seem both reasonable and not knowing, why their luck enters exactly at the false premise, and why removing the falsehood removes the luck (P1). A redefinition that explained nothing would not do that.
4. Call it a fourth condition and the substance is unchanged.
Grant that in the sense Ayer and Chisholm intended, Gettier showed that justified true belief is not enough. Then the claim simply becomes a claim about the fourth condition: the premises a belief's support runs through must be true. That requirement can be filed under "justification" or listed beside it. Either way, the real question is whether it handles the cases, and that is argued on its merits (P1, D1, D2, D11).
+Requiring true premises asks too much. People know things through false beliefs all the time: a speaker who counts 53 in his audience when there are 52 still knows his 100 handouts are enough.
1. The handout case is a real challenge to (P2).
Ted Warfield (2005) pressed it. A careful speaker counts his audience, gets 53, and reasons that his 100 handouts will be enough. One listener changed seats during the count and was counted twice, so there are really 52. His premise is false, and almost everyone says he knows. If any false premise in the reasoning destroyed knowledge, he would not.
2. His support does not run through the false number.
The count supports "a little over fifty people are here," which is true, and that is all the handout conclusion needs. The exact figure is idle.
- The objector's best reply is that the speaker never believed "a little over fifty." He reasoned from 53. But a premise is relied on whether or not it is put into words (P1), and that same principle settled the careful Nogot case (D2). There the support ran through an unspoken falsehood. Here it runs through an unspoken truth.
- The test in (P1) shows the difference. Tell the speaker he miscounted by one, and his confidence about the handouts does not move. Tell Smith that Jones will not get the job, and his belief about the ten coins collapses.
3. Only the premises the support depends on matter.
A person can hold some false beliefs and still know many things, because most of her beliefs do not rest on them. Strict justification asks about the premises a particular belief's support runs through, not about every belief the person holds.
- Think of a recipe that calls for flour, sugar and eggs. A mistake about the kind of salt in the cupboard does not spoil a cake that uses no salt.
- Redundant support survives a false premise too. A detective with fingerprints, footage, and a witness who turns out to have lied still has a case. Remove the witness, and the fingerprints and footage still stand.
4. The cases where it bites are the right ones.
The view denies knowledge exactly where the believer's support runs through a falsehood and nothing else. That is just where the Gettier intuition says knowledge is missing. (An intuition, in this sense, is the immediate, unargued sense most people have that the believer does not know.)
+Whether you are Gettiered depends on whether silent premises about your surroundings are true, which you cannot check from the inside. So you can never tell whether you know, which is the very complaint internalists make against externalism.
1. Whether you know does depend on the world.
That much should be granted. Knowledge requires truth, so on every account whether you know turns partly on facts outside your mind.
But the internalist complaint against externalism is about justification. Externalism makes whether a belief is reasonable depend on facts the believer may be unable to reach. Legitimation stays fully within the believer's reach. Only knowledge waits on the world, as it always has.
2. Certainty that you know is not required.
Knowing does not require being certain that you know. A person can know her friend's phone number without being certain she has not misremembered it.
3. The challenge is answered the ordinary way.
The McGrews argue that if someone challenges a belief by suggesting it might be Gettiered, the answer is simply an argument that one's premises are likely to be true. That is no special burden. It is the same question that arises for any reasoning, answered by the same means.
- Think of a builder asked whether the foundation will hold. He does not need to prove that no crack exists anywhere in the world. He shows his inspection reports, and they are either good or not.
4. Externalism faces the same question with less to offer.
On externalist accounts, whether you are Gettiered depends on outside facts, such as whether your process was safe in your surroundings, and you may have no way to check them. The internalist at least points to something you can investigate: the truth of your own reasons.
+Gettier intuitions are not universal. Studies have found people in different cultures, and ordinary people within one culture, judging the cases differently, so the cases may show only what Western philosophers happen to think.
1. The early findings were striking.
In 2001 Jonathan Weinberg, Shaun Nichols and Stephen Stich reported that participants of East Asian background were more likely than Western participants to say that a Gettier subject knows. If that held up, the intuition behind the whole debate would look like a local habit.
2. Later studies did not confirm it.
A larger study by Edouard Machery and colleagues, published in 2017, tested Gettier cases across several countries and languages. It found the intuition that Gettier subjects do not know to be widely shared. The early result has not held up well.
- Studies within a single culture are more mixed, and that should be granted. Christina Starmans and Ori Friedman (2012) found ordinary people crediting knowledge in some Gettier cases, especially where the believer formed the belief on genuine evidence and the facts changed only afterward. Other studies find ordinary people denying knowledge much as philosophers do.
3. The answer does not rest on intuition alone.
The false-premise account explains why the cases look as they do: the believer's reasons point to one fact, and the truth comes from another. Even someone without the intuition can see that structure, and see why a chain with a false link is not a sound case.
+On this view, a Christian whose faith rests on even one mistaken argument does not really know the gospel is true. That makes knowledge of God hostage to the weakest apologetic argument.
1. A belief resting only on a falsehood is not knowledge.
Suppose a person believed in the resurrection solely because of a relic that later proved to be a forgery. Her belief would be true, but she would not, on that basis, know it. That is the honest verdict, and Christians have no reason to fear it.
2. Christian belief rarely rests on one line.
Most believers have many independent reasons: the testimony of the apostles, the empty tomb, the transformation of the disciples, the fulfilled hopes of Israel, the witness of the church, and their own experience of God. A false premise in one line does not touch the others, and strict justification rests on the lines that are sound (D7).
3. Scripture invites the believer to have true reasons.
Luke wrote his Gospel to Theophilus, drawing on what eyewitnesses had handed down (Luke 1:2), "that you may have certainty concerning the things you have been taught" (Luke 1:4). The remedy for a weak argument is a stronger one, and the case for the faith does not depend on any single apologist getting everything right.
See also:
Philosophy / Epistemology: Evidentialism
+The silent false premise is found only after the verdict is in. "My situation is normal" is false in every Gettier case by construction, so the account can always find one and explains nothing.
1. An account that cannot fail would be empty.
This is the sharpest charge against the false-premise view, and it is fair to press. Suppose any falsehood at all could be named as the silent premise, chosen after we already know the believer lacks knowledge. Then the view would only be restating the verdict. "Smith relied on the false premise that he was not in a Gettier case" explains nothing.
2. The silent premise has to pass a test fixed in advance.
A premise counts only if it meets two conditions, and both can be checked before anyone asks whether the believer knows:
- It is a claim about the subject matter (Jones, the clock, the barns in this county), not about knowledge, luck, or whether the case is normal.
- The support depends on it: tell the believer it is false, and her reason for the conclusion goes with it (P1).
"Most barn-shaped things in this county are barns" passes both. It is about barns, and telling Henry otherwise takes his reason away. "My situation is not a Gettier case" fails the first.
- Think of a doctor working out which test result a diagnosis rested on. She does not get to name whichever finding later turned out wrong. She asks which finding, had it come back differently, would have changed her conclusion.
3. The view makes predictions that could fail.
It says knowledge survives a false belief the support does not depend on (D7), and fails wherever every route of support runs through a falsehood. A single case in which every premise the support depends on is true, and the belief is still true only by luck, would refute it. Cases like Skyrms's match are attempts to build exactly that (D3), and they are where the debate now stands.
4. The rivals face the same charge.
Every account has to say which facts count, and each has been accused of choosing them to fit the verdict.
- Reliabilism must say which of the many ways of describing a believer's process is the one whose reliability matters, a difficulty known as the generality problem. Henry's process could be "seeing," "seeing a barn shape from a moving car," or "seeing a barn shape from a car in this county," and the verdict changes with the choice.
- Safety must say which possibilities are near enough to count.
None of this shows the false-premise view is right. What it shows is that the charge is not one the false-premise view alone must answer.