(●)Skepticism doubts that we know what we think we know.
+Maria wakes at six in a cold sweat. She has just sat an exam in a subject she never studied, in a building she has never seen, and the clock on the wall was running backward. For a moment, lying in the dark, she is not sure which world is real. Then the alarm goes off, the cat jumps on the bed, and the dream drains away.
Every one of us has had a dream that felt real while it lasted. Philosophers have asked, for more than two thousand years, what that ordinary experience shows. If a dream can feel exactly like waking life, how do you know you are not dreaming now? And if you cannot know that, what can you know?
- Skepticism, in philosophy, is the view that we know much less than we think, or nothing at all. A skeptic is someone who holds it, or who presses its arguments to see whether they can be answered.
- Local skepticism doubts one area of knowledge: other minds, the past, the future, morality. Global skepticism doubts it all.
- Epistemology, from the Greek word for knowledge, is the part of philosophy that studies knowledge and justification, the good standing a belief needs, besides being true, to count as knowledge. If you believe the store opens at nine because you read the sign on the door, your belief is justified; if you merely guessed nine and happened to be right, it is not.
The ancient skeptics came in two schools.
- The Pyrrhonists were named for Pyrrho of Elis (about 360-270 BC), and their arguments survive in the writings of Sextus Empiricus (about AD 200). They set every argument against an equally strong counter-argument and suspended judgment, believing neither side. What they hoped to find on the far side of all that suspending was peace of mind.
- Their sharpest weapon is the regress argument, credited to a skeptic named Agrippa and built from three of the five standard lines of attack, or modes, that Sextus records. Anyone who has fielded a four-year-old's string of "But why?" has met it. Every reason needs a reason of its own, and so the chain of reasons has only three ways to go: on forever, round in a loop back to where it started, or to a stop at something simply assumed. None of the three seems to give a real reason.
- The Academic skeptics, who taught in Plato's old Academy in the third and second centuries BC, argued that nothing can be known for certain, though some things are more persuasive than others.
Christians answered them early. Augustine, before he became a bishop, wrote a book against the Academic skeptics (386) to show that some things are known. Later he put his answer in three words: si fallor, sum, "if I am mistaken, I exist." The point is simple once you see it. A person who doubts his own existence must exist to do the doubting.
(●)Modern skepticism asks how we know there is a world outside our minds.
+René Descartes, a Catholic philosopher and mathematician, gave skepticism its modern form in a short book of six meditations (1641).
- He noted that he had often dreamed he was sitting by the fire when he was in bed, and asked how he could tell, now, that he was not dreaming.
- He imagined an evil demon, a being of great power and cunning, who fed him every experience he had, with no world behind them at all.
- He found one thing the demon could not touch: "I am, I exist" is true whenever he thinks it. He had already put the point, four years earlier, as the famous "I think, therefore I am." Doubting is a kind of thinking, so even the doubt shows that the doubter exists. From there he tried to rebuild knowledge. He argued that God exists and is no deceiver, and he used that to vouch first for whatever he grasped so plainly that he could not doubt it, and then, in the last meditation, for a material world that causes what his senses report.
David Hume, a Scottish philosopher of the next century, saw the problem at its starkest. Whether our perceptions are produced by outside objects, he wrote in 1748, is a question of fact, the kind settled by looking rather than by reasoning alone. The only way to answer a question of fact is by experience, "but here experience is, and must be entirely silent." We never see an object apart from our seeing of it, so we can never compare the two. It is like being asked whether a photograph is faithful to a scene you have only ever seen in that photograph.
Today the demon wears a scientific costume. Imagine a brain kept alive in a vat of nutrients, wired to a computer that feeds it exactly the signals a body walking down a street would send. From the inside, life would feel just as yours does. Hilary Putnam gave the case its best-known form in 1981. As it happens, Putnam himself thought the hypothesis undermines itself: a brain that had only ever been fed signals, he said, could not even mean a real vat when it used the word "vat." The film The Matrix (1999) made the picture famous. In its modern form the argument runs in three steps.
- You do not know that you are not a brain in a vat.
- If you know you have hands, then you know you are not a handless brain in a vat.
- So you do not know you have hands.
The second step rests on a principle called closure: if you know one thing, and know that it rules out another, then you know the other is false. It sounds almost too obvious to need a name. If you know your car is in the garage, and you know that rules out its being stolen from the street, then you know it was not stolen from the street.
(●)Classical foundationalism answers the skeptic by reasoning from certain foundations.
+Philosophers have not answered the skeptic with one voice. The main replies each take aim at a different step of his argument.
- Common sense. G. E. Moore held up his hands in 1939 and said, in effect, "Here is one hand, and here is another." He was surer of that than of any premise in a skeptic's argument, so the argument must have a false premise somewhere (D6). Thomas Reid, a Scottish minister, had answered Hume in a similar spirit in 1764: the principles of common sense are trusted before any argument and need none.
- Externalism. On this view, a belief can be justified by facts outside the believer's mind, such as the reliability of the senses. If your senses are in fact reliable, you know you have hands whether or not you can show it, much as a thermometer reads true without knowing anything about how thermometers work. Christian philosophers such as Alvin Plantinga hold a version of this view (D13).
- Denying closure. Fred Dretske (1970) and Robert Nozick (1981) argued that you can know you have hands without knowing you are not a brain in a vat, so the second step fails (D3).
- Contextualism. Keith DeRose, David Lewis and Stewart Cohen argued that the word "know" sets a higher bar in a philosophy seminar than at the kitchen table, so the skeptic and the ordinary speaker can both speak truly (D7).
- Dogmatism. James Pryor (2000) argued that how things look gives you a reason to believe they are that way, all by itself, unless you have a reason to doubt it (D14).
- The best explanation. Bertrand Russell (1912) argued that a real world is the simplest explanation of our experience, such as a cat seen in one corner of a room and later, hungry, in another. Laurence BonJour (1985) and Jonathan Vogel (1990) developed the argument in detail.
Timothy McGrew takes the last of these roads, and he travels it as a defender of classical foundationalism. Classical foundationalism is the view that knowledge rests on foundations that are certain, the way a house rests on a footing poured on bedrock, with everything else built up from there. On McGrew's view the skeptic is right about one thing and wrong about another.
- Right: beliefs about the outside world are not certain. They are reached by inference, which simply means reasoning from one thing to another, and they could be false.
- Wrong: knowledge does not need certainty about the world. It needs certain foundations, and those are within reach.
- Here is what a certain foundation looks like. Suppose your shoulder aches and you stop to attend to the ache. You can think "this is like this," where the "this" simply points at the experience you are having. The belief does not describe the ache from a distance, because the ache itself is part of what the belief says. With no gap between the belief and what it is about, there is nowhere for error to get in. Take the ache away and there is no belief left to be mistaken, so it cannot be false, any more than Augustine could be mistaken that he existed.
- Self-evident truths, like two plus two making four, are foundations too.
- From those foundations, beliefs about the world are reached by inference to the best explanation, reasoning to the account that makes the most sense of everything observed. Here "best" means most probable on all the evidence, which is not the same as simplest (D4).
Detectives reason this way. They cannot deduce the culprit like a proof in geometry, but they can show that one suspect explains all the evidence and the others do not. Philosophers call this kind of reasoning abductive: lay out the facts to be explained, set the rival explanations side by side at full strength, and ask which one explains the facts best. The case for a real world runs the same way. It starts from three plain facts about experience, and then asks which story makes the best sense of them.
(P1)Our experiences come in stable and orderly patterns.
+The first fact is one every skeptic grants: whatever lies behind experience, experience itself is orderly.
(1) Things stay where they were left.
- Leave your keys on the counter, go to work, and come home. The keys are on the counter. Nobody moved them, and the experience of seeing them again fits neatly with the experience of putting them there nine hours earlier.
- Leave a cup of coffee on the table for an hour, and it is colder when you come back, in exactly the way a real cup cools.
- Or watch the cat. You see it in one corner, then later in another, and now it is hungry, just as if it had walked across the room and gone without food in between.
(2) Events follow one another in regular ways.
- Strike a match and it lights. Drop a glass on a tile floor and it breaks. Turn the key and the car starts, or else fails to start in ways a mechanic can predict.
- These regularities are so dependable that whole sciences are built on them. A physicist can tell you where a ball thrown at a certain speed will land before it lands.
(3) Experience fits together across the gaps.
- A fire left burning in the fireplace while you sleep is embers in the morning. A candle burns down by just the amount it would have burned in the hours you were away.
- Here is the thing worth noticing. Nothing in the experience of going to bed, or of waking, contains the burning. You were asleep for all of it. Yet the two experiences fit together exactly as if the burning happened in between.
Therefore, our experiences come in stable and orderly patterns, including patterns that reach across the times we are not experiencing anything at all.
(P2)Our senses agree with one another and with what other people tell us.
+The second fact is that experience is not one stream but many, and the streams agree.
(1) Different senses report the same things.
- You see an apple on the table, reach for it, and feel it exactly where you saw it: round, cool and firm. You bite it and hear the crunch you expected.
- Sight, touch, sound, smell and taste are separate channels, and any one of them could have told a different story from the rest. They almost never do.
- Think of five witnesses to a car accident who did not speak to each other before the police arrived. If their stories line up, the best explanation is that they saw the same accident.
(2) Other people report what we experience.
- You hear thunder in the night, and your neighbor mentions the thunder the next morning. You see a comet, and it is in the newspaper.
- The skeptic will say, "But the neighbor and the newspaper are only more of your experiences." That is true, and it is exactly the point. Here is one more stream of experience, with no obvious reason to agree with the others, and it agrees with them.
(3) Instruments agree with the senses.
- A thermometer reads 100 degrees on the day you feel the heat. A camera records the comet you saw. A scale confirms the weight you felt when you lifted the box.
- None of these machines is trying to agree with you. Each is one more independent witness, and each tells the same story your senses told.
Therefore, our senses agree with each other, and with the reports of others and of instruments, far more often than chance would allow.
(P3)Experience often surprises us and resists what we want.
+The third fact is that experience does not behave like something we are making up.
(1) We cannot will experience to change.
- Wish as hard as you like that the rain will stop, and the rain goes on. Wish that the toothache would end, and it throbs.
- A person who controlled their own experience, as one sometimes can in a dream, would not stub their toe on the same chair twice.
(2) Experience confirms predictions no one could have guessed.
- In 1846 astronomers reasoned from small wobbles in the orbit of Uranus to an unseen planet. They calculated where it should be, pointed a telescope there, and saw Neptune.
- A world that really contains a planet where the mathematics says it is explains why the telescope showed it. Now suppose instead that our experience came from some source that only had to keep matching what had already been seen. It would have had no reason to supply a new planet in exactly that spot.
(3) Experience corrects us.
- You expect the step to be there in the dark, and it is not. You expect the milk to be fresh, and it is sour. Experience regularly overrules what we believe.
- That is just how experience behaves if there is a world we did not make, with facts of its own, pressing on us.
Therefore, experience often surprises us, corrects us, and refuses to change at our command.
(P4)Every skeptical rival explains these patterns only by copying them.
+A skeptical rival is any story on which our experience does not depend on a world that is largely as it seems: a dream, a demon, a vat, or order with no cause at all. Each can be made to fit experiences like ours, but only by being built to reproduce the very order a real world explains.
(1) The rivals must be detailed to fit the facts.
- Start with the plainest version: "some deceiver is feeding me experiences." On its own, that predicts nothing in particular. A deceiver could feed random noise, or a world that changes every second, just as easily as an orderly one.
- The same goes for saying the order has no cause at all. That is less a rival explanation than a refusal to explain, and it leaves the order as unexpected as a long lucky streak at the roulette table.
- So to fit (P1), (P2) and (P3), the story has to be filled in. The deceiver must feed experiences arranged exactly as if there were stable objects, agreeing senses, and a planet where the mathematics says it is.
(2) The details are borrowed from the real-world explanation.
- Once it is filled in, the story contains a complete copy of the world, built into the deceiver's plan. It explains the order only because it copies the order.
- Think of two explanations for a line of footprints across a beach. One is that someone walked there. The other is that someone carved footprints into the sand, one by one, exactly as a walker would have left them. The second fits the evidence too, but only by copying the first.
(3) A real world makes orderly experience just what we should expect.
- Two questions decide how well any story explains what we see. How strongly does the story lead us to expect it? And how probable was the story to begin with, before any of the evidence came in? Philosophers call that second figure the story's prior probability, its starting point.
- On the first question, a world of stable objects under fixed laws makes orderly, agreeing, stubborn experience just what we should expect.
- A deceiver, by contrast, could have produced any of countless streams of experience, nearly all of them chaotic.
- Think of a shuffled deck of cards. Almost every way it can fall is a jumble, and the deals that come out in perfect suit order are a tiny handful. A bare deceiver gives no more reason to expect the orderly deal than the shuffle does.
(4) Filling in the story does not escape the cost.
- You might think the skeptic can simply fill in his story until it predicts order as well as a real world does. He can, but that only writes the cost into the story itself, into its prior probability. "A deceiver who chose to mimic exactly this world" starts out as improbable as the lucky choice it describes.
- Nor does it help to give the deceiver a motive, say, a wish to make you believe in a world. A deceiver with exactly that aim, aimed at exactly this world, is one more thing to assume, with nothing in its favor but the fit.
- Think of a monkey at a typewriter that turns out Hamlet. Saying "it was a monkey that was going to type Hamlet" explains nothing, because a monkey like that is exactly as improbable as the typing.
- Of course, the real-world story has its details too, from where the keys lie to the orbit of Neptune. But the rival must carry every one of them in its plan and then add a deceiver besides. That is why the skeptical rivals are called ad hoc, Latin for "for this purpose": they are tailored to fit after the fact rather than predicting anything.
Therefore, the skeptical rivals fit our experience only by borrowing, detail for detail, from the explanation they are meant to replace, and they pay for the borrowing in probability.
(C)Therefore, a real world outside our minds is the best explanation of our experience.
+A real world outside our minds is the best explanation of our experience. A real world, in this sense, means things that exist and keep their order whether or not we perceive them, that our experience depends on, and that are largely as they seem. What those things are ultimately made of is a further question (D8).
Two measures show why this explanation comes out ahead. The first is scope, which is simply the range of facts an explanation covers. A real world has the widest scope, because it takes in all three facts at once: the order of experience (P1), the agreement of the senses (P2), and the way experience resists us (P3). The second is power, an explanation's ability to make the facts expected rather than merely possible. Here too a real world has the most, since it leads us to expect exactly the experience we have, while every skeptical rival fits that experience only by copying it (P4). So we are justified in believing that the world is there, and that it is largely as it seems.
(1) The conclusion is probable, not certain.
- Inference to the best explanation yields probability, not proof. A person could, in principle, be dreaming, and no argument can make that bare possibility vanish.
- But knowledge does not need the bare possibility to vanish. A jury convicts beyond reasonable doubt, not beyond all conceivable doubt, and it is right to.
- Whether this answer fully succeeds is still debated among philosophers, as every answer to the skeptic is. Its most disputed step is how far reason alone can judge which stories start out more probable (P4, D4).
(2) The skeptic asks for more than knowledge needs.
- The skeptic is right that beliefs about the world could be false. Where he goes wrong is in concluding that they are not justified. Justification comes in degrees, like a dimmer switch rather than an on-off switch, and a belief made highly probable by the evidence is justified to a high degree.
- What must be certain is the starting point, and it is. A belief that simply points at your present experience, "this is how the room looks," rests on no inference and cannot be false.
- The strongest modern form of skepticism grants all of this. Its complaint is a different one. Since the skeptic's stories lead us to expect exactly the experience we have, it says, the evidence does not favor the world over them even a little. That is the claim (P4) answers.
(3) Skepticism still has something to teach.
- For all its errors, skepticism shows something true. Our knowledge of the world rests on reasoning rather than on direct sight of things in themselves, and that reasoning can be examined.
- Christians have reason to welcome that. The apostles staked their message on what they had seen and heard, and a person who can explain why the senses deserve trust is better placed to weigh their testimony.
See also:
Philosophy / Epistemology: Classical FoundationalismPhilosophy / Epistemology: InternalismPhilosophy / Epistemology: ExternalismPhilosophy / Logic: Inductive Arguments
On skepticism and its answers:
BonJour, Laurence. The Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press, 1985.
Cohen, Stewart. "Contextualism, Skepticism, and the Structure of Reasons." Philosophical Perspectives 13 (1999): 57-89.
DeRose, Keith. "Contextualism and Knowledge Attributions." Philosophy and Phenomenological Research 52 (1992): 913-929.
DeRose, Keith. "Solving the Skeptical Problem." Philosophical Review 104 (1995): 1-52.
Dretske, Fred. "Epistemic Operators." Journal of Philosophy 67 (1970): 1007-1023.
Huemer, Michael. "Serious Theories and Skeptical Theories: Why You Are Probably Not a Brain in a Vat." Philosophical Studies 173 (2016): 1031-1052.
Lewis, David. "Elusive Knowledge." Australasian Journal of Philosophy 74 (1996): 549-567.
McGrew, Timothy J. The Foundations of Knowledge. Lanham, MD: Rowman & Littlefield, 1995.
McGrew, Timothy, and Lydia McGrew. Internalism and Epistemology: The Architecture of Reason. London: Routledge, 2007.
Moore, G. E. "Proof of an External World." Proceedings of the British Academy 25 (1939): 273-300.
Nozick, Robert. Philosophical Explanations. Cambridge, MA: Harvard University Press, 1981.
Pryor, James. "The Skeptic and the Dogmatist." Noûs 34 (2000): 517-549.
Putnam, Hilary. Reason, Truth and History. Cambridge: Cambridge University Press, 1981.
Russell, Bertrand. The Analysis of Mind. London: George Allen & Unwin, 1921.
Russell, Bertrand. The Problems of Philosophy. London: Williams and Norgate, 1912.
Stroud, Barry. The Significance of Philosophical Scepticism. Oxford: Clarendon Press, 1984.
Vogel, Jonathan. "Cartesian Skepticism and Inference to the Best Explanation." Journal of Philosophy 87 (1990): 658-666.
White, Roger. "Problems for Dogmatism." Philosophical Studies 131 (2006): 525-557.
Critics and rival views engaged:
Bergmann, Michael. "Epistemic Circularity: Malignant and Benign." Philosophy and Phenomenological Research 69 (2004): 709-727.
Berkeley, George. A Treatise Concerning the Principles of Human Knowledge. 1710.
Bostrom, Nick. "Are You Living in a Computer Simulation?" Philosophical Quarterly 53 (2003): 243-255.
Fumerton, Richard A. "Induction and Reasoning to the Best Explanation." Philosophy of Science 47 (1980): 589-600.
Fumerton, Richard. Metaepistemology and Skepticism. Lanham, MD: Rowman & Littlefield, 1995.
Montaigne, Michel de. "An Apology for Raymond Sebond." In Essays, Book 2, chapter 12. 1580.
Plantinga, Alvin. Warrant and Proper Function. New York: Oxford University Press, 1993.
van Fraassen, Bas C. Laws and Symmetry. Oxford: Clarendon Press, 1989.
van Fraassen, Bas C. The Scientific Image. Oxford: Clarendon Press, 1980.
Historical sources:
Augustine. Against the Academics. 386.
Augustine. The City of God. Book 11, chapter 26. 426.
Descartes, René. Discourse on the Method. 1637.
Descartes, René. Meditations on First Philosophy. 1641.
Hume, David. An Enquiry Concerning Human Understanding. 1748. Section 12.
Hume, David. A Treatise of Human Nature. 1739-1740. Book 1, part 4, section 2.
Pascal, Blaise. Pensées. 1670.
Reid, Thomas. An Inquiry into the Human Mind on the Principles of Common Sense. 1764.
Sextus Empiricus. Outlines of Pyrrhonism. About AD 200.
Biblical citations:
Luke 24:39; 1 John 1:1. Scripture quotations are from the English Standard Version.
+Agrippa's regress: every reason needs a reason, so the chain either never ends, runs in a circle, or stops at an assumption. The inference to the world rests on something unjustified.
1. The regress is the oldest problem in the theory of knowledge.
Philosophers have wrestled with it for more than two thousand years, and it has never been dismissed lightly. The worry is easy to state. If the inference to a real world starts from beliefs that themselves need reasons, and those need reasons, the skeptic wins by default.
2. The chain stops at what is certain.
The regress assumes that every starting point is a mere assumption. Some are not. A belief that simply points at your own present experience rests on nothing else and cannot be false, because the experience itself is part of what it says, and neither can truths seen to be true just by understanding them, like two plus two making four.
- Think of a headache. You do not infer it from evidence. The pain is present to you, and there is nothing further to ask for.
- The skeptic's next question, how any belief can need no further reason, is the very question classical foundationalism was built to answer.
3. The inference starts from those foundations.
The facts of (P1), (P2) and (P3) are facts about experience. The rules of probability that weigh one explanation against another are seen by reflection, simply by thinking them through, the way you see that two plus two is four. Which stories start out more probable is harder to settle, and (D4) takes it up. But the inference to the world does not rest on an assumption the skeptic is entitled to call arbitrary.
See also:
Philosophy / Epistemology: Classical Foundationalism
+Descartes's dream argument: there are no certain marks by which to tell waking from dreaming. So you cannot know you are awake now.
1. The evidence makes waking far more probable than dreaming.
Descartes was right that no single feature of an experience proves it is waking. But the question is not whether there is a mark that guarantees waking. It is whether the evidence makes waking far more probable than dreaming, and it does.
2. Dreams fail the tests that waking life passes.
Dreams are disconnected. They begin in the middle, skip from place to place, and do not link up with yesterday or tomorrow. Waking life connects: today's experiences fit yesterday's memories and tomorrow's plans. Descartes himself made this point in the last of his six meditations, noting that memory never links dreams to the rest of life as it links waking experiences.
- Think of Maria waking from her exam dream. The building she had never seen, the clock running backward, and the subject she never studied all failed to fit everything else she knew. That is how she knew it was a dream.
3. Feeling sure is not the test.
The strongest form of the objection says the test is useless, because inside a dream we accept a backward-running clock without a second thought and feel quite sure we are awake. That is true, and worth granting plainly. But the test is not how sure you feel. It is whether your experience, examined, actually hangs together with yesterday, with your other senses, and with other people, and a dream that is examined fails.
- A dream that passed every such examination, day after day, would be a perfect dream, and a perfect dream is one more skeptical rival that explains the order of experience only by copying it (P4).
+The brain-in-a-vat argument: you cannot know you are not a brain in a vat, and if you cannot know that, you cannot know you have hands. Closure carries the doubt from the one to the other.
1. One of the argument's premises has to go.
An argument is valid when its conclusion must be true if its premises are, and this one is. So anyone who rejects the conclusion has to reject a premise. Some philosophers reject closure, the second step; others say the word "know" shifts its meaning between the steps (D7). The better target is the first premise.
2. You have good reason to believe you are not a brain in a vat.
The first premise assumes that you have no reason to reject the vat. But you do. Think about the two ways the vat story can be told. A bare vat, wired to nothing in particular, would make the order, agreement and resistance of experience (P1, P2, P3) a fluke. A vat built to produce exactly that experience fits it, but it was improbable from the start, because it explains the experience only by copying a world (P4). And a belief that is highly probable on all the evidence is justified; when it is also true, and true not by luck, it is knowledge.
- Think of a letter in your mother's handwriting that mentions last Sunday's dinner, asks after your cough, and signs off with the nickname only she uses. You cannot prove a forger did not write it. But a forger would have to copy every one of those details, while your mother writing it explains them all at once, and you are right to believe she did.
3. Denying closure costs too much.
Fred Dretske and Robert Nozick gave up closure, and their reason deserves respect. Nozick's test for knowledge asks about a different situation: if the belief were false, would you still hold it? On his account, you know something only if the answer is no. Try the test on your hands. If you lacked them, you would not believe you had them, so that belief passes. Now try it on the vat. If you were a brain in a vat, everything would seem exactly as it does now, and you would still believe you were not one, so that belief fails. So, Nozick concluded, you know the first and not the second.
- Dretske reached the same verdict by a different route. At the zoo, he said, you know the animal in the pen is a zebra, yet you do not know it is not a mule cleverly painted to look like one, because you have never checked.
- The price is a sentence nobody can say with a straight face: "I know I have hands, but I don't know I'm not a handless brain in a vat." Keith DeRose called sentences like this abominable conjunctions.
- Nor does the zebra case require giving up closure. You know the zoo is a zoo, that zoos display real animals, and that painting a mule well enough to fool visitors is a costly fraud with no motive. That evidence makes "painted mule" improbable, and so you know the animal is not one, without ever having checked its coat.
4. Closure can stay.
Since you can know you are not a brain in a vat, you can know you have hands, and closure carries the knowledge in the right direction. There is no need to give up a principle as natural as "if you know your car is in the garage, you know it was not stolen from the street."
+As Bas van Fraassen argues, simplicity and explanatory power are preferences, not signs of truth, and the best explanation anyone has thought of may still be false. The skeptical stories fit every experience by design.
1. This is the deepest form of the challenge.
The strongest form of this objection comes from van Fraassen's account of how science works. Simplicity and explanatory power, he holds, are practical virtues: they make a theory easier to use, not more likely to be true. He adds a second worry. Inference to the best explanation only picks the best of the explanations we happen to have thought of, and that may be the best of a bad lot.
- Van Fraassen does not doubt the table in front of him. His target is the claim that explaining well is a sign of truth, and if he is right, the inference to the world rests on a preference the skeptic need not share.
- This is a serious position held by serious people, and it presses on the very step (P4) where the argument is most exposed.
2. The argument runs on probability, not taste.
Nothing in (P4) prizes elegance. It uses the two things probability theory weighs for any hypothesis whatever: how strongly the hypothesis leads us to expect the evidence, and how probable the hypothesis was before the evidence came in, its prior probability. "Explanatory power" is the first, and "ad hoc" names a story that scores well on the first only by paying for it in the second.
- Think of a teacher with two essays that match word for word, typos included. "They wrote independently" leaves the matching typos a staggering coincidence. "One copied the other" makes them just what you would expect. The teacher is not indulging a taste for tidy stories. The typos are evidence.
3. The bare skeptical story predicts nothing.
The unfilled skeptical hypothesis, "something feeds me experiences," gives no reason to expect order rather than chaos. The probability of the orderly experiences we actually have, given only that bare hypothesis, is tiny. Given a real world, it is high. That difference is exactly what makes something evidence: the facts are far more to be expected on one story than on the other.
4. Fit by design is not free.
The filled-in story fits every experience, but only because the fit was written into it, and what is written in must be paid for at the start (P4). It assumes everything the world story assumes, transferred into a plan, plus a deceiver who exists and chooses exactly that plan.
- Think of a night watchman's log showing him on his rounds every hour. "He did his rounds" explains it. "Someone forged the log" fits it too, but the forger had to know where the watchman would have been at every hour, and then also exist, and want to forge it.
5. The lot is not a bad one.
The bad-lot worry bites when the truth may lie among explanations no one has thought of. Here the choice is between (C) and its denial. Either our experience depends on a world that is largely as it seems, whatever that world is made of (D8), or it does not, and every skeptical story ever told or yet to be told falls on the second side.
- Think of a juror asking whether the defendant was in the room. However inventive the theories of the crime, each one puts him there or does not.
+Hume: we infer causes only from things we have seen joined together. Nobody has ever seen an experience joined to an outside object, so no inference from experience to objects is possible.
1. Hume's point hits one kind of inference.
Hume was right about enumerative induction, reasoning from observed cases to unobserved ones of the same kind: every swan seen is white, so the next will be. That kind of inference needs past pairings to go on.
- Hume looked hard at the very patterns of (P1). In an earlier book he noticed that a fire left in his room has burned down by the right amount when he returns, and he called this kind of fit coherence. Yet he concluded that it produces our belief in the world through habits of imagination, not through reasoning.
- The strongest modern form comes from Richard Fumerton, who shares the view that knowledge rests on certain foundations. He has argued that inference to the best explanation, examined closely, rests on ordinary induction from observed cases. If he is right, Hume's point returns, since no case of an experience paired with an outside object has ever been observed.
2. Inference to the best explanation reaches unobserved causes.
Science reasons constantly to things no one has seen joined to anything: atoms, electrons, the planet Neptune before anyone saw it, the inside of the sun. In each case the unseen thing is believed because it explains the patterns of what is seen.
- Think of a detective who has never seen this suspect at any crime scene. He still reasons from the fingerprints, the receipt and the footage to the conclusion that the suspect was there.
3. The inference needs no track record.
Fumerton's challenge assumes that explanations earn their credit by a history of being checked. The support in (P4) comes from somewhere else: when the evidence is far more to be expected on one story than on its rival, and that story did not start out far less probable, the evidence favors it. That is a truth about probability, seen by reflection, not a lesson learned from past pairings.
- Think of a die that comes up six twenty times in a row. You suspect it is loaded without ever having opened a loaded die or kept a record of one. The arithmetic alone tells you a fair die almost never does that.
4. Hume's standard would end science.
If only observed pairings could justify a belief in a cause, belief in atoms, in the earth's core, and in the distant past would all be unjustified. Those who accept a version of the restriction today, like van Fraassen (D4), apply it to atoms and electrons, not to the table in front of them. And Hume himself granted that nobody can live by such doubts: the business of ordinary life, he wrote, is what overturns them.
+G. E. Moore's answer is better. You are more certain that you have hands than you could be of any philosopher's premise, so you may reject the skeptic's argument without any theory of explanation.
1. Moore's point is a serious one.
Moore compared certainties. Any argument against his hands has premises, and he was surer of his hands than of those premises. Many philosophers find this compelling, and it captures something true: a clever argument against an obvious truth is more likely to have a hidden flaw than to be sound.
2. Moore gives a verdict, not a reason.
The strongest form of Moore's answer adds that knowledge does not need proof. Moore freely granted that he could not prove "here is one hand," and held that he knew it all the same. That much is right: almost nothing we know about the world is proved. But "here is a hand" is not a foundation either. It could be false, in a dream for instance, and it is believed because of how things look and feel. So the skeptic's question stands: what makes that belief justified? "I am very sure" is a report of confidence, not a reason for it.
- Think of a witness who is completely sure of what he saw. His sureness matters, but a jury still asks what he saw and how.
3. The best explanation supplies what Moore's answer lacks.
The confidence Moore felt has a basis. Experience is orderly, agreeing and resistant (P1, P2, P3), and a real world explains it. That turns a certainty felt into a conclusion reached.
+Contextualism dissolves the problem: in ordinary talk "know" sets a low bar that we meet, and the skeptic raises it to one we fail. Both are right in their own contexts.
1. The view captures something real about language.
Keith DeRose, David Lewis and Stewart Cohen noticed that what counts as knowing can shift with the conversation. Deciding on a Friday afternoon whether to wait in a long line at the bank, you may say "I know it is open Saturday, I was there two weeks ago," and that may be fine. In a philosophy seminar, the same sentence may seem false. That observation deserves respect.
2. The skeptical question is not about a word.
Even if the word "know" shifts, the question underneath does not, and contextualists themselves agree that the strength of your evidence stays the same from one conversation to the next. Is it more probable that there is a real world than that there is not? Is a person justified in believing it, and how strongly? Those questions have answers in every context, and the answer is the same in the seminar as at the bank.
3. The skeptic is not satisfied by a change of subject.
Telling the skeptic that his standards are unusual does not answer him. He wants a reason to think our beliefs about the world are more likely true than false. Inference to the best explanation gives one.
+George Berkeley: God gives us our experiences directly, in a stable order, with no material world behind them. That is simpler than matter and fits the evidence just as well.
1. Berkeley was a Christian bishop defending common sense.
Berkeley held that physical things are collections of ideas (his word for anything perceived: a color, a shape, a sound, a feel), and that God gives them to us in the regular order we call the laws of nature. When no human being perceives them, he said, they still exist in the mind of God. He thought this defended common sense against the skeptic, because on his view the tables and trees we see are real and we see them directly. Christians who find his view attractive are brothers and sisters in a debate about metaphysics, the study of what things ultimately are.
- Grant at once that his view fits the evidence as well as the belief in matter does. Berkeley built it to fit every experience anyone has, and no experiment could tell the two apart.
2. Berkeley's God is no deceiver.
It can look as if (P4) should count against Berkeley too, since his God, like the demon, gives us ideas directly. It does not, because his view is not a skeptical rival in the sense of (P4): on it, our experience depends on an order outside us that is largely as it seems. Two differences show it.
- The demon's plan mimics a world that is not there, so our beliefs about tables and planets come out false. Berkeley's God hides nothing. On his view the table really is where you see it, Neptune really is where the mathematics put it, and the order of nature is a real order outside us that our beliefs largely get right.
- A bare deceiver gives no reason to expect order (D4). A wise and good God does: Berkeley argued that he gives our ideas fixed rules so that we can foresee what will happen and plan our lives. And the order is not tailored to mimic something else; on his view it is the world.
- Think of the difference between a forged letter and a letter dictated to a secretary. The secretary's hand is not your friend's, but the words are truly his.
3. "Simpler" counts kinds of thing, not details.
Berkeley removes matter, but he keeps every detail matter was invoked to explain. The fire that burns down overnight and the planet circling the sun before anyone found it must all be there in God's ideas, or in the fixed rules by which he gives them. So his view has one fewer kind of thing, not fewer things to account for. Whether one fewer kind makes a view more probable is a question of metaphysics, and serious philosophers have argued it both ways.
4. The debate is about what the world is, not whether it is there.
The conclusion (C) is that a real world outside our minds is the best explanation of our experience: things that exist and keep their order whether or not we perceive them, that our experience depends on, and that are largely as they seem. Berkeley affirms every part of that. It is true that on his view the tables and planets themselves cause nothing, since only minds act. God gives us our ideas by the fixed rules that make up the world's order. But our experience still depends on that order, and the order is there whether or not we perceive it. Whether those things are matter upheld by God or ideas given by God is a question for another argument, and (C) does not need it settled.
+The answer covers the present world at most. It does nothing for the past, which could have been created five minutes ago with false memories, or for other minds, which we never observe.
1. The same reasoning applies.
Bertrand Russell raised the five-minute hypothesis: the world might have sprung into being five minutes ago, complete with memories of a past that never happened. Like the vat, it explains the evidence only by copying the past into the present (P4). Fossils, ruins, tree rings and memories agree with one another as if there had been a past, and a real past explains that.
- Think of a house with worn stairs, faded wallpaper, and pencil marks on the kitchen doorframe showing children growing. Someone could have faked all of it yesterday. The best explanation is that a family lived there.
2. Other minds are reached by inference too.
Other people's words and actions are explained best by their having thoughts and feelings like ours. A friend who winces when he cuts his finger, says "ouch," and reaches for a bandage behaves exactly as a person in pain would. These are among the hardest cases, and the answers are probable, not certain, which is all the view ever claimed for them.
See also:
Philosophy / Epistemology: Classical Foundationalism
+The Pyrrhonist asserts nothing. He only notes that every argument has an equal counter-argument and suspends judgment, so there is no claim of his for you to refute.
1. Suspension is itself a response to reasons.
The Pyrrhonist suspends judgment because the arguments on each side seem to him equally balanced. A careful Pyrrhonist will not even assert that much. Sextus says he only reports how things strike him, the way a person reports that honey tastes sweet to him.
That is fair, and it should be granted plainly: there is no thesis to refute. But a report of how the arguments strike one person is an open invitation for everyone else to look at the arguments. If the reasons on one side are in fact stronger, suspension is not the fitting response, and anyone who sees that has reason to judge.
2. The reasons are not balanced.
On the question of the outside world, they are not equal. A real world explains the order of experience; the rivals only copy it (P4). A person who has seen that has no reason to suspend judgment and good reason not to.
- Think of a juror who refuses to vote because "both sides made arguments." If one side had fingerprints, footage and a confession, and the other had a theory that someone faked them all, refusing to vote is not caution. It is a failure to weigh the evidence.
3. The Pyrrhonist lives as if the world is real.
Sextus Empiricus advised following appearances in daily life: eating when hungry, drinking when thirsty, keeping the laws and customs of one's country. That is wise, and the best explanation of why it works is that the appearances are appearances of a real world.
+Faith answers the skeptic, not argument. Christian thinkers from Montaigne to Pascal used skepticism to humble reason and make room for faith.
1. The tradition is real and serious.
Michel de Montaigne, a Catholic, used the ancient skeptics' arguments to show the weakness of unaided reason and to commend humble faith. Blaise Pascal wrote that nature confounds the Pyrrhonists and reason confounds the dogmatists (those who claim to know with certainty), and he turned the human condition between them toward Christ. Christians who hold that faith must finish what reason cannot are part of a long and honorable line.
- The strongest form is Pascal's. We know we are not dreaming, he wrote, however powerless reason is to prove it. That knowledge comes from the heart, his word for the direct grasp by which we know first principles such as space, time, motion and number, and reasoning tries in vain to attack it.
2. The heart's verdict has reasons behind it.
Pascal is right that a person knows he is awake without first building an argument, and right that some first principles are grasped directly rather than proved. Classical foundationalism agrees about the second: self-evident truths and present experience are known without argument. Where it parts company is over whether "I am not dreaming" is one of those first principles or something known from them.
- Ask a person how he knows he is awake, and he will soon point to the evidence: the day hangs together, the room stays put, other people agree. The heart's verdict was answering to that evidence all along, and the argument (P1 through C) simply spells it out.
- Think of a seasoned nurse who knows at a glance that a patient is in trouble. She could not give the reasons in the moment, but they were there, in the color of his skin and the rhythm of his breathing.
3. Skepticism about the senses would undercut the faith.
Christianity rests on public events reported by people who saw them. The risen Jesus said, "See my hands and my feet, that it is I myself. Touch me, and see" (Luke 24:39). John writes of what "we have heard, which we have seen with our eyes, which we looked upon and have touched with our hands" (1 John 1:1). A thoroughgoing skeptic about the senses cannot accept that testimony either.
4. Faith is trust grounded in reasons.
Biblical faith is trust in a God who has shown himself, not a leap made because reason has failed. A Christian answer to the skeptic that shows the senses deserve trust also shows why the apostles' testimony deserves a hearing.
See also:
Philosophy / Epistemology: Evidentialism
+Only God can answer the skeptic. Descartes saw that the senses are trustworthy only if a good God made them, so without proving God first, the inference to the world fails.
1. Descartes's route has a well-known problem.
It is an honored route, and it is worth seeing how it goes. Descartes argued for God from ideas he was certain of, then used God's goodness to vouch for clear and distinct perception, his name for an idea grasped so plainly that, while you attend to it, you cannot doubt it, as with two plus two making four. Critics in his own day, the theologian Antoine Arnauld among them, noticed the Cartesian circle: he seemed to need clear and distinct perception to prove God, and God to vouch for clear and distinct perception. Whether he escaped the circle is still debated. Only in the last meditation did the same guarantee reach the senses: God had given him a strong inclination to believe that a material world causes his sense experiences, and a God who is no deceiver would not have done that if there were none.
2. The inference to the world does not need God as a premise.
The argument for a real world (C) uses only the facts of experience and the principles of good reasoning. An atheist can follow every step of it, and Bertrand Russell, no friend of Christianity, gave a version of it himself (1912). That is a strength, because it means the skeptic can be answered on common ground.
3. Belief in God adds support.
Christians have a further reason to trust the senses: a good God who made human beings to know his world would not leave them hopelessly deceived. That reason does not replace the argument from experience. It confirms it, and it explains why the world is orderly enough to be known at all.
See also:
Philosophy / Epistemology: Externalism
+Externalism needs no argument against the skeptic. If your senses are in fact reliable, you know you have hands, and demanding proof of their reliability first is the mistake that makes skepticism look strong.
1. This serious view is held by many Christians.
Alvin Plantinga and other Christian philosophers hold that a belief produced by faculties working as they were designed to, in the setting they were designed for, can be knowledge whether or not the believer can show it. The view explains how a toddler knows her mother's face, and it refuses the skeptic the demand that started all the trouble. Christians who hold it are brothers and sisters, and the disagreement is about the theory of knowledge, not the faith.
2. The answer is conditional.
The externalist says you know you have hands if your senses are reliable. The skeptic grants that cheerfully. Whether the "if" holds was his question, and the externalist answer leaves it where it found it.
- Think of a man who asks whether his smoke alarm works and is told, "If it works, it will warn you of a fire." True, and no help at all.
3. Checking the senses by the senses runs in a circle.
If the believer asks for a reason to think his senses reliable, the externalist can offer only more looking and listening, which is to say the senses vouching for themselves. A test that could not have come out the other way tells us nothing.
- Think of asking a witness whether he is honest and taking his word for it.
- Some externalists answer that a circle of this kind is harmless for our most basic sources. Its best defenders grant, though, that it cannot help someone who has begun to doubt the source, and the skeptic's question is that person's question.
4. Evidence answers the skeptic without a proof.
The alternative is not a demand for proof. It asks for evidence the believer actually has, and finds it in the order, agreement and resistance of experience (P1, P2, P3), which a real world explains best (P4). That gives the reason externalism declines to give, and it does so without a circle.
See also:
Philosophy / Epistemology: ExternalismPhilosophy / Epistemology: Internalism
+James Pryor's dogmatism: how things look justifies belief all by itself, unless you have a reason to doubt it. A child believes in tables without reasoning to any best explanation, so the inference asks too much.
1. Dogmatism gets ordinary life right.
Nobody reasons their way to the chair before sitting in it. The look of a red wall is a reason to think it red, and Pryor's view honors that: the look is reason enough, he holds, until someone tells you the wall is lit by a red bulb. It is a serious position, a close cousin of the view that how things seem gives some reason to believe they are that way. And it should be granted that people do not consciously run through the inference to the world.
2. Evidence need not be argued to be had.
The inference to the world is not a speech anyone makes before breakfast. A person has the evidence of (P1), (P2) and (P3) and responds to it, the way an experienced reader responds to letters. The look of the wall is real evidence on both views; the difference is that the classical view counts the look itself as the foundation, and the belief about the wall as resting on it.
- Think of reading a word on a sign. You do not sound out the letters, but the letters are still what you are reading, and a misprint would stop you.
3. A look cannot rule out the vat by itself.
Take the story "I am a handless brain in a vat, fed exactly the experience of having hands." That story predicts the experience of hands with complete certainty. So when you look down and the experience arrives, it cannot count against the story, because it is exactly what the story said you would see. It fits perfectly, and if anything it raises the story's probability a little.
- So whatever lets a person reject the vat must already be in place before the look. It is the story's low starting probability, its prior probability, and that is what (P4) supplies.
- Dogmatism asks the look to do work it cannot do, and so it makes knowing the world too easy.
- Think of a suspect whose defense is that his identical twin did it. The camera footage showing his face does not count against the twin story, because the twin story predicted that face. What counts against it is everything else, beginning with whether there is a twin.
See also:
Philosophy / Epistemology: Classical Foundationalism
+Nick Bostrom's simulation argument: if advanced civilizations run many simulations of minds like ours, simulated minds far outnumber real ones, so you are probably simulated. That makes the vat probable, not ad hoc.
1. The argument gives a real reason for the vat.
Bostrom's actual conclusion (2003) is a three-way choice: almost all civilizations like ours die out before they can run such simulations, or almost none that could would want to, or we are almost certainly living in one. He does not claim to know which. Grant what deserves granting: this is the form of the vat most readers meet today, and it gives a reason for the vat rather than a bare possibility, which is exactly what the stories in (P4) lacked.
2. It assumes the inference to the world.
The argument's premises are facts about computers, brains and physics, learned by observing the world. It works only if our observations are, on the whole, a good guide to how things are. Used as a reason to distrust the senses wholesale, it would remove its own premises; used as Bostrom uses it, it relies on the very trust in experience that (C) defends.
- Think of a man who reads in the newspaper that every newspaper is fabricated. If he believes it, he has to stop believing the paper he read it in.
3. A simulation is a much smaller doubt.
A simulation would not leave us without a world. It grants most of (C): our experience would depend on a real, lawful order that exists whether or not we perceive it, namely a computer in a real universe, built by real beings, following fixed rules. What it disputes is the last part of (C), that things are largely as they seem, and there it is a genuine rival. How much it disputes is itself debated. Some philosophers argue that a simulated person's beliefs about tables would still be largely true, because the tables would be real, made of computation rather than wood, a view that in this respect resembles Berkeley's (D8). Others think a simulated table is no table at all. Either way, what remains in doubt is far less than the skeptic set out to doubt, and so far as a simulation does deny that things are largely as they seem, its case is only as strong as the guesses it rests on.
- Think of a pilot in a flight simulator who forgets where he is. The instruments, the controls and the rules they obey are all real; what he has misjudged is where they are.
4. Its key numbers are guesses.
The whole argument assumes that a computer running a program could be conscious at all, and many philosophers doubt that outright. The third option then wins only if the first two fail, and those turn on things nobody knows: whether civilizations survive, and what they would choose to do. The popular version of the argument, which drops Bostrom's caution, turns those unknowns into a confident-sounding probability by assumption, and a probability built from guesses is no stronger than the guesses.
- Think of an office pool that works out its expected winnings from a guess at how many tickets were sold. The arithmetic can be flawless and the answer worthless.