Justified True Belief

Mapping the Landscape of Good Reasons for the Truth of Christianity

Christian Evidences

Evidence & Arguments for Christian Theism

Divinity of Christ

Biblical Evidence for High-Christology: Jesus is God

TBD

Old Testament Criticism

Evidence for the Reliability of the OT Bible

Common Objections

Objection Analyses to Christian Theism

World Religions

Critical Analyses of Non-Christian Religions

Philosophy

Phileō Sophia - to Love Wisdom.

Epistemology

Study of Knowledge

A Priori Knowledge

(●) A priori knowledge is knowledge justified without relying on experience. + Helen's daughter phones on a Thursday to say that both of her boys and all three of her girls are coming for Sunday dinner. Helen hangs up and takes five extra plates down from the cupboard. Nobody told her the number five. She heard about two boys and three girls, and she worked out the rest standing at the counter, without looking at anything. On Sunday she counts heads at the door, and of course it comes to five. If it had come to six, she would have assumed that somebody brought a friend. It would never have crossed her mind that two and three had stopped making five. So Helen's belief that five grandchildren are coming rests on two very different kinds of knowledge. - What she learned on the phone, that two boys and three girls are coming, is a posteriori knowledge, Latin for "from what comes after." Its justification, whatever makes it reasonable for her to believe, depends on experience: on seeing, hearing, touching, remembering what was seen, or being told by someone who saw. If she had missed the call, she would have no reason to believe it. - That two and three make five is a priori knowledge, "from what comes before," meaning before any checking of the world. Its justification does not depend on experience. Helen does not need to count to know it, and no count at the door could overturn it. "Without relying on experience" needs one clarification, because nobody is born knowing arithmetic. - Experience may be needed to learn the ideas. Helen learned what "five" means by counting on her fingers as a child. - But once she has the ideas, experience is not what makes the belief reasonable. She sees that two and three make five by understanding what the numbers are. Experience opened the door. It did not supply the reason. - Philosophers mark this as the difference between experience's enabling role, which gives us the ideas, and its evidential role, which gives us reasons. A priori knowledge needs the first and not the second. Immanuel Kant put the point this way in 1787: "though all our knowledge begins with experience, it by no means follows that all arises out of experience." The standard examples come from a few families. - Arithmetic and geometry: two and three make five, and a triangle has three angles. - Logic: if every dog is a mammal and Rex is a dog, then Rex is a mammal. - Truths about colors and relations: nothing is red all over and green all over at the same time, and if Ann is older than Ben and Ben is older than Cal, then Ann is older than Cal. - Principles of evidence, which say what counts as a reason for what. The laws of logic are the strictest of them, saying what must follow from what. Others say what makes what probable: if nine of the ten tickets in a hat are winners and you know nothing else, the ticket you draw is probably a winner. Two other distinctions travel with this one, and it helps to keep them apart. - A necessary truth is one that could not have been otherwise, like two and three making five. A contingent truth could have been otherwise, like Helen's daughter having five children rather than four. - An analytic truth is one whose truth comes from the meaning of its terms, like "every bachelor is unmarried." Kant described it as a judgment whose predicate, the thing said, is "contained (though covertly)" in its subject, the thing it is said about. Being unmarried is already part of being a bachelor. A synthetic truth adds something its terms alone do not contain, like "every bachelor in this room is tall." - So the three distinctions answer three different questions: how a truth is known (a priori or a posteriori), whether it could have been otherwise (necessary or contingent), and what makes it true (analytic or synthetic). Much of the long argument about the a priori is an argument about how the three line up.

(●) Philosophers have long disagreed over how much reflection alone can teach. + The question goes back to Plato. Christian thinkers took it up early. - In Plato's dialogue Meno, from the early fourth century BC, Socrates draws squares in the dust. By asking questions rather than giving answers, he leads a slave boy who has never studied geometry to see how to draw a square with twice the area of a given one. Plato took this as a sign that the soul remembers what it knew before birth. Few accept that explanation now, but the puzzle stuck: the boy came to know a truth that no one told him and that he never measured. - Augustine, writing in the early 390s, observed that seven and three make ten, not only now but always, for every mind that understands them, and that the bodily senses could never have taught anyone that. He took such unchanging truths as a road toward God, who is Truth itself (On Free Choice of the Will, Book 2). In the seventeenth and eighteenth centuries the question divided two camps. - The rationalists, among them René Descartes and Gottfried Leibniz, held that reason alone can reach substantial truths. Leibniz divided truths into two kinds: truths of reasoning, which are necessary and whose opposite is impossible, and truths of fact, which are contingent (Monadology, 1714). - The empiricists, among them John Locke and David Hume, held that all our ideas come from experience. Hume drew the line sharply in 1748: "All the objects of human reason or enquiry may naturally be divided into two kinds, to wit, Relations of Ideas, and Matters of Fact." Relations of ideas, like the truths of geometry, are "discoverable by the mere operation of thought," whether or not anything they describe exists: Euclid's theorems would hold if there were no circle or triangle anywhere in nature. Matters of fact are learned by experience, and "the contrary of every matter of fact is still possible." • In Helen's kitchen, that two and three make five is a relation of ideas. That her daughter has five children is a matter of fact, since she might have had four. - Kant tried to steer between the camps. He held that some a priori knowledge is synthetic, telling us more than its terms contain. The truths of arithmetic were one example ("7 + 5 = 12 is not an analytical proposition," he wrote), and the principle that every event has a cause was another. These, he argued, are the shape the mind itself gives to experience, the way a mold gives its shape to whatever is poured into it. In the nineteenth and twentieth centuries the empiricists pressed their case hard. - John Stuart Mill argued in 1843 that even arithmetic is a generalization from experience, learned much as we learn that fire burns. - The logical positivists, a movement of philosophers and scientists centered in Vienna, held the opposite of Kant. Every a priori truth, they said, is analytic: true by the meanings we give our words, and so silent about the world. A. J. Ayer brought the view to English readers in a book of 1936. - Then in 1951 W. V. Quine challenged the line between analytic and synthetic truths itself. He pictured our beliefs as a single web that meets experience as a whole, a web in which "no statement is immune to revision," not even the laws of logic. Many who followed him concluded that there is no a priori knowledge at all, a view sometimes called radical empiricism. Saul Kripke unsettled the picture from another direction, in lectures given in 1970 and published as a book in 1980. He pulled apart two ideas that had usually been run together: being necessary and being known a priori. - His first example runs one way. "Water is H2O," he argued, is necessary, and yet it was discovered by experiment: a necessary truth known a posteriori. - His second runs the other way. Picture whoever first fixes the length of a meter by pointing to a particular rod and declaring that a meter is that long. That person, Kripke argued, knows without measuring that the rod is a meter long. Yet the rod itself could have been longer than it is. So, Kripke concluded, here is a truth known a priori that could have been otherwise. The debate is still live. - On one side, Laurence BonJour, writing in 1998, defends a priori justification as "rational insight" into necessary truths. He argues that giving it up leads straight to skepticism, the view that we cannot know anything beyond what we directly observe. - On the other, Michael Devitt, writing in 2005, argues that there is no a priori knowledge, and that everything we know answers in the end to experience.

(●) Classical foundationalism holds that the principles of good reasoning are known by reflection alone. + Think of Helen again, on that Thursday. Two things were simply in front of her. One was her own present experience: the sound of her daughter's voice in her ear. The other was a truth she could see just by understanding it: two and three make five. Classical foundationalism is the view that knowledge rests on foundations of these two kinds: a person's awareness of her own present experience, and simple truths she sees to be true just by understanding them. Those foundations, it holds, are certain. Everything else is reached by reasoning from them, and the reasoning yields probability rather than certainty. That five grandchildren will actually walk through the door on Sunday is very likely, but it is not guaranteed. The certainty is claimed only for those simple cases, not for everything known by reflection. - A long proof can hide a slip, and a complicated principle can seem obvious and still turn out false. A priori justification of that wider kind can be partial and fallible, open to error (D1, D3). - What is certain is the simple truth a person actually sees, such as that two and three make five, the way Helen saw it, standing at the counter. The reasoning that builds on the foundations runs on principles of evidence. A principle of evidence is a general truth about what supports what. Here are three. - If nine of ten tickets are winners and you know nothing else, the one you draw is probably a winner. - A conclusion drawn from a rumor is no better than the rumor. - If one explanation makes what we observe far more likely than its rivals do, the observation favors that explanation. The philosophers Timothy and Lydia McGrew hold that principles like these are known a priori. In a 2007 book they go a step further and argue that the principles are analytic, true in virtue of what their terms mean. The same book takes up two kinds of reasoning. In deduction the conclusion must follow from the premises: if every dog is a mammal and Rex is a dog, Rex is a mammal, and there is no way around it. In induction we reason from cases observed to cases not yet observed, as when every sunrise so far leads us to expect one tomorrow. The book has a chapter on what it calls the problem of deduction, and it closes by defending a ground for induction. Why does it matter how the principles are known? The McGrews' answer is an argument they call the metaregress. - A regress is a chain of reasons that runs backward without end, like a child who asks "but why?" after every answer. The metaregress is that chain one level up, among the principles of reasoning themselves. - Here is how it would start. Suppose the principles of evidence were contingent, true of our world only the way it happens to be true that your town gets more rain in April than in August. Then knowing whether they hold would need evidence. Weighing that evidence would need principles of evidence, and knowing those would need more evidence, and so on without end. The McGrews argue that this has devastating consequences for every view on which the principles are contingent. - Principles known a priori end that regress. They are not checked against the world, any more than "two and three make five" is checked by counting guests at the door. Others have reached the same point by different roads. Bertrand Russell wrote in 1912 that "all knowledge which, on a basis of experience tells us something about what is not experienced, is based upon a belief which experience can neither confirm nor confute." (To confute is to disprove.) BonJour, who holds that many a priori truths are synthetic rather than analytic, makes a similar argument (D5). The case can be set out as a deductive argument, one whose premises work like a locked chain: if each link holds, the conclusion cannot slip out. This one has the form called modus tollens, Latin for "the way of denying." If one thing is true, a second thing is true; the second is not true; so the first is not either. "If the dog had barked, the neighbor would have woken; the neighbor slept through the night; so the dog did not bark." Here (P1) is the "if" sentence, (P2) denies its second half, and (C) denies its first half. Each premise need only be more plausible than its denial, not certain beyond all doubt.

(P1) If no principle of evidence is known a priori, nothing about the world beyond present experience is known. + Everything a person knows about the world beyond her present experience is reached by reasoning. And reasoning needs principles that experience cannot supply on its own. (1) Knowing beyond present experience means reasoning from it. - What is present to you right now is a set of sights, sounds, and feelings: the look of these words, the hum of a refrigerator, an ache in your back. That your car is still where you left it is not among them, and neither is tomorrow's sunrise. You reach those from what is present. • Memory works the same way. What is present is your seeming to remember parking on Elm Street. That you really parked there is a further step (D11). - Think of a woman who opens the curtains in the morning and sees the street shining wet. The wet street is what she sees. That it rained in the night is what she concludes. (2) Every such inference relies on a principle of evidence. - An inference is a step from evidence to a conclusion, and it is a good step only if the evidence really supports the conclusion. That it does is a further truth, not part of the evidence. - The woman at the window relies on something like this: if a wet street is far more likely after rain than without it, then a wet street makes rain more likely. - She need not be able to recite it. A child who sees that every cookie in the jar has been eaten knows at once that none is left for her, though she could not state the rule. What a reasoner needs is to see that her evidence bears on her conclusion, and seeing that, in the case in front of her, is grasping the principle. - The demand that the reasoner see this for herself comes from internalism, the view that justification depends only on what lies within the believer's own mind. Externalists, who let outside facts such as reliable eyesight count, reject it. So (P1) rests on internalism, which (D6) defends. (3) Experience cannot supply the principle without assuming it. - Suppose the woman tried to learn from experience that wet streets are evidence of rain. Her record covers only the mornings she checked. To get from those mornings to a general rule, she needs another principle: that observed cases are evidence about unobserved ones. No record can teach her that one, because every lesson drawn from a record already uses it. - A teacher is no shortcut. A student who takes a rule of logic on her teacher's word is reasoning from what she hears, and she needs a reason to trust the teacher, drawn from experience of how often teachers are right. So testimony, and the memory that keeps the lesson, hand on principles only to someone already reasoning with other principles. - Think of a witness called to testify that witnesses can be believed. Whatever he says, the court has to trust a witness before it can learn anything from him. (4) Specific links are learned only through general ones. - Many links between evidence and conclusion really are learned: smoke means fire. But the learning runs on more general principles, such as that a large sample tells us about the population it was drawn from (the larger whole it came out of), the way one spoonful tells the cook about the pot. - So beneath every chain of learned links lie principles whose justification does not come from experience. If they are not known a priori, they are not known at all, and nothing reasoned from them is known either. Therefore, if no principle of evidence is known a priori, nothing about the world beyond present experience is known.

(P2) We know some things about the world beyond our present experience. + Almost everyone believes this already. The only question is whether any argument could overturn it. (1) Knowledge of the world does not require certainty. - To know something about the world, you must believe it, it must be true, and you must have strong support from evidence. You do not have to rule out every bare possibility, such as the possibility that you are dreaming right now. - Beliefs about the world are probable rather than certain. Probable is enough. - Think of a jury that convicts beyond a reasonable doubt. It is not asked to convict beyond every possible doubt, and it would never convict anyone if it were. (2) Ordinary life is full of such knowledge. - You know that your kitchen did not vanish while you slept, that the bridge you cross each morning will bear your car, and that the world existed before you were born. - Astronomers predict a solar eclipse to the minute, years in advance, and the moon arrives on time. - Well-established history and science are knowledge of what no one now experiences: that Rome burned in AD 64, that blood circulates through the body. (3) Denying it undercuts itself. - A skeptic who argues that nothing beyond present experience is known has to rely on things beyond his present experience. He trusts that his words still mean what they meant a minute ago, that his listeners exist, and that the earlier steps of his argument, now past, were sound. - Think of a man who mails out letters to persuade people that no one can know whether anyone else exists. The stamps on the envelopes give the game away. (4) The burden lies with the denial. - Like every premise, (P2) need only be more plausible than its denial. - That we know nothing of yesterday, of other people, or of tomorrow's sunrise is far less plausible than that we know some of it. Therefore, we know some things about the world beyond our present experience.

(C) Therefore, some principles of evidence are known a priori. + If no principle of evidence were known a priori, nothing about the world beyond present experience would be known (P1). But some such things are known (P2), so some principles of evidence are known a priori, by reflection alone. (1) Which principles these are is a further question. - The likeliest candidates are these: • the laws of logic, such as that if every dog is a mammal and Rex is a dog, then Rex is a mammal; • the basic rules of probability, such as that a seventy percent chance of rain is a thirty percent chance of none; • the principle that a large sample probably resembles the population it came from, which is why one spoonful can tell the cook about the pot; • the principle that evidence favors the explanation that makes it most likely, which is why a wet street favors a night of rain. - Which of these are known a priori, and how securely, is argued case by case. (P1) and (P2) show only that the list is not empty. (2) They are the second kind of foundation. - Classical foundationalism names two kinds of foundation: awareness of present experience, and truths seen to be true by understanding them. Principles of evidence belong to the second kind. They are also what lets the first kind support anything beyond itself. - Think of fuel and an engine. Experience supplies the fuel, and the principles of evidence are the engine. Neither one gets the car out of the driveway alone. (3) The principles that end the regress are necessary truths. - Some philosophers hold that a few truths known a priori could have been otherwise. Their standard case is the meter rod: on their view, whoever first fixed the length of a meter by pointing to a particular rod knew without measuring that the rod was a meter long, though the rod could have been longer. Even if they are right, truths like that come from how a word was tied to the world. They say nothing about what supports what. - A principle of evidence is different. On either account of it, analytic or synthetic (D5), it could not have been otherwise, and that is what stops the regress, the endless chain in which each principle waits on evidence and the evidence waits on more principles. A principle seen to be true by understanding it does not wait on further evidence. - A principle that was only a contingent fact about how some faculty, such as memory or eyesight, happens to perform would be another matter. It would have to be checked against the world, and the regress would start again. - Think of a kitchen clock. Whether it keeps good time has to be checked against another clock. That sixty minutes make an hour does not have to be checked against anything. See also: Philosophy / Epistemology: Classical Foundationalism Philosophy / Epistemology: Internalism Philosophy / Epistemology: Externalism Philosophy / Logic: First Principles

On a priori knowledge: Ayer, A. J. Language, Truth and Logic. London: Victor Gollancz, 1936. Benacerraf, Paul. "Mathematical Truth." Journal of Philosophy 70 (1973): 661-679. BonJour, Laurence. In Defense of Pure Reason: A Rationalist Account of A Priori Justification. Cambridge: Cambridge University Press, 1998. BonJour, Laurence. "In Defense of the A Priori." In Contemporary Debates in Epistemology, edited by Matthias Steup and Ernest Sosa. Malden, MA: Blackwell, 2005. Carroll, Lewis. "What the Tortoise Said to Achilles." Mind 4 (1895): 278-280. Casullo, Albert. A Priori Justification. New York: Oxford University Press, 2003. Devitt, Michael. "There Is No A Priori." In Contemporary Debates in Epistemology, edited by Matthias Steup and Ernest Sosa. Malden, MA: Blackwell, 2005. Donnellan, Keith S. "The Contingent A Priori and Rigid Designators." Midwest Studies in Philosophy 2 (1977): 12-27. Field, Hartry. Realism, Mathematics and Modality. Oxford: Basil Blackwell, 1989. Frege, Gottlob. The Foundations of Arithmetic. Translated by J. L. Austin. Oxford: Blackwell, 1950. First published 1884. Frege, Gottlob. Grundgesetze der Arithmetik. 2 vols. Jena: Hermann Pohle, 1893-1903. Kitcher, Philip. "A Priori Knowledge." Philosophical Review 89 (1980): 3-23. Kitcher, Philip. The Nature of Mathematical Knowledge. New York: Oxford University Press, 1983. Kripke, Saul A. Naming and Necessity. Cambridge, MA: Harvard University Press, 1980. Lectures given in 1970; first published 1972. Mill, John Stuart. A System of Logic, Ratiocinative and Inductive. London: John W. Parker, 1843. Book 2, chapters 5-6. Neurath, Otto. "Protokollsätze." Erkenntnis 3 (1932-33): 204-214. Putnam, Hilary. "There Is at Least One A Priori Truth." Erkenntnis 13 (1978): 153-170. Quine, W. V. "Truth by Convention." In Philosophical Essays for Alfred North Whitehead, edited by Otis H. Lee. New York: Longmans, Green, 1936. Quine, W. V. "Two Dogmas of Empiricism." Philosophical Review 60 (1951): 20-43. Russell, Bertrand. "Letter to Frege." 1902. In From Frege to Gödel: A Source Book in Mathematical Logic, 1879-1931, edited by Jean van Heijenoort, 124-125. Cambridge, MA: Harvard University Press, 1967. Russell, Bertrand. The Problems of Philosophy. London: Williams and Norgate, 1912. Chapters 6-7. Tierney, John. "Behind Monty Hall's Doors: Puzzle, Debate and Answer?" New York Times, July 21, 1991. vos Savant, Marilyn. "Ask Marilyn." Parade, September 9, 1990. On principles of evidence and classical foundationalism: Alston, William P. "Level-Confusions in Epistemology." Midwest Studies in Philosophy 5 (1980): 135-150. Boghossian, Paul. "What Is Inference?" Philosophical Studies 169 (2014): 1-18. Cohen, Stewart. "Justification and Truth." Philosophical Studies 46 (1984): 279-295. DePoe, John M. "Defeating the Self-Defeat Argument for Phenomenal Conservativism." Philosophical Studies 152 (2011): 347-359. Goldman, Alvin I. "A Causal Theory of Knowing." Journal of Philosophy 64 (1967): 357-372. Goldman, Alvin I. "What Is Justified Belief?" In Justification and Knowledge, edited by George S. Pappas, 1-23. Dordrecht: D. Reidel, 1979. Goodman, Nelson. Fact, Fiction, and Forecast. Cambridge, MA: Harvard University Press, 1955. Chapter 3. Huemer, Michael. "Compassionate Phenomenal Conservatism." Philosophy and Phenomenological Research 74 (2007): 30-55. Maher, Patrick. "The Hole in the Ground of Induction." Australasian Journal of Philosophy 74 (1996): 423-432. McGrew, Timothy J. The Foundations of Knowledge. Lanham, MD: Rowman & Littlefield, 1995. McGrew, Timothy. "Direct Inference and the Problem of Induction." The Monist 84 (2001): 153-178. McGrew, Timothy, and Lydia McGrew. Internalism and Epistemology: The Architecture of Reason. London: Routledge, 2007. Chapters 3 and 5-7. Plantinga, Alvin. Warrant and Proper Function. New York: Oxford University Press, 1993. Chapter 12. Pryor, James. "The Skeptic and the Dogmatist." Noûs 34 (2000): 517-549. Rawls, John. A Theory of Justice. Cambridge, MA: Harvard University Press, 1971. Stich, Stephen P., and Richard E. Nisbett. "Justification and the Psychology of Human Reasoning." Philosophy of Science 47 (1980): 188-202. Stove, D. C. The Rationality of Induction. Oxford: Clarendon Press, 1986. Swinburne, Richard. Epistemic Justification. Oxford: Clarendon Press, 2001. Van Cleve, James. "Reliability, Justification, and the Problem of Induction." Midwest Studies in Philosophy 9 (1984): 555-567. Williams, Donald Cary. The Ground of Induction. Cambridge, MA: Harvard University Press, 1947. Historical sources: Augustine. On Free Choice of the Will. Book 2. Calvin, John. Institutes of the Christian Religion. Translated by Henry Beveridge. Edinburgh: Calvin Translation Society, 1845. Book 2, chapter 2, sections 15-16. Hume, David. An Enquiry Concerning Human Understanding. Section 4. First published as Philosophical Essays Concerning Human Understanding. London: A. Millar, 1748. Kant, Immanuel. Critique of Pure Reason. 1781; 2nd ed. 1787. Translated by J. M. D. Meiklejohn. London: Henry G. Bohn, 1855. Leibniz, Gottfried Wilhelm. Monadology. 1714. Section 33. Plato. Meno. 82b-85b. Biblical citations: Acts 17:2-3; Romans 1:21; 1 Corinthians 1:20; 1 Corinthians 15:13-14; Colossians 2:8. Scripture quotations are from the English Standard Version.
+ W. V. Quine showed that no statement is immune to revision, not even the laws of logic. If anything can be given up in the light of experience, nothing is known a priori.
1. Quine's picture has real force. Quine argued in 1951 that our beliefs face experience as a whole, like a spider's web, where a tug on one strand moves all the rest. Experience touches the web at its edges, while logic and mathematics sit deep inside it. When experience surprises us, he held, we may make adjustments anywhere in the web, even there at the center. He noted that some had proposed revising a law of logic to simplify quantum mechanics. This is a serious position held by serious people, and it reshaped philosophy for half a century. 2. Revisable does not mean learned from experience. A priori justification can be fallible, which means open to error. A mathematician can believe a theorem on the strength of a proof and later find a flaw in the proof. Her belief was held a priori, and it was revised. So the question is not whether the principles of evidence could be revised. What would show them to be a posteriori is that experience is what supports them, and (P1) gives the reason it cannot be. Some press a sharper demand, that a priori justification be safe from defeat by any experience whatever. That demand mistakes what might take a belief away for what supports it (D14). - Think of a long sum on a grocery receipt that you correct after adding it again. The correction came from more arithmetic, not from weighing the groceries. 3. Quine's own method rests on principles no experiment taught him. When experience clashes with the web, Quine advised revising wherever the change disturbs the least and keeps the whole simplest. Those are principles of evidence. They decide what experience shows, so experience cannot be what shows them. - Hilary Putnam, who shared much of Quine's outlook, argued in 1978 that at least one truth is a priori after all: that not every statement is both true and false. Give that up, and no revision would mean anything, since every statement and its denial could stand together. See also: Philosophy / Epistemology: Externalism Philosophy / Epistemology: Coherentism
+ Arithmetic is learned from experience. Children learn that two and three make five by counting blocks, and John Stuart Mill argued that it is simply our best-confirmed generalization about collections of things.
1. Children do learn arithmetic with blocks. Mill argued in 1843 that "two and three make five" is an inductive generalization, confirmed every time anyone counts, and that its certainty is only the certainty of an enormous track record. About how children learn, he was right. 2. Blocks teach the idea, not the reason. Learning a truth from an example is not the same as believing it because of examples. Once a child grasps what two and three are, one carefully considered case is enough for her to see that the same must hold in every case. Russell made exactly this point: nobody's confidence that two and two make four grows with each new count. - Compare the old European belief that all swans are white. It grew stronger with every white swan, right up until black swans turned up in Australia. Arithmetic has never been held that way. 3. No count could overturn it. Suppose Helen counted her guests at the door and got six. She would look for the extra guest. She would not for a moment doubt that two and three make five. A belief that no count could overturn is not resting on counting. - Mill's view also struggles with numbers no one has ever counted. Nobody has gathered a trillion and three objects to check that adding one more makes a trillion and four, yet everyone who understands the numbers knows that it does. 4. Mill's account has had few defenders since Frege. Gottlob Frege pressed points like these against Mill in 1884. Some naturalists, philosophers who take the methods of science as the model for all knowledge, still defend descendants of Mill's account (D13, D14).
+ Euclid's geometry seemed self-evident for two thousand years, yet space turned out not to obey it. If reason's clearest insights can fail, a priori insight cannot be trusted.
1. The history is real. Euclid's geometry is the one taught in school, where two lines that start out parallel never meet and a triangle's angles always add up to 180 degrees. Kant counted it as synthetic a priori knowledge of space itself. Then Einstein's general relativity described a space that is not Euclidean, and observations since the eclipse of 1919, when starlight was seen to bend as it passed the sun, have borne it out. Kant was wrong about physical space. - Logic has a case of its own. Gottlob Frege built into his system a natural assumption: that every condition picks out a collection of the things that meet it, the way "is a dog" picks out all the dogs. In 1902 Bertrand Russell wrote to him that the assumption leads to contradiction. Take the collection of all collections that are not members of themselves. It would be a member of itself if, and only if, it were not. 2. The discovery was about space, not about geometry. A pure geometry is a system of theorems, conclusions proved from its starting assumptions, which are called its axioms. In the nineteenth century Gauss, Bolyai, Lobachevsky, and Riemann worked out geometries with different assumptions about parallel lines, and later work showed them to be just as consistent, as free of contradiction, as Euclid's. On the surface of a globe, for instance, lines of longitude start out parallel at the equator and meet at the poles. What experiment settled was which geometry describes physical space. That is a matter of fact, and it was always going to be learned by experience. That each geometry's theorems follow from its axioms is still known a priori. - Think of a chess problem. Whether a sequence of moves forces checkmate is settled by reasoning. Whether anyone actually played those moves last night is settled by asking who was at the board. 3. Errors of reason are corrected by better reasoning. Not every principle that feels obvious has really been seen to be true. A priori justification, like justification from experience, can be partial. Classical foundationalism claims certainty only for the simple truths a person actually sees. - Frege's principle was never one of those. When he published it, he wrote that it was the one principle over which a dispute might arise. And his error was found by reasoning, in a letter, with no experiment anywhere in it. - Here is a famous case. On a game show you pick one of three doors, and one hides a car. The host, who knows where the car is, always opens another door to show a goat and always offers to let you switch. Switching wins two times in three. When the columnist Marilyn vos Savant gave that answer in Parade magazine in 1990, thousands of readers wrote to tell her she was wrong, many of them with doctorates. - Some were persuaded only by playing the game over and over. But the answer can be seen by reflection. Your first pick is right one time in three, and switching wins every time your first pick was wrong. Once that is seen, no run of games could change it. See also: Philosophy / Epistemology: Classical Foundationalism
+ A priori truths are true only by definition or convention. They record how we have agreed to use words, so they say nothing about the world, and principles of evidence are just rules we have chosen.
1. Some a priori truths are true by meaning. "Every bachelor is unmarried" is one of them, and the logical positivists were right that such truths cost nothing to know. Ayer argued in 1936 that all a priori truths are like that. 2. Convention cannot make logic true. Quine himself found the flaw, in 1936. Logic yields infinitely many truths, so nobody could have agreed to them one at a time, and the conventions would have to be general ones. But getting from a general convention to any particular truth already takes a step of logic. So logic cannot be true merely because we agreed to it. - Think of a note on the fridge: "Everyone in this house clears their own plate." The note covers Sam only if someone reasons that everyone in the house clears a plate, Sam is in the house, and so Sam clears his. The note cannot supply that step. It depends on it. - Conventionalists reply that the conventions need never be written down, since they live in how we use our words. Quine considered that reply too. A convention no one ever stated, he answered, adds nothing to the plain fact that we firmly accept the truths of logic, so calling them conventions explains nothing. 3. Chosen rules would be arbitrary. If principles of evidence were rules we chose, other rules would be just as good. Counter-induction, expecting the opposite of whatever has happened before (the dog has bitten me every time, so this time it surely won't), would be as reasonable as induction, only less popular. Nobody believes that. - Think of the rules of chess. The bishop moves diagonally because that is how the game was set up, and the game could have been set up otherwise. That a wet street makes rain more likely could not have been otherwise. 4. Analytic does not mean trivial. We choose which word names an idea. We do not choose what follows from the idea once it is named. If the principles of evidence are analytic, as the McGrews hold, they are true in virtue of the concepts of evidence and probability, and what follows from those concepts was never put to a vote. - A truth reached by reasoning alone can still be very hard to see. Fermat's Last Theorem, a claim about whole numbers, was stated in 1637 and not proved until 1994, more than three and a half centuries later.
+ Kant and Laurence BonJour hold that the deepest a priori truths are synthetic, not analytic. If so, the claim that principles of evidence are analytic fails, and the classical foundationalist account fails with it.
1. The disagreement is real. BonJour, the most thorough recent defender of the a priori, holds that much of what we know by reason is synthetic. That nothing is red all over and green all over at once is not true by the mere meaning of "red" and "green," he argues, and yet it is seen to be necessary. The McGrews hold that the principles of evidence are analytic. Philosophers have not settled which description is right. 2. (C) does not depend on it. (C) says that some principles of evidence are known a priori. It does not say that they are analytic. Analytic or synthetic, a principle seen by reflection to be necessary ends the regress, because it is not a contingent fact waiting on further evidence. - Think of a house key. Whether it was cut from brass or from steel, it opens the same door. 3. Both sides share the point that matters. BonJour argues, as (P1) does, that without a priori justification no inference from experience could be justified. Externalists hold that what justifies a belief can lie outside the believer's view, such as whether the faculty that formed it, like memory or eyesight, works reliably. On the externalist view the principles of reasoning are contingent facts about our faculties, and BonJour and the McGrews both reject it. So their dispute is about how to describe the a priori, not about whether it exists or does the work.
+ Most people reason well without knowing any principles of evidence. What makes an inference good is that it reliably leads to truth, and a priori knowledge of principles is an armchair invention.
1. The observation is correct. Hardly anyone can state a principle of evidence, and children reason long before they can name a rule. Externalists, who hold that what justifies a belief can lie outside the believer's view, build on that fact. Alvin Goldman holds that a belief is justified when a reliable process, such as normal eyesight in good light, produced it, whether or not the believer knows anything about the process. On that view a good inference needs only a reliable habit of moving from evidence to conclusion, and the reasoner need grasp nothing. - If that is right, (P1) is false, because a reliable habit could give knowledge of the world with no principle known by anyone. (P1) rests on internalism at exactly this point, and externalism is a serious position held by serious people. 2. Seeing a principle at work is knowing it. Externalists often charge internalists with a level confusion: mixing up being justified with knowing that one is justified, as if every believer had to be a philosopher. (P1) asks for nothing of the kind. The child at the cookie jar has no thoughts about justification at all. She sees that an empty jar means no cookie for her, and what she sees is about jars and cookies. That is grasping the principle in the case at hand, the way a person can recognize a friend's face without being able to describe it. 3. Reliability alone is not a reason the reasoner has. A process can be reliable without the believer having any reason to trust it. Think of a man who acts on a stock tip from a stranger who, unknown to him, really is an insider. The tips are reliable. But from where he stands, he is gambling. Justification is about what the believer has to go on. - Run it the other way. Take two reasoners with the same experiences and the same care, one in our world and one fooled at every turn by a deceiver. They are equally reasonable, since everything they have to go on is the same. Yet externalism, at least in its simplest form, has to say that only the first reasons well. 4. A reliable rule cannot vouch for itself. Which ways of reasoning are reliable is a question about the world, and it can be answered only by reasoning from evidence. The strongest externalist reply is that a reliable rule may be used to show its own reliability: induction has worked so far, so, reasoning inductively, it will keep working. James Van Cleve argued that such a rule-circular argument, which uses a rule to vouch for itself without listing that conclusion among its premises, is no vice. - But a fortune-teller who reasons by counter-induction, expecting the opposite of whatever has happened before, can certify her method the same way. It has failed every time so far, so, by counter-induction, it will succeed from now on. An argument that would certify any rule at all cannot be what tells a good rule from a bad one. - Think of a man who buys a second copy of the morning paper to make sure the first one was right. See also: Philosophy / Epistemology: Internalism Philosophy / Epistemology: Externalism
+ Lewis Carroll showed that inference cannot depend on principles. If every inference needed a principle as an extra premise, that premise would need another principle to connect it to the conclusion, and so on forever.
1. Carroll's puzzle is real. In 1895 Lewis Carroll, the author of Alice's Adventures in Wonderland, published a short dialogue between Achilles and a tortoise. The tortoise accepts two premises, A and B, but will not accept the conclusion Z until Achilles writes down a third premise, C: "If A and B are true, Z must be true." Then the tortoise wants a fourth premise, saying that if A, B, and C are true, Z must be true. Achilles keeps writing, and the tortoise never reaches the conclusion. - A sharper form aims at (P1) itself. If the reasoner must grasp that her evidence supports her conclusion, that grasp looks like one more thing she believes, and so one more premise. Then the tortoise's regress starts again. 2. A principle of evidence is not an extra premise. Carroll's lesson is that a rule of inference cannot do its work by being added to the premises. It has to be seen in the step itself. The child at the cookie jar does not add "if a jar is empty, none is left" to her premises. She sees that her conclusion follows from what she sees. 3. Seeing the connection needs no premise of its own. (P1) asks only that the reasoner grasp that her evidence supports her conclusion. That grasp is not one more premise sitting in a list, waiting for something to connect it to the conclusion. It comes from her awareness of the connection itself, in the very step she is taking. And awareness of a link needs nothing further to link it. It is knowledge of the principle in the case in front of her. It is a priori, but it is not something she reasons from. - Classical foundationalism holds that this grasp rests on direct awareness of the connection, which philosophers call acquaintance. It is like your awareness of a headache while it throbs. What rests on acquaintance is not inferred from anything else, so it needs no premise of its own. - Critics reply that such awareness is really a belief in disguise, and a belief would start the regress again. The internalist answers that contact with what is present is not the kind of thing that can be done well or badly. Nobody is acquainted with a headache carelessly. So there is nothing in it for a further premise to back. Whether that answer holds is still argued over in the debate about internalism. - The tortoise's trouble is a failure to see, and no premise could cure it. - Think of explaining a joke to someone who does not get it. You can add more and more explanation, but sooner or later he either sees it or he does not. See also: Philosophy / Epistemology: Internalism
+ Knowing something requires being causally connected to it. Numbers and logical truths cause nothing, so a priori "insight" would be a mysterious sixth sense.
1. The puzzle is famous. Paul Benacerraf posed it in 1973. Suppose mathematical truths are about abstract objects, things like numbers that are not anywhere in space or time and cannot be touched. Suppose too that knowledge requires a causal connection with what is known, the way light from a tree reaches the eye of someone who sees it. Then mathematical knowledge looks impossible. Benacerraf offered it as a dilemma for philosophers of mathematics, and it is still discussed. - Hartry Field later sharpened the puzzle so that it no longer leans on any theory of causation. If mathematical truths are about objects like these, he argued, those who claim to know them owe an explanation of how their beliefs come to match the facts. Without one, the match looks like a coincidence. This is the puzzle in its strongest form. 2. The causal theory was built for knowledge through the senses. Alvin Goldman, who proposed the causal theory of knowledge in 1967, applied it only to knowledge of empirical facts, the kind learned through the senses, and within a decade he had moved to a different account. A test written for knowledge of the world cannot be turned against knowledge of necessary truths without assuming what is in dispute. 3. The requirement is not known by experience. "All knowledge requires a causal connection" is a claim about every possible case of knowledge, and no survey of cases could establish it. If it is known at all, it is known by the kind of reflection it forbids. 4. Insight is the most ordinary thing there is. Anyone who sees that if Ann is older than Ben and Ben is older than Cal, then Ann is older than Cal, has done it. Nothing was detected by a sixth sense. The truth was understood. - That is also the answer to Field. A belief about a contingent fact needs a link to explain why it matches the fact, such as light from the tree or a reliable witness, because the fact could have been otherwise while the belief stayed put. A necessary truth cannot drift away from the belief like that. To understand what "older than" means is already to see why the conclusion must hold, so the match is explained by understanding the claim, not by a signal from somewhere else. - Think of a girl who sees that a square cut corner to corner makes two triangles. No message reached her from the land of shapes. She looked at what a square is. - A Christian has a further reason to expect minds to fit such truths: the same God made both the world and the minds that study it (D9).
+ Our sense of what is self-evident was shaped by evolution for survival, not for truth. A mind built to stay alive has no special claim to see necessary truths.
1. The question is fair. Our capacity to reason has a history, and it is reasonable to ask whether that history gives us any assurance that reason gets things right. The strongest form of the objection is not that reason has an origin. It is that its origin would have given us the same convictions whether or not they were true. Evolution rewards what keeps creatures alive, and quick rules of thumb often do that well enough. Our snap judgments about chance really are unreliable, as the game-show puzzle shows (D3). 2. The argument uses what it doubts. The case for an evolutionary history reasons from fossils, genes, and anatomy to conclusions about the distant past, using principles of evidence at every step. If those principles deserve no trust, neither does the argument that says so. 3. Where a capacity came from does not settle whether it sees truly. Judging a belief by its origin rather than its grounds is the genetic fallacy. An origin story undercuts a belief only when nothing else is available to check it by, and principles of reasoning can be checked by understanding them. The way to test "two and three make five" is to look at it, not at the history of the one looking. - That is how the bad rules of thumb were caught in the first place: by careful reasoning with principles that pass the check. A debunking argument that cannot tell the sound principles from the shortcuts has proved too much. - A girl who learned to read from cereal boxes can still read the newspaper correctly. To check her reading, you look at the newspaper. 4. Christians have a further reason for trust. On a Christian view, reason is not an accident of survival but the gift of a God who is himself rational and truthful. Augustine took the unchanging truths of number as pointing toward God. Alvin Plantinga has argued that the naturalist, who holds that nature is all there is, faces a real problem here that the theist, who believes in God, does not. That argument is debated. See also: Philosophy / Logic: Common Fallacies: Genetic Fallacy
+ David Hume showed that the most important principle of evidence, that the future will resemble the past, cannot be known a priori, because its denial involves no contradiction.
1. Hume's point is correct as far as it goes. "That the sun will not rise to-morrow is no less intelligible a proposition," Hume wrote, "and implies no more contradiction than the affirmation, that it will rise." No reasoning proves that the future must resemble the past. 2. A priori principles can be about probability. The principle needed is not that the future must resemble the past. It is the weaker principle that a large sample, absent any reason to think it unusual, probably resembles the population it came from. Donald Williams set out this answer in 1947, David Stove refined it in 1986, and Timothy McGrew defended it in 2001. It rests on a mathematical fact: of all the large samples that could be drawn from a population, the great majority closely match the whole. - Think of a poll of a thousand voters. It works because most samples that size come close to the true proportion, and that is a matter of counting, not of how elections went in the past. - Critics answer that every sample we have comes from the past, so it may be unusual in just the respect that matters. Defenders reply that a bare possibility like that counts against a sample only if there is some evidence that it is unusual. - Patrick Maher pressed a more technical objection. The last step, from "most samples are like this" to "this sample is probably like this," needs starting probabilities, how likely each possibility looked before the sample was drawn, and the argument does not supply them. Whether the sampling answer succeeds is debated. 3. (C) needs only some principles. Even if no a priori ground for induction succeeded, (C) would stand on the laws of logic and the basic rules of probability. What would shrink is how far our knowledge reaches into the future. See also: Philosophy / Epistemology: Externalism Philosophy / Logic: Inductive Arguments
+ Perception and memory give us knowledge directly, with no inference at all. Seeing a tree is not reasoning to a tree, so (P1) fails from the start.
1. It does not feel like inference. Nobody experiences seeing a tree as drawing a conclusion. Modest foundationalists (Richard Swinburne, James Pryor, and Michael Huemer among them) keep the foundations but do not require them to be certain. They hold that if it seems to you that something is so, you thereby have a reason to believe it, unless something defeats that reason. Philosophers call such an experience a seeming, and the principle phenomenal conservatism. If it seems to you that a tree stands outside the window, that seeming is itself a reason to believe one does. - On this view the seeming does the work by itself, and no principle needs to be known first, so (P1) is false. - Its defenders add two sharper charges. Huemer argues that rejecting the principle defeats itself, since anyone who rejects it does so because rejecting it seems right to him. And classical foundationalism, they say, trusts how things seem to reason while refusing to trust how they seem to the senses, which looks like special pleading, a double standard. 2. Inference need not be slow or conscious. Reading a friend's mood from her face takes no time at all, yet it is still a step from what you see to what you conclude. What matters is how the support runs, not how fast the step feels. 3. A seeming supports what it presents only with a reason to trust it. Classical foundationalism answers that how a thing looks is certain, and how it is goes beyond that. "It looks like an apple" cannot be wrong. "It is an apple" can, because the apple might be wax. The step from the first to the second needs a reason to trust looks of that kind in conditions like these. 4. Seeing a truth of reason is not a seeming. The charge of special pleading assumes that rational insight is one more appearance. But the two cases differ where it matters. The look of an apple and the apple are two things, and the look could be there with no apple behind it. When a person sees that if Ann is older than Ben and Ben is older than Cal, then Ann is older than Cal, she has the claim itself before her mind and sees how its parts fit. There is no second thing out in the world for the insight to come apart from. - Think of the difference between hearing someone describe a letter and reading the letter yourself. - Whether every sound judgment of reason is seeing of this kind, rather than a strong seeming, is the point in dispute, and phenomenal conservatives deny it. 5. The self-defeat charge misses. The same difference between seeing and seeming answers it. The classical foundationalist does not reject phenomenal conservatism because rejecting it seems right. He rejects it for a reason he can lay out, the wax apple, and that reason is built from truths about evidence that he sees. - And suppose the self-defeat argument worked. It would show at most that some seemings about truths of reason carry weight. It would do nothing for seemings about trees and parked cars, which is exactly where this objection to (P1) needs them. - A man who trusts his arithmetic has given no reason to trust his horoscope. 6. Modest foundationalism still needs a principle known a priori. "If it seems to you that something is so, you have a reason to believe it" is itself a principle of evidence. It cannot be learned from experience without being used along the way, and its defenders argue for it by reflection, as the self-defeat argument does. So a modest foundationalist who rejects (P1) still holds (C): if he knows his own principle, he knows it a priori. See also: Philosophy / Epistemology: The Principle of Credulity Philosophy / Epistemology: Classical Foundationalism
+ Scripture warns against trusting human reason. Paul says that God "made foolish the wisdom of the world," and he warns against being taken captive "by philosophy and empty deceit."
1. The warnings are real. Paul writes, "Where is the one who is wise? Where is the scribe? Where is the debater of this age? Has not God made foolish the wisdom of the world?" (1 Corinthians 1:20). And to the Colossians: "See to it that no one takes you captive by philosophy and empty deceit, according to human tradition, according to the elemental spirits of the world, and not according to Christ" (Colossians 2:8). Christians should take both seriously. 2. Paul's target is a message, not logic. In 1 Corinthians, Paul is answering those who judged the cross by what seemed wise to their culture, to whom a crucified Messiah looked like weakness and folly. In Colossians the philosophy he warns against is defined by its source, "human tradition" and "not according to Christ." Neither passage is about whether two and three make five, or whether a conclusion drawn from a rumor is any better than the rumor. - Even the darkened thinking Paul describes in Romans follows a refusal to honor the God people already knew (Romans 1:21). It is a turning away of the heart, not a loss of arithmetic. Calvin, who pressed the damage sin does to the mind as hard as anyone, drew the line the same way. He counted logic and mathematics among the gifts God's Spirit gives even to unbelievers, and told Christians to use them (Institutes 2.2.15-16). 3. Paul himself reasons, and relies on logic. At Thessalonica "he reasoned with them from the Scriptures, explaining and proving that it was necessary for the Christ to suffer and to rise from the dead" (Acts 17:2-3). And to the Corinthians he argued step by step: "But if there is no resurrection of the dead, then not even Christ has been raised. And if Christ has not been raised, then our preaching is in vain and your faith is in vain" (1 Corinthians 15:13-14). That argument works only because its logic holds for Paul and for the Corinthians alike. 4. The truths of reason are not a human tradition. Augustine saw that the truths of number are the same for every mind and at every time, and he took them as pointing beyond every human mind toward God. Christians have long seen the principles of reason as belonging to the Creator's world, not as rivals to faith.
+ Michael Devitt argues that the a priori is obscure, since no one has explained how reflection alone could yield knowledge. He adds that it is unnecessary, because logic and the methods of science earn their place by the success of science as a whole.
1. This is naturalism's best case. Devitt is a naturalist about knowledge. He holds that there is only one way of knowing, the empirical way of the sciences, by observation and experiment. So he asks what a priori insight actually is, and how it could connect a mind to the truth, and he holds that its defenders have never given a satisfying answer. He adds that it is not needed anyway. If beliefs face experience together, as Quine held, then logic, mathematics, and the rules of good method are supported by experience along with everything else, through the success of the whole body of belief they serve in. And if someone objects that this leaves inference itself unsupported, he gives the same answer: our rules of inference are justified by their place in the theory that experience confirms. - Naturalists often borrow Otto Neurath's picture of sailors who must rebuild their ship at sea, plank by plank, with no dry dock to stand in. - This is a serious position held by serious people, and the demand for an account of a priori knowledge is fair. 2. Support from the whole of science is still inference. To say that logic and method are supported by the success of science as a whole is to reason from a record: science has predicted well, so it is probably true, and its methods are probably reliable. That step, from success so far to probable truth, is an inference from evidence. It runs on principles of evidence, such as that a theory whose risky predictions keep coming true is probably on the right track. And those principles cannot draw their support from the success of the whole, because they are what anyone uses to read that success as support (P1). Quine's own rules for revising the web are a case in point (D1). - A web wide enough to vouch for its own methods would vouch just as well for one built around counter-induction, which reads every failure as a sign of success to come (D6). - Think of a company that audits its own books. The numbers may be right, but the audit cannot be what tells you so. 3. Understanding is not obscure to anyone who has done it. The charge of obscurity asks what kind of process a priori insight is, the way vision is light striking the eye. But understanding is not a mechanism, and it is not mysterious to anyone who has followed a proof until she sees why it works. Whoever has worked through the game-show puzzle until switching plainly wins knows what it is to see a truth by reflection (D3, D8). - Nor does the empiricist have a mechanism for justification. Light striking the eye explains how a belief about a tree is caused, not why the sight makes the belief reasonable. That a sight supports a belief is a truth about evidence, and no account of the eye supplies it (P1). - The naturalist's own thesis is no laboratory result either. No observation, and no survey of the sciences, could show that every possible kind of knowledge is empirical. If the thesis is supported at all, it is by an inference from the success of science, and that inference runs on the very principles it would retire. See also: Philosophy / Epistemology: Externalism Philosophy / Epistemology: Coherentism
+ Philip Kitcher argues that most of what anyone knows of logic and mathematics was learned from teachers and books, and so is known by testimony, not a priori. He adds that a priori justification would have to survive any experience whatever, and none does.
1. Kitcher is right about how most people learn. Kitcher argued in 1980 that a priori justification, if there were any, would have to hold up in the face of any experience whatever. In 1983 he argued that mathematical knowledge is handed down through a community of teachers and students, resting in the end on experience. Both observations have force. - Most people believe the Pythagorean theorem because a teacher told them. A truth that can be known a priori is often in fact known through experience, as when a calculator reports that a large number is prime. - Experience can also undermine a belief reached by reflection. A mathematician who hears that the experts have found a flaw in her proof should doubt her result until she has looked again. 2. Testimony hands on principles only to reasoners. A student who believes a rule because her teacher said so is reasoning from what she heard, and she needs a reason to trust the teacher (P1). Testimony can be how she came to believe this rule. It cannot be how she came to every principle she reasons with, since trusting the teacher already uses some of them. Somewhere in her own mind there have to be principles she sees for herself, or her trust in the teacher is a guess. - Think of banks that lend on one another's guarantees. However long the chain of promises, somewhere a bank has to hold the money. 3. Experience can undercut a belief it does not support. Kitcher's condition runs together two jobs experience can do. One is to support a belief, the way a record of mornings supports a forecast. The other is to give a reason to suspect a slip in the reasoning, the way the report of a flaw does for the mathematician. That report does not show her theorem false. It shows that her proof might have a gap, and she answers it by going back to the proof, not by running an experiment. - Think of a smoke alarm. It can send you running to the kitchen, but what settles whether dinner is burning is what you find when you open the oven. - The condition also proves too much. Turn the case around and suppose a recipe card says that 2 cups of flour and 3 of sugar make 6 cups in all. Arithmetic tells you the card is wrong somewhere, so a belief that rested on reading has been overturned by reasoning alone. Yet nobody concludes that what you believed about the recipe never rested on reading it. - Classical foundationalism never claimed that a priori justification is beyond all undercutting. It claims certainty only for simple truths a person actually sees (D3). - What makes a justification a priori is where its support comes from, not what might take it away. Many who have written on the a priori since Kitcher reject his condition.
+ Nelson Goodman argues that rules of inference are justified by fitting the particular inferences we accept, and those inferences by fitting the rules. Principles of evidence earn their standing by that mutual adjustment, not by a priori insight.
1. This is how rules are actually refined. Goodman argued in 1955 that deduction and induction are justified in the same way. A rule is amended when it licenses an inference we are unwilling to accept, and an inference is rejected when it breaks a rule we are unwilling to give up. John Rawls later named the balance this reaches reflective equilibrium, and it describes much of how logicians, statisticians, and ethicists actually work. A coherentist, who holds that beliefs are justified by how well they fit together, sees in it the whole story. The description is accurate. Nobody receives the rules of evidence complete, and they are sharpened against cases. 2. The verdicts on cases are a priori. When a rule is amended because it licenses an inference we are unwilling to accept, something makes us unwilling. We see that in this case the evidence does not support that conclusion. That is a judgment about what supports what, and it is not learned from a record either (P1). So mutual adjustment does not replace a priori knowledge. It refines it, moving back and forth between principles seen in particular cases and principles stated in general, and (C) asks for nothing more. - Think of tuning a guitar by ear. You adjust each string against the others until they sound right together, and the adjusting works only because your ear can hear when two notes are in tune. Take away the ear, and there is nothing to adjust toward. 3. Fit alone creates no support. If the verdicts on cases had no standing of their own, bringing them into agreement with the rules could not give them any, just as two witnesses who saw nothing add nothing by agreeing. A gambler whose rules and verdicts fit together perfectly ("red has come up five times, so black is due") is still wrong about the roulette wheel, because the fit was built on a bad verdict. Equilibrium describes a system that fits together. Whether it fits the truth depends on whether what went into it was worth something. See also: Philosophy / Epistemology: Coherentism

Metaphysics: Free Will

Whether Our Choices Are Up to Us

Public Theology

Theological Analyses of Societal Issues

Theology of the Family

Where Faith and Family Intersect

Biographies

Notable Works & Great Quotes from Key Figures

Contemporary Voices

1960 AD – present
Playlist
Justified True Belief