Devitt and BonJour on Rationalism and Empiricism: Notes

BonJour's Defense of A Priori Knowledge

Why does BonJour write about non-propositional insights concerning modus ponens?

The term 'proposition' can be traced back to Aristotle. Aristotle used it for the basic unit of a deductive argument, or syllogism. A syllogism has premises and a conclusion. The premises and the conclusion are propositions. The syllogism is valid if and only if it is absurd (irrational) to assert the premises and deny the conclusion. So, propositions are things that can be asserted and denied. They are frequently described as being things that can be true or false, since normally asserting something is saying that it is true, and denying it is saying that it is false.

So, here is an argument that is clearly valid:

If it is raining, there are puddles.
It is raining.
Therefore, there are puddles.

When I say that the argument is valid, I am not saying that either of the two premises are true. It may not be raining when you read this. Furthermore, there could be situations in which it rains and there are no puddles. To say that the argument is valid is not to say that one must be committed to either of the premises. To say it is valid means that one cannot be committed to the premises and also deny the conclusion.

The argument follows a certain pattern. Take any two propositions (things that can be asserted and denied. Call one of them "P" and the other "Q". Then assert the following:
"If P, then Q." "P." Therefore "Q".
What you have is an argument that follows the same pattern as the argument about rain and puddles. Arguments that follow this pattern are said to have the form modus ponens.

Lewis Carroll was the author of Alice In Wonderland. His real name was Charles Dodgson, and he was a mathematician. In 1895, he published a famous philosophical article in the journal Mind, a copy of which you can find here. Carroll wondered whether, in order to make arguments of the form modus ponens complete, we should add an extra premise, the premise that arguments of the form modus ponens are valid. In other words, he asked whether we should not perhaps argue:

If it is raining, there are puddles.
It is raining.
But if it is true that it is raining, and it is true that if it is raining there are puddles, then it is true that there are puddles.
Therefore, there are puddles.

He demonstrated that this would lead to an infinite regress. So, the fact that an argument is valid is not itself a premise of a valid argument. Remember, 'propositions' are things that serve as the premises and conclusions of deductive arguments, but when we recognise that the argument is valid, it seems that we cannot be recognising an extra premise. BonJour's suggestion is that we know the argument is valid a priori - we see that accepting the premises would force us to accept the conclusion as well - but this knowledge is not the knowledge of some true proposition that serves as a premise in the argument.

BonJour's point then is that we do not wake up one day and realize "Aha, if someone were to say "If P then Q, and then say P, they would be bound to admit the truth of Q." Rather, when presented with people saying "If this is a Sunday, then it isn't a schoolday" and then "This is a Sunday", we respond with incredulity when they say "So, today is a schoolday", because that conflicts with what they have just said. The a priori insight comes when we reject the absurdity by saying "But it can't be a schoolday! You just said that today is Sunday, and that Sunday isn't a schoolday." If we didn't respond to such a combination as absurd, how could we ever learn anything about anything?


BonJour refers to 'Dialectical arguments', what does he mean?

BonJour states at the start that he is defending a priori justification. He believes that justification is a necessary ingredient for knowledge, but he does not defend that proposition in this article. On p. 102, he observes that an externalist about knowledge would be immune to some of his arguments. You should now be in a position to understand why he says this.

The externalist argues that knowledge need not be justified true belief, where justification is understood to be something internal. It need only be true belief that is arrived at in a suitable way - for example, by a reliable means. So a five-year old can know that his father is coming home because he arrives at that belief by a reliable method (listening for a car at four p.m.) even though the four year old cannot give reasons. BonJour is saying that if we give reasons for everything we know, then we would have to include some a priori reasons. The externalist replies that there are lots of things we know for which we do not need to give any reasons at all, so even if BonJour is correct that if we did give reasons they would have to be a priori reasons, it does not follow that there must be a priori reasons, because in fact we do not have to give reasons at all. (Imagine this analogy. You argue that I owe you money. Furthermore, you argue that I should pay you in U.S. dollars, since that is the only legal tender in Panama. I reply that it does not matter what the legal tender is, since I don't owe you any money anyway).

BonJour is doing something that is very common in philosophy, something that you would do well to imitate. He is narrowing down the scope of the question. BonJour has devoted a whole book to the defense of rationalism, but on this occasion, he is only writing a brief article. He knows that it is impossible to say everything necessary to defend rationalism, so he places clear limits on what he will defend - he will confine himself to showing that insofar as we put forward reasons for many things we believe, those reasons must include appeals to things that can only be known a priori.

Later in the article, he refers to 'dialectical arguments'. Dialectic is the process that Socrates engages in in Plato's dialogues - a discussion where people exchange their views and criticize each other, hoping to arrive at the truth. Paying attention to dialectics means being aware of the different range of positions that people may be committed to, and bearing in mind the different responses that they have available to arguments you offer. In the process, it might well emerge that certain positions are incompatible. In this case, for example, BonJour wants to persuade us that either we must accept the existence of a priori reasons, or else we must accept either externalism or scepticism. But, according to BonJour, I couldn't maintain that knowledge is justified true belief (where justification is internal), that  that there are no a priori reasons, and that I know that I once purchased a casio electronic piano in Albrook Mall. If I accept BonJour's conclusion, I could still deny a priori reasoning, but only at a price. This constitutes dialectical progress: we have succeeded in narrowing down the range of acceptable philosophical positions, even if we haven't managed to narrow it down to one philosophical position, which we would then know is the correct answer.

So, how would this apply to, for example, beliefs based on memory?

Begin with the question "What reason do I have for trusting my memory?" I might answer this by pointing out that my memory has served me well in the past. But how do I know this? Well, I remember cases where my memory was useful. There's the problem - my memory is being called into doubt - a good reason for accepting the reliability of memory cannot simply be another memory. That would be circular.

If I am an externalist about knowledge, this need not bother me. I remember that I purchased a casio electronic keyboard in Albrook Mall. It just so happens that my memory is reliable. So, according to the externalist, I know that I purchased a casio keyboard in Albrook Mall. The fact that my memory is reliable cannot serve as a reason for believing my memory is reliable, because circular arguments are not good reasons. However, if I am an externalist, I am not worried about the fact that I do not have a reason to trust my memory.

What kind of reason might be offered for trusting my memory? I think my mother's name is Pauline because I remember her name. But maybe this is a false memory, implanted in my brain by aliens. Or maybe I am suffering from a newly discovered disease. Or maybe I am just a prisoner in the matrix...

We are now considering various explanations for the existence of my (apparent) memory that "My mother's name is Pauline." There are a whole set of rules that we generally appeal to when analysing explanations - for example, simple explanations are more likely. It is certainly much more simple to explain my apparent memory of my mother's name being Pauline by supposing that she told me this was her name than to explain the apparent memory by supposing it was implanted by aliens. So, I have reason to trust my memory if I accept that simple explanations are more likely to be true than complicated explanations. And this is something that scientists often appeal to  - 'simplicity is a sign of truth'. But then, why do we believe that simplicity is a sign of truth? The rationalist would say that this is an example of a priori knowledge.


BonJour is hoping that many of his readers will be perturbed by the thought that they have no good reason for trusting their memories. He expects many of his readers to say "Well, if these philosophers who go round calling themselves externalists go round saying that they trust their memory without any reason, that's all very well for them, but it sounds to me as though they've given up too easily. There must be some good reason for trusting one's memory, what could it be?" BonJour thinks that if there is a good reason for trusting one's memory, it could not be empirical (because the empiricist relies on what we have experienced, that is what we remember that we've experienced, that is what we remember...), and so it must be something a priori.

Moderate and Radical Empiricism: Hume, Kant, Frege and Quine

Hume argued that all of our ideas are derived from experience. Once we have ideas in our mind, we can compare them and examine the relationship between them without needing any more experience. I see a square and a triangle. I can then compare the properties of squares and triangles. Indeed, when I observe that a triangle has three sides and a square has four sides, I realize that one could keep adding extra sides to a shape to come up with a never-ending series of shapes. I can figure out some of the properties that a shape with a thousand sides would have if it existed, but to know whether such a shape actually does exist, I have to do more empirical research. So, according to Hume, mathematicians study only the relations between ideas. They do not study what exists and does not exist in the world - matters of fact.

Hume divides knowledge into relations between ideas and matters of fact. Kant divided knowledge into that which is analytic and that which is synthetic. This division had a similar purpose.

Kant thinks that all propositions have a subject and a predicate. For example, if I say "The cat is black", the subject is "The cat" and "is black" is the predicate. In order to make such a statement, we must have some concept of "The cat" and "is black." An analytic statement is one where the concept of the predicate is contained in the concept of the subject, for example "All bachelors are unmarried." Analytical knowledge is not based on experience, but it has never been seen as a problem for empiricists - like Hume's "relations between ideas." The empiricist is arguing that we can only know what is out there in the world by actually examining the world, by having experiences. But knowing what our words mean isn't knowing something about the world. The words have the meaning they do because we give them those meanings, so of course we know what those words mean. And knowing an analytic truth is just putting that knowledge into effect.

Synthetic statements, on the other hand, are statements where the concept of the subject does not contain the concept of the predicate. "All bachelors are messy" is an example. "Every effect has a cause" is analytic, because "effect" means "the result of a cause". But "Every event has a cause" is different - the concept of being an "event" does not contain the concept of a cause. After Kant, the defining characteristic of empiricism was the rejection of synthetic a priori truths.

Kant believed that truths of arithmetic are both synthetic and a priori. They are a priori, he argued, because the concept of the subject does not contain the concept of the predicate. However, this is hard for him to establish. When we say '2+2=4', what is the subject, what is the predicate, and what is the concept of '+', for example? Kant did not really have a satisfactory answer.

In the 19th Century, Gottlob Frege argued that to understand the structure of statements of arithmetic, it is not sufficient to divide them into subject and predicate. He offered an alternative way of analyzing such statements, one that is now generally accepted. He was then in a position to argue that Kant was wrong: arithmetic was not synthetic, but analytic.

This was eagerly accepted by empiricists. It was now possible for empiricists to say that everything we know is either a priori but analytic, or else a posteriori and synthetic. The analytic truths require no empirical study, but they do not tell us anything about the world we live in: analysis is just exploring our own concepts. Synthetic truths tell us about the external world, but they can only be known if we engage in empirical research. This is what BonJour means when he refers to 'moderate empiricism'. It is moderate, because it accepts a priori knowledge, but still a form of empiricism, because it treats such knowledge as 'merely analytic'.

(Frege, as it happens, was not an empiricist. Although he thought that Kant was wrong about arithmetic, he did believe in synthetic a priori truths. But, at the time, most philosophers were only interested in the argument that arithmetic is analytic, since arithmetic had been seen as the big problem for empiricists).

Then, in the 20th Century, Willard van Ormand Quine argued that Kant's division of truths into analytic and synthetic was itself an error.

According to those who accept the analytic/synthetic division, analytical truths are independent of experience because they are just truths about our concepts. However, Quine pointed out that our concepts have altered over time as we have encountered new experiences. For, me, it is part of the concept of water that water is H₂0. This was no part of Shakespeare's concept of 'water' however. The chemical properties of water were discovered by scientists engaged in empirical research, and as a result, the concept of water altered. So, the moderate empiricist thought that at least when we study concepts, we can avoid empirical research. Quine argues that even concepts must be adjusted in the light of experience. The rejection of a realm of analytic truths makes Quine a radical rather than a moderate empiricist.

In that case, what does Quine say about arithmetic? His answer involves holism. As a holistic empiricist, Quine believes that when we examine a belief in the light of experience, we examine not just one individual proposition, but our whole set of beliefs, our complete theory about the world.

Suppose that I tell you  my shoes are  radioactive. Being an empiricist, you want to test this scientifically. You use a gieger-counter to measure the radiation, but it reveals no radiation. You say "You were wrong. The geiger-counter reveals your shoes are not radioactive." I might reply, "No, what this reveals is that your geiger-counter isn't working. Or I might reply, "This is a new form of radiation, one that cannot be detected by a conventional geiger-counter. You need to buy my New! Improved! geiger-counter, which can detect this new, special kind of radioactivity."

So, there are different ways you could revise your beliefs in this situation. You could abandon your belief that professors are trustworthy. You could abandon your belief that the particular geiger-counter you have is trustworthy. Or, finally, you could accept that I have discovered a new form of radioactivity, and that we must revise our very concept of radioactivity.

Of these possible changes to your belief system, the one that is hardest to accept is the revision of your idea of radioactivity. Occasionally, someone makes a scientific discovery so great that we have to alter the very meanings of our words, but this doesn't happen every day.

So, for a pre-Quinean empiricist, that is a moderate empiricist, when I announce that my shoes are radioactive, and you announce that your geiger counter does not detect any radiation, either I am wrong, or else your geiger-counter is not functioning properly. There is no alternative. After all, a geiger-counter is, by definition, a device that detects radiation when it is properly functioning. It is an analytic truth that a properly functioning geiger-counter would detect the radioactivity of my shoes if my shoes were radioactive. But for a holistic empiricist like Quine, the whole of our thinking about the world is tested by every experience, including our concepts and definitions. Although it is very unlikely I might, in the light of experience, alter my defintions of 'radioactive' and 'geiger-counter'.

This is why Quine, as a holist, is able to be radical about his empiricism - he does not accept any a priori knowledge, not even knowledge that is 'merely analytic'.

As I hope was clear from the example, Quine does not think that we should revise our concepts every day. The moderate empiricist divides truths into those that are analytic, and those that are synthetic. This is meant to be a strict and absolute division. Quine says that we are, quite rightly, more likely to give up some beliefs than others. I might very well have to revise my current belief that Manchester United will win the Premiership in the light of future events. I am less likely to revise my belief that Manchester United is the name of a soccer club. I am even less likely to revise my belief '2+2=4' - but I cannot be sure that I will not have to do so.

According to Quine, some of our beliefs are very closely connected to one particular experience. I feel water on my head and say 'It is raining.' Then I see the water came from a water-sprinkler and say 'It is not raining after all.'  A single experience gave me the belief, and a single experience was enough to take it away. Other beliefs rest on a much wider range of experiences. There is not one, single, experience I can point to that leads me to say 'Water is H₂0,' nor would I be likely to revise that belief based on one single experience. But the belief that water is H₂0 depends on a whole history of experimentation - it was arrived at as a result of thousands of experiences. The belief that '2+2=4' is even more firmly entrenched. It is hard to imagine what could shake this belief. But, Quine argues, one reason we have for believing that '2+2=4' is our respect for science and technology. We know that engineers use mathematics, and engineers are pretty good at making bridges that don't collapse and airplanes that don't fall out of the sky. So if engineers rely on math, then math must be pretty trustworthy. Quine thinks that such beliefs appear to be a priori because when we are asked 'What experience provides the basis for saying '2+2=4', we cannot think of an experience. But Quine's response is that the belief that '2+2=4' rests not on a single experience, but thousands of years of technological achievement by the human race.

BonJour against Quine

BonJour argues that Quine's arguments are directed not against rationalism, but moderate empiricism. Quine he says, assumed that empiricism was correct in some form. He then argued that moderate empiricism fails -  definitions can be revised in the light of experience. Fine, says BonJour, maybe moderate empiricism cannot be sustained. But a rationalist is not a moderate empiricist, so what does it matter if moderate empiricism fails? BonJour's strategy here is to concede that even if Quine is right on the issues that separate the radical empiricist from the moderate empiricist, this does not make him right about the issue that separates the rationalist from the empiricist - the existence of a priori knowledge.

So, one question: why shouldn't a holist be a rationalist? I might believe that when I encounter something unexpected, I must consider revising my whole theory, perhaps even revising my concepts. BonJour could point to the fact that sometimes, concepts have been revised not just because we encounter a new experience, but because of some new mathematical discovery. One could argue that the concept of infinity changed as a result of Cantor's work, for example.

Also, BonJour has already explained why he thinks empirical knowledge must be supplemented by a priori principles if we are to avoid scepticism. If he was right to say that without a priori knowledge we have no reason to trust our memory, then, so he argues, it is also right to say that a holist such as Quine relies on a priori principles (such as 'simplicity is a sign of truth') when choosing between theories.

Devitt's Attack on the A Priori

Devitt is defending the position that BonJour calls 'radical empiricism'. Devitt, for reasons of his own, chooses to call this position 'naturalism'. He explains this in a footnote. He also adds, in the footnote, that many philosophers use the word 'naturalism' to mean that there are no supernatural beings. This is potentially confusing, but the confusion is easily solved. Whenever Devitt refers to 'naturalism', just remember that he could also have written 'radical empiricism'.

Devitt begins by observing that rationalists accept that we have a posteriori knowledge, but think we have a priori knowledge as well. So the empiricist does not have to persuade the rationalist that we have a posteriori knowledge - and if one admits that at least some of our knowledge is a posteriori, would it not make things simpler to say that all of it is a priori?

So, says Devitt, even a rationalist will admit the following:

Sometimes we believe things because of experiences, and sometimes experiences lead us to change our beliefs. This can be direct or indirect. People in Europe sees lots of white swans and no swans of any other color. So they start to believe that all swans are white. Then they travelled to Australia and saw black swans. As a direct result they abandoned the belief that all swans are white. Sometimes, the process of revising beliefs is indirect. Doctors used to believe that good health was a matter of balancing four humours, blood, phlegm, black bile and yellow bile. As modern medicine advanced, the theory of the four humours was gradually abandoned. There was not one single dramatic observation that overturned the theory - it was just that as new discoveries were made, people stopped worrying about the balance of the four humours.

The empiricist thinks that our knowledge of mathematics can be explained in this way as well. Devitt admits that we do not know very much about what numbers are, and we do not know very much about how indirect empirical justification works. But unless we have good reason to suppose that indirect empirical justification will not explain our knowledge of mathematics, we should stick to the simpler theory - empiricism. He is putting the burden of proof on the rationalist: "We both agree there is such a thing as empirical justification. Show me that we need to believe in something else."

Devitt does consider a possible objection. He is aware that many philosophers have argued that knowledge of necessary truths cannot be empirical. The reasoning runs as follows:

I am confident that the sum of any two odd numbers, however big, is an even number. If this knowledge were based on the odd numbers that I, and other human beings, have actually added, I would not have grounds for this confidence. There is no limit to the odd numbers, but there is a limit to how many numbers humans have added (and even to how many numbers computers could add). It would be as though I confidently pronounced that 'all swans are quite definitely white', even though I was aware that there exists a place, Australia, where there are swans that I have never seen, and aware that there are more swans in Australia than in Europe.

However, I can prove that the sum of any two odd numbers has to be even, and this makes all the difference. I know a necessary truth - 'The sum of two odd numbers must be even', and this gives me confidence in stating a universal truth 'The sum of two odd numbers is always even.' So, runs the argument, empirical knowledge cannot extend to infinity, because our observations are always limited. But knowledge of necessary truths does enable us to extend our gaze to infinity. So knowledge of necessary truths must be based on something non-empirical, and, after all, it is part of the definition of a priori knowledge that it is non-empirical.

Devitt responds with an example of a necessary truth that was an empirical discovery - the truth that water is H₂0. This was certainly an empirical discovery. It also seems to be a necessary truth - Socrates might have said that water has the form H₂0. The importance of this kind of example was first established by Saul Kripke in his book Naming and Necessity, a copy of which can be found in our library. (I had great fun reading that book when I was still a fresher, by the way). Most philosophers now accept Kripke's thesis that this is a necessary truth that can be known a priori.

So, Devitt is like a good chess player. He challenges the rationalist to provide a reason why there must be something more than a posteriori knowledge. He anticipates a likely response: "Because necessary truths cannot be known through a posteriori means", and he has his response planned, "Oh yes they can: or do you disagree with Saul Kripke?"

Devitt then considers BonJour's argument that empiricism leads to scepticism. He discusses ampliative and deductive rules - but what exactly is that all about?

Deduction and Induction

I explained above Aristotle's account of deductions (syllogisms). The aim of a deduction is to produce an argument such that it is absurd to accept the premises and deny the conclusion. But many arguments are not like this, and we have already seen some examples. The simplest explanation of my having an apparent memory that my mother's name is Pauline is that she really told me that her name is Pauline. Devitt is arguing that it is simpler to accept one form of justification  - empirical justification - than to suppose two kinds of justification. Evolution is widely accepted as the simplest explanation for the diversity of life-forms.

In this kind of argument, we begin with some evidence, something that has been observed. We then look at competing theories that explain the evidence. We then apply various criteria to select the best theory. One such criterion is simplicity. Another is conservativeness. In class, we spoke about Tycho Brahe's geo-heliocentric theory. This theory was more complicated than either of it's rivals. Copernicus thought everything revolved around the Sun, Ptolemy thought everything revolved around the Earth. Brahe proposed two centers of motion, the Earth and the Sun. This was not good for the credibility of his theory, and he admitted this. However, he thought that his theory had other advantages. For example, it is more conservative than Copernicus' theory. A 'conservative' theory leaves most of our current beliefs intact when we accept it. A theory that requires us to change our thinking about many issues is not conservative. Copernicus' theory was not very conservative at all, as we discussed in class. Accepting that the earth was in motion raised a lot of issues, and meant rejecting widely accepted answers to common questions (e.g. why do objects fall down?) When Devitt refers to 'ampliative rules', he is referring to rules like this 'Prefer conservative theories', 'Prefer simple theories' and so on. Most knowledge that we call 'empirical' requires us to apply such ampliative rules - that is how we use observations to arrive at knowledge about things we have not observed - to predict future events, or, indeed, come to conclusions about events in the distant past, like the Big Bang. Devitt admits that there is a lot of uncertainty among philosophers about these ampliative principles. However, he argues that whatever these rules are, and however we know them, they do not come from BonJour's 'rational insight'.

Remember, BonJour characterized 'rational insight' in two ways. One was negative: rational insight is not empirical. The other is positive: when you have a rational insight, you sense that 'This has to be true. It couldn't be otherwise. It is just self-evident.' However, Devitt argues, if ampliative principles are so hard to understand, how could they strike us as self-evident? Many students object, at first, to the idea that simplicity is a guide to the truth. Devitt describes these principles as 'tacit'. Tacit knowledge is, literally, silent knowledge. It is knowledge that we know, but do not and perhaps cannot put into words.

In a wild-life documentary, I saw a group of hunting dogs chase an animal, something like a deer, to a river. The prey jumped in the water and began to swim away. The hunting dogs waited by the river bank. The narrator, Sir David Attenborough, explained that the dogs knew the prey could not swim very far, and would soon have to return to the shore. They knew that if they waited, there was a good chance they would be able to kill it. The dogs did not say this, and could not say it. We ascribe the knowledge to them based on the way they behave. That is tacit knowledge.

Devitt is saying that scientists often proceed in the same way. Some theories seem right, others seem wrong. Good scientists follow promising ideas. We cannot always explain why some hypotheses seem so much better than others. As we explain our tacit knowledge, as we find words to express our feelings of approval or disapproval for certain theories, we come to understand the ampliative rules. But this, Devitt argues, is not at all like BonJour's description of rational insight.

Whereas BonJour thinks knowledge of the ampliative rules must be based on rational insight, Devitt thinks that not only are the ampliative rules based on indirect empirical knowledge, the deductive rules (such as our old friend Modus Ponens) might as well be based on empirical knowledge as well.

But what of BonJour's charge that if we try to justify everything we know empirically, we end up with circular reasoning, and we know that circular reasoning is bad? Devitt admits this is a worry. However, he thinks it can be overcome. Circular reasoning is bad, he thinks, because it traps you in your mistakes. However, he argues that scientific progress reveals we are not trapped by past errors. Not only have scientists produced better theories, they have produced better methodologies for analyzing theories. Furthermore, he offers a tu quoque argument (a Latin phrase meaning 'You too'.) Suppose that our repetoire of techniques for acquiring beliefs does include rational insight, as BonJour suggests. In that case, just as BonJour asks the empiricist "Why trust memory?" we could ask BonJour "Why trust rational insight." So Devitt admits that there are reasons to worry about circularity when we consider justification. However, he argues that rationalists have just as much reason to worry about this as empiricists.

Finally, Devitt attacks the obscurity of rational insight.

Since Gettier's attack on the definition of knowledge as true, justified belief, many philosophers have turned to causal theories of knowledge. In one of Gettier's examples, a job candidate has a true belief about the number of coins in the pocket of the successful candidate. But what led to this true belief was an examination of the coins in the pocket of the wrong candidate. Let's call the candidates Brown and Blue. What made the belief that 'There are seven coins in the pocket of the successful candidate' were the seven coins in Brown's pocket. But the belief was formed on the basis of the seven coins in Blue's pocket. The difference between knowledge and a lucky guess comes down, in this case, to there being the right kind of causal connection between the belief that there are seven coins in the pocket of the successful candidate, and the seven coins that make that belief true. This is hardly a complete explanation; what is the 'right kind' of causal connection? However, it does point us in a clear direction - we need to examine causal connections more carefully, to see what distinguishes those that produce knowledge from those that do not.

But how could this apply to 'rational insight', as described by BonJour? Rational insight is supposed to lead us to knowledge of necessary truths - but what are the objects involved? I can come to know about the coins in someone's pocket because I put my hands in the pocket where they are - this is the start of a chain of causes, at a particular time and place, that results in my having knowledge. Rationalists sometimes talk of 'rational insight' as a perception of timeless truths - but perception always involves something like touching - responding to an object that exerts certain effects on the body at a particular time and place. We cannot apply this model to rational insight - so then what is rational insight? Devitt admits we have much still to learn about empirical knowledge, but at least we know what to look at. But with rational insight, where could we look?

Back to PHI 2010 Home.