Kant says, in the first sentence of the introduction to the Critique of Pure Reason, that "all our knowledge begins with experience." However, he later claims that some knowledge we have is "absolutely independent of all experience." This type of knowledge he calls a priori knowledge, compared to empirical knowledge, which is based on experience. In this paper I would like to show how the possibility of there being a priori knowledge does not conflict with the claim that all our knowledge does in fact begin with experience. I will also explain what this type of knowledge is.

In order to see how a priori knowledge is possible and why it does not contradict the notion that all our knowledge begins with experience, it will first be helpful to talk about what a priori knowledge is not. Kant says that it experience is what "awakens" our faculty of knowledge. This means that before we can have any knowledge whatever, a priori or not, we have to have some experience first. This means that a priori knowledge, if it exists, is not temporally prior to experience. Moreover, Kant does not think that any knowledge is causally independent of experience, since experience not only precedes all knowledge, it makes any knowledge possible by giving our minds something to work with, so to speak. From this we can also see that, by a priori knowledge, Kant does not mean some kind of innate knowledge that we are born with.

So what does it mean for knowledge not to "arise" out of experience even if it is both causally dependent on and comes temporally after experience? First, it might be helpful to see how anything at all can be caused by something but not arise from it. Barry Stroud gives a useful analogy using a Television and its plug: In order to get an image, we must first plug the television in. Moreover, plugging the television in is necessary in order to get an image. Thus, no television image temporally precedes its being plugged in, nor is the image causally independent of our plugging it in. However, we can safely say that the image does not arise out of the plug.

There is one more thing Kant warns against mistakenly defining as a priori knowledge. There are some things we know will happen without having to experience them happening. Thus, we know some things prior to experiencing them and this may lead us to believe that we knew that thing a priori, but this is not necessarily true. For example, Kant gives the curious example of someone undermining the foundations of his house. This person may make the following argument to himself: "All unsupported bodies fall and my house will soon be unsupported. Therefore, my house will soon fall." The conclusion logically follows from the premises, and the type of step made from the premises to the conclusion is not an empirical one. However, Kant says the knowledge of the house falling is not a priori knowledge, since before making such an inference, we must know that unsupported bodies fall. Thus, Kant says that a priori knowledge is not "knowledge independent of this or that experience, but knowledge independent of all experience."

Now that we see what Kant does not mean by a priori knowledge, we can try to see what he does mean. In a word, when Kant says that certain knowledge does not arise out of experience, he is saying something about its justification. The justification for an a priori statement is not to be found in experience. But why does Kant think that we have any statements we can justify independently of experience? Kant thinks that there are certain things we know that we think "could have been otherwise." In other words, we can imagine a state of affairs where a particular truth did not hold. For example, we know that Berkeley is in California, but we can imagine that if the state lines had been drawn differently, Berkeley could easily have been in another state.

However, there are certain things we know which we think "couldn't possibly have been otherwise." For example, we don't think it would be possible to meet a bachelor who is unmarried. Without getting into the reasons why, we can say that the fact that all bachelors are unmarried is a necessary truth. However, Kant points out that no amount of experience can tell us enough to justify a necessary truth. This is because experience never yields necessity: it tells us what is so, but not what could or could not be so. Hence, we have the first positive criterion for what an a priori truth is: necessity.

Kant gives another criterion which in the end boils down to the same thing as necessity, but is nonetheless useful in our understanding of a priori truths. That is, a priori truths are universal, meaning that they are true in all cases. Just as in necessity, no amount of experience can give us universality. If we use again the case of the unmarried bachelors, if we were to rely solely on experience to say that "all bachelors are unmarried" we would never be entirely satisfied. For we would never know if there might be some bachelor out there whom we haven't met yet who is married.

But the question remains as to why necessary and universal truths are absolutely independent of all experience. In other words, if we grant that experience is not enough to justify necessary truths, how do we know we don't need experience at least as part of their justification? But this is not hard to see. Again, taking our example of the bachelors, we do not even need to go out and experience one bachelor to know that he is unmarried. In other words, we know just based on the meaning of the word "bachelor" that he is unmarried: the justification for such a statement is absolutely independent of any experience of any bachelors. In fact, even if there were no actual bachelors in the world, we would be right in saying that all bachelors are unmarried.

In conclusion, if it weren't for experience, we would not have any knowledge at all. However, once we have had experience, there are certain truths whose justification is based on experience and certain truths whose justification could not possibly be based on experience.

December 7, 1997