For my final piece of philosophical writing in my four year academic career here at Berkeley [not so, I did another semester…], I have chosen to write on a rather broad topic, and in a way I find myself in a more difficult position than Descartes himself when he speculated on the idea that, "Whatever I have up till now accepted as most true I have acquired from the senses or through the senses." For Descartes not only had an intellect far superior to mine, but he had the added advantage of approaching this question by sitting in his oven and clearing his thoughts of all preconceived opinions. I, on the other hand, having reread in two long sittings the entire reader prepared by Professor Stroud, have the task of approaching the issue faced with the overwhelming specter of responding to the plethora of writings that have occurred in the four hundred years since Descartes' initial musings. The various writers since then have complicated the issue to an extent probably unimaginable to Descartes.

Although I find myself a bit at a loss as to how to begin, I think it might be convenient to reverse the order in which we have dealt with the problem in this class. I will first address the problem of whether a priori knowledge exists and whether we in fact have such knowledge. Once we have reached any conclusions concerning the distinction between knowledge that is and knowledge that is not acquired by the senses, we can then ask how it is that the senses actually provide us with any useful or reliable information. Finally, there remains the problem of whether such information is at all applicable to knowledge we seem to have about the unobserved, including knowledge about the future.

Although the a priori/empirical distinction can be traced to Hume and his differentiation between relations of ideas and matter of fact, Kant was the first to provide what seemed to be a complete account of its relationship to the way we attain knowledge. He stated at the very beginning of his Critique of Pure Reason by saying that we have two types of knowledge, one whose justification is to be found in experience, or what for our purposes can be called information attained from the senses, and another whose justification is independent of sense data. Although the latter knowledge may "begin" with experience, insofar as experience temporally and even causally precedes any and all knowledge, Kant argued that this knowledge does not "arise out of experience." This is because the justification for such knowledge, being necessary and universal, cannot be provided by sense data, which only provides contingent truths. In this class of knowledge Kant included, among other things, all mathematical truths, as well as relations of cause and effect and the spatio-temporal nature of our experience.

With this very brief and inadequate account of Kant's philosophy we have nevertheless an initial set of terms and concepts with which we can begin to address Descartes' claim. Depending on how we interpret the above quote, we can either say that Descartes agrees or disagrees with Kant's division. If the quote says that all that we take as true is justified solely by sense data, then we have a clear discrepancy with Kant for the reasons stated above. However, we can fit the quote into Kant's scheme if we interpret it loosely by saying that knowledge that comes "from the senses" is empirical and knowledge that comes "through the senses" is not necessarily so (in the way that Kant's a priori knowledge "begins with experience.)

However, we can avoid this clever interpretation and assess the truth of the claim more easily if we move forward a few hundred years and adopt a "pragmatic" perspective, question the a priori/empirical distinction, and deal with the issue more "holistically". Kant himself admitted that "thoughts without content are empty, intuitions without concepts are blind," meaning that neither mental processes nor sense data by themselves can provide knowledge. Instead, there is some sense in which the mind and the material provided by the external world (if it exists) come together to form knowledge. Lewis points out that "Until the criteria of our interpretation have been fixed, no experience could be the sign of anything or even answer any question." This is because the sense data themselves do not constitute knowledge. Knowledge, to put it broadly, depends on what we do with the chaos of perception, and more specifically, how we divide and categorize the elements of perception.

How we in fact do this, and whether or not this categorization is fixed or stable, is beyond the scope of this paper. However, with this in mind, the idea that what we accept as true "comes from the senses" is just too vague. Even Quine, who wishes to do away with the a priori altogether, must admit that there is some kind of "back and forth" going on when we decide what we will or will not accept as true. There is a complicated and constant readjustment, he points out, of which sense data fit into our understanding of the world. Some sense data are unproblematic, but when we are presented with something that doesn't conform, we have to either change our conceptualization or say that a particular sense datum is illusory or mistaken.

Granted that there we have a complicated relationship with our sense data, we must now ask what it is that perception actually consists of. Descartes, after admitting that all he had hitherto accepted as true came from the senses, then conceded that he may have been deceived about all of it. In other words, whether or not we have a priori knowledge that affects our interpretation of sense data, if we are asking whether or not what we accept as true is acquired by the senses, the time has come to ask whether or not we have any reason to trust what we perceive.

Again, Descartes realized that there is no way of categorically disproving the hypothesis that there is an evil demon deceiving us about all our sense impressions. In other words, when it comes down to it there can be no logical denial of skepticism, for, as Hume pointed out, matters of fact are never true necessarily. This is because a judgment is necessary only if its negation is inconceivable. However, none of the judgments we make based on information acquired by the senses is of this sort.

Descartes came to the conclusion that there is one thing of which there can be no doubt: his own existence. However, with regard to the senses, there is another class of judgments that is just as indubitable, but seems to have escaped Descartes. These are statements about what it seems to us, about our own sense experience. If, for example, it seems to me that there is a glass in front of me, I may be mistaken about that glass existing, but I cannot be mistaken about what seems to me to be true. The fact that there is a glass in front of me is questionable, but the fact that it seems to me to be so seems irrefutable (granted, you may doubt that I'm telling the truth, but the person making the statement can have no doubt.)

Based on this fact, a whole theory of knowledge can be constructed, defining statements of truth wholly in terms of statements about what is perceived. Without getting into it, I will say only that Lewis sets up a system whereby any statement of fact is translated into a series of what are called terminating judgments in the form of "if it seemed to me that I did such and such, then in all probability, it would seem to me that such and such." Each of these terminating statements is thus a hypothetical test that could confirm or deny the truth of a particular statement.

Chisholm points out the difficulties that arise when we ask ourselves how we can establish the probability part of the terminating judgments, which seems to be a new statement of objective fact, not having to do with our sense perception. However, whether or not these objections hold is not of importance here. Rather, what is important is that this view attempts to dissolve the gap that exists between our perception and the external world. It does this by defining that world in terms of our experience. In other words, there is a sameness of meaning between objective statements and subjective statements of experience, and this is the most radical affirmation of the quote we are discussing in this paper: all that we accept as true is acquired from the senses insofar as the only things we know of are our senses. Lewis claims that if Chisholm is right and this sameness of meaning does not hold, then all we are left with is skepticism.

Thus, there are three alternatives regarding the information we get from our senses. First, one we haven't directly discussed: What we see is really out there, and our senses are for the most part reliable indicators of an external world. This is the common sense view assumed by most human beings who are not engaged in philosophical queries, but, as we have seen, it is subject to skeptical doubts. Second, meaning may boil down to facts about sense experience, but, this too is subject to skepticism by Chisholm's objection. Third, we can be complete skeptics and doubt all the information provided by the senses.

All three of these cases, though, are compatible with the claim that what we accept as true comes from the senses. They are just different interpretations of why we do this. In the first case, we do so because of the way the world is. In the second case, we do so because what it means to be true has something to do with our sense data themselves. And, in the third case, we might do so just because we can't help it: we accept things as true, it just happens to be the case that they are not true, or, rather, we really have no way of knowing whether or not they are.

Let us now consider what we can say, based on what I have written so far, about whether or not what we accept as true comes from the senses. We may or may not know things that do not come from the senses, a priori. If we grant that we do know some things based on our senses, we can't be sure about whether or not these actually coincide with the common notion of truth: that is, that it is something about the world, not just about ourselves.

If that isn't enough confusion, there is one more complication to discuss. Even if we just grant that our senses in some sense give us truths about what we are perceiving, our actual beliefs encompass much more than that. We think we know things about matters that are not immediately observed. These include beliefs about things that are going on at a specified moment, but in places our senses do not have access to, as well as beliefs about the past and the future, both of which we cannot immediately perceive. So we must now ask whether or not the senses provide us with the necessary information for this class of things we generally accept as true.

The above problem can be discussed under the label of "induction", for the question stated more precisely is, on what basis can we make an inference from the observed to the unobserved, even if we have no doubts as to the information provided by our senses. I will discuss this topic in terms of Hume's doubts about our inferences about future events. However, everything I say can be extended to any unobserved matter of fact.

Hume said that all our inferences from past observations to beliefs about the future "proceed upon the supposition that the future will be conformable to the past." He then went on to show that we have no reason to believe this supposition. As I have already mentioned, Hume divides all human knowledge into two exhaustive categories: "relations of ideas" and "matters of fact". The former consists of propositions whose justification lies outside of experience and whose truth is therefore "demonstratively certain." All other propositions, those which are justified by empirical evidence, are termed matters of fact. The opposite of a matter of fact is always conceivable and hence never implies a contradiction.

Hume said that all inferences or expectations about the future are necessarily based on matters of fact, and not relations between ideas. This is because nothing that conflicts with our "common sense" expectations is inconceivable. In other words, we can at least imagine, when we have an expectation about a future event that something else could occur.

This means that any inferences we make about the future, or any beliefs at all we have about the future, have to be based on experience, or empirical evidence. So the next step is to try to see how Hume thinks we use experience to make such inferences. He says that the way we do this is by noticing that certain events are always coupled with others, and that although all events are distinct, when we see similar events following each other time after time, we imagine that the two are connected and we form the idea of cause and effect. He says that all our arguments concerning existence, or matters of fact, are based on the relation of cause and effect. And, as we already pointed out, this relation is based on experience, not on reasoning alone. However, this is where Hume discovers the problem at hand: the only way we can use the notion of cause and effect to predict future events is to assume that the future will resemble the past. But any attempt to prove this supposition, Hume argues, is necessarily circular. Thus, all our inferences from past observations to beliefs about the future are based on an unfounded supposition.

The difficult thing to see is why Hume believed that this supposition, which we all hold, is unfounded. The supposition that the future will resemble the past cannot be shown demonstratively, since it does not imply a contradiction to state that the future will not conform to the past. Therefore, the only option left is reasoning based on experience. However, all our experience tells us is about things we have experienced in the past, and any attempt to use this experience to tell us something about the future will necessarily lead us in a circle.

We can see why this is so by observing the following line of thought: suppose we reason that we know that the future will conform to the past, because, in the past, the future has always resembled the past. This is true. In the infinite number of moments we have experienced in our lives, the following moment has resembled the preceding moment in all the ways we consider important: things have continued to exist, like events have been followed by like events, which we have termed their "causes", and, most importantly, we have been able to make successful predictions based on this uniformity. However, this doesn't prove anything, since, again, all this was in the past and tells us nothing about the future. As Russell points out, just because "past futures" have resembled "past pasts" how do we know that "future futures" will resemble "future pasts". Of course, we could try and go a step further and say that past past pasts have resembled past past futures, but this would plainly lead nowhere.

But what, you might ask, does say about our original question of whether or not what we take as true comes from the senses? If Hume is right, the principle of induction, on which all our beliefs about the unobserved are based, can neither be proved nor disproved by the information we get from the senses. However, as Edwards pointed out, this depends entirely on what we take to be a valid proof. If we are looking for deductive reasons, we won't find any, since if induction could be reduced to deduction, we wouldn't need induction to begin with.

Strawson pointed out that a belief in induction is in some way a belief about the constitution of the universe, for if the universe changed to where nothing is predictable, then the question of whether induction would be valid would clearly be no. However, this belief about the universe is itself based on induction. It is hard to say, then, whether the type of things we accept as true, which are based on induction, can be said to be acquired by the senses.

The only conclusion I can seem to reach, and it is a conclusion that is frustratingly empty, is that the things we accept as most true must have something to do with our senses, but what that is, I have no idea.

December 15. 1997