Lecture Notes, UC Davis Philosophy 102, Theory of Knowledge
Concluding Remarks
G. J. Mattey, Senior Lecturer
Most current work in epistemology is done ahistorically. That is, it is done on the basis of our present use of concepts such as knowledge, justification, belief, etc. I have for a long time believed that the best way to approach these concepts is by examining the history of their usage. The reason is that the concepts we use now have been encrusted with many layers of meaning. This makes any "analysis" of "S knows that p" suspect from the very start. And, I will argue, it accounts for the vast disagreement in the accounts of knowledge that I highlighted in my first lecture and which is evident in our readings. So this last look at the theory of knowledge will be place its history front and center.
Disagreement in Ancient Accounts of Knowledge
We began our examination of knowledge by reading Plato, who was the first Western philosopher to write extensively about it. An important feature of Plato's treatment was his assumption that knowledge has a nature which it is the task of the philosopher to uncover. In this way, it is no different from virtue or a triangle.
The nature of knowledge is to be found through the use of dialectic, an activity of question and answer, from which the truth gradually emerges. A problem with the kind of dialectic practiced by Plato was that it was controlled by the interlocutor, Socrates. To be sure, Socrates explored many alternatives, but they were often not granted much of a hearing and were dismissed lightly. If Socrates got agreement, then the point of agreement was accepted as the truth without further ado.
Contrast the procedure of Socrates with that of the Pyrrhonian skeptics. They cast their eye over the whole range of disputes (including many that occurred after Plato's time), and they discovered massive disagreement. The only thing people could really agree about was the way things appear. It is not as if the disagreement was irrational or due to stubbornness; each party to a given dispute advanced reasons for their positions.
For example, the account of knowledge given by Aristotle is in many respects inconsistent with that given by Plato. The most glaring differences are Aristotle's rejection of the doctrine of recollection and his view that we can have knowledge, not merely opinion, about the material world. And he backed up these claims by arguments that for centuries were considered to be conclusive (though, thanks in large part to Descartes, they are now mostly rejected).
We Are Still No Closer to Agreement
Now think about the situation in theory of knowledge at present. We have looked at the claims of many philosophers about what knowledge is, what kind of structure it has, and whether we in fact have any. There is widespread disagreement on nearly every point. There are fallibilists and infallibilists, foundationalists and coherentists, internalists and externalists. What prospects are there for coming to any kind of agreement among them?
The Pyrrhonians would no doubt nod their heads and tell us that they warned us of this almost two thousand years ago. Their official prescription would be to keep looking, but in their hearts they would be murmuring that it will never work out. They could point to the fact that the current trends are toward even further disagreement. In this course, we have not looked at all the more recent points of view. Some of them are very radical. Most significantly, many "post-modern" thinkers claim that what counts as knowledge is relative to cultures, sub-cultures, or even individuals.
More optimistic philosophers would look for some kind of convergence on an agreed-upon account of knowledge. It appears that there are several ways this could take place. The first is that one or another of the current theories or its descendants would win out over the others. Another possibility is that an entirely new theory of knowledge will emerge and win everyone over. Yet a third is that there will be some kind of synthesis of existing theories, with perhaps some new elements added.
I doubt that any of these are going to happen, though like the Pyrrhonian skeptics, I will keep searching for the truth. The reason is that, in my opinion, there is no single concept of knowledge; on the contrary, there are at least two conceptions of what knowledge is. Different theories of knowledge tend to be directed at different of these conceptions. I will try to back this claim up in what follows.
Before going on, I will note a final possibility. It may be that someone has actually given a true theory of knowledge, or that someone will do so in the future, without there being any agreement about it. The problem here is that most people, maybe everyone, would not know that it is the true theory. And if it does not generate agreement, what reason would there be for most people to think that it is the true theory?
Fallibilism vs. Infallibilism
Theory of knowledge in the twentieth century and in the present century is overwhelmingly fallibilist in its orientation. That is, the vast majority of philosophers agree that it is possible for one to have knowledge without being able to rule out the possibility of being wrong. Of the philosophers we have studied, only Plato and Descartes were infallibilists, and both are long-dead.
Why was infallibilism abandoned by almost every theorist of knowledge? It was not because someone demonstrated that the traditional infallibilist conception was a mis-characterization of what knowledge is. Instead, it was because after repeated attempts, philosophers were unable to show that we have much infallible knowledge, if any. Maybe we have infallible knowledge in the areas of mathematics and logic, but that is about it. We are fallible about everything else.
In the ancient period as well as the period immediately after Descartes, infallible knowledge was distinguished from probable opinion. David Hume put the distinction in the strongest possible way: "But knowledge and probability are of such contrary and disagreeing natures, . . . [that they] must be either entirely present, or entirely absent" (A Treatise of Human Nature, Book I, Part 4, Section 1). Descartes thought that outside of infallible knowledge, there is only "moral certainty," which is enough assurance to regulate our behavior (Discourse on Method, Part Four; Principles of Philosophy, Part Four, Article 205).
As I have stated, philosophers reacted to the repeated failure to extend infallible knowledge past some limited domain by becoming fallibilists. They broke down the distinction between knowledge and probability and moral certainty and instead treated knowledge as including infallible knowledge and fallible knowledge. (A notable exception is G. E. Moore, who insisted that he had "certain knowledge" of matters about which nearly everyone else was uncertain.)
Security
The issue between the infallibilists and the fallibilists comes down to security. The infallibilist view is that safety in the relevant sense is binary: our beliefs are either safe or unsafe. In Plato's image, either the statues are tied down so that they cannot fly away, or they are free to fly away. The importance of knowledge, for Plato, rested in its security. Knowledge is something we can count on. I believe that this is the core of the understanding of knowledge from Plato to the present day.
In the Meno, Plato's concern was the transitory nature of opinion. Knowledge is secure in the sense that it is tied down in such a way as to last. Later epistemologists, such as the Stoics, downplayed or ignored the temporal aspect of security in favor of a modal aspect relating to truth. Security is understood in terms of the possible falsehood of belief.
Fallibilists believe that there are degrees of security, and in particular that we can count on our beliefs to a greater or lesser degree. When we can count on them well-enough, they can be called knowledge. In my view, what fallibilists have done is to allow that Cartesian "moral certainty" is good enough for knowledge. Of course, determining the "tipping-point" between moral uncertainty and moral certainty is a difficult problem, which has given rise to much disagreement. It is not a problem afflicting infallibilist accounts of knowledge.
Security issues can generally be handled in an "externalist" way. If a belief is the result of rational intuition of the truth or a logically valid deduction from what is rationally intuited, it is secure. If a belief is caused by its object in an "appropriate" way, it is secure, though perhaps not infallibly so. The same holds for a belief formed by a reliable belief-forming process. Generally, the way in which a belief is produced might by itself provide the kind of security that is demanded for knowledge.
As far as fallible knowledge or "moral certainty" is concerned, attention to issues of security favor naturalism. Aristotle held the view that human beings (and other animals) are naturally suited to detect crucial features of the environment in which they exist. Descartes himself had held that humans (though not other animals) are equipped by God with sense-organs which provide us with very secure (though not absolutely secure) information vital to the well-being of the body (Sixth Meditation).
For this reason, I think that modern "naturalist" epistemologists are returning to the classical roots of the theory of knowledge. They are interested in the scientific study of how human beings, primates with highly-developed brains, receive and process information from their surrounding environment. If we understand how this is done successfully, that is, how features of the environment are correctly detected, then we understand why the information encoded in our beliefs is as secure as it is.
Animal Knowledge
You may have noticed that in the discussion of knowledge as secure belief I did not once mention "justification." Most epistemologists these days think that "justification" is a necessary condition of knowledge. Why not say, as Goldman did in his 1963 paper, that some of our knowledge is simply the product of the interaction of objects with our sensory perception? In a way, this is more faithful to Plato than most modern theories of knowledge, despite its fallibilism.
Sosa distinguished "animal knowledge," which is a "direct response" to the impact of the world upon us "with little or no benefit of reflection or understanding" from "reflective knowledge" (p. 440). I think that animal knowledge is an indispensible part of any comprehensive account of fallible knowledge of the material world. More specifically, animal knowledge seems well-suited as the proper "foundation" for more distinctively human, "reflective," knowledge.
My own inclination is to think that animal knowledge is truly distinct from other things that have been called "knowledge," due to the fact that it has no reflective element. For this reason, I claim that it is misleading to say that there is justification in animal knowledge, because I think that justification as we ordinarily understand it involves reflection.
If I am right about this, then much of the motivation for coherentism disappears. Much of the appeal of coherentism lies in the view that justification cannot emerge, like a rabbit out of a hat, from what is not justified. This claim stems, I think, from the idea that justification is a product of reasoning: you justify something by appeal to something else. That something else has to be some other belief, the coherentist maintains.
I think all this talk about justification obscures the real issue, an issue that Aristotle recognized very clearly. What is important is the dependence of one item of knowledge on another. So if there is animal knowledge, it may serve as a foundation for other, "reflective" knowledge.
Nobody seems to want to take a hard line on animal knowledge, however. Goldman has backed off from his causal theory to the point where he regards reliable belief-formation as a kind of justification, "strong" justification. Sosa tells us that "No human blessed with reason has merely animal knowledge of the sort attainable by beasts" (p. 440). He tells us that reason is a "silent partner" in all belief. Further, he regards animal knowledge as being the product of intellectual virtue, and as such as being justified in something like Goldman's "strong" sense.
It seems to me that the distinction between strong and weak justification is just like the distinction between different kinds of knowledge (i.e., relatively secure belief). They are not two specific types of the same kind. Their functions are very different. Strong justification is perfectly suited for the security aspect of knowledge. Weak justification does another job.
Alvin Plantinga (Warrant: The Current Debate, 1993) has come up with a good way of dealing with the situation. He uses the term "warrant" to stand for whatever it is that yields knowledge when added to true belief. In that case, animal knowledge has warrant, either through the causal connection with the object, through the exercise of intellectual virtue, or by being the result of a reliable process. There is only one kind of justification, and it is "weak."
Another promising approach is due to Timothy Williamson (Knowledge and its Limits, 2000). He thinks that knowledge cannot be analyzed into true belief with justification, warrant, or anything else added to it. His reasons for this claim are too complex to go into here. The important thing is that he does not hold that all knowledge requires "justification." (Williamson's thinking is behind my remarks about animal knowledge as a foundation for reflective knowledge.)
Justification
What is the proper role of justification in the context of knowledge? As I said before, I think that it central to "reflective" knowledge. Gettier had held that the "traditional" analysis of knowledge included justification along with true belief. Scholars have noted the resemblance between that analysis and an account of knowledge examined by Plato in the Theaetetus. Knowledge, Plato suggested, is true opinion accompanied by an "account" (logos).
The chief importance of having an account is that it gives us reason to feel secure in what we believe. Moreover, being able to give that account to others helps us to evaluate the level of our security. And it gives other people assurance that our own belief is secure. The account, the justification, "internalizes" the security that is fundamental to knowledge by placing our belief in what Sellars called "the logical space of reasons."
Much of modern epistemology was influenced heavily by Descartes. His aim, in the Meditations, was to provide an account of his own beliefs such that he would be able to recognize the extent to which he could trust them. This required him to examine the major possible ways in which his beliefs could turn out to be false. Having found a set of infallibly secure beliefs, he then tried to build a superstructure of knowledge upon them.
Descartes held that until this was done, he could not feel completely secure in what he believed. He recognized that whatever he believed could be overturned by some new considerations if he had not established, from the inside, the fact that God exists and is not a deceiver.
Descartes made a further claim: until these two facts have been established, one fails to have knowledge at all. His reason is that "no act of awareness that can be rendered doubtful seems fit to be called knowledge" (Second Set of Replies to the Objections to the Meditations, AT 141). Not only must knowledge be secure in that the "act of awareness" involved cannot yield a falsehood, but it must also be secure in the sense that it cannot ever be called into question by any new information.
This requirement is, I think, the main basis for the modern view that knowledge requires justification. (I shall call it "internal security," but compare Sosa's "project of validation, in "The Coherence of Virtue and the Virtue of Coherence," 1985). Descartes laid down the requirement in the context of an infallibilist account of knowledge, but it can easily be transposed into a fallibilist format. This comes out negatively the "no-defeaters" clause that is worked into most accounts of knowledge. Goldman builds such a clause even into his notion of "strong" justification, which otherwise is "externalist" (p. 395).
Positively, the claim comes out as a requirement that one have a story to tell one's self about why the belief is true. Keith Lehrer, for example, says that in order to know, you must be able to answer the question "how do you know?" This calls for reflection, and in effect it reduces all knowledge to reflective knowledge.
Skepticism
One of the main effects of the "internal security" demand is that it brings skepticism into play. To assure one's self of the security of one's beliefs, one must be able to "answer the skeptic." But because the resources for giving the answer is purely internal, answering the skeptic regarding the security of beliefs about external matters seems problematic, if not impossible.
Reflective Knowledge
I have indicated that I think justification goes hand-in-hand with reflective knowledge. As Sosa defines it, "one has reflective knowledge if one's judgment or belief manifests . . . understanding of its place in the wider whole that includes one's belief and knowledge of it and how these come about" (p. 440). Quite rightly, I think, Sosa recognizes that any account of the understanding of the place of belief in the wider whole Sosa describes is going to have to be a coherence account. (See, for example, "Intellectual Virtue in Perspective," 1991. "Reflective justification, our best reflective intellectual procedure, is a matter of perspectival coherence -- and necessarily so.")
An argument to this effect due to Lehrer is as follows. Reflective knowledge of matters external to our minds, requires the use of "bridge" principles (as Stroud calls them) which connect the internal contents to these external affairs. The justification of these principles will require appeal to information about the very kind of external matters which are supposed to be justified by the principles.
I do not think that such appeals are viciously circular. The reason is that no particular piece of information about the external affairs is essential to the justification of the general principle. You can remove it, like a leg on a table, and the general statement would remain justified. So no particular belief is required for its own justification.
A second criticism of coherentism is that justification understood as coherence does not require any external input. If there is animal knowledge, however, it could function as input into one's belief system. Granted that one may not believe that one has animal knowledge, but just having animal knowledge generates beliefs, and these beliefs will figure into the coherence of the whole belief set.
A final point is that I find the notion of "reflective knowledge" not to be as cut-and-dried as Sosa presents it. The range covered by reflection can vary considerably. I do not think that anyone is fully reflective, i.e., takes the entire "wider whole" of one's belief-system into account when reflecting on the truth of a belief. We ordinarily cherry-pick what we think is relevant.
Ampliative Justification
Meeting the Cartesian "internal security" demand is not the only motivation for justification of belief. Its main purpose, I think, is to give us a way to "amplify" our evidence. If I am planning what to wear tomorrow, I might form a belief that it will be hot outside. This belief will be justified by my past experience as well as other beliefs I have about the way that experience can be projected onto the future. Scientific beliefs that go beyond mere observation similarly require justification if they are to be knowledge. But this kind of justification is not a condition for animal knowledge.
Epistemic Responsibility
Justification as the establishment of "internal security" is at the heart of the view that justification is a normative concept which involves the further concepts of responsibility and irresponsibility. This comes out in various ways. One demand might be that one is not entitled to hold a belief unless its internal security has been established. In practical terms, this seems far too much to ask for all justification, and I don't even think Descartes ever required it.
A weaker demand is that one is entitled to hold a belief so long as the person has not found it to be insecure, or perhaps that there is nothing in the person's beliefs and experiences that would make the belief insecure. This is tantamount to the "no-defeater" clause, and it is characteristic of "prima facie" justification.
I don't subscribe to this as a blanket principle, because I think it applies only to animal knowledge. Perhaps epistemic responsibility in reflective knowledge should be assigned in degrees, just as we assign degrees of responsibility to actions. (That was a very irresponsible thing to do.) For reflective knowledge, we seem to require some notion of "due diligence." What this is in practice would probably vary widely from case to case. Some epistemic contexts, such as highly theoretical investigations in physics, would demand great diligence, while other contexts, such as counting how many pens are on the desk, would not require much.
As I have intimated, I think that degree of epistemic responsibility we demand depends on considerations of security. We do not raise the bar very high when we think that we are quite secure. We are quite content to let our guard down and undertake actions that in other circumstances would be considered quite risky.
Conclusion
I have argued that the main problem with the theory of knowledge is that a single concept is supposed to explain too many things. This is a main reason why there is no agreement at this point, despite 2,500 years of reflection on knowledge.
The cause that I believes accounts for the protean character of the concept of knowledge is that it is ultimately one that is created for human purposes. Here I disagree with Plato's primary assumption that it denotes something independent of human thought. Knowledge is not a concept like that of food or water. We could get along without it, I believe. I see no reason to suppose that the concept denotes any mind-independent reality, except for a general claim that all general concepts do so.
What most people have been interested in when they have described knowledge is some kind of security that opinion or mere belief lacks. Security is a relative concept: things may be more or less secure. What counts as secure in one context does not count as secure in another. We are are sometimes willing to attribute knowledge and sometimes willing to deny knowledge of the same person with all the same information and relation to the environment. It may simply depend on why the attribution of knowledge is being considered. Security is relative to purposes.
There are times when it is important to discover just how secure one is. This is when we call for justification. Not all situations require this. In many cases, it is important simply that belief be secure, not that one be able to determine one's level of security. This is why externalism has a good deal of appeal. It may well be that justification must ultimately be understood in the internalist way, but I have argued that justification is not a necessary condition of knowledge.
The End