Lecture Notes, UC Davis Philosophy 102, Theory of Knowledge


G. J. Mattey, Senior Lecturer

Picture of Barry Stroud

Barry Stroud

Text: Barry Stroud, "Understanding Human Knowledge in General" (1989)

Barry Stroud, professor at UC Berkeley, is one of the very few living philosophers who believes both that skepticism should be taken seriously and that it remains a threat to the success of the theory of knowledge. In the reading for today, Stroud puts forth a conception of what the theory of knowledge is supposed to accomplish and mobilizes a general skeptical argument against the possibility of its success.

The Goal of Epistemology

The study of human knowledge, Stroud tells us, is supposed to yield an understanding of "what human knowledge is and how it comes to be." But Stroud voices pessimism about whether we can ever get "satisfaction" on these points. There are two ways in which we will be unsatisfied. The first is if skepticism is right, and "we do not know the things we thought we know." The second is that even if skepticism is wrong, and we do know those things, "we cannot see how the state we find ourselves in is a state of knowledge." The entire paper is an extended argument in support of this dilemma.

This line of thinking is resisted by those who espouse naturalism in epistemology. The naturalists ignore skeptical considerations, on the reasonable grounds that human knowledge is the proper object of study by psychology, "cognitive science," and the like. Once we get "a successful theory that explains how we do in fact know the things we do," the skeptic can be defeated.

Stroud's contention is that the study of human knowledge is different from the study of natural processes such as "human digestion." We can learn a great deal about "how we know things," or "this or that aspect of human knowledge." But this may not satisfy us with respect to the main point, which is to "understand human knowledge in general, and to do so in a certain special way." This is not merely to say that the task is very difficult and will never be finished, which is the case.

Skepticism, "the view that we do not, or perhaps cannot know anything," is important to epistemology because it seems to be an unavoidable consequence of the attempt to do epistemology in the way that Stroud will describe. If we think skepticism is incorrect, then that casts doubt on the way of pursuing the account of knowledge. Yet philosophy does not question the goal when it runs into trouble, as "we feel that human knowledge ought to be intelligible in that way." For this reason, we do not acknowledge that there is no satisfactory solution to the problem, and "so we deny the force, and even the interest, of skepticism."

Now Stroud explains in more detail what this understanding of human knowledge is supposed to look like. He had stated that it should be "completely general." This could be understood in two ways. The first is to "understand how anything we currently accept amounts to knowledge." The second, less ambitious, goal is to understand in a completely general way "how we come to know anything at all in a certain specified domain."


Three examples of specified domains are given:

We want to know in each case how anyone can know anything at all about bodies, other minds (what they are thinking and even whether they exist), and what lies beyond the pale of observation, respectively.

One way of getting at the generality of the questions is to see what would not serve as an answer. In the case of knowledge of other minds, it would not do to find out what X is thinking by asking X's good friend Y. This is, to be sure, a way of knowing, "but in fact we seek a more inclusive description of all our ways of knowing that would explain our knowledge in general."

The problem is not the fact that asking a friend is the only way of knowing what someone else thinks. It lies in the fact that it appeals "to knowledge of some other fact in the same domain." In the example, an appeal is made to what is going on in Y's mind. But our knowledge in general of the contents of others' minds cannot be explained by appeal to the contents of another's mind. The same holds for explaining our knowledge of one fact about bodies by reference to another fact about them. "It could not answer the philosophical question as to how I know anything about any objects around me at all."

The Basic Problem

When we want to explain knowledge in one domain, we do so by appeal to "information available to us for knowing things" in that domain. There are "apparently simple, problem-generating moves" that confront us from the start, as we realize that the information is not by itself sufficient for knowledge. Descartes showed that sensory information is not an adequate basis for knowledge of material objects. Our knowledge of other minds could only be based on information about bodily behavior, but this does not by itself amount to knowledge, either. And how could information about what is observed necessarily yield knowledge of the unobserved?

What is common to these three cases is that the information appealed to comes from outside the domain which is supposed to be explained. We must not assume that we have knowledge of what lies within the domain, since then there would be no explanation. So we have to appeal to other knowledge from outside the domain. And this can help us in our explanation only if there is some kind of "bridge" between the two, an "inference or transition."

Herein lies the problem. The bridge either lies entirely within our evidential base or it goes beyond the base. If it is restricted to the base, then it fails in the goal of getting us beyond the base. But if it goes beyond the base, then it must not intrude into the domain in question, since that would violate the generality of the explanation. (A third alternative Stroud does not mention is that it goes beyond the evidential base in a way that does not presuppose knowledge of the domain.)

This familiar quandary arises from the demand for a general account of our knowledge. We get this generality, it is assumed, only by appeal to some "prior kind of knowledge that does not imply or presuppose any of the knowledge we are trying to explain." But Stroud thinks that this appeal to "epistemic priority" is what dooms epistemology to be unsatisfying.

It is fashionable to blame the Cartesian "quest for certainty" as the root of skeptical problems. If we abandon that quest in favor of fallibilism, it is argued, the skeptical problem will go away. But the search for a "foundation" for knowledge is based on the attempt to answer the general question of epistemology, not to provide some kind of security against error.

As an illustration, Stroud notes that the usual evidence for the contents of other minds, behavior, does not have to be taken as certain knowledge. It only must not presuppose knowledge of the contents of those minds. The same holds for the evidence that is advanced to explain our knowledge of physical objects. Call it "sensations," "sense-data," or "experience," it must be "prior" to our knowledge of physical objects, and in that way it serves as a "foundation" for it.

Having stated this, Stroud returns to his original skeptical claim. When we restrict our "foundation" in this way, the question arises as to how we possibly could move out from it. "It would seem possible only if we somehow knew of some connection between what we are restricted to in observation and what is true in the wider domain we are interested in." In that case, we would have overstepped the boundary, "knowing something about that wider domain after all, not just about what we are restricted to in observation." And generality would then be violated.

So we are back to the dilemma noted at the beginning. If we take the generality of the explanation seriously, we are left with skepticism. The only way we can avoid it is to abandon the project of finding a satisfactory general account of knowledge in the given domain.

Epistemic Principles

One way to deal with the problem is to turn it on its head, so to speak. Admit that we need bridge principles to get us from one domain of knowledge to another, but claim that this relation is precisely what justifies the principles. "The fact that certain 'postulates' or 'principles' can be shown to be precisely what is needed for the knowledge in question is somehow taken to count in their favor." The reason is that they are needed for us to know what we think we know.

Stroud objects that this strategy is of no value as a solution to the epistemological problem he is presenting: the problem of explaining knowledge in general. To solve that problem, it must be shown that the "epistemic principles" in question are themselves known to be true, and how this is known. And this thrusts us back on the earlier dilemma for epistemic priority.

Independent Knowledge

Epistemic principles are supposed to tie together two kinds of things that we know. But Stroud asks how we know that they are true. He has generated a "regress" that he thinks cannot be satisfactorily resolved. One way to try to stop the regress is to claim that there is some way of knowing that the epistemic principle is true that is independent of any knowledge with respect to the domain to which it is a bridge.

Hume, for example, had held that the inductive principle which forms a bridge between the observed and the unobserved can only be justified by facts about the relation between the observed and unobserved. So, in Stroud's terms, there is no relation of epistemic priority here, and the inductive principle could not explain our knowledge of the unobserved. Kant's solution was to claim that we can have a priori knowledge of the required bridge principles.

Appeal to independent knowledge only generates a new regress, however. Anyone proposing this kind of solution must, to satisfy our desire for general explanation, show how it is that we know that the epistemic principle is true. "It has to be explained how we know anything at all a priori, and how in particular we know those very things we need for empirical knowledge."

Answering this question thrusts us back on the original dilemma. If there is an evidential base for our knowledge of the a priori, then either it contains that knowledge or not. If it does contain that knowledge, we violate generality, and if it does not, then it seems that skepticism is unavoidable. So no appeal to an evidential base is made, and the knowledge is taken as a brute fact. But this is less than satisfying.


A different, so-called "enlightened" approach is to lower the bar and require less in our understanding of knowledge. To move from evidence to knowledge in a domain, we need "principles" connecting the evidence to information in the domain. The skeptical problem arises when questions are raised about whether we have knowledge that the principles are true or good reasons to believe that they are. The problem can be avoided by dropping the requirement one must be able to support the principles.

According to this view, requiring knowledge of the truth of the principles is tantamount to requiring that in order to know something, one must know that he knows it. But one can know many things without knowing that he knows it, the objection goes. "The theory of knowledge asks simply whether and how people know things. If that can be explained, that is enough."

Stroud agrees that it "seems right to allow that someone can know something even when we recognize that he does know that he knows it." An example from Colin Radford is given. Someone recalls the answer to a question without being able to say whether he had actually remember it. In that case the person knows more than he thought he did.

This apparent "fact about our everyday assessments of people's knowledge" is then applied to the philosophical quest for understanding of our knowledge. Specifically, it implies that there only need be a "bridge" connecting our evidence to the domain in question, and not any good reason to think that there is such a bridge. The "bridge" is built right into the definition of knowledge in the given domain.

For example, Peter Strawson has argued that in the case of induction, the possession of a principle that correlates past observations with future events is just what it is to have reasonable beliefs about those future events when the observations have been made. "You do not have to find some additional reason for thinking that what you have observed in the past gives you good reason to believe something about the future." An external observer might have good reason to believe that the bridge principle is true, but the person using induction may gain knowledge of future events without such a reason.

Criticism of Externalism

Stroud finds that externalism "is not all smooth sailing." He lists some specific problems for Strawson's proposal. To satisfy the philosophical demand for explanation of our knowledge, it must establish that we really know that these events will occur in the future. It must be able to distinguish knowledge from lack of knowledge.

But it can do so only if it can distinguish "law-like" correlations from "accidental" correlations. That is, it must establish when the past observations indicate an enduring connection and when they do not. This is a difficult task. Moreover, it faces the "new riddle of induction" (due to Nelson Goodman) which makes it plausible to think the task cannot be completed.

A similar criticism can be given of externalist accounts of knowing that p, say as "equivalent to something like having acquired and retained a true belief that p as a result of the operation of a properly functioning, reliable belief-forming mechanism." It is difficult to spell out what this means.

A deeper problem, though, is that even if one were to find an externalist analysis that matched up precisely with what it is we think we know, it would be unsatisfying. This is due to the most profound aspect of the generality we are seeking in epistemology. It is the reflexive character of the investigation of human knowledge: it is we humans who are both the objects of investigation and those undertaking it. "We want to be able to apply what we find out about knowledge to ourselves, and so to explain how our own knowledge is possible."

Back to Descartes

If we apply our results about knowledge to ourselves, can we get a satisfactory outcome? Stroud invites us to return to Descartes, who tried to pull off just such a project. His strategy was to prove that God exists and is no deceiver. Since God created us in such a way as to know the truth, we can count on our clear and distinct perceptions. "If I cam careful and keep God and his goodness in mind, I can know many things, and the threat of skepticism is overcome."

An externalist can criticize Descartes's strategy on the grounds that it requires knowledge of a "bridge" principle, that whatever I perceive clearly and distinctly is true. This in turn requires knowledge of God's existence, etc. Descartes himself denies that an atheist can have "true knowledge," while he, who knows that God exists, can have it.

The fact that an atheist can be "clearly aware that the three angles of a triangle are equal to two right angles" is something I do not dispute. But I maintain that this awareness of his is not true knowledge, since no act of awareness that can be rendered doubtful seems fit to be called knowledge. Now since we are supposing that this individual is an atheist, he cannot be certain that he is not deceived on matters which seem to him to be very evident (as I fully explained). And although this doubt may not occur to him, it can still crop up if someone else raises the point or if he looks into the matter himself. So he will never be free of this doubt until he acknowledges that God exists." (Second Set of Replies to the Meditations, AT 141)
The externalist holds that requiring this pre-requisite knowledge is a requirement that in order to know, one must know that one knows.

There is also a problem of internal coherence for Descartes. It would seem that he has knowledge of God's existence only insofar as he clearly and distinctly perceives that God exists and knows that whatever he clearly and distinctly perceives is true. But he knows this latter fact only because he knows that God exists. This is the problem of the "Cartesian circle."

Stroud rightly points out that the problem of circularity does not arise for an externalist. One can know what one knows without knowing that whatever he perceives clearly and distinctly is true, and so without knowing that God exists. "As long as God did in fact exist and did in fact make sure that his clear and distinct perceptions were true, Descartes would have the knowledge he started out thinking he had, even if God's existence and nature remained eternally unknown to him."

In fact, we can generate a Cartesian version of reliabilism. God's guarantee of the truth of clear and distinct perceptions makes intellectual perception a super-reliable source of true beliefs. Moreover, it explains why the source of our belief is reliable, unlike generic reliabilist theories. So it seems that this Cartesian externalism would as acceptable than generic reliabilism if not more so.

The glitch is that most of us do not buy into the story. Many people are atheists, and even believers may not grant that God guarantees the truth of any particular kind of human perceptions. But this is hardly the only defect in the "externalized" Cartesian account of human knowledge.

The problem from the standpoint of traditional epistemology is that we want to know that the account of the reliability of our beliefs is a true one. Without this knowledge, we are no closer to understanding how human knowledge is possible. But Stroud thinks that no version of externalism can bring us any closer.

Descartes's real account of knowledge, as we saw, appears to get him into a circle from which he cannot emerge. The "externalized" Descartes escapes the circle, but he does so at a price: not giving a satisfactory explanation of human knowledge. If the story about God is false, then nothing is explained. So in order for the explanation to be satisfactory, he must have good reason to think that his explanation is a good one. But this is just what the externalist denies he must have.

The Naturalistic Epistemologist

A modern "naturalized" epistemologist is no better off than the "externalized" Descartes. The modern epistemologist would study the human being scientifically to discover those mechanisms of belief-formation which are reliable. With this information in hand, he would proclaim to have explained human knowledge.

But our naturalized epistemologist has failed in fact to have provided a satisfactory explanation of human knowledge, according to Stroud. The reason is that the scientific knowledge which is the basis for our claims of reliability is itself subject to investigation for its own reliability.

Of course, externalism can be applied to this demand as well. So long as he has carried out his scientific investigations in a reliable way, the investigator's beliefs formed on its basis are reasonable. After all, he has found similar beliefs to be reliably formed in other people, and he thinks that he is in the same position as they are. "He is one of the human beings that his theory is true of." He can also claim that his own theory is correct and that of Descartes is wrong. "In accepting his own explanation he claims to filfill the conditions his theory asserts to be sufficient for knowing things."

Stroud counters that this naturalized epistemologist would be no better off than the "externalized" Descartes. Both are in a position where if their own theories are true, and they meet the conditions of the theory, then they have the knowledge they claim. But this is of no help to them. Both need to have a good reason to believe that their theories are correct, but this is precisely what externalism says they do not need in order to have knowledge.

It does no good for the externalist to move up a level and say that he has good reason to think his natural or supernatural theory is correct because he arrived at his belief that the theory is correct in a reliable or God-guaranteed way. This would just drive the problem up to a higher level. It is hard to say just what the externalist needs here, except that it is something that externalism cannot provide.

Wondering Whether we Understand

The bottom line is that the externalist in the end has to admit that he does not know whether he understands human knowledge or not. If his theory is correct, he does, while if it is incorrect, he does not. "I wonder whether I understand human knowledge or not." Stroud concludes that such a statement is not a satisfactory conclusion of any study.

Perhaps it can be said that there is some kind of "unwitting" understanding going on in the mind of the externalist. If there were such a thing, it would be the most that an externalist theory could produce. "But even if there is such a thing, it is not something it makes sense to aspire to, or something to remain content with having reached, if you happen to have reached it." We want "witting" understanding, which requires something that we cannot provide, since something would always be left unexplained (if skepticism is avoided).

There is nothing wrong with the every-day presumption that we can explain "how people know certain things about the world by assuming that they or we know certain other things about them. We do it all the time." But this will not do for "the general epistemological enterprise." We need a reason to accept the externalist account as being true of ourselves if we are to give a general account of human knowledge. "It is only with self-understanding that the unsatisfactoriness or loss of complete generality makes itself felt."

The ultimate question is whether we can take an "external" position of an observer of ourselves and our knowledge, while at the same time giving a satisfactory general explanation of how we know the things we do. "That is where I think the inevitable dissatisfaction comes in." If we seek to detach ourselves from the domain in the effort to attain generality, "it seems that the only thing we could discover from that point of view is that we can never know anything in that domain," for lack of "bridge" principles. "This is the plight the traditional view captures. That is the truth in skepticism."

The truth in "externalism" is that we can explain other peoples' knowledge. But to apply this knowledge to ourselves requires that we place knowledge of ourselves in the domain to be explained, which prevents complete generality in our explanations.

In conclusion, Stroud urges us "to go back and look more carefully into the very sources of the epistemological quest." If we discover that there is a flaw in the project of giving a general account of human knowledge, we will see that we cannot have a satisfactory explanation of how human knowledge in general is possible. "And that too, I believe, will leave us dissatisfied."


In a 1994 paper, "Scepticism, 'Externalism', and the Goal of Epistemology", (Proceedings of the Aristotelian Society: Supplementary Volume LXVIII, 1994 pp. 291-307), Stroud elaborates a bit on his advice to look back to the sources of the project of epistemology. He suggests that it

lies somewhere within the familiar and powerful line of thinking by which all of our alleged knowledge of the world gets even temporarily split off all at once from what we get in perception, so we are presented with a completely general question of how perception so understood gives us knowledge of anything all in the physical world. (p. 306)
The gist of this comment seems to be that if there is no split to begin with, there will be nothing to explain or understand about where our knowledge of physical objects comes from. An explanation is called for only if we take knowledge of perception to be "epistemologically prior" to knowledge of physical objects. But even if we take knowledge of perception and of physical objects to be cut from whole cloth, there still seems to be a question of understanding this unified kind of knowledge.

[ Course Home Page | Menu of Lectures ]