Tetsuji Iseda

Abstract

Frederick Suppe argues that skeptical arguments from Descartes to Feyerabend are based upon the KK thesis, and therefore are not worth worrying about. It seems to me that his argument is unwarranted, and the purpose of this paper is to reconstruct skeptical arguments without the KK thesis. Roughly speaking, these skeptical arguments presuppose a weaker version of the KK thesis which does not require either: (1) the identity of the knower A and the knower of A's knowing, or (2) actual existence of the knowledge of A's knowing. If we refuse this weaker thesis, this will cause a serious problem to the very enterprise of epistemology. Surely Suppe should take skeptical doubts more seriously.

1. Introduction

In his forthcoming book, Facts, Theories, and Scientific Observation, Frederick Suppe dismisses traditional skepticism as follows:

"The collapse of the Cartesian and Empiricist epistemological programs into pervasive skepticism has been one long series of dismal corollaries to the level confusions inherent in Descartes' uncritical acceptance of the KK Thesis. Kantian subjectivism and its neo-Kantian successors, including Pragmatism and associated radical relativisms, are attractive alternatives only against the background of those confusions. These level confusions naturally led epistemologists to conflate the evidential basis for knowledge with reasons given in justification for knowledge and philosophers of science to conflate the reasoning given in scientific articles with the belief formation process used to obtain scientific knowledge." (Suppe forthcoming, Part II chapter 5, section 5)

In short, Suppe claims that the majority of traditional philosophers have kept making level confusions by accepting the KK thesis, and this is the only reason skepticism seemed so forceful to them. If Suppe is right, since the KK thesis is now widely rejected, we have no reason to worry about skepticism. But I think that Suppe's claim is unwarranted.

2. The KK thesis and what is wrong with it

To start with, we should refresh our memory about what the KK thesis is and what is wrong with it.

the KK thesis is is supported by several epistemologists. Jaakko Hintikka (1962) is the most famous recent defender, and he formulates the thesis as follows:

(1) Kap -> KaKap,

namely, "if a knows that p, then a knows that a knows that p" .

There are several arguments against the KK thesis. First of all, even though Hintikka 'proved' the thesis from several axioms of epistemic logic (Hintikka1962, 104-105), his proof was circular because the axioms used in the proof implicitly presupposed the KK thesis (Chisholm, 1963, 784-787). Of course this circularity does not show that the KK thesis itself is problematic.

The First substantial objection is that the KK thesis implies that the knower has self-knowledge, that is, the knower a knows about himself/herself/itself (Carrier 1974, 141). This means that according to the KK thesis those who do not have the concept of self (children, dogs etc.) have no knowledge whatsoever. This sounds counterintuitive.

Hilpinen's (1970) argument is also illuminating. Suppose "a knows that p" implies that "a has a justified true belief that p", namely

(2) Kap -> Bap & Jap & p,

where Bap means "a believes that p" and Jap means "a is justified in believing that p". Suppose also that the distribution law about knowledge holds, namely

(3) Ka(A&B) -> KaA & KaB.

(2) and (3) are widely admitted among philosophers. Then from (1), (2) and (3), we can infer as follows:

(4) Kap -> KaKap

-> Ka (Bap & Jap & p)

-> KaBap & KaJap & Kap

-> BaBap & JaBap & Bap & BaJap & JaJap & Jap &Bap & Jap & p.

There are several dubious conjuncts on the right side of (4), but the most dubious one is the term BaJap, namely "a believes that a is justified in believing that p". Does this follow from "a knows that p" (this is what (4) implies)? Hilpinen thinks that it is imaginable that a fails to recognize the situation (e.g. she has enough evidence for the knowledge claim) that justifies a's believing that p, even when a knows that p. This is an uncomfortable consequence of the KK thesis.[1]

Hintikka's reply to these challenges is that he is talking about a strong notion of 'knowing' (Hintikka 1970). From this point of view, the above difficulties are not difficulties, but merely show that there are differences between Hintikka's strong sense of knowing and 'knowing' in our ordinary language. For example, the first difficulty with self knowledge disappears by asserting that children and dogs do not have this strong sense of knowledge. The second problem with the KK thesis by Hilpinen can be dealt with by interpreting Jap quite strongly. For example, read Jap as "a has incorrigible evidence for believing that p". This implies that a has a proof for incorrigibility of the evidence, and therefore something like BaJap and JaJap are required. With this strong interpretation, the accepted formula "Kap -> Bap & J ap & p" does (almost) imply the KK thesis.[2]

Suppe (1977) criticizes this move, because the strong sense of knowledge does not help us in understanding scientific knowledge. He enumerates four features of scientific knowledge which are not suitable for the KK thesis. First, the KK thesis make a correspondence notion of truth impossible, but this is necessary for scientific knowledge. Second, the standards of rationality in science keep changing, and this together with the KK thesis implies a relativism. Third, science's procedure to assess knowledge claims is fallible, therefore the incorrigibility does not apply to science. Finally, the KK thesis make sophisticated observational knowledge impossible, because scientific observations depend on background knowledge, which is not incorrigible.

I totally agree with Suppe that Hintikka's strong sense of knowing does not help us much in philosophy of science (though I disagree with him in that the rejection of the KK thesis eliminate the skeptical doubt about correspondence notion of truth and the relativism on the scientific rationality). I think that this is a good reason to reject the KK thesis, so if the KK thesis is essential for skepticism, we also have a reason to reject skeptical doubts.

3. Suppe's argument against skepticism

In chapter 10 of his 1989 book, Suppe argues that skeptical arguments by Berkeley, Hume, Kuhn, Feyerabend and others "depend essentially on the KK thesis" (332). What is the argument for this claim? In the first section of this chapter Suppe provides an overview of the history of skeptical arguments by empiricists up to the logical positivism, but Suppe does not mention the KK thesis here. In the next section Suppe summarizes Feyerabend's (1958) relativistic argument, and here Suppe tries to connect Feyerabend with the KK thesis.

Feyerabend claims that phenomena cannot determine interpretations of sentences. Feyerabend's argument goes as follows. If an observer O utters a sentence S because he observes a phenomenon P and recognizes that S 'fits' P, this relation of phenomenological adequacy between P and S is another phenomenon, P'. But O can recognize P' by relating it to S', and this calls for P'', S'', and so on infinitely. Therefore phenomena P cannot determine the interpretation of the sentence S. Rather, we need somewhat arbitrary assignments of interpretations. (Feyerabend 1958, 155; see also Suppe 1989, 312). Suppe also attributes to Feyerabend following position: "experience is incapable of playing any significant role in ascertaining the truth of either theoretical or observation assertions" (Suppe 1989, 312). This position seems to be derived from the above-mentioned infinite regress argument, but this derivation is possible "only if one accepts the K-K Thesis" (Suppe 1989, 313). His argument for this claim is worth quoting:

"For if this thesis is accepted, then from the fact that experience cannot show that such a correspondence holds and from the vaguely plausible assumption (tacit in his discussion) that such a correspondence could not be known except if displayed in experience, it would follow that one cannot know that P is true; hence one cannot know that he knows that P; hence via the K-K Thesis, one cannot know that P. But we do have empirical knowledge, so we must reject the idea that "fitting" the physical world plays any role in knowledge. But this argument (and I can think of no other one compatible with what he says that will serve his purposes) depends essentially on the K-K Thesis, for if it is denied, the requirement of truth by correspondence is totally compatible with Feyerabend's other claims: One can experientially know that P, where P " fitting" or corresponding to the physical world is a condition of knowledge, but one cannot know that one knows that P." (Suppe 1989, 313; italics in original)

This is the most elaborate argument to connect the KK thesis and skepticism (or relativism) that I can find in Suppe's book. What does this argument show? First, Suppe reconstructs a plausible line of Feyerabend's argument using the KK thesis (thus the KK thesis is a part of the sufficient condition for Feyerabend's relativism). The argument goes as follows:[3]

F1. Suppose that a correspondence between a sentence and physical objects play an essential role in knowledge.

F2. Experience cannot show that a correspondence between P and physical objects holds (because of infinite regress).

F3. Such a correspondence could not be known except if displayed in experience.

F4. If S knows that P, then S knows that S knows that P. (the KK thesis)

F5. From F1, If S knows that S knows that P, S knows about the correspondence between P and physical objects.

F6. From F2 and F3, S does not know about the correspondence between P and physical objects.

F7. From F5 and F6, S does not know that S know that P.

F8. From F4 and F7, S does not know that P.

F9. Since we do have knowledge, assumption F1 is false.

Then he shows that in this line of argument the KK thesis is essential. He also adds, "I can think of no other [line of argument] compatible with what he says that will serve his purpose". That is all. This is supposed to show that the KK thesis is a necessary condition for Feyerabend's relativism! Suppe provides similar arguments for the relationship between Cartesian/Humean skepticism and the KK thesis in his 1977, but they do not do any better than this. He constructs lines of argument for Cartesian skepticism and Hume's skeptical argument against induction using the KK thesis, and shows that the KK thesis is essential in these lines of argument.[4]

Despite this obvious gap in inference, he asserts that "the KK thesis has showed up at virtually every crucial juncture as empiricism has proceeded toward complete capitulation to the skeptic" (Suppe 1989, 332). From this, he arrives at this conclusion: "a viable epistemology of science, where experience plays an essential role in knowledge, must begin by denying the K-K thesis, thereby rendering the Cartesian skeptic's challenges question-begging" (332).

Thus, to undermine Suppe's claim, all I have to do is to show that there are plausible alternative interpretations of skeptical arguments which do not appeal to the KK thesis. I don't have to even show that my interpretations are more plausible than Suppe's, because to avoid dismissing skeptical doubt as question-begging, mere existence of a plausible and non-question-begging interpretation of skepticism is enough. This is what I try to show in the next section.

4. Alternative interpretations of traditional skeptical arguments

The purpose of this section is to propose alternative interpretations of skeptical arguments, ones by the KK thesis and others by what I call the HEKK thesis (I will explain the details of the thesis later). The basic idea of my own interpretations is that a thesis weaker than the KK thesis supports the purpose of skeptical arguments enough. To illustrate, take an example from the above Feyerabend's argument. My alternative interpretation goes as follows:

F'1. F'2. F'3. The same as F1, F2 and F3

F'4. If S knows that P, then someone can know that S knows that P.

F'5. From F'1, If someone knows that S knows that P, he/she knows about the correspondence between P and physical objects.

F'6. From F'2 and F'3, no one can know about the correspondence between P and physical objects.

F'7. From F'5 and F'6, no one can know that S know that P.

F'8. From F'4 and F'7, S does not know that P.

F'9. Since we do have knowledge, assumption F'1 is false.

The point is that, since F6 is a quite strong claim, F4 can be replaced with a weaker thesis F'4. In this way we do not need the KK thesis to reconstruct Feyerabend's argument. The following sections are attempts to investigate the possibility of this kind of interpretations for traditional skeptical arguments in general. I use Descartes, Berkeley and Hume as examples.

4-a. Descartes' methodical skepticism

First of all, we should look at Descartes's methodical skepticism. Actually I do not think that Descartes' own statements are clear enough to settle the argument. Here is the statement of his methodical skepticism.

"...reason already convinces me that I must withhold assent no less carefully from what is not plainly certain and indubitable than from what is obviously false; so the discovery of some reason for doubt as regards each opinion will justify the rejection of all" (Descartes [1642] 1971, 61).

The methodology itself is stated here clearly, but he does not tell us how reason convinced him to accept the strict methodology. Here is a room for interpretations. First, let us assume "Descartes' uncritical acceptance of the KK Thesis" (Suppe forthcoming, Part II chapter 5, section 5). Then surely we can understand why Descartes set the criteria so high. To know something, I should know that I know. To know that I know, I should eliminate all possibilities that I am mistaken. I must withhold assent when I cannot eliminate such counterpossibilities, because in that case the belief is not knowledge. Therefore methodical doubt is necessary (Suppe 1977, 717-718). To put slightly more formally, the interpretation goes as follows (using Cartesian Demon as an example):

D1. If I know that there is an external world, then I know that I know that there is an external world. (the KK thesis)

D2. I know that I know that there is an external world only if I eliminate all counterpossibilities to the existence of external world.

D3. I cannot eliminate the counterpossibility that a Cartesian Demon is deceiving me.

D4. Therefore I do not know that I know that there is an external world.

D5. Therefore I do not know that there is an external world.

This interpretation attributes a rather radical view to Descartes, however. If the reason to adopt methodical skepticism is the KK thesis, then actually going through the methodical doubt is a necessary condition for making a belief a piece of knowledge. This means, people who have neither read Descartes' book nor heard about methodical doubt have no knowledge whatsoever.

Maybe Descartes was that radical, but here is another interpretation. Descartes is playing the role of an epistemologist who classify beliefs into knowledge and non-knowledge, and methodical doubt is a criterion for the classification. Under this interpretation, to go through the skeptical doubt does not make a belief a piece of knowledge, any more than an investigation of an ornithologist does not make a raven black. Thus those who never think about methodical doubt can have knowledge, whether or not an epistemologist come to them and confirm that their beliefs are knowledge.

From these considerations, I would like to propose a weaker version of the KK thesis to which Descartes might have committed himself:

(5) if a knows that p, then someone can know that a knows that p.

This thesis explains Descartes motivation equally well. We might have knowledge but we do not know which beliefs are knowledge. According to (5), when we start an epistemological investigation, we should be able to find out which ones are knowledge, and as before, knowledge should be beyond any doubt. Therefore, until epistemologists can eliminate the possibility of the existence of the Cartesian demon, they should withhold classification of knowledge and non-knowledge. This is why epistemologists should obey the strict methodology. To put in the same way as above D1-D5:

D'1. If I know that there is an external world, then someone can find out that I know that there is an external world. (a variation of (5))

D'2. Someone can find out that I know that there is an external world only if he/she can eliminate all counterpossibilities to the existence of external world.

D'3. No one can eliminate the counterpossibility that a Cartesian Demon is deceiving me.

D'4. Therefore no one can find out that I know that there is an external world.

D'5. Therefore I do not know that there is an external world.

Some comments on (5) are in order here. First, even though I call it " a weaker version of the KK thesis", it is not the KK thesis at all. The KK thesis requires the identity of the knower a and the knower of a's knowing, while the weaker thesis does not.[5] The KK thesis requires the actual presence of knowledge about knowledge, while the weaker thesis requires only the possibility of such knowledge about knowledge. Another comment is necessary about this "possibility". When I say "someone can know", what kind of modality do I mean? Since Descartes' program requires that we actually can find out which beliefs are knowledge, 'logical possibility' or 'causal possibility' is not suitable for the purpose (logically speaking I can be omniscient, and causally speaking, there can be a world in which I know about the Cartesian demon without violating any causal law). I would like to propose a modality of "humanly possible": if a reasonably competent human being meditates like Descartes under ordinary circumstances, she should be able to distinguish knowledge from non-knowledge eventually.

With this last consideration, it is appropriate to rewrite (5) as follows:

(6) If a knows that p, then it is humanly possible that someone knows that a knows that p.

Or, formally,

(6') Kap -> <H> Ex KxKap,

where <H> stands for "humanly possible" and Ex stands for a quantifier "there is at least one x such that". I shall call the thesis (6) "the HEKK thesis" henceforth.

To sum up, in this subsection I have proposed an alternative interpretation of Cartesian skepticism using the HEKK thesis instead of the KK thesis. If all our epistemological efforts fail to show that there is no such thing as the Cartesian demon, then the HEKK thesis is enough to destroy most of our knowledge (and this is what happened in Descartes' case). In this sense, Cartesian skepticism does not depend on the KK thesis.

4-b. Berkeley's rejection of substance

Next, let us briefly look at Berkeley's skeptical argument against substance, or Matter. Here is the base line of his argument:

But, though it were possible that solid, figured, moveable substances may exist without the mind, corresponding to the ideas we have of bodies, yet how is it possible for us to know this? Either we must know it by Sense or Reason. As for our senses, by them we have the knowledge only of our sensations, ideas, or those things that are immediately perceived by sense, call them what you will: but they do not inform us that things exist without the mind, or unperceived, like to those which are perceived. This the materialists themselves acknowledge. -- It remains therefore that if we have any knowledge at all of external things, it must be by reason inferring their existence from what is immediately perceived by sense. But what reason can induce us to believe the existence of bodies without the mind, from what we perceive, since the very patrons of Matter themselves do not pretend there is any necessary connexion betwixt them and our ideas? (Berkeley [1710] 1965, section 18)

Thus, basically, Berkeley says that we cannot know about substances from either of our two sources of knowledge, sense or reason. Therefore we have no knowledge about substance. The question is, is this argument dependent on the KK thesis? Of course if we assume that Berkeley accepts the KK thesis, we can understand why Berkeley thinks that the "necessary connexion" between substances and ideas is essential for the inference to substance; it is because, to know that we know about substances, we should be absolutely certain that we know about substance. Thus the argument can be reconstructed as follows:

B1. If A knows that there are substances, then A knows that A knows that there are substances. (the KK thesis)

B2. Our evidence comes from one of two sources, sense or reason.

B3. Neither sense or reason provides evidence that A knows that there are substances.

B4. Therefore A does not know that A knows that there are substances.

B5. Therefore A does not know that there are substances.

But this is by no means the only way to understand Berkeley. The HEKK thesis serves exactly the same function as the KK thesis. Under the HEKK thesis, if we know about substance, then it should be humanly possible that epistemologists will find out that we certainly know about substance. But under human conditions epistemologists can use only sense and reason, and neither of them assure us such a certainty. Thus the interpretation goes as follows:

B'1. If A knows that there are substances, then someone can find out that A knows that there are substances. (the HEKK thesis)

B'2. Our only ways of inquiry are by sense or by reason.

B'3. Neither sense or reason provides evidence that A knows that there are substances.

B'4. Therefore no one can find out that A knows that there are substances.

B'5. Therefore A does not know that there are substances.

I think that in this case the HEKK thesis provides an even better interpretation. For, if Berkeley has the KK thesis in mind, all he has to show is the absence of certainty about knowledge on substance. But his proof is intended to be much stronger: impossibility of certainty about the knowledge, by any human being. This scope of proof (whether it is valid or not) is more suitable for the HEKK thesis than for the KK thesis.

4-c. Hume's skepticism on induction

Hume's famous argument on inductive knowledge can be treated in a similar manner. For Hume, the sole source of our knowledge is what he calls "impressions". Then how is the knowledge about causation possible? Hume argues that all we know about causation from impression is a "constant conjunction" of cause and effect in our memory (Hume [1739] 1878, Bk. I, Pt. III, sect. VI). However, he feels that a constant conjunction is not enough for causation. Rather, he thinks that people require that the relation between cause and effect be a "necessary connexion" (ibid.). How can we know about such a necessary connection? One way is to assume the following principle: "instances, of which we have no experience, must resemble those, of which we have had experience, and that the cause of nature continues always uniformly the same" (ibid. italics in original). Call this the principle of uniformity. Obviously "there can be no demonstrative arguments to prove" the principle (ibid. italics in original). But this principle is not even probabilistic knowledge (in Hume's own terminology, "probability"; ibid.). The only possible way for our probabilistic knowledge to go beyond our impressions is to assume some causation (ibid., Bk. I, Pt. III, sect. II), and the principle apparently goes beyond our impressions. But the principle to be proved is necessary to establish such causation. Therefore, since "the same principle cannot be both the cause and effect of another", the principle itself cannot be a probabilistic knowledge (ibid., Bk. I, Pt. III, sect. VI).

Now we are not interested in the details and validity of Hume's argument, but in the overall structure of it. Like Berkeley's argument, this argument can be reconstructed using the KK thesis, but again like Berkeley's argument, Hume argues against not existence but possibility of knowledge about knowledge. One characteristic feature of Hume's argument is his explicit appeal to circularity or infinite regress in the denial of the principle of uniformity. This seems favorable for the KK thesis interpretation because the KK thesis implies an infinite regress (Kap -> KaKap -> KaKaKap -> ...). On the other hand, my HEKK thesis does not imply such an infinite regress. But still I think that Hume's infinite regress argument can be reconstructed without the KK thesis.

Let us compare following formulatons:

(a) The KK thesis interpretation

H1. If A knows that x causes y, then A knows that A knows that x causes y. (the KK thesis)

H2. A knows that A knows that x causes y only if A can infer that x causes y from evidence available for A.

H3. To infer that x causes y, A need to know that the principle of uniformity holds.

H4. (repeat H1~H3 replacing "that x causes y" with "that the principle of uniformity holds")

H5. Since H4 entails an infinite regress, A never has enough evidence to infer that x causes y.

H6. Therefore A does not know that A knows that x causes y.

H7. Therefore A does not know that x causes y.

(b) The HEKK thesis interpretation

H1. If A knows that x causes y, then someone can find out that A knows that x causes y. (the HEKK thesis)

H2. Someone can find out that A knows that x causes y only if it is possible to infer that x causes y from evidence available for human beings.

H3. To infer that x causes y, the principle of uniformity is necessary.

H4. The only (humanly possible) way to establish the principle of uniformity is to infer it from itself.

H5. Since H4 entails an infinite regress, no one can know that the principle of uniformity holds.

H6. Therefore no one can find out that A knows that x causes y.

H7. Therefore A does not know that x causes y.

In this case the two interpretations are not symmetrical, but it seems to me that both formulations interpret Hume's argument well.

5. The HEKK thesis and the fate of epistemology

As my analyses in the previous sections show, the KK thesis has played almost no role in skeptical arguments. Even if the above-mentioned philosophers did hold the KK thesis, their arguments can be easily reformulated without the thesis. Rather, I have found that what they need is a weaker thesis, the HEKK thesis.

What is the point of replacing the KK thesis with the HEKK thesis? Can we not dismiss the HEKK thesis as easy as the KK thesis? First of all, the HEKK thesis is immune from most difficulties raised against the KK thesis. The HEKK thesis does not require self-knowledge of the knower, and it does not imply that the knower believes that she is justified in believing. Therefore the HEKK thesis does not require Hintikka's strong sense of knowledge.

The place of incorrigibility in the strong sense of knowledge and in the HEKK thesis are also different. The strong sense of knowledge requires that the incorrigibility should be attained with the evidence available to the knower. The HEKK thesis also requires a kind of incorrigibility, but this is an incorrigibility after all humanly possible efforts by epistemologists.

This point is related to another point, namely the HEKK thesis is even compatible with a kind of externalist notion of knowledge. The externalism is a position that the justification of a belief has nothing to do with evidence available to the knower herself. For example, existence of an appropriate causal relation between the knower and the state of affairs expressed by p may be enough to say that 'a is justified in believing that p'. The HEKK thesis is compatible with this view as far as the 'appropriate causal relation' is knowable to epistemologists. Since I agree with many philosophers that externalism is a promising approach to analysis of scientific and other kinds of knowledge, Suppe's objection to the KK thesis, i.e. the notion of knowledge associated with the KK thesis do not help us in understanding scientific knowledge, does not apply to the HEKK thesis.

But, one may object, two of Suppe's points had to do with the fact that if we accept the KK thesis, we are led to skepticism/relativism. This is exactly what I have proved about the HEKK thesis. Then why do we not dismiss the HEKK thesis for this reason? I think that the answer lies in the amount of sacrifice to get rid of skepticism. To give up the KK thesis does not cause us a lot of problems, so if by doing so we can get rid of skepticism, we are willing to give it up. But giving up the HEKK thesis has a grave consequence. The HEKK thesis is a corollary of the following thesis:

(7) (x) (x -> <H> Ey Kyx)

The thesis (7) states that no truth is out of our reach. I would like to call it the 'epistemological optimism' thesis. To reject the HEKK thesis leads to rejection of the 'epistemological optimism' thesis, which is a widely accepted thesis, especially by both sides of the realism - anti-realism debate (e.g. Putnam 1981,134; Fine 1986, 131; Devitt 1997, 232 ).

To see the result of the rejection in epistemology, let us consider a negation of the HEKK thesis:

(8) ~ (Kap -> <H> Ex KxKap)

This is equivalent to

(9) Kap & ~(<H>Ex KxKap).

with some easy transformations,

(10) Kap & [H] (x) ~KxKap,

namely, "a knows p, and it is humanly necessary that, for all x, x does not know that a knows that p". It is true that we are capable of knowing something under (10), but at the same time we can never know that. This sounds too much a sacrifice to avoid skepticism.

Instead of discarding the HEKK thesis, if we adopt "Kantian subjectivism" or "its neo-Kantian successors, including Pragmatism and associated radical relativisms" (Suppe forthcoming, Part II chapter 5, section 5), we can avoid skepticism to some extent. In this sense, despite Suppe's claim quoted in the introduction, these are still attractive alternatives.

6. Conclusion

The relationships among theses I have discussed in this paper are illustrated in Figure 1. In this paper, first I have discussed that the KK thesis with Hintikka's strong sense of knowledge is unhelpful for the inquiry of scientific knowledge. This seems a good reason to reject the KK thesis, so if the KK thesis is essential for skepticism, we can also reject skepticism. However, Suppe's argument to connect skepticism and the KK thesis is weak, and I have argued that the HEKK thesis, a weaker version of the KK thesis, is enough for skeptical arguments. Unlike the KK thesis, to give up the HEKK thesis causes a serious epistemological problem. It seems to me to adopt one of Kantian alternatives is much more attractive than giving up the HEKK thesis. Certainly Suppe should take skepticism more seriously.References

Barense, J.G. (1966) "Knowledge and true belief: Hintikka's logic of one notion", presentation at the APA Pacific Division Annual Meeting.

Berkeley, G. ([1710] 1965) Principles of Human Knowledge, reprinted in Berkeley's Philosophical Writings. ed. by D.M. Armstrong. New York: Collier Macmillan.

Carrier, L.S. (1974) "Skepticism made certain" in The Journal of Philosophy 71, 140-150.

Chisholm, R.M. (1963) "The logic of knowing" in The Journal of Philosophy 60, 773-795.

Descartes, R. ([1642] 1971) Meditations on First Philosophy, translation in E. Anscombe and P. Geach (trans., ed.) Descartes; Philosophical Writings. Prentice Hall.

Devitt, M. (1996) Realism and Truth, second edition with new afterword. Blackwell.

Fine, A. (1986) Shaky Game; Einstein, realism and the quantum theory . University of Chicago Press.

Feyerabend, P. (1958) "An attempt at a realistic interpretation of experience" in Proceedings of the Aristotelian Society 58, 143-170.

Hall, M. (1976) "Skepticism and knowing that one knows" in Canadian Journal of Philosophy 6, 655-663.

Hilpinen, R. (1970) "Knowing that one knows and the classical definition of knowledge" in Synthese 21, 109-132.

Hintikka, J. (1962) Knowledge and Belief: An introduction to the logic of the two notions. Ithaca: Cornell University Press.

--. (1970) "'Knowing that one knows' reviewed" in Synthese 21, 141-162.

Hume, D. ([1739] 1878) A Treaties on Human Nature, a reprint edited by T.H. Green and T.H. Grose. London: Longman, Green and co.

Putnam, H. (1981) Reason, Truth and History. Cambridge: Cambridge University Press.

Suppe, F. (1973) "Facts and empirical truth", in Canadian Journal of Philosophy 3, 197-212.

--. (1977) "Afterword", in F. Suppe (ed.) The Structure of Scientific Theories, 2nd ed. University of Illinois Press.

--. (1989) The Semantic Conception of Theories and Scientific Realism. University of Illinois Press.

--. (forthcoming) Facts, Theories, and Scientific Observation.