METHODISM AND HIGHER-LEVEL EPISTEMIC REQUIREMENTS Brendan Murday bmurday@ithaca.edu Draft: Please do not cite without permission Abstract Methodist solutions to the problem of the criterion have often been alleged to entail some kind of KK principle. An argument that methodism is committed to some kind of higherlevel requirement on epistemic states has never been satisfactorily presented, however. This paper demonstrates that the methodist solution to a particular formulation of the criterion problem must endorse a JJ principle according to which a subject cannot have first-order justification for a belief without also having higher-order justification of that belief. The methodist cannot soften the impact of this result by limiting the application of the JJ principle to those who possess the concept of justification, or to those who have second-order beliefs about their first-order justified beliefs; nor can they appeal to process reliabilism to reconcile methodism with the rejection of a JJ principle. Methodism is thus unattractive to those unwilling to espouse a strong version of the JJ principle. Higher-level requirements make frequent entry into epistemic debates. Perhaps the most notorious higher-level requirement is the KK principle, which in its simplest form states that one cannot have knowledge without knowing that one knows. 1 Several epistemologists have suggested that there is an important relationship between the problem of the criterion and such a higher-level epistemic requirement; it is often presumed that methodist solutions to the problem of the criterion are committed to a KK principle, while the rival particularist solutions are not. 2 1 More nuanced versions of a KK principle may suggest that if one knows that p, one is in a position to know the one knows, or that if one knows that p and believes that one knows, then one knows that one knows that p. Influential discussions of how to formulate the KK principle can be found in Ginet (1970) and Hintikka (1970); Kelp and Pedersen (2010) present an overview of the literature. For the moment, however, the precise formulation of a higher-level requirement principle need not concern us; we will return to the issue in 4. 2 The most influential statement of this relationship is in Chisholm (1973); he suggests that methodist solutions to the problem of the criterion are inferior to particularist solutions because: 1
It is curious, however, that the only two discussions in the literature (at least, that the author is aware of) that explicitly argue for or against a relationship between methodism and the KK principle both contend that there is no connection between the two at all. 3 This paper argues that methodism does indeed entail a higher-level requirement on epistemic justification. Specifically, we will find that methodists are committed to a JJ principle according to which one cannot be justified in believing p unless one is justified in believing that one is justified in believing p. Before that argument is presented, however, we will be wellserved to carefully formulate the problem of the criterion, and the methodist and particularist solutions. In 4, after showing that methodists are committed to a JJ principle, we will consider whether such commitment should constitute a reason to reject methodism. 1 Formulating the Problem of the Criterion Contemporary discussion of the problem of the criterion owes a great deal to Chisholm (1973); a clarification of the problem is best begun by reviewing his formulation, which formulates the problem in terms of a pair of questions: one question concerns which beliefs are instances of knowledge, and the other concerns how one can identify those beliefs as instances of knowledge. This formulation, however, is only one type of criterion problem, and it will be useful to view the problem in its most general form. We should begin by observing the form of the two questions. [I]n order to find out whether you know such a thing as that this is a hand, you don t have to apply any test or criterion. Spinoza has it right. In order to know, he said, there is no need to know that we know (Chisholm 1973: 22) Chisholm is not the only one to suggest that methodism entails a KK principle; see also Flores (2005), Hetherington (2001: 185-190), Landesman (2002: 57-59), and Lemos (2004: 106). 3 See Amico (1993: 81-82) and Cling (1997). 2
(A) questions ask which elements of a certain category are members of some further sub-category. In Montaigne s formulation 4 of the criterion problem, the category at issue is the class of appearances, and the sub-category is the class of true appearances. In Chisholm s formulation mentioned in the previous paragraph, the category is the class of beliefs and the subcategory is the class of beliefs that are instances of knowledge. The present point is simply that any (A) question will identify a category and sub-category. We might identify Chisholm s formulation as the problem of the knowledge-criterion, to distinguish it from formulations which focus on a different sub-category. (B) questions ask for a criterion of that sub-category. However, we must distinguish two ways to construe criterion ; the import of the (B) question depends on which construal we adopt. An evidential criterion of knowledge is a feature of a belief that would give a subject reason to think that the belief qualifies as knowledge. A conceptual criterion of knowledge, on the other hand, is the collection of features which provide a conceptual analysis of knowledge. For instance, the feature is a warranted and true belief might be taken to provide a conceptual criterion of knowledge this feature is what makes an instance of knowledge count as knowledge. An evidential criterion of X, however, need not be the feature that provides the conceptual analysis of X; an evidential criterion is any feature that is a reliable indicator of X. 5 If feature F is an evidential criterion of knowledge, then in recognizing that a belief exhibits feature F, one would be justified in thinking that the belief is an instance of knowledge. In the literature on the problem of the criterion, some have interpreted criterion evidentially and 4 5 As presented in Chisholm (1973: 3). Thus there may be more than one evidential criterion of X, though there is only one conceptual criterion. 3
others have interpreted it conceptually. 6 It would be a mistake to think that only one interpretation is correct; the important point is to avoid conflating the two problems, which differ with respect to the construal of the (B) question. We will be concerned here solely with the evidential construal of the criterion problem. 7 The (A) and (B) questions thus take these forms: (A) Which category-members are sub-category members? (B) Which feature of sub-category members serves as a criterion of the sub-category? Since we are focusing on the evidential notion of a criterion, the (B) question would be answered by identifying some feature of X that would give one evidence or reason to believe that X is a member of the sub-category. Having observed the form of the (A) and (B) questions, and the different ways to interpret the (B) question, we need now to say something about the sense of priority at issue. The crux of the problem of the criterion does not concern the answers to the (A) and (B) questions; it is instead a matter of which of these questions is epistemically prior to the other. 8 6 DePaul (1986) and Lemos (2004), for instance, both formulate the problem of the criterion as the conceptual problem, while Amico (1993) and Cling (1997) formulate it as the evidential problem. The distinction is rarely noted; Cling (1997: 110-111, 114-115) is a welcome exception (though the labels evidential and conceptual are mine, not his). 7 There are interesting things to be said about the problem of the conceptual criterion as well, but those points would take us too far afield. 8 Formulating the criterion problem in terms of epistemic priority is controversial; DePaul (1986) argues that the problem of the criterion is best understood as a problem about how to construct a theory, not a problem about epistemic justification. It is clear, however, that there is a problem of the criterion that can be formulated in terms of epistemic priority. DePaul is interested in the exegetical question whether Chisholm intended to be raising a 4
In asking whether (A) is epistemically prior to (B), we are not asking whether the answer to (A) will be discovered before the answer to (B) that would be a question of temporal priority. How should we explicate epistemic priority? Let us construe the notion as follows: the (A) question is epistemically prior to the (B) question iff one can justifiably answer the (A) question without relying on a justified answer to the (B) question to secure the justification for the (A) answer, and conversely one cannot justifiably answer the (B) question without relying on a justified answer to the (A) question. Epistemic priority is thus an asymmetric relation in which the justification for one answer depends on the justified answer to the other question. The solutions to the problem of the criterion are distinguished by their views on the epistemic priority of the (A) and (B) questions. Particularists think that (A) is epistemically prior to (B), while methodists think that (B) is epistemically prior to (A). Criteria skeptics say that neither question is epistemically prior, and they conclude from that fact that we cannot give a justified answer to either the (A) or (B) question. 9 2 How Not to Tie Methodism to a Higher-Level Epistemic Requirement We have now characterized the form of the problem of the criterion. Our focus is on the putative connection between a methodist solution to the evidential problem of the criterion and a problem about epistemic priority or about theory construction; since our concern is not with Chisholm per se, we will ignore any questions about the criterion problem construed as a question about theory construction. 9 Notoriously, there is some controversy about whether there are other ways to respond to a problem of the criterion a coherentist solution might hold that neither question is epistemically prior, but that we can nevertheless give justified answers to both questions. Given the definitions of particularism, methodism, and criteria skepticism offered above, it would seem that this would constitute a fourth solution. Since our concern here is with methodism, we need pay only the briefest attention to prospective fourth solutions. 5
higher-level requirement (embodied in a KK or JJ principle) for first-order knowledge or justification. We have noted above that there are different problems of the criterion corresponding to different sub-categories. Since Chisholm focuses on knowledge as the sub-category, we might expect that the methodism solution to the problem of the knowledge-criterion will reveal a commitment to a KK requirement. We should thus consider the following formulations of the (A) and (B) questions: (A-k) Which beliefs are instances of knowledge? (B-k) Which feature of a belief serves as a criterion of knowledge? Is there any connection between the methodist solution to this problem and a KK thesis? Probably not. But if one were confused about the kind of priority that is at issue in the methodist/particularist debate, one might think that such a connection could be established. It may be helpful to see the flaws in an argument that knowledge-methodism entails the KK thesis. That argument might run as follows: Suppose that X has the answer to question (B-k); let us stipulate that feature F turns out to be a criterion of knowledge. In recognizing that F is a criterion of knowledge, X thereby has a procedure for identifying an instance of knowledge X merely has to ask whether an arbitrary belief exhibits F; if the belief does exhibit F, X is justified in concluding that the belief is an instance of knowledge. Consider some proposition p that is known by X. X is now in a position to know that she knows that p, since she merely needs to ask herself whether p exhibits feature F. We have begun with the assumption that a person has the answer to (B-k), and arrived at one version of a KK principle specifically, the claim that whenever X knows that p, she is in a position to know that she knows. One might conclude on the basis of this reasoning that 6
methodism, with its emphasis on starting with an answer to the (B-k) question, entails a KK principle. The argument is flawed. The first premise is that the subject has an answer to the (B-k) question, but that supposition has little to do with methodism. Methodism is the view that the (B-k) answer is epistemically prior to the (A-k) answer; the supposition that we have an answer to the (B-k) question has no bearing on the matter of epistemic priority. If a KK principle were entailed merely by having an answer to the (B-k) question, then particularists would be equally committed to that KK principle, since particularists and methodists both hold that we can justifiably answer question (B-k). 10 Particularists and methodists do not differ with respect to whether one can answer (B-k); where they differ is with respect to the epistemic priority of (Bk). We have yet to see why that might make a difference in entailing a KK principle. 3 How to Tie Methodism to a Higher-Level Epistemic Requirement We have seen how not to establish a connection between methodism and a KK principle; let us turn to the positive case. To draw the connection, we need to focus on a different problem of the criterion the problem of the truth-criterion, characterized by these (A) and (B) questions: 10 Given the definition of epistemic priority presented in 1, this statement is not quite accurate. Half of the particularist view is the statement that the (B) question cannot be justifiably answered without relying on a justified answer to the (A) question. If one thought that the (A) question can be justifiably answered but that the (B) question cannot be justifiably answered, one would be a particularist who denies that we can justifiably answer (B-k). On the other hand, every methodist will hold that the (B) question can be justifiably answered. Still, the typical particularist will think that the (B-k) question can be justifiably answered, reasoning that we can identify a common feature of items of knowledge by extrapolating from the particular instances of knowledge, so most particularists will agree with methodists that we can justifiably answer the (B-k) question. 7
(A-t) Which of one s beliefs are true? (B-t) Which feature of a belief serves as a criterion of a true belief? The argument in broad strokes will proceed as follows: to have a justified answer to the (A-t) question is to have first-order justification. To have a justified answer to the (B-t) question is to have second-order justification, i.e., justification that one has first-order justification. 11 If the (Bt) question is epistemically prior to the (A-t) question, as methodists believe, then first-order justification is epistemically dependent on second-order justification, which constitutes a kind of JJ requirement on justification. The argument can be stated as follows: (1) Methodism holds that (B-t) is epistemically prior to (A-t). (2) By the definition of epistemic priority, methodists thereby hold that: any justified answer to (A-t) depends for its justification on a justified answer to (B-t); justification for an answer to (B-t), on the other hand, can be given without relying on a justified answer to (A-t). (3) To have a justified answer to (A-t) is to have justification that beliefs p 1 -p n are true. This is first-order justification for p 1 -p n. (4) A criterion of G is a feature whose presence gives evidence to an agent that the bearer of that feature is G; thus, that feature can provide justification that the bearer is G. A feature that is a criterion of truth is thereby a first-order justifier. (5) If one is justified in thinking that F is a criterion of truth, then one is justified in believing that a belief with feature F is justified. A justified belief that F is a criterion of truth will 11 It will be convenient at times to speak of second-order justified beliefs, a term that might be interpreted in different ways. For present purposes, a second-order justified belief is a justified belief that one has a [first-order] justified belief. 8
thus constitute second-order justification that beliefs with feature F [namely, p 1 -p n ] are true. (6) Methodism thus holds that first-order justification for p 1 -p n depends on second-order justification for p 1 -p n. (7) Methodists are thereby committed to a JJ principle. Since (1) and (2) are mere definitions of methodism and epistemic priority, we need not say more about them here. (3) asserts the relationship between the (A-t) question and first-order justification. That relationship is straightforward; to have justification for an answer to (A-t) is to have justification for thinking that beliefs p 1 -p n are true. That is just another way of saying that p 1 -p n are justified beliefs. Step (4) claims that a criterion of truth is itself a first-order justifier. When criterion is interpreted evidentially, F is a criterion of G just in case the instantiation of F provides evidence that G is also instantiated. If F is a criterion of truth, then, the fact that a belief exhibits feature F provides evidence that the belief is true. But there is a clear connection between providing evidence of truth and providing justification; if F is a criterion of the truth of p, then the fact that p exemplifies F can itself justify the belief that p. The point made here in step (4) sets up the claim made in step (5) that a justified answer to the (B-t) question will constitute second-order justification; we should turn our attention to that claim. Methodism requires that a justified answer to (B-t) is used to justify the answer to (A-t); the answer to the (B-t) question, which states a criterion of truth, is a first-order justifier, as noted in the previous paragraph. To have a justified answer to the (B-t) question is to be justified in thinking that feature F is a first-order justifier that p, so to have a justified (B-t) answer is to have second-order justification for the truth of p. And as stated in (6), methodists, in insisting that the 9
(B-t) question is epistemically prior, are thus requiring that a justified (A-t) answer [i.e., a set of beliefs with first-order justification] is derived from X s justification for thinking that F is a justifier. This is placing a second-order justification requirement upon first-order justification X needs justification for the claim that she has justification for p in order to generate justification for p. 12 The argument here presupposes that methodists are committed to the claim that we have a justified answer to the (B-t) question, and that this justified answer is used to justify the (A-t) answer. If methodism merely required that one use a (B-t) answer whether or not it is justified to justify the (A-t) answer, there would be no basis for claiming that methodism entails a JJ principle. Thus the argument relies upon formulating the epistemic priority principle in such a way that the epistemically prior answer is itself a justified answer. Is this the right way to formulate epistemic priority? 13 It is. The methodist should not attempt to weaken the definition of epistemic priority simply to avoid the conclusion of the argument. An unjustified answer to the (B) question is easy to come by simply assert that being formed on a Monday is a criterion of true beliefs, and then use this criterion to arrive at an opinion which beliefs are true. But if this would constitute an acceptable solution to the problem of the truth-criterion, why would criteria skepticism have ever seemed tempting? Surely there is some force to the criteria skeptic s position; that force derives from the fact that we demand justified answers to both the (A) and (B) questions, and so 12 The argument tying methodism to a JJ principle is applied for the problem of the truth-criterion; what about the problem of the knowledge-criterion, though? When the sub-category is the class of beliefs that count as knowledge, then the kind of justification that applies to an answer to the (A) question is already second-order justification, and hence a methodist solution to the problem of the knowledge-criterion will entail a more nuanced higher-level epistemic requirement, namely that JJ requires JJJ. For this reason it is better to think of methodism in general as entailing some higher-level epistemic requirement or other, as opposed to saying that all methodist solutions to criterion problems are committed to the JJ principle specifically. 13 Acknowledgement. 10
the methodist and particularist must insist that we use a justified answer to one question to justify the other answer. What value would an answer to the (A) question have if it were putatively supported merely by an unjustified claim? The formulation of epistemic priority originally proffered is the right way to construe the principle. 4 Strong and Weak Higher-Level Requirements Thus far we have argued that methodism is committed to a JJ principle. The significance of our efforts will be clearer after we scrutinize the JJ principle itself. One version of the principle has deeply counter-intuitive implications. Another version is more plausible, but as we will see, methodists will be committed to counter-intuitive implications even if they argue that they 3 shows that their view entails merely the more plausible JJ principle. Ultimately, we will find that methodism is less attractive than we might otherwise have supposed. A Strong version of the JJ principle asserts outright that a subject cannot have a firstorder justified belief without in addition having the corresponding second-order justified belief. The Strong JJ principle is vulnerable to familiar counterexamples. Consider, for instance, a child who lacks the concept of justification; 14 she is intuitively capable of having justified beliefs, but since she does not form higher-order beliefs about the justificatory status of her own beliefs, she does not have second-order justified beliefs. For a second counterexample to Strong JJ, consider a subject who fails to realize that he has a justified belief. 15 Because he is not aware of his firstorder justified belief, he lacks a second-order justified belief. 14 15 Hetherington (2001: 184) presents a representative example of this objection. See Ginet (1970: 166), Segal (1979: 183-184). 11
Intuitively, we want to say of both of these cases that the subject can have first-order justified beliefs despite the lack of a corresponding second-order justified belief; the Strong JJ principle is implausible because it prohibits that claim. The fundamental problem with the Strong JJ principle is that it applies to all subjects. A Weak JJ principle restricts the higherlevel requirement to certain subjects, namely those who satisfy certain conditions (such as possessing the concept of justification and recognizing upon reflection when she possesses a justified belief). 16 For ease of exposition, call a subject unsophisticated just in case she fails to satisfy the conditions stated in Weak JJ, and sophisticated when the subject does satisfy such conditions. [Thus, according to Weak JJ sophisticated subjects will have second-order justified beliefs whenever they have first-order justified beliefs; Weak JJ imposes no higher-level requirement upon unsophisticated subjects.] This suggests a strategy for the methodist. 3 suggests that methodists are committed to a JJ principle. If the methodist were forced to endorse Strong JJ, there would be cause for concern the implications of Strong JJ are hard to accept. If instead methodists are merely committed to Weak JJ, they might be content to bite the bullet. We will see presently, however, that the methodist cannot evade criticism by embracing Weak JJ and rejecting Strong JJ. The problem with such a strategy will be revealed when we ask whether the methodist solution to the criterion problem is true of unsophisticated as well as sophisticated subjects. The methodist has four options; (i) methodism is true of all subjects, whether they are sophisticated or unsophisticated. The higher-level requirement on first-order justification only applies to sophisticated subjects, so that Weak JJ but not Strong JJ is affirmed. Methodism itself, 16 See Ginet (1970: 164) for a proposal that the KK principle should be restricted in roughly this sort of way. Kelp and Pederson (2010: 590) have more to say about such weakened principles. 12
however, is not likewise restricted in application to sophisticated subjects. The remaining three 17 options, by contrast, all involve claiming that methodism is true only of the sophisticated subjects. Those options can be distinguished as follows: (ii) particularism is true of unsophisticated subjects; (iii) criteria skepticism is true of the unsophisticated; (iv) a coherentist account of the relationship between justified (A-t) and (B-t) answers is true of the unsophisticated. Consider (i) first: if methodism is true even of the unsophisticated, but we concede that the JJ constraint on first-order justification does not apply to those subjects, then the methodist would seem forced to say that unsophisticated subjects cannot have justified beliefs. Suppose, for instance, that the feature that makes a subject sophisticated is simply possession of the concept of justification. Methodism would require that to have justified beliefs, the subject would have to have higher-order justification, but without a concept of justification, unsophisticated subjects could not satisfy such a requirement. This is precisely the result of the 17 Strictly, there are more than three ways of developing that strategy methodism comprises two claims: that the (B-t) question can be justifiably answered independently of the (A-t) answer, and the denial of the converse claim that the (A-t) question can be justifiably answered independently of the (B-t) answer. To deny that methodism is true of unsophisticated subjects leaves open which of those two claims is false of those subjects. But if the methodist is trying to avoid the results of the Strong JJ principle, she should affirm that unsophisticated subjects can have justified beliefs, since Strong JJ is disturbing precisely because it suggests that unsophisticated subjects fail to have any justified beliefs. Thus it will do no good to say that methodism is false of unsophisticated subjects merely in virtue of the falsehood of the first claim; the methodist needs to give unsophisticated subjects a path to having justified beliefs, and not simply deny those subjects the path that is granted to sophisticated subjects. So the remaining three options available to the methodist will be three different ways of granting unsophisticated subjects a path to having justified beliefs, ways that do not require that unsophisticated subjects have second-order justified beliefs. 13
Strong JJ principle those without the resources to possess higher-order justification are unable to possess first-order justification either. So in opting for (i), the methodist finds herself with the same unpalatable result that one would get from endorsing the Strong JJ Principle. The methodist might opt instead for (ii), arguing that particularism is true of unsophisticated subjects. This is an uncomfortable proposal. The methodist who adopts (ii) holds that unsophisticated subjects can justify an (A-t) answer without relying on the (B-t) answer; we might expect that sophisticated subjects ought to be able to justify an (A-t) answer through the same means, since they differ from unsophisticated subjects merely in having something more. The methodist might not surrender immediately; sometimes additional beliefs or information defeats beliefs or information that would otherwise constitute justification. So the methodist who adopts (ii) might say that unsophisticated subjects can have an independently justified (A-t) answer, but that for sophisticated subjects any such attempt to justify the (A-t) answer would be defeated. While this is an available path to pursue, it is unpromising. Why would possessing the concept of justification defeat an otherwise successful attempt to justify an (A-t) answer independently of the (B-t) answer? The (ii) strategy is entirely unmotivated, and as such, the methodist should not embrace it. The methodist might opt for (iii): criteria skepticism is true of unsophisticated subjects, and hence those subjects are unable to give justified answers to either the (A-t) or the (B-t) question. But if those subjects are unable to justify any answer to the (A-t) question, they simply lack justified beliefs. Once again, we would find that the methodist would be embracing the 14
unpalatable result generated by the Strong JJ Principle those who lack the concept of justification would turn out to lack justified beliefs entirely. We have finally option (iv) to consider; on this view, for conceptually impoverished subjects the (A-t) and (B-t) answers are mutually justifying. This strategy exhibits exactly the same flaw we witnessed with (ii) why would the methodist think that unsophisticated subjects are able to have justified answers to the (A-t) question through a route that is unavailable to sophisticated subjects who differ only in possessing some additional epistemic feature? If option (iv) were tempting, we should simply abandon methodism entirely and say that a coherentist solution is true of all subjects. Furthermore, if a justified answer to (A-t) is even partially dependent on a justified answer to (B-t), and conceptually impoverished subjects lack the latter, they will lack the former as well. This again is the unpalatable result that many would have prefer to avoid intuitively, subjects who lack the concept of justification could nevertheless have justified beliefs, and option (iv) [like (i) and (iii)] fails to sustain that result. A similar line of reasoning presents itself if we consider reliabilist theories of justification. 18 Reliabilists have reason to deny a JJ principle, since the fact that a first-order belief was formed through a reliable process is no guarantee that a corresponding second-order belief would be formed through a reliable process. But reliabilism seems perfectly compatible with methodism; although a reliabilist could be a particularist, her view is compatible with the methodist claim that the (B-t) question is epistemically prior to the (A-t) question. We have thus exposed a tension; reliabilists seem to be able to embrace methodism and deny the JJ principle, yet we have suggested that methodism entails a JJ principle. What do we make of this? We could think of reliabilism as expanding the category of those for whom the JJ principle fails to obtain; in addition to the unsophisticated subjects, we would include those who 18 Acknowledgement. 15
form second-order beliefs through a different process than the process by which the corresponding first-order belief was formed. But should our reliabilist say that methodism is true of those subjects? Nothing in the previous paragraph identifies an error in the argument presented in 3, so if she endorses methodism across the board, she is embracing contradictory claims in saying that the JJ principle both does and does not apply to the subjects in question. If she denies that methodism applies to those subjects, though, we can ask the same questions that we asked of the unsophisticated subjects. There is a tension between embracing methodism and denying the JJ principle, and reliabilists will hence have to either eschew methodism or identify the problem with the argument in 3. We have argued that methodism entails a JJ principle. We have seen in this final section that this is no trivial commitment; methodists, on pain of ad hoccery, are forced to embrace the implausible consequence of Strong JJ according to which subjects who do not possess and deploy the concept of justification lack justified beliefs altogether. For those unwilling to bite this bullet, methodism will have lost a great deal of its appeal. 19 19 Acknowledgements. 16
References Amico, Robert P. (1993). The Problem of the Criterion. Lanham, MD: Rowman & Littlefield. Chisholm, Roderick. (1973). The Problem of the Criterion. Milwaukee: Marquette University Press. Cling, Andrew. (1997). Epistemic Levels and the Problem of the Criterion. Philosophical Studies, 88(2), 109-140. DePaul, Michael R. (1986). Reflective Equilibrium and Foundationalism. American Philosophical Quarterly, 23(1), 59-69. Flores, Tito Alencar. (2005). The Problem of the Criterion, Knowing that One Knows and Infinitism, Veritas: Revista de Filosofia, 50(4), 109-128. Ginet, Carl. (1970). What Must be Added to Knowing to Obtain Knowing that One Knows? Synthese, 21, 163-86. Hetherington, Stephen. (2001). Good Knowledge, Bad Knowledge: On Two Dogmas of Epistemology. Oxford: Clarendon Press. Hintikka, Jaako. (1970). Knowing that One Knows reviewed. Synthese, 21: 141-62. Kelp, Christoph, and Pedersen, Mikolaj J.L.L. (2010). Second-Order Knowledge. In Duncan Pritchard and Sven Bernecker (Eds.), The Routledge Companion to Epistemology, (pp. 586-596). New York: Routledge. Landesman, Charles. (2002). Skepticism: The Central Issues. Oxford: Blackwell. Lemos, Noah. (2004). Common Sense: A Contemporary Defense. Cambridge: Cambridge University Press. Segal, Robert. (1979). Knowing that One Knows. Auslegung, (6), 178-191. 17