/data/papers/buckwalter2010knowledgeisntclosed/out/fulltext.mdKnowledge Isn't Closed on Saturday: A Study in Ordinary Language
Wesley Buckwalter
Department of Philosophy The City University of New York Graduate Center
365 Fifth Ave. 10016 New York NY USA, Department of Philosophy The City University of New York Graduate Center
365 Fifth Ave. 10016 New York NY USA
Published on March 19, 2010
Recent theories of epistemic contextualism have challenged traditional invariantist positions in epistemology by claiming that the truth conditions of knowledge attributions fluctuate between conversational contexts. Contextualists often garner support for this view by appealing to folk intuitions regarding ordinary knowledge practices. Proposed is an experiment designed to test the descriptive conditions upon which these types of contextualist defenses rely. In the cases tested, the folk pattern of knowledge attribution runs contrary to what contextualism predicts. While preliminary, these data inspire prima facie skepticism for the contextualist hypothesis regarding folk knowledge claims, as well as challenge certain predictions made by recent theories of subject-sensitive invariantism. It is further argued that such results raise methodological questions concerning the practice of relying on an assumption of intuitions, with respect to ordinary language practices, as evidence for philosophical conclusions regarding knowledge.
In the last part of the twentieth century, philosophers have debated whether knowledge claims fluctuate relative to conversational factors. In turn, many contemporary debates in epistemology are split between those theorists who think that knowledge assertions or denials are somehow sensitive to conversational contexts, and those who think they are not. Let us call those philosophers who hold the former view variantists. They argue that various features of the conversational context can affect the truth value of knowledge claims. While positions of this type vary, one such view, epistemic contextualism, claims that knowledge is contextsensitive. More specifically, contextualists argue that the truth conditions for sentences of the kind "S knows that p" are relative to the context in which the utterances of those sentences are made.
Alternatively, let us call those philosophers who traditionally hold the opposing view invariantists. These theorists deny that features of conversational context determine the standards for knowledge claims. Instead, they argue for a fixed set of truth conditions that govern knowledge attribution regardless of the circumstances in which that attribution is expressed. A recent kind of invariantist theory, subject sensitive invariantism (SSI), claims that practical interests of a subject can influence knowledge attribution. Specifically, while denying that knowledge is relative to the conversational contexts in which knowledge is attributed, SSI theorists argue that the truth conditions of a sentence like "S knows that p" fluctuate according to what is at stake for the subject.
SSI and contextualism are rival theories that disagree about the nature and semantics of knowledge attribution. However, both contextualists and subject sensitive invariantists often make appeals to semantic intuitions about everyday conversational contexts. Each provides thought experiments that advance a particular analysis of knowledge, citing our intuitions regarding ordinary language practices in these cases as evidence to support their respective views. It is thought that if, through our intuitions, contextualism or SSI accurately describe the knowledge practices of normal agents in everyday speech, and if it is agreed that our conceptual analysis of 'knowledge' should reflect those practices, then there is good reason to accept that analysis of knowledge. However, given the results of related work in experimental philosophy (eg., Weinberg et al. 2001; Machery et al. 2004), the practice of parlaying analytic assumptions of folk epistemological intuitions into philosophical conclusions should indeed inspire caution. 1 This is especially true when such assumptions take the guise of what seem to be a series of descriptive claims. Instead, the question of empirical predictions made by contextualism or SSI regarding our ordinary language practices deserves an empirical answer.
This paper presents evidence, taken from probes modeled after specific cases from the prevailing literature, that the pattern of folk knowledge attribution among ordinary English speakers is not significantly altered by the relevant factors that contextualism or SSI predict. Having shown this, I will then go on to discuss the philosophical implications of these findings in defending an account of knowledge by appeal to an assumption of intuition. It is argued that these results continue to motivate methodological fatigue for various philosophical positions which rely considerably on folk intuitions in their arguments, but which fail to find empirical evidence to support them.
Knowledge and Context-Sensitivity
There are a variety of different epistemic theories that might be considered contextualist. 2 Generally speaking, all of these views have something to do with 1 Machery et al. 2004 shows cross-cultural variation among semantic intuitions regarding theories of reference, suggesting "it is wrong for philosophers to assume a priori the universality of their own semantic intuitions." Famously, Weinberg et al. 2001 shows that such divergence of intuition extends to the epistemic domain.
2 While these theories have been developed a variety of ways, the most widely referenced epistemic contextualist theories today are perhaps those of Keith DeRose (1992), Stewart Cohen (1986), and David Lewis (1979, 1996).
the way knowledge or justification relates to the relevant conversational situation in which they are imbedded. They argue that the truth conditions of sentences that attribute or deny knowledge are context sensitive. Contextualism then, is primarily a thesis about language. The claim is that 'knows' functions like an indexical. For example, to understand any sentence containing such familiar indexical operators as 'I', 'here' or 'now', we must consult the contexts in which they are used. It seems obvious that the meaning of these types of sentences can differ drastically depending on situation in which they are uttered. Similarly, contextualists often cite analogies between knowledge-sentences and sentences with terms like 'tall' or 'flat', which mean different things in different contexts. In some conversational situations a sentence attributing tallness to me will be true, and in other cases the same sentence will be false. The same is said to apply for knowledge.
When contextualists argue that the truth conditions for a knowledge claim depend on conversational context, they often mean that 'know' exhibits just the same sort of context sensitivity. If 'knows' is an indexical, then like the indexicals above, we will have to consult the context of its use to arrive at its meaning. To the extent in which the contexts of a knowledge-sentence vary, different sentences are expressed. The claim is that these differing conversational contexts affect the changing standard for 'knows', and in turn the truth conditions for knowledge-sentences.
Opposed to this view are the recent SSI theories of John Hawthorne (2004) and Jason Stanley (2005). In defending SSI, they deny that 'know' is an indexical and consequently that the truth conditions of knowledge-sentences fluctuate with conversational context. However, they too depart from the traditional invariantist position by arguing that those conditions can be influenced by how important it is to the subject that the knowledge claim is true. They argue that the practical interests and concerns of a subject can influence attributions of knowledge to that subject. The idea is that there are certain personal facts pertaining to the subject of the knowledge-sentence that can affect knowledge attribution. The claim is that the practical stakes for the subject will have a direct influence on whether it is said that the subject knows. For the contextualist then, the relevant epistemic standards are determined by the attributor's context, while for the SSI-er, they are determined by the situation of the subject.
But now we may ask: what determines the particular standards of the conversational context? While contextualists are free to disagree on the particular standards that affect the truth conditions of an assertion or denial between contexts, two factors are usually considered (see DeRose 1992; Cohen 1999). First is what we generally think of as stakes, or the importance of being right. For instance, take the assertion, "I know those berries are not poisonous." When the previous statement is uttered while looking at a picture in a botanical textbook, a lot less hinges on its truth than say, when trapped on a desert island where food is scarce. In this illustration, SSI and contextualism both claim that knowledge is sensitive to a factor of stakes. They predict that in the former circumstance, you will be more inclined to consider your beliefs about berry-toxicology as knowledge than when the stakes are significantly higher. Both contextualism and SSI rely on this intuition, whether it is driven by context or by practical interests, when they claim that as the stakes go up, knowledge goes down.
The second, and perhaps the most prominent contextual factor said to affects epistemic standards, is the salience of error possibilities. This often involves giving counterevidence, proposing other possible explanations, or raising skeptical challenges that question the veracity of a knowledge claim. Take the assertion, "I know they are harboring weapons of mass destruction." Now, suppose that someone challenges your assertion by saying, "You may be basing your claim on what is faulty intelligence," or "Perhaps you are allowing a religious or political bias to cloud your judgment. Do you really know they are harboring weapons?" Given these alternatives, many may be inclined to doubt the truth of the former knowledge claim. Again, the contextualist predicts that the truth of the same knowledge claim in a situation with a low epistemic standard will change in cases where error possibilities become salient. Similarly, as the possibility of error is raised for the subject of the knowledge sentence, SSI-ers also claim that error possibilities will raise the epistemic standard for the subject. Thus in such cases, contextualism and SSI both predict that when error possibilities become salient, knowledge goes down.
Evidence for Contextualism
Contextualism has given elegant and sophisticated solutions to perennial philosophical puzzles-particularly Gettier cases (Cohen 1998) and skeptical challenges (DeRose 1995; Cohen 2000)-that often strain our otherwise best conceptual theories of knowledge. Similarly, SSI has also claimed to capture many of these results. While such theories are often propagated by appeal to the barrage of explanatory power this type of analysis can provide in practice, we still need arguments to motivate the theory itself. In this way, contextualism and SSI are often defended by citing their support in ordinary language. Keith DeRose (2005) has since championed such a strategy:
The best grounds for accepting Contextualism concerning knowledge attributions come from how knowledge-attributing (and knowledge-denying) sentences are used in ordinary, non-philosophical talk: what ordinary speakers will count as 'knowledge' in some non-philosophical contexts they will deny is such in others. This type of basis in ordinary language provides…the best grounds we have for accepting contextualism concerning knowledge attributions (p. 172).
The appeal to intuition in such cases has become common practice in the prevailing literature. As in the examples above, thought experiments are often employed to advance the intuition that contextualism or SSI are consistent with ordinary language practices. This approach usually involves presenting two adjacent scenarios in a philosophical paper where the same knowledge claim is uttered, but in distinct situations. The reader is then asked to evaluate the same knowledge claim in both cases, often with conflicting results. The goal is to show that particular factors of the relevant contexts or characters are what play the central role in assigning the truth conditions of the relevant knowledge claim. These intuitions about ordinary language are then used as evidence to support the relevant philosophical analysis.
A paradigmatic example of this practice comes from the contextualist literature surrounding the DeRose bank cases (1992). In these much-discussed cases, a husband and wife stop at the bank on their way home from work on a Friday afternoon, when they notice that lines inside the bank are very long. In the following situation, let's call it Bank Case A, it is not very important that they deposit their paychecks that day, and they decide, rather than waiting in line, to come back the following morning. Because the husband has visited that bank on Saturday mornings in the past, he is confident that this option will be available to them. He says, "I was just there 2 weeks ago, I know the bank will be open on Saturday."
Now, compare the situation above with the subsequent Bank Case B, where the stakes of the situation are raised. In this case, imagine that the husband has also written a very large check that day. If the money from his pay is not deposited by Monday, it will bounce, leaving them in a very bad situation. Since the bank is closed on Sunday, it is very important that the check is deposited as soon as possible. Again, the husband says, "I was just there 2 weeks ago, I know the bank will be open on Saturday." When these two cases are presented, contextualists generally argue that most people will agree that he knows the bank is open on Saturday in Bank Case A. However, since a lot is at stake in Bank Case B, he better go check just in case. In other words, the knowledge claim is false in the latter case, relative to the higher stakes of the situation.
Similarly, we can imagine an additional Bank Case C, where error possibilities become salient. Here, the husband still insists that he knows the bank will be open on Saturday. Upon hearing this, his wife challenges the claim. She says, banks change their hours all the time," or "perhaps since you were here last this branch has changed its management." In this situation, the same knowledge claim meets a higher epistemic standard. As before, the contextualist argues that these skeptical challenges will influence the truth conditions of the knowledge claim. The driving intuition that arises from this case, claims the contextualist, is that the higher epistemic standard in Case C will make the same knowledge claim false.
These bank cases, or other cases like them, are often presented together in philosophical prose, and when they appear it is almost always assumed that the folk intuition about knowledge will conflict between cases. It is taken for granted that in Case A the given knowledge assertion will be true (a clear case of knowledge), and in situations like Case B and Case C, it will be false (a clear absence of knowledge). Then, the contradiction is explained by taking a contextualist view of knowledge attribution. While such asymmetry has presented philosophical questions for such an analysis in and of itself, the contextualist intuition between cases is still generally thought to obtain. 3 As DeRose (2005) says:
When the cases are considered individually, without worry about the other case, the intuitions are quite strong, and, in any case, the linguistic behavior displayed in the cases quite clearly does accurately reflect how "I know" / " I don't know" is in fact used (p. 177). 4 However, philosophers often have a bad habit of assuming folk intuitions based on their own training, positions, or commitments. Consistent with recent trends in experimental philosophy, if folk intuitions about these cases are going to be used to advance metaphysical, ethical, or epistemic claims, then we ought to be sure that they are, empirically speaking, accurately represented. 5 Since arguments for epistemic contextualism and SSI claim to be supported by ordinary language or everyday linguistic behavior, then the way 'knowledge' works in normal, non-philosophical speech should be collected and tested. If it is true that the best grounds for accepting contextualism are reflected by our intuitions, then surveying those intuitions should be uncontroversial. Indeed, one would even think it advantageous for such a theory to include a very specific empirical component, if in fact a particular kind of intuition is relied on as evidence for the view at hand.
Since the claims that these theories make about ordinary language practices are primarily descriptive in nature, they too should fall into this category. The following experiment examines whether folk knowledge claims in ordinary language actually do support the relevant philosophical analyses. Through the use of probes modeled after Keith DeRose's original bank cases, this study was designed to test the predictions contextualism and SSI make about the ways stakes and error possibilities influence folk knowledge attribution. The hypothesis tested was that either by contextualism or SSI, the truth conditions of knowledge claims in the following cases will fluctuate when the stakes or error possibilities of those cases are raised.
The Experiment
This study is comprised of two distinct tests of the contextualist hypothesis that the truth of knowledge claims will vary between low and high stakes contexts and between contexts of low and high error possibilities. Since participants in the experiment are asked to make judgments about a subject's knowledge, we can also extend this test to the predictions made by SSI. To test for the effect, if any, that stakes or error possibilities have on the truth of knowledge claims, three survey types were devised with vignettes modeled directly from Keith DeRose's Bank Cases (labels added here, emphasis in original):
Bank Sylvie and Bruno are driving home from work on a Friday afternoon. They plan to stop at the bank to deposit their paychecks, but as they drive past the bank they notice that the lines inside are very long. Although they generally like to deposit their paychecks as soon as possible, it is not especially important in this case that they be deposited right away. Bruno tells Sylvie, "I was just here last week and I know that the bank will be open on Saturday." Instead, Bruno suggests that they drive straight home and return to deposit their paychecks on Saturday. When they return to the bank on Saturday, it is open for business.
High Stakes Sylvie and Bruno are driving home from work on a Friday afternoon. They plan to stop at the bank to deposit their paychecks. Bruno has written a very large check, and if the money from his pay is not deposited by Monday, it will bounce, leaving Bruno in a very bad situation with his creditors. As they drive past the bank, they notice that the lines inside are very long. Bruno tells Sylvie, "I was just here last week and I know that the bank will be open on Saturday." Instead, Bruno suggests that they drive straight home and return to deposit their paychecks on Saturday. When they return to the bank on Saturday, it is open for business.
High Standards Sylvie and Bruno are driving home from work on a Friday afternoon. They plan to stop at the bank to deposit their paychecks, but as they drive past the bank they notice that the lines inside are very long. Although they generally like to deposit their paychecks as soon as possible, it is not especially important in this case that they be deposited right away. Bruno tells Sylvie, "I was just here last week and I know that the bank will be open on Saturday." Instead, Bruno suggests that they drive straight home and return to deposit their paychecks on Saturday. Sylvie says, "Banks are typically closed on Saturday. Maybe this bank won't be open tomorrow either. Banks can always change their hours, I remember that this bank used to have different hours." When they return to the bank on Saturday morning, it is open for business.
Subjects, Design and Procedures
The population group for this study consisted of 544 undergraduate students at The State University of New York at Buffalo. Participants were selected from large lecture courses satisfying University wide general curriculum requirements. Such courses represent a wide variety of college majors and are usually taken by students as they begin their collegiate studies. Of the participants, approximately 75% were between the ages of 18 and 20 (range 18-38), 55% were male, and over 70% reported that they were originally from the New York State area.
In a classroom setting, participants were randomly given one and only one of the three probes listed above in which a subject of the story makes a knowledge claim about the operating hours of the bank. Participants were then asked to judge whether the knowledge statement that the subject made within the dialogue of the vignette was true.foot_4 Specifically, participants were asked to specify their level of agreement with the target sentence regarding the truth of Bruno's knowledge assertion as follows:
On a scale of 1 to 5, circle how much you agree or disagree that Bruno's assertion, "I know the bank will be open on Saturday" is true.
Data was collected using a five-point scale anchored with strongly disagree, disagree, neutral, agree, and strongly agree, respectively. The responses from the Bank probe were compared between groups with the two additional variations of the baseline Bank case. 7 In what follows in Test 1, a comparison between Bank and High Stakes was used to test for the folk pattern of agreement between groups given a low and a high stakes probe. In Test 2, a comparison between Bank and High Standards was used to test for the same pattern between probes of low and high error possibilities. For the hypothesis to be confirmed, we should expect that the mean level of agreement with the target question will decrease between groups in both tests.
Results of Test 1
This test was designed to examine the contextualist claim that the truth conditions of knowledge claims will fluctuate between low and high stakes conditions. It also tests the claim in SSI that subject stakes will influence knowledge attribution. For the hypothesis to be supported, we would expect that from the following prompts, ordinary speakers are more likely to agree that the subject's knowledge assertion is true in low stakes conditions and more likely to disagree that the subject's knowledge assertion is true in high stakes conditions.
Given that a score above three constitutes agreement, 136 (or 74.3%) in the base Bank probe agreed that Bruno's knowledge assertion was true, as opposed to 124 (or 68.5%) of participants given the High Stakes condition. 8 However, statistical analysis indicates that there is no significant difference between the means of these two groups, and that both means are significantly above the midpoint. 9 We should conclude that when comparing participants' mean level of agreement with the target question between Bank and High Stakes groups on this occasion, the contextualist and SSI pattern of knowledge attribution was not found.
Results of Test 2
This test was designed to examine the contextualist claim that the truth conditions of knowledge claims will fluctuate between contexts of low and high epistemic standard. Since we are raising the salience of error possibilities for the subject, this will also test the similar predictions made by SSI. If these theories accurately reflect 7 This scoring method did not prevent participants from writing their own responses in addition to using the scale provided. For instance, on a substantial number of surveys in each condition, participants often underlined Bruno's statement, "I was just here last week" and "I know that the bank will be open on Saturday," possibly signaling that they simply accept the subject's evidence and testimony as warranting across the board. Furthermore, on many surveys participants also wrote things like, "He's [Bruno] just being lazy" and "He just doesn't want to wait in line." These write-ins were often accompanied by responses below 3 (regardless of the survey type), perhaps suggesting that surprising non-epistemic factors can influence ordinary subjects' inclinations to attribute or deny knowledge to others (see Beebe and Wesley Buckwalter forthcoming). 8 In Bank, 27 (or 14.8%) disagreed with the statement and 20 (or 10.9%) remained neutral. In High Stakes, 30 (or 16.6%) disagreed with the statement and 27 (or 14.9%) remained neutral. 9 Differences are not significant between groups, t (362)=.987, p=.243, as well as significantly above 3, p<.01.
ordinary language practices, we would expect speakers to be more likely to attribute knowledge in low error conditions and less likely to attribute knowledge when the possibility of error is high.
Compared in this test are the responses between groups from the baseline Bank probe to High Standards. Given that a score above three constitutes agreement, 119 (or 66.1%) of participants given them High Standards prompt agreed that Bruno's knowledge assertion was true, as opposed to 136 (or 74.3%) in the base Bank probe. 10 Nonetheless, statistical analysis shows that there is no significant difference between group means, while both means are significantly above the midpoint.foot_6 Again, we should conclude that when comparing participant responses between Bank and High Standards prompts, the contextualist or SSI pattern of knowledge attribution was not found (Table 1).
Discussion
Two central contextualist and SSI hypotheses were tested from probes taken from the prevailing literature. In a rather surprising turn of events, this experiment has shown the pattern of folk knowledge attribution detected here to be at odds with the outcome predicted.foot_7 When comparing the pattern of participant responses between probes of higher and lower stakes, as well as higher and lower epistemic standards, any observed differences between the means of these groups on this particular occasion is likely due to chance alone. Most importantly however, mean scores are all significantly above three, signally participant agreement with the target sentence in all survey types. Therefore in the particular bank cases tested we have reason to doubt the contextualist hypothesis; the truth conditions of the knowledge claims tested did not fluctuate between contexts, and the SSI hypothesis; nor were they influenced by the practical interests of the subject.
While such data may conceivably constitute prima facie evidence for import into a philosophical argument for traditional invariantism, it is by no means conclusive. Several questions remain unanswered in matters of both content and experimental pragmatics. The former is clearly one of robustness. Will the pattern of folk knowledge attribution shown here generalize to more than just the specific bank probes tested? Additionally, are college students a suitably representative sample of folk epistemic intuitions, and if so, are probe-style vignettes given in the classroom the best way to collect them? These are exciting research questions that empirically minded epistemologists and philosophically minded psychologists will indeed address.
Despite possible pragmatic concerns for these initial results, data reflect strong folk intuitions about knowledge claims in the bank cases incongruent with contextualism and SSI. However, it is important to note that this paper has not produced certain claims. First and foremost, these results should not be viewed as a refutation of any particular variantist view. Simply testing a few isolated cases from the literature will not yield definitive arguments against these positions, just one major source of evidence said to support them. Also, this paper takes no stance on whether an ordinary language defense-or any other strategy-constitutes the best grounds for accepting or defending contextualism or SSI. Instead, this approach is simply inherited as the antecedent in a larger conditional argument in which such a strategy is applied. Similarly, I remain cautiously agnostic regarding the relationship between vignette-style experimental philosophy and the conceptual analysis of terms. By invoking the ordinary language defense, contextualism has already bridged the divide for us. Given the previous commitments by these theorists regarding the relation to everyday knowledge practices, this paper simply seeks to provide empirical support for what is seemingly a series of empirical claims about ordinary language. Unfortunately, it was not found here.
Nevertheless, the results presented in this study do yield some interesting philosophical and methodological conclusions. The latter will bear directly on the former. First, the divergence of epistemic intuition between the experiments given here (and similarly those presented in the prevailing philosophical literature) calls into question the conventional wisdom that contextualism and SSI really have folk knowledge attribution on their side. This is troubling if the relevant intuitions were meant to provide evidentiary support for the respective analyses. Alternatively, data suggest that the intuitional arguments given above cannot be used to support these analyses of knowledge. Furthermore, in the absence of other independent arguments for the views at hand, these results also threaten to seriously undermine those epistemological projects which rely solely on ordinary knowledge practices significantly different than the ones people actually have.
Secondly, the more troubling discovery seems to be the degree in which our assumptions about folk knowledge attribution have been mistaken. Recall, vignettes presented in the epistemological literature very similar to high stakes and high standards are usually considered clear cases in which a subject's knowledge claim is false. Yet, in the bank cases tested, the pattern of folk knowledge attribution was considerably divergent from what these theories assume or predict. Of course, future testing will be decisive. For instance, perhaps it might be shown that certain particularities in our ordinary language practices, since obfuscated by the sole use of philosophical thought experiments, actually do support the philosophical analysis at hand. Yet, such revelations will only occur as a result of more studies in experimental epistemology investigating folk knowledge practices, and not from analytic assumptions made from the armchair. Thus, results of this kind should inspire general caution regarding the practice of appealing to the evidentiary status of ordinary language by simply intuiting compatible ordinary language. Philosophers who rely on this kind of evidence by continuing to endorse this practice must perilously risk asking themselves the following question: if it is possible that these epistemological assumptions are mistaken, what else could we be wrong about?
Conclusion
In the matter of folk knowledge attribution and certain conceptual analyses of knowledge, much is at stake. If certain descriptive claims concerning our ordinary language practices are at work to support contextualism or SSI, it is at least conceivable that these arguments proceed by describing the ways in which they are empirically privileged. Yet, it is far from clear that this work has actually been done. By simply surveying folk epistemic intuitions among a few relatively simple cases we have discovered something of the opposite. These data both seriously question this kind of evidentiary support for contextualism and SSI among our ordinary language practices, as well as inspire the need for more research in experimental epistemology. Those philosophers who continue to rely on descriptive claims about folk intuitions involving ordinary knowledge practices can no longer afford to ignore the need for empirical testing. Perhaps genuine philosophical progress in epistemology will be the result. 13
References
[1] The epistemic side-effect effect James Beebe and Wesley Buckwalter Mind and Language (2010). [2] Knowledge and context Stewart Cohen Journal of Philosophy (1986) 83. [3] Contextualist solutions to epistemological problems: Skepticism, gettier, and the lottery Stewart Cohen Australasian Journal of Philosophy (1998) 76. [4] Contextualism, skepticism, and the structure of reasons Stewart Cohen Philosophical Perspectives (1999) 13 pp. 58. [5] Contextualism and skepticism Stewart Cohen Philosophical Issues (2000) 10. [6] Contextualism and knowledge attributions Keith Derose Philosophy and Phenomenological Research (1992) 52 (4). [7] Solving the skeptical problem Keith Derose Philosophical Review (1995). [8] Now you know it, now you don't Keith Derose (2000) V. [9] The ordinary language basis for contextualism and the new invariantism Keith Derose The Philosophical Quarterly (2005) 55 (219). [10] As a matter of fact: Empirical perspectives on ethics John Doris et al. (2005). [11] Knowledge and lotteries John Hawthorne (2004). [12] Elusive knowledge D Lewis Australasian Journal of Philosophy (1996) 74. [13] Scorekeeping in a language game David Lewis Journal of Philosophical Logic (1979). [14] Semantics, cross-cultural style Edouard Machery et al. Cognition (2004) 92 (9). [15] Practical interests, relevant alternatives, and knowledge attributions: An empirical study Joshua May et al. Review of Philosophy & Psychology (2010) 1 (3). [16] Folk concepts and intuitions: from philosophy to cognitive science Joshua Nichols Trends in Cognitive Science (2004) 8. [17] Knowledge and practical interest Jason Stanley (2005). [18] Normativity and epistemic intuitions Jonathan Weinberg et al. Philosophical Topics (2001) 29.