Subject-Sensitive Invariantism and Isolated Secondhand Knowledge
- First Online:
- Cite this article as:
- Kasaki, M. Acta Anal (2014) 29: 83. doi:10.1007/s12136-013-0215-3
- 155 Downloads
Jennifer Lackey challenges the sufficiency version of the knowledge-action principle, viz., that knowledge that p is sufficient to rationally act on p, by proposing a set of alleged counterexamples. Her aim is not only to attack the knowledge-action principle, but also to undermine an argument for subject-sensitive invariantism. Lackey holds that her examples are counterexamples to the sufficiency version of the knowledge-action principle because (a) S knows the proposition in question, but (b) it is not rational for S to act on it. In this paper, first, I argue against (a) on intuitive and on theoretical grounds. Second, I point out that (b), even if combined with (a), is not sufficient to make for counterexamples to the knowledge-action principle of the relevant kind. Third, I offer two alternative explanations of the intuition Lackey relies on. If either one of them is right, (b) may not be satisfied in her examples.
KeywordsSubject-Sensitive Invariantism Knowledge-Action Principle Isolated Secondhand Knowledge Testimonial Knowledge Jennifer Lackey
One prominent movement in recent epistemology is to theorize a close link between knowledge and practical reasoning or rationality. Let us call the principle that states such a link the ‘knowledge-action principle’. While some deny that any version of the knowledge-action principle is correct, a lively debate is ongoing over which version is correct. For example, Jason Stanley (2005) and John Hawthorne (2004) each propose a biconditional version of the principle, and they jointly defend the following: “it is appropriate to treat the proposition that p as a reason for acting iff you know that p” (Hawthorne and Stanley 2008, p. 578). On the other hand, Jeremy Fantl and Matthew McGrath, in a series of writings (2002, 2007, 2009), endorse (different but similar) sufficiency versions of the principle, such as “S knows that p only if S is rational to act as if p” (2007, p. 559).
The knowledge-action principle is of interest in epistemology for at least two reasons. First, if it is true, it reveals the importance of knowledge for practical reasoning or rationality. Second, it is often used as a premise in an argument for subject-sensitive invariantism (SSI).1 The sufficiency version of the knowledge-action principle is relevant to SSI, not only because it is less controversial than the necessity version, but also because it is enough for the argument for SSI.
Jennifer Lackey (2010), however, challenges the sufficiency version of the knowledge-action principle by proposing a set of alleged counterexamples.2 In this paper, I argue against Lackey that an arsenal of theoretical resources is available in defence of the sufficiency version of the knowledge-action principle ( ‘the knowledge-action principle’ below refers to the sufficiency version, unless otherwise noted). In Section 2, I explain the knowledge-action principle and its significance for SSI in detail, and distinguish the SSI-friendly version of the knowledge-action principle from other versions. Only the SSI-friendly version directly entails SSI. Then, in Section 3, I describe one of Lackey’s examples and its setting. Lackey holds that her examples are counterexamples to the sufficiency version of the knowledge-action principle because the following are true of them: (a) S knows the proposition in question, but (b) it is not rational for S to act on the proposition. In Section 4, I argue against (a) on both intuitive and theoretical grounds. I proceed, in Section 5, to point out that given the proper understanding of the SSI-friendly version of the knowledge-action principle expounded in Section 2, (b) is not enough to make Lackey’s examples genuine counterexamples to the principle; what is required is (b′) that S’s epistemic position with regard to the proposition in question is too weak to rationally act on it. I argue that with plausible assumptions, (b′) does not hold in Lackey’s examples, and hence they are consistent with the SSI-friendly version of the knowledge-action principle, even though they may not be consistent with the knowledge-action principle Lackey formulates. In Section 6, I argue that Lackey even fails to establish (b); she purports to do so by explaining our intuitions about her examples in a certain way, but two other explanations that do not imply (b) are readily available. My aim is not to settle what the best response is against Lackey’s alleged counterexamples; indeed, I doubt that a single counter-resource can handle all of them. They have or at least suggest different settings and backdrops, and which is the best response may differ from example to example. I only aim to show that with many counter-resources available, proponents of the knowledge-action principle need not worry about them.
2 The Knowledge-Action Principle
(Purism) For any subjects S1 and S2, if S1 and S2 are just alike in their strength of epistemic position with respect to p, then S1 and S2 are just alike in whether they are in a position to know that p.
Both purists and SSIists (anti-purists) accept the working hypothesis that S knows that p iff (i) p is true, (ii) S believes that p, and (iii) S is in a strong enough epistemic position with regard to p (or S’s warrant for p is strong enough).3 How strong an epistemic position S is in is determined by one or a set of epistemic factors—epistemic in the sense of being truth-conducive or error-averse. Condition (iii) is satisfied iff the strength of S’s epistemic position regarding p is strong enough relative to the relevant epistemic standard—the standard determining how strong S’s epistemic position must be to be strong enough for S to have knowledge. The factor(s) that set the relevant epistemic standard are epistemic in a different sense: they are factors relevant to the satisfaction of condition (iii) but not to that of the truth condition (i) and the belief condition (ii); put more simply, they are factors to make a difference between true belief and knowledge.
Knowledge-Action: If you know that p, then p is warranted enough to justify you in φ-ing, for any φ. (ibid., p. 66)6
(They call this principle ‘KJ,’ perhaps meaning a general link between knowledge and justification. It is indeed misleading to call it “knowledge-action,” as I do here, for the obvious reason that φ is not restricted to actions. I will come back to this point in Section 4.) From this principle and condition (iii), it follows that S’s epistemic position regarding p is strong enough relative to the relevant epistemic standard only if it is strong enough to justify S in φ-ing, for any φ. This entails the claim that pragmatic factors, though neither truth-conducive nor error-averse, still count as epistemic in the second sense. That is, they set the epistemic standard and hence contribute to determining whether a true belief is a case of knowledge by meeting condition (iii). Compare this version of the knowledge-action principle with Hawthorne and Stanley’s (2008): “it is appropriate to treat the proposition that p as a reason for acting iff you know that p.” Their version of the knowledge-action principle is not SSI-friendly in the sense defined here. For it is a principle to the effect that S’s knowledge of p is necessary and sufficient for p to be S’s reason for action, i.e., an epistemic factor in the truth-conducive or error-averse sense. The denial of purism does not directly follow from this principle.7
3 Lackey’s Alleged Counterexamples
KNPR-S*: It is epistemically appropriate for one to use the proposition that p in practical reasoning, to act as if p, and to act on p, if one knows that p. (362)
ONCOLOGIST: Eliza is an oncologist at a teaching hospital who has been diagnosing and treating various kinds of cancers for the past twenty years. One of her patients, Lucas, was recently referred to her office because he has been experiencing intense abdominal pain for a couple of weeks. After requesting an ultrasound and MRI, the results of the tests arrived on Eliza’s day off; consequently, all of the relevant data were reviewed by Anna, a competent medical student in oncology training at her hospital. Being able to confer for only a very brief period of time prior to Lucas’s appointment last week, Anna communicated to Eliza simply that her diagnosis is pancreatic cancer, without offering any of the details of the test results or the reasons underlying her conclusion. On the basis of the reliable and trustworthy testimony that she accepted from Anna—combined with her background knowledge, that if a patient has pancreatic cancer, a highly aggressive combination of radiation and chemotherapy is the necessary course of action—Eliza decided to schedule this treatment for Lucas, which she began administering to him this morning. (364)
For the time being, assume with Lackey that Eliza knows that Lucas has pancreatic cancer. Eliza’s knowledge is an instance of what Lackey dubs ‘isolated secondhand knowledge’; it is secondhand because “the subject in question knows that p solely on the basis of another speaker’s testimony that p”; and it is isolated because “the subject knows nothing (or very little) relevant about the matter other than that p” (365). Eliza believes that Lucas has pancreatic cancer merely on the basis of Anna’s sincere testimony to that effect. Eliza has a plethora of knowledge about symptoms of various types of cancers, and, after the brief meeting with Lucas, she also knows that he is in abdominal pain; but none of this knowledge is enough for her to know that Lucas has cancer of a specific type. Thus, the source of Eliza’s knowledge, if any, is nothing other than Anna’s testimony.
(a) Eliza knows that Lucas has pancreatic cancer.
(b) Eliza is not rational or justified in scheduling or administrating a highly aggressive combination of radiation and chemotherapy for Lucas.
(b′) Eliza’s epistemic position regarding Lucas has pancreatic cancer is too weak to rationalize or justify her in acting on it.
Lackey argues for (a) on the ground that no account of non-testimonial knowledge prevents Anna from knowing that Lucas has pancreatic cancer, and no account of testimonial knowledge prevents Eliza from gaining knowledge about Lucas’s cancer from Anna’s testimony. Granted, Anna knows that Lucas has pancreatic cancer; she satisfies virtually all conditions for non-testimonial knowledge existing in the literature: she is adequately justified in believing in, has no defeater for, and is globally and locally reliable with regard to, her diagnosis of the type of cancer that Lucas is suffering. In addition, Lackey claims, Eliza satisfies virtually all conditions for testimonial knowledge existing in the literature, if her belief that Lucas has pancreatic cancer is formed on the basis of Anna’s sincere testimony. Crude though the following characterizations are, there are mainly two varieties of accounts of testimonial knowledge: non-reductionism and reductionism. Non-reductionism only requires that S’s testifier be in fact reliable and S have no defeater for her reliability, whereas reductionism in addition requires that S herself have adequate justification for the testifier’s reliability. Eliza is adequately justified in believing Anna’s testimony, has no defeater for Anna’s reliability as a testifier, and Anna is in fact a reliable testifier. Eliza meets conditions imposed by both reductionism and non-reductionism, and therefore neither of them disqualifies her as a knower.
Lackey’s argument for (b) is based on the intuition we supposedly have about ONCOLOGIST, that Eliza’s scheduling or administrating a highly aggressive combination of radiation and chemotherapy for Lucas is to be criticized and is therefore inappropriate. Lackey holds that the inappropriateness involved in the intuition is epistemic: “Eliza lacks the appropriate epistemic credentials to schedule and begin administering radiation and chemotherapy to Lucas” (374). Eliza is socially expected to fulfil a certain explanatory duty; she must be able to explain her diagnosis and treatment for Lucas if questions are raised about them. To do this, Lackey claims, it is required that Eliza, on her own, has a specific kind of evidence for Lucas’s condition, e.g., data from an ultrasound and MRI, not merely evidence for the reliability of Anna’s diagnosis.
4 Arguments Against (a): Eliza Knows
If Lackey successfully refuted the sufficiency version of the knowledge-action principle, it would have significant ramifications for the current debates on knowledge and practical rationality. However, Lackey’s arguments for both (a) Eliza’s knowing that Lucas has pancreatic cancer and (b) her lack of rationality or justification for her action are inadequate or uncompelling. In what follows, I show that there is an arsenal of theoretical resources to counter her cases for (a) and (b). To be fair, I do not doubt that Lackey’s examples are interesting on their own; what I deny is their relevance to the knowledge-action principle.
Let us begin with (a). Lackey strengthens her case for (a) by pointing out that the essential features of her examples remain the same even when very little is at stake on whether the proposition in question is true or S’s evidence for the reliability of the testifier is perfectly good. This move is meant to block the response from proponents of the knowledge-action principle that S does not know, either due to high stakes or to the weakness of S’s epistemic position. Although it is questionable whether these responses are really blocked this way, that is not the line of response I pursue here.8 For there are two more positive reasons to deny that S knows the relevant proposition in the examples.
First, Lackey does not defend the presence of S’s knowledge by appeal to intuition at all for any of her examples, the ground instead being that all existing accounts of testimonial knowledge allow for S to have knowledge. On the other hand, her case for the epistemic inappropriateness of S’s action is defended merely on the intuitive basis. There is no established theory of testimonial knowledge and all of them face one or another intuitive counterexample. So it is worth examining whether and to what extent (a) is supported by intuitions.
Well, it may sound odd if a colleague of Eliza claims, with no peculiar emphasis: “Eliza, you didn’t review the data yourself? Well, either way, you should administer the highly aggressive combination of radiation and chemotherapy to the patient.” And yet, at least to my ears, it equally sounds odd if Eliza responds to criticism of her action: “No, I didn’t review the data. Well, either way, I do know that he has pancreatic cancer.”9 Of course, this does not show that (a) is false. Notice, however, that no account of testimonial knowledge has been developed with an eye on the Lackey-type situations in which S is expected but unable to answer explanatory questions on the relevant topic because all S has is isolated and secondhand information. Hence, her examples may be regarded as further data for modifying accounts of testimonial knowledge, not for denying the knowledge-action principle.10
Second, Lackey’s case for (b) can easily turn into a case against (a). Lackey holds that it is because of S’s lack of non-testimonial evidence for p that S is not rational or justified in acting on p in her examples. For the sake of argument, now, let us assume that Lackey is right about this. One question to be asked, then, is whether this lack of rationality or justification in the practical realm carries over to the theoretical realm. Even those who deny the knowledge-action principle would not want to deny the link between knowledge and theoretical rationality or justification. Though it may be controversial what precisely that link is, I simply formulate it as this: if S knows that p, then p is warranted enough to justify S in believing q, for any q.
Fantl and McGrath (2009, pp. 69–76) argue for a “unity thesis” that theoretical and practical justification or rationality do not come apart, on the ground that they are subject to rebuke and defence in the same way.11 For example, on Lackey’s analysis, Eliza is not rational to schedule or administrate radiation and chemotherapy on the basis of the information provided by Anna, because she cannot properly defend her scheduling or administration, due to her failure to review the data from the ultrasound and MRI. At the same time, she would form many beliefs on the same basis, e.g., Lucas needs a highly aggressive combination of radiation and chemotherapy, he may lose appetite, he may have his urine and stool changed in color, the results of the ultrasound and MRI were positive, and so forth.12 At least, insofar as Eliza is subject to epistemic criticism for her scheduling radiation and chemotherapy for Lucas, there is little reason to doubt that she is subject to the same type of criticism for her beliefs about Lucas’s condition and treatment. In addition, if Eliza’s being subject to criticism for her action is the intuitive basis for Lackey’s analysis that Eliza lacks justification for her action, Eliza’s being subject to criticism for her beliefs should count as an intuitive basis for Eliza’s lack of justification for her beliefs. Given the link between practical and theoretical justification or rationality, as spelled out in the unity thesis, it is difficult to maintain both that Eliza is not practically rational to act on the proposition that Lucas has pancreatic cancer and that she nonetheless is theoretically rational to believe the same proposition.13
5 Arguments Against (b′): Eliza's Epistemic Position is Strong Enough
As described in Section 3, Lackey’s case for (b)—Eliza’s lack of rationality or justification for acting on the isolated secondhand information—is based on the intuition that Eliza’s action is epistemically criticisable or inappropriate. However, to get Lackey’s intended conclusion that ONCOLOGIST is a counterexample to the version of the knowledge-action principle in support of SSI, Lackey needs to show both (b) that the relevant sense of epistemic inappropriateness is non-rationality or unjustifiedness and (b′) that the weakness of Eliza’s epistemic position regarding Lucas has pancreatic cancer is responsible for the non-rationality or unjustifiedness of her action. In this section, I point out that given plausible assumptions about ONCOLOGIST, Lackey fails to establish (b′). Then, in the next section, I question (b) by offering two ways of explaining the intuition that Eliza’s action is epistemically inappropriate, without denying Eliza’s having knowledge of Lucas’s condition or her being justified for her action.
The requirement for showing that the relevant sense of inappropriateness is lack of rationality or justification is obvious. Lackey’s KNPR-S* is formulated in terms of epistemic inappropriateness, but, as she is well aware, this notion has a purchase on the knowledge-action principle only as non-rationality or unjustifiedness. I concede to Lackey that there is a sense in which it is epistemically defective or inappropriate for S to act on isolated secondhand information in her examples. Whatever it is, call this sense ‘L-epistemic’—which stands for the type of epistemic deficiency Lackey is concerned with. Then, the requirement here is to show that ‘L-epistemic’ coincides with the truth-conducive or error-averse sense of ‘epistemic.’
Lackey may respond that imposing this requirement too strictly jeopardizes the SSIist’s defence strategy for the knowledge-action principle; the SSIist often defends the knowledge-action principle by appeal to our practice of defending one’s action by citing knowledge and denying knowledge by rebuking one’s action as epistemically inappropriate. Lackey claims that this strategy is based on the intuitive sense of epistemic inappropriateness, and hence if a worry is raised against the sense of epistemic inappropriateness in her examples, the same worry is raised against the sense of epistemic inappropriateness in the SSIist’s favored examples.14
(1) S knows that p.
(2) S knows that if p, then A is the thing to do.
(3) S knows that A is the thing to do.
(4) S is rational to do A. (Fantl and McGrath 2002, pp. 72–3)
This is a two-tiered argument; the first sub-argument has (1) and (2) as premises and (3) as conclusion, and the second is a one premise argument from (3) to (4). The first sub-argument is an application of the epistemic closure principle. Fantl and McGrath take the logical relation between (3) and (4) to be entailment.15
In ONCOLOGIST, for instance, Eliza knows that Lucas has pancreatic cancer, knows that if Lucas has pancreatic cancer, then a highly aggressive combination of radiation and chemotherapy is the necessary course of action, and therefore knows that a highly aggressive combination of radiation and chemotherapy is the thing to do. But because she is Lucas’s oncologist and her knowledge is only isolated and secondhand, it can still be argued that it is not rational for her to conclude that a highly aggressive combination of radiation and chemotherapy should be scheduled for him or to begin administering this treatment. Thus, (4) does not follow from (3), and so my arguments do not involve a denial of closure in this sense. (372)
Lackey seems a little uncomfortable with this reply, but accepts it after all, since she holds that other options are all counterintuitive. The options she examines are denying (1), denying (2), and denying closure. She accepts (1) as the consequence of any existing account of testimonial knowledge and endorses (2) as “unquestionable.” The denial of closure surely leads to some counterintuitive consequences, though Lackey concedes that closure does not hold in every case.16 Be that as it may, the move from (3) to (4) does not rely on the kind of intuition that Lackey appeals to in her argument against the knowledge-action principle, viz., that it is appropriate for S to do A. If any intuition supports it at all, it must be to the effect that necessarily, if S knows that A is the thing to do, then S is rational to do A. Note that this intuition is different in content from the intuitions appealed to in the SSIist’s intuition-based defense strategy and in Lackey’s argument. Most importantly, the former intuition directly concerns rationality, whereas the latter intuitions do not. Hence, the burden of proof is on Lackey’s shoulders; her examples and accompanying intuitions do not undermine Fantl and McGrath’s argument-based defense strategy, unless Lackey shows that the intuitive inappropriateness in her examples really amounts to lack of rationality or justification for action.
Even worse for Lackey, it is not clear at all that SSIists accept that (2) holds in ONCOLOGIST (and other Lackey-type examples). Fantl and McGrath (2002, p. 72) note that “the thing to do” in (2) is construed as “what would be the best thing one can do in light of all one’s goals”. They specify this further in terms of rational preference, and (2) can be replaced with the claim: for any state of affairs B, S rationally prefers A & p to B & p. Now, according to Lackey, what makes (2) true in ONCOLOGIST is that “if a patient has pancreatic cancer, a highly aggressive combination of radiation and chemotherapy is the necessary course of action” for Eliza. This course of action is necessary for treating Lucas, but it is compatible with Eliza’s diagnosing Lucas herself before scheduling radiation and chemotherapy—the course of action Lackey regards as appropriate for Eliza. Which course of action is rationally preferable for Eliza, first diagnosing and then scheduling, or scheduling without her own diagnosis, given that Lucas has pancreatic cancer? The answer is clearly the first option. Rational preference is a function of what values or goals S invests in and how likely, conditional on S’s evidence or reasons, it is for S to gain the values or achieve the goals; or more simply, it is a function of preference and epistemic probability. I see no reason that Eliza, qua doctor, can value or prefer anything more than executing her job well. Her job, in normal circumstances, involves both answering explanatory questions about patients’ diseases and treatments and treating them well. Failure of either of these is a loss for Eliza, and it is greater than the loss Eliza may incur from executing her job. Hence, the second option is rationally preferable to the first option. As a result, the SSIist is not committed to the truth of (2) (and (3) and (4)) in ONCOLOGIST.
Now, for illustrative purposes, consider the following example: Eliza is an emergency doctor at a small ER, and she and her reliable assistant, Anna, are only night shift workers there. One night, many patients from a severe traffic accident are sent to the hospital, and they are all going to die without immediate treatment. Since nothing is more valuable than time, after diagnosing the first patient with Anna, Eliza has decided to let Anna diagnose other patients and started operating on them one by one in the order Anna diagnoses. Now, she has finished operating on the first patient and asked Anna “How is the second patient?” Anna, then, sincerely reports the result of her diagnosis: “He has visceral damages, and the critical damages are to kidney and liver.” Eliza decides to operate on the patient’s kidney and liver and to ignore the minor damage to other organs.
I have no inclination to claim of this example that Eliza’s decision and following action are inappropriate in any sense, even though they result from the isolated secondhand information passed from Anna. This example, as opposed to ONCOLOGIST, implies that Eliza’s action is concerned with the patients’ lives much more than with explaining their conditions, and the goal of saving their lives is more valuable for Eliza. Therefore, the appropriateness of her action is easily accounted for: it is rational because it is most likely to lead to the achievement of this goal, provided that Eliza knows that critical damages are to kidney and liver.
If I am right that Eliza’s action in ONCOLOGIST is not rational because her goal is not merely to save Lucas’s life and her action is in conflict with her other goals, then it can be maintained that Eliza’s epistemic position is strong enough to justify her action. Thus, even if (b)—Eliza’s lack of practical rationality or justification—obtains in ONCOLOGIST, it is merely because it is Eliza’s preference rather than the weakness of her epistemic position that prevents Lucas has pancreatic cancer from rationalizing or justifying her action. This might be enough to refute some version of the knowledge-action principle, such as Lackey’s KNPR-S*, but the version of the knowledge-action espoused by Fantl and McGrath is consistent with such cases in which S’s epistemic position regarding p is strong enough to rationalize or justify S in acting on p but other factors block p from rationalizing or justifying S in acting on p. ONCOLOGIST is one of these cases, and it falls short of a counterexample to the SSI-friendly version of the knowledge-action principle.
As a matter of fact, Lackey seems careless about the difference between versions of the knowledge-action principle. She requires for any counterexample to the knowledge-action principle (b) that S be not rational to act on isolated secondhand information. However, what is truly required for a counterexample to the SSI-friendly version is more specific: (b′) that S’s epistemic position regarding isolated secondhand information is too weak to rationalize or justify S in acting on it. The difference between (b) and (b′) is of crucial importance. Remember that, according to Lackey, what makes S’s action epistemically non-rational or unjustified in her examples is S’s lack of non-testimonial evidence. Though she does not develop this point in full detail, Lackey suggests that rationality of action might be a matter of the “quality of evidence” as opposed to the quantity thereof. In other words, S’s lack of non-testimonial evidence for p prevents S from rationally or justifiably acting on p, however strong an epistemic position S is in regarding p on account of good testimonial evidence for p. This is an interesting idea, but her examples cannot be counterexamples to the SSI-friendly version of the knowledge-action principle, if having evidence of poor quality does not undermine S’s good epistemic position; if it does, then, as opposed to what Lackey claims, S’s epistemic position regarding isolated secondhand information cannot be nearly maximal without S’s action on such information being L-epistemically appropriate. The SSI-friendly version of the knowledge-action principle specifies when S’s epistemic position is strong enough to rationalize or justify S’s action, and the quantity of S’s evidence is, at least in light of traditional evidentialism, reckoned to be a determinant of the strength of S’s epistemic position. Hence, if L-epistemic inappropriateness of S’s action is merely a matter of S’s failure to have evidence of good quality, it is possible that S’s epistemic position is so strong that (b′) is not satisfied. I have argued above that this is not merely possible but also plausible for ONCOLOGIST.
6 Arguments Against (b): Eliza’s Action Is Epistemically Inappropriate in Other Ways
The discussion in the last section suggests that rationality, as Lackey conceives it, may not be the same kind as that which figures in the version of the knowledge-action principle the SSIist favors. She may contend that her examples still raise a problem for the version of the knowledge-action principle because it fails to capture the full-blooded sense of practical rationality. Obviously, to establish this problem requires more than putting forth intuitive examples. Nevertheless, it seems more charitable to read Lackey’s concern as being with (b) itself; she is concerned with whether Eliza is not rational simpliciter to act on isolated secondhand knowledge rather than whether the weakness of Eliza’s epistemic position is responsible for the non-rationality of her action. Be that as it may, note that Lackey concludes (b), that Eliza’s action is not rational, by cashing out our intuition that her action is L-epistemically inappropriate in terms of non-rationality. Lackey’s conclusion does not follow if there are other ways to explain our intuition. There are at least two ways of explaining our intuition of L-epistemic inappropriateness without equating it with lack of rationality or justification.
Although we can sometimes disclaim responsibility by saying “I did not know” or “I did not realize”, we often counter that disclaimer with the retort “Well, you should have known”. (Lucas 1993, p. 52).
In English there is a commonly-used locution to describe the purely negligent agent. We say that although she did not realize at the time that she was violating a norm she should have realized it. (Sverdlik 1993, p. 141; emphasis is original)
Lackey suggests that the social expectation for Eliza to fulfil her explanatory duty is constitutive of the L-epistemic inappropriateness of her action. I don’t find it implausible at an intuitive level that one’s expectation for Eliza implies that she should have known the relevant norm concerning her explanatory duty.17 If Eliza violates the norm she knows or is expected to know that she ought to follow, the L-epistemic inappropriateness of her action can be explained as her being epistemically irresponsible for her action.18 Epistemic responsibility in this sense is far from being a condition for practical justification or rationality; for example, a slightly illegal action, e.g., turning off one’s car lights during the day, can be rational or justified for S even though S knows or should know that it is illegal.19
If my union calls a strike, I may be within my epistemic rights in believing that I should strike, based on the testimony of trusted others, and I may be within my rights in acting upon that belief, but if I do not grasp the moral reasons for striking, I lack an important epistemic good. If my Church teaches that abortion is wrong, I may be within my epistemic rights in believing that it is wrong, but if I do not grasp the reasons for the wrongness, I am in an epistemic position inferior to the one I would be in if I did grasp the reasons. But this does not show anything peculiar to moral testimony, since I have argued that testimony in general cannot give us understanding. (ibid., p. 147)
Zagzebski focuses only on moral understanding, but her point applies to understanding in general. On her diagnosis of the examples here, S knows the relevant proposition by testimony and is within her “epistemic rights” in acting on that knowledge. Nevertheless, it is hard to deny that S’s action is still intuitively epistemically inappropriate, at least to some extent. To follow Zagzebski’s lead, it is possible to construe L-epistemic inappropriateness as lack of understanding.
Here is no place to get into the difficult issues of what understanding is and how it differs from knowledge.20 It is enough for my purposes here to point out that the essential features of understanding, as Zagzebski analyzes them, are all present in Lackey’s examples. Zagzebski argues that understanding has three essential features: first, understanding requires the mastery of a techné, a practical human art or skill to be found in fields ranging from the highly complex and professional, such as medicine and ship-building, to the mundane, such as cooking and game-playing. Second, understanding requires a grasp of explanatory relations among things. Hence, “[t]he person who has mastered a techné has a kind of understanding… He is able to explain features of the techné and to answer questions related to its practice” (ibid., p. 143). On Zagzebski’s account, understanding requires explanation, and it is distinguished from knowledge in that knowledge is possible if justification, as opposed to explanation, is gained. Third, but most importantly, unlike knowledge, understanding cannot be transmitted by testimony. For understanding requires S herself to grasp the relevant explanatory relations by putting her mind to work. This does not exclude cases in which a testifier transmits understanding after dwelling on the relevant explanatory relations with S. The point is that there is no case of isolated secondhand understanding.
All these features are important in Lackey’s examples. They are all cases in which S is unable to answer the request for explanation about some technical topic, even though the relevant piece of knowledge is delivered to S by testimony. Again, going back to ONCOLOGIST, Eliza knows but does not understand that Lucas has pancreatic cancer. Of course, qua oncologist, she understands what it is for a patient to have pancreatic cancer, and what symptoms and treatments are associated with it. What she lacks, I presume, is local understanding about Lucas’s cancer; after all, Eliza has no clue what is going on in Lucas’s body and how it is related to his pancreatic cancer, independently of the testimony. Thus, Eliza’s L-epistemic inappropriateness may be explained on account of her lack of understanding.
If Zagzebski is right that knowledge is possible where understanding is not, then it is possible that Eliza’s scheduling or administrating a highly aggressive combination of radiation and chemotherapy for Lucas is epistemically appropriate in the sense that it is rational or justified, while Eliza’s doing so is L-epistemically inappropriate.21 Zagzebski grants that in many cases, understanding is not so important. Perhaps, Lackey’s examples are designed so as to require that one value understanding even more than usual and to give rise to the strong sense of epistemic inappropriateness for acting without understanding. Lackey holds that Eliza qua doctor incurs a duty to explain why she has done what she does, and Eliza’s failure to fulfill this duty is responsible for the L-epistemic inappropriateness of her action. It is natural that the duty can be fulfilled only if she understands and is thereby able to explain Lucas’s health condition on her own. Furthermore, Lackey’s distinction between the quantity and quality of evidence may be better cashed out in terms of the distinction between knowledge and understanding, or justification and explanation. The latter distinction is already powerful enough to capture what is at stake in Lackey’s examples.22
I have argued that Lackey’s examples fail to be genuine counterexamples to the knowledge-action principle for a variety of reasons: (i) alleged knowledge in her examples may be denied on intuitive grounds, or (ii) it may be denied by appeal to the link between theoretical and practical justification or rationality; (iii) the knowledge-action principle in support of SSI allows for cases where S’s action is non-rational because it does not square with S’s preferences; and L-epistemic inappropriateness has little to do with rationality or justification, either because (iv) it may amount to epistemic irresponsibility or because (v) it may consist in lack of understanding. In the face of this arsenal of counter-resources, Lackey needs a stronger weapon to fight against the knowledge-action principle.23
There is more than one argument for SSI. Many regard the argument from the knowledge-action principle as the main one.
Lackey (2011), in much the same way, challenges the sufficiency version of the knowledge norm of assertion—the principle that states, roughly, that if S knows that p, then it is permissible for S to assert that p. Some of my arguments against Lackey’s alleged counterexamples to the knowledge-action principle, mutatis mutandis, work against those to the knowledge norm of assertion, but my main focus here is on the knowledge-action principle. Note that the knowledge-action principle and the knowledge norm of assertion are different principles, and usually need to be defended or objected to on different grounds. I agree with Brown (2012a) that they bear no straightforward relation to each other.
If anti-Gettiering conditions are not responsible for the strength of S’s epistemic position, it can be added (iv) that S is not Gettiered. Adding (iv) will make the relations between different senses of ‘epistemic’ discussed below more complex.
A way to maintain both the knowledge-action principle and purism about knowledge is to espouse infallibilism about knowledge. Infallibilism about knowledge arguably entails global skepticism, viz., that the epistemic standard is always that relative to which, for any S and p, S’s epistemic position regarding p is too weak to be good enough for knowledge. The knowledge-action principle, then, is trivially true because its antecedent is false. For more on the relation between SSI and skepticism, see Fantl and McGrath (2009).
On Fantl and McGrath’s definition, p being warranted enough to justify means that the weakness of S’s epistemic position with regard to p does not prevent p from being a justifier. It is, then, possible that p, even if warranted enough to justify φ-ing, fails to be a justifier for it, e.g., because p has no inferential relation to φ-ing (I discuss this point in more detail in Section 5). Throughout the paper, I follow this definition and apply it to other similar locutions, such as ‘warrant or epistemic position is strong enough.’
Although the knowledge-action principle, whether SSI-friendly or not, is formulated differently, and their differences are not marginal, I ignore most of them in much of the paper. This is because Lackey’s aim is not to attack the details of particular formulations of the knowledge-action principle. The only difference that is crucial for my purposes is between SSI-friendly formulations and others. I aim to defend the former from Lackey’s alleged counterexamples. Ichikawa (2012) defends the latter from many alleged counterexamples including Lackey’s. Ichikawa’s point is not incompatible with mine. Although he does not put his point this way, his main idea can be easily modified as follows: Lackey’s and others’ putative counterexamples to the knowledge-action principle are consistent with the possibility that S’s epistemic position regarding p is strong enough for S to justify or rationalize the action S does, but the degree of justification or rationalization p confers on the action is weak. This possibility remains because p may not be a good enough reason to justify or rationalize S’s particular action. I pursue a different line of response, since the relevant propositions in the putative counterexamples seem to be good enough reasons for what S does therein; though I agree with Ichikawa that to settle what interpretation of the examples is correct, there needs to be a principle determining what counts as good reason for what. I thank Jonathan Jenkins Ichikawa for helpful discussion.
Lackey seems to assume that the inductive strength of the evidence S has for the reliability of the source of isolated secondhand information p determines S’s epistemic position regarding p. This is perhaps why she holds that when Eliza knows that Anna is extremely reliable, Eliza is in a nearly perfect epistemic position with regard to Anna’s diagnosis. One may object to this assumption for the following reason: S’s epistemic position is significantly improved if S gains independent evidence for p, and discharges the isolated status of p, even when S has nearly perfect inductive evidence for the reliability of the information source. This may suggest that inductive evidence, however good, may not put S in a very strong epistemic position.
As a matter of fact, my intuitions about Lackey’s examples significantly vary in terms of content and strength. For some examples, I have strong intuitions that S knows and that it is appropriate for S to act. For ONCOLOGIST, my intuition is that Eliza knows but her action is somehow inappropriate. In Section 6, I attempt to articulate the sense of ‘somehow’ involved in my intuition.
MacFarlane (2005) argues that non-reductionism about testimonial knowledge, as it stands, is inconsistent with the knowledge-action principle for a principled reason: non-reductionism does not take account of interpersonal differences in stakes. Unless some further argument is given, it is an open question whether the knowledge-action principle or non-reductionism is to be modified. Lackey’s examples, even if granted to be legitimate, just extend this question so as to apply to reductionism as well. I argue below that the knowledge-action principle is theoretically more secure than Lackey claims it is.
Fantl and McGrath formulate the unity thesis in terms of a reason to believe rather than justification or rationality: “If p is warranted enough to be a reason you have to believe that q, for any q, then p is warranted enough to be a reason you have to φ, for any φ” (Fantl and McGrath 2009, p. 73). Again, details of the thesis do not matter much here. Hawthorne and Stanley (2008, p. 577) allude to, but do not argue for, the unity thesis.
Note that the results of the ultrasound and MRI were positive is not a reason for which Eliza believes that Lucas has pancreatic cancer. Rather, she has inferred the former from the latter.
My arguments in this section are modeled on Fantl and McGrath’s (2009) arguments against some alleged counterexamples to their knowledge-action principle, including one offered by Brown (2008). To counter the argument from the unity thesis, however, Brown (2012b) argues that her counterexample indeed goes against the unity thesis as well. I am convinced by Fantl and McGrath’s (2012a, 2012b) responses to Brown, and take it to be safe to assume the unity thesis.
This is one of the points Lackey makes in response to the objection that the intuitive inappropriateness in her examples is merely social rather than epistemic.
Fantl and McGrath, in their (2007), defend (3) entailing (4) by arguing that the conversational implicature from (3) to (4) is not cancelable.
Lackey holds that by closure, Eliza knows and hence rationally or justifiably believes that scheduling or administering a highly aggressive combination of radiation and chemotherapy for Lucas is the thing to do, whereas she is not rational or justified in doing the thing to do. As I argued in Section 4, unless more arguments are given, it is not easy to sever the link between theoretical and practical rationality.
Sher (2009) examines various analyses of ‘you should know’ involved in an agent’s responsibility for her wrong-doing. He refuses the analysis that merely situational factors constitute the relevant sense of epistemic responsibility, on the ground that it makes epistemic responsibility too foreign to the agent’s epistemic abilities to render it worth calling ‘epistemic’. Sher’s own analysis implies that the agent should know her wrongness or foolishness if her failure to recognize evidence for her wrong or foolish action (a) “falls below some applicable standard” and (b) “is caused by the interaction of some combination of his constitutive attitudes, dispositions, and traits” (ibid., p. 88). These conditions may not be applicable to some of Lackey’s examples. Notwithstanding the theoretical difficulty of factoring out precise conditions for epistemic responsibility, it is enough for my purposes here that there is some intuitive sense of ‘S should know the norm’ in the relevant examples in which S doesn’t know it; I only intend to give an account of what is behind our intuition of L-epistemic inappropriateness. It is not surprising if some of our intuitions conflict with a well-grounded account in the final analysis.
The epistemic irresponsibility for action I am suggesting may carry over to the epistemic irresponsibility for belief via a variant of the unity thesis. This is indeed how Van Woudenberg (2009) argues for epistemic responsibility for belief, oblivion, and ignorance. Van Woudenberg’s notion of epistemic responsibility for belief is a far cry from the notion so called in the traditional epistemological literature, and there is no way to make the former a necessary condition for knowledge. For this reason, even if the unity thesis about epistemic responsibility in his or my sense is correct, it does not imply that S does not know p when S’s acting on p is epistemically irresponsible in this sense. Kornblith (1983) tries to defend epistemic responsibility in the traditional sense as a necessary condition for knowledge by making use of a variant of the unity thesis.
When I wrote this paper, I lived in Canada, where Federal Motor Vehicle Safety Standard 108 requires that all vehicles made since 1 January 1990 turn on lights even in the daytime.
For these issues, see, e.g., Grimm (2006), Kvanvig (2003), and Zagzebski (2001, 2009). Zagzebski’s view on understanding as a whole may be controversial, but all I need to run the argument below is that understanding has essential features that are missing in Lackey’s examples, and the absence of such features is explanatory of L-epistemic inappropriateness. I believe that any available view on understanding can satisfy these conditions, but I pick Zagzebski’s because only she explicitly mentions how understanding is (not) related to testimony.
Hills (2009, 2010) holds, roughly, that moral beliefs lead to morally worthy action only if they are held on the basis of moral understanding, and that moral beliefs are not rational unless they are held on the basis of moral understanding. Some might argue from here that subjects in Zagzebski’s examples are not rational to believe or act on the proposition in question, and so her cases are counterexamples to the knowledge-action principle after all. What Hill means by ‘epistemic rationality,’ however, is much broader than the sense of rationality that matters in the knowledge-action principle (see Hill 2010, pp. 223–30). So Hill’s position does not exclude the point I’m making here, that the knowledge-action principle is compatible with epistemic inappropriateness of action in a broad sense. I thank Jennifer Lackey for suggesting comparing my view with Hill’s.
The same kind of diagnosis of Lackey’s examples as offered here, that L-epistemic inappropriateness is better explained by lack of understanding, is also offered by Carter and Gordon (2011), though their focus is on the knowledge norm of assertion attacked by Lackey (2011). They argue that the kind of understanding S lacks in Lackey’s examples is “atomistic understanding” of why p is evaluated as being as it is. Beyond what I claim here, I don’t commit myself to any substantive thesis about understanding in general and the kind of understanding relevant to Lackey’s examples in particular. Carter and Gordon claim that Lackey succeeds at denying what they call the “quantitative view,” assumed by many proponents of the knowledge norm of assertion: “The question “What is the norm of assertion?” is best answered by determining how much epistemic support is required to warrant assertion” (Carter and Gordon 2011, p. 619). What they mean by ‘quantity of support’ is close to epistemic in the sense of being truth-conducive or error-averse. The quantity view is denied because, they argue, Lackey’s examples suggest that at least in some cases, understanding rather than knowledge is the norm of assertion. As they are aware, whether the denial of the quantitative view entails the denial of the knowledge norm of assertion depends on what sense of propriety or appropriateness of assertion is relevant to the knowledge norm of assertion. The SSI-friendly version of the knowledge-action principle does not involve this kind of ambiguity, since the relevant sense is determined to be quantitative, as Carter and Gordon use the term. For this reason, the SSI-friendly version of the knowledge-action principle is compatible with the claim that epistemically appropriate action in this sense is not L-epistemically appropriate due to S’s lack of understanding.
This does not mean that Lackey’s examples are not interesting for any epistemological reason. Epistemology covers other cognitive states than knowledge. I have no doubt that L-epistemic appropriateness provides fuel for further epistemological theorizing.
I thank Roger Clarke, Jeremy Fantl, and Carrie Ichikawa Jenkins for helpful discussions and comments on earlier drafts of the paper. A short version of the paper was presented at the 34th International Wittgenstein Symposium. I appreciate discussions there, especially with Jennifer Lackey.