Scepticism and Reliable Belief

José Zalabardo takes an unusual approach to skepticism in this book. Most epistemologists acknowledge the force of skepticism but aim to mitigate its threat. Zalabardo tells us that he undertook to show that “our philosophical outlook was incompatible with some of our most cherished knowledge claims” (vii). He found that reliabilism presented “the main obstacle to a cogent sceptical argument” (vii), and so he set out to discover some important type of knowledge that was not skeptic proof. He thinks he's found it. More exactly, he argues that reflective beliefs, or cognitive self-approvals (CSAs), are unknowable on what is otherwise the most plausible, antiskeptical theory of knowledge.According to Zalabardo's original version of reliabilism, there are three types of knowledge: foundational, truth-tracking knowledge; evidence-based inferential knowledge; and knowledge where one simply has a true standing belief, that is, when one has an “innate predisposition to form [a belief] that is largely independent of input, but [one is] not equipped to detect its truth value” (137). Zalabardo's epistemology is novel both in its specific tripartite structure and in its development of each part.Zalabardo argues that tracking is sufficient for basic knowledge and that tracking consists substantially in a version of Nozickean sensitivity, which says that S knows that p only if, were p false, S would not believe that p. Zalabardo's version of tracking also accommodates the insights underlying both Nozick's appeal to adherence (if p were true, S would believe it) and the increasingly popular safety condition (in close worlds where S believes that p, p). Rather than construing these well-known conditions in a possible worlds framework, Zalabardo follows Sherrilyn Roush's (2005) lead and explicates them in terms of conditional probabilities. On Roush's presentation, sensitivity is satisfied for S's belief that p when the probability that S does not believe that p, given that not-p, meets a threshold. (Adherence is handled similarly.) Zalabardo, however, rejects probability thresholds and favors high likelihood ratios: S tracks that p only if the ratio of the probability that S believes that p, given that p, to the probability that S believes that p, given not-p, is sufficiently high. Having articulated well-known problems with adherence while acknowledging its purpose, Zalabardo elegantly shows how the likelihood ratios capture the importance of sensitivity while speaking to the point of an adherence condition. Finally, S knows that p via tracking iff S's belief has a high tracking ratio and a high level of safety.Zalabardo rejects Nozick's (1981) claim that we must hold constant S's actual method of belief formation when assessing whether the belief satisfies the sensitivity condition. The claim was well motivated, but appeals to methods have encountered numerous problems in the literature. Nozick himself was inconsistent in his application of methods to handle problem cases. Zalabardo thinks we can do without methods and that we can handle the cases for which they were designed—cases where S knows that p but his or her belief fails to satisfy non-method-relativized sensitivity—primarily by treating them as cases of evidence-based knowledge. In Nozick's famous example, Granny knows that her grandson is well because her visual evidence supports that belief. In chapters 4 and 5, bookended by his treatment of tracking, Zalabardo articulates his theory of inferential knowledge based on evidence. Here again, likelihood ratios play a starring role. E supports H just in case the ratio of the probability of E given H to the probability of E given not-H is sufficiently high. S knows that H inferentially via E iff S knows that E and that E supports H.Zalabardo's accounts of tracking and inferential knowledge permit responses to three common developments of external world skepticism—those based on the regress of reasons, skeptical hypotheses, and the problem of the criterion. Foundational, tracking knowledge blocks the regress of reasons, and it blocks the skeptic's claim that our inability to rule out radical skeptical scenarios undermines everyday knowledge. What about the claim that we can't know we're not in one of the scenarios? And how do we know that our empirical beliefs based on sense perception are reliably formed? (This question arises from the problem of the criterion.) To these questions, Zalabardo offers standing knowledge, saying that, “For standing beliefs, truth is the only form of epistemic excellence that we can reasonably demand” (138). Thus, he claims, true standing belief suffices for knowledge. Of course, the skeptic might well appeal to the quoted sentence and say, “And that's why you don't know, because all that can be reasonably demanded is woefully inadequate.”For present purposes, let us accept knowledge of standing beliefs. Where does that leave us? We still cannot know that our CSAs are true. Suppose I believe truly that p. Do I know that my belief that p is true? The by-now familiar problem is that my belief that I don't falsely believe that p cannot track the truth. If I did falsely believe that p, I would believe that I don't falsely believe that p. Zalabardo also argues that there is no adequate evidential support for CSAs and that they're not standing beliefs. The upshot is that Zalabardo has located the one type of important knowledge that his theory, otherwise skeptic-proof, cannot accommodate. Zalabardo offers a tentative line on knowledge of CSAs in the final chapter but acknowledges that its further development is a task for future work.I noted earlier the peculiarity of Zalabardo's approach—aiming to locate some kind of knowledge-proof skepticism (as opposed to the usual skeptic-proof knowledge). Given this, it's also peculiar that he subsequently seems hell-bent on defeating every possible form of skepticism. To that point, I'd like to add a few more specific questions and concerns.First, before developing his theory of knowledge, Zalabardo responds to Bonjour's (1985) qualms about externalism generally, qualms based on the thought that knowledge requires evidence-based reasons because believing without adequate evidence is irrational and irresponsible. Zalabardo argues that epistemic rationality and responsibility do not always require believing on the basis of adequate evidence. One can easily understand why a reliabilist would want not merely to dismiss Bonjour's concerns but also to answer them, and yet, given Zalabardo's commitment to tracking as sufficient for basic knowledge, it's hard to imagine how all such cases of knowledge satisfy an epistemic responsibility condition, no matter how the condition is construed. (Think: Truetemp, and add whatever is needed to the standard version to make him epistemically irresponsible.)Second, Zalabardo's revisions of the general Nozickean approach raise some questions. (a) Discussing problems with appeals to methods, which Nozick introduced to solve the problem of insensitive knowledge, Zalabardo rejects the approach based primarily on a case of method-sensitive belief without knowledge (61). But few think sensitivity is sufficient for knowledge. (b) Zalabardo doesn't always give the method-based version of sensitivity the credit it deserves. In my view, it can handle the well-known Grandma, Dack the dachshund, and Williamsonian margins-for-error cases pretty well. It is only through having already rejected the method-based approach (chapter 3) that Zalabardo can claim that Nozickean sensitivity flunks these cases (chapter 6). (c) With Zalabardo's addition of standing-belief knowledge, his theory countenances fewer closure violations than Nozick's: I do know that I'm not a BIV. But Zalabardo admits that his theory allows closure violations nonetheless. I'm on board with this, but for many this is the central problem with Nozick's epistemology. Will critics find Zalabardo's theory a genuine improvement here?Finally, perhaps the most original aspect of Zalabardo's book is his identification of a knowledge-proof skepticism, namely, that one cannot know one's CSAs, which he initially characterizes as believing that one's belief is true. Zalabardo later defines CSAs as believing that one does not believe falsely. How, precisely, ought we to gloss these reflective beliefs? As Zalabardo originally does? Salerno (2010) and I (2006) have argued, independently, that these are knowable. If my belief that my belief that p is true were false, perhaps that would be because I wouldn't have that first-order belief, or because the first-order belief would be false, but in either case, I might well not believe that I have that true belief. On a similar analysis, I can also know that I know. It's true that I cannot know that I don't falsely believe, but not all CSAs take this form. A problem remains, but some CSAs are trackable.In closing, I want to acknowledge that there are genuine problems with method-based tracking, that various CSAs really do present difficulty for tracking, and that Zalabardo's treatment of these and other issues is often original and inspired. Take my critical responses as raising questions for further reflection, not as attempts to refute

Appeal · Argument (complex analysis) · Epistemology · Political science · Relation (database) · Set (abstract data type) · Skepticism · Computer Science · Epistemology, Ethics, and Metaphysics · Law · Logic, Reasoning, and Knowledge · Philosophy · Psychology

• The structure of empirical knowledge

  Laurence Bonjour•The structure of empirical…•1985

  Add to list

• Tracking Truth

  Sherrilyn Roush•TRACKING TRUTH: KNOWLEDGE,…•2005

  Add to list