Although I’ve not been around for Brain Pains discussions this semester, that doesn’t mean I haven’t been getting some work done. It’s just that whatever work I get done, I get done in a remote location. Typically, it’s a moving train. I’ve dashed together a draft of a paper responding to some arguments of Comesana [...]
In the last post I tried to raise a couple of questions meant to clarify Stalnaker’s stance on the knowledge argument. In this post I want to be slightly more critical. In particular, I want to raise two potential problems for parts of Stalnaker’s answer the the knowledge argument. (Note: just as I was finishing [...]
In Our Knowledge of the Internal World, Robert Stalnaker offers his answer to the knowledge argument. Unfortunately, I find it difficult to determine which of the premises he denies: the epistemic premise, the metaphysical premise, or both. This post is basically a request for clarification.
There are several different ways to formulate the knowledge argument. One simple way to see the argument is in terms of an epistemic step and a metaphysical step.
1. (Epistemic Step) Mary knew all the physical truths when she was in the room, but she learned something new when she left the room.
2. (Metaphysical Step) If Mary learned something new when she left the room, the physical truths are not all the truths.
Conclusion: The physical truths are not all the truths, so physicalism is false.
In Our Knowledge of the Internal World, Robert Stalnaker offers his answer to the knowledge argument. Unfortunately, I find it difficult to determine which of the premises he denies: the epistemic premise, the metaphysical premise, or both. This post is basically a request for clarification.
There are definite signs that Stalnaker denies the second, metaphysical premise. In chapter 4, Stalnaker seems to characterize the type of information Mary gets as “essentially contextual” information (ECI), which one gets, I take it, when one gets new information in virtue of gaining a belief the content of which “cannot be detached from the context in which it is expressed or thought.” (OKIW, 81) (In fact, Stalnaker says that cases involving ECI are analogous to Mary’s case. I don’t know whether that means he ultimately thinks her case is one that involves ECI or not.) ECI is gained, for example, when John Perry looks at the aircraft carrier Enterprise when its middle is obscured by a building. He might first think the bow and the stern are not part of the same ship, until someone points out that “This ship” (pointing at the stern) “is this ship” (pointing at the bow). If Mary’s new knowledge is like this, it seems no metaphysical inferences can be drawn, because no real possibilities are being eliminated.
Elsewhere, though, Stalnaker seems to be denying even the epistemic step. In chapter 4 section 4, he denies what he calls The Principle of Phenomenal Indistinguishability: If a possibility is an epistemic alternative for a knower at a time…then it is phenomenally indistinguishable from the actual world to the knower at that time. (OKIW 88) Stalnaker then presents an example that supposedly conflicts with this principle. Leaving the details of that example aside—I intend to talk about them more explicitly in my next post—it is worth asking how this relates to the Mary case. One natural answer is that when Mary leaves the room, though she enters a new phenomenal state, she has not actually narrowed her epistemic alternatives. This, I take it, is to deny the epistemic step—she really didn’t learn anything new. She entered a new phenomenal state, but that doesn’t eliminate any epistemic alternatives for her. Notice that this is different from saying that she could rule out epistemic alternatives that are not really metaphysical possibilities. That is the traditional way of denying the metaphysical step.
So does Mary learn anything according to Stalnaker? In Chapter 5 section 5, Stalnaker starts to address this question head on. “Just having the experience of seeing the red star was not enough to give Mary the evidence she needed to eliminate the alternative possibility that she was having an experience with a very different qualitative character (the qualitative character she would have had if she had seen a green star.) But what were the epistemic consequences, for Mary, of leaving her room and seeing color for the first time—what did she learn?” (OKIW 105). Unfortunately, I don’t ever find a clear answer to this question. Stalnaker dismisses several possible answers—that she acquires a new concept or that she becomes acquainted with a phenomenal essence—but I don’t quite get his positive response. Thus, I’m left a little unsure—does he reject the epistemic premise, the metaphysical premise, or both?
Here’s one way Stalnaker might respond, based in part upon the semantic model he developed in chapter 3: When we ask “does Mary learn anything new?” we are really asking an underdefined question. Or rather, the answer is yes, in a sense, and no, in a sense. She does enter a different belief state—that is, her total epistemic circumstance has changed. She does not, however, come to have a justified belief with a content that she did not know. Put another way, she does gain essentially contextual information, but she does not gain new information in a way that can be specified in a context-independent fashion. The anti-physicalist claim that she gains new phenomenal information is a result of conflating these two senses in which she gains information.
That’s just me trying to don the hat I’ve been learning to wear from reading Stalnaker’s book. I’d be happy to hear about whether this strikes readers—or even better, Stalnaker himself—as a good interpretation. If it is a good interpretation, I have one more question. How is this an improvement on Perry’s solution in terms of reflexive content? In Chapter 2, Stalnaker’s main criticism of Perry’s solution is that Mary’s gain in knowledge can—under some circumstances—be divided into two parts. In the first part, Mary learns what it is like to experience red. In the second part, she learns that it was red she experienced. I agree with Stalnaker that the first part is what is most important, but I don’t quite see what Stalnaker says is happening in the first part. Perhaps Mary is entering a different state, but that doesn’t seem epistemically relevant if there is nothing like acquaintance that results from entering that state. So, I don’t quite see the advantage of Stalnaker’s solution over Perry’s.
-->In his recent book, Our Knowledge of the Internal World, Robert Stalnaker offers a novel resolution to the Sleeping Beauty problem (pp. 59-64). This resolution depends heavily upon Stalnaker’s view of possible worlds as being somewhat more numerous than many (especially in a Lewisian tradition) have thought (Chapter 3). Other contributors will likely [...]
The Brainpains group has started reading Robert Stalnaker’s new book “Our Knowledge of the Internal World” recently released by OUP. We will have read chapters 1-4 by the middle of next week, and hope to have a post up soon after that. Stalnaker is one of those philosophers you can’t help but learn from, so [...]
This week’s Brain Pains reading was Alex Byrne’s paper, “Introspection” (Philosophical Topics 33 (2005)). Much of Byrne’s paper concerns the “inner sense” model of self-knowledge. According to the “inner sense” model, I have knowledge of my own mental states by means of “some sort of mechanism (perhaps more than one) for detecting my own mental [...]
This week’s Brain Pains reading was Alex Byrne’s paper, “Introspection” (Philosophical Topics 33 (2005)). Much of Byrne’s paper concerns the “inner sense” model of self-knowledge. According to the “inner sense” model, I have knowledge of my own mental states by means of “some sort of mechanism (perhaps more than one) for detecting my own mental states—something rather like visual, auditory, and gustatory systems, although directed to my mental life. That is, I have knowledge of my mental life by a special kind of perception,” or perception-like faculty for detecting my own psychological states (2). Byrne devotes much of the paper to defending the inner sense model from various prominent criticisms, and distinguishing it from various alternative accounts (such as Moran’s “self-constitution” view).
There are a lot of interesting ideas in his discussion of the inner sense model and in his analyses of competing views. But since this is going to be a longish post, I’ll skip straight to Byrne’s positive account.
Byrne states that there are two remarkable facts about self-knowledge that need to be explained. First, we seem to have a kind of privileged access to our own mental lives. I know (at least many of) my own mental states—e.g., that I believe that the cat is indoors—in a way that is less prone to error than the way in which I know empirical facts (like the fact that the cat is indoors), or facts about the mental states of others (like the fact that Fred believes that the cat is indoors) (2-3).
Second, Byrne states, “knowledge of one’s mental states is peculiar in comparison to one’s knowledge of others’ minds. One has a special method or way of knowing that one believes that the cat is indoors, that one sees the cat, that one intends to put the cat out, and so on, which one cannot use to discover that someone else is in the same mental state,” (3). Byrne notes that we know others’ mental states in a way that is in important respects similar to the way in which we know empirical facts about our environment: by perceiving how things are in our surroundings. Just as I can see that the cat is indoors, it is by seeing Fred’s behaviors that I can infer that he believes that cat is indoors, or wants the cat to go out, etc. Knowledge of one’s own beliefs, desires, and other mental states is typically not acquired by inferences drawn from observation of one’s own behavior. Of course, one sometimes does come to know what one believes (desires, etc.) by observing one’s own behaviors—e.g., in a theraputic context. But typically, one can know one’s own mind in a some other, “peculiar” way.
So how are these two features of self-knowledge—privileged access and peculiar access—to be explained?
Byrne’s own account is mostly focused on knowledge of one’s own beliefs. He begins by noting that much of our knowledge is acquired by following epistemic rules—namely, conditionals of the form
R If condition C obtains, believe that p.
An example of an epistemic rule, according to Byrne, would be
DOORBELL If the doorbell rings, believe that there is someone at the door.
What does it mean to “follow” an epistemic rule? Part of the answer to this question is that for one to follow a rule like DOORBELL, one must form the relevant belief (the one specified in the rule’s consequent) because one recognizes that the antecedent condition obtains. If Mrs. Hudson knows that there is someone at the door, and if her knowledge is to be explained in terms of her having formed the belief that there’s someone at the door by following DOORBELL, then Mrs. Hudson “believes that there is someone at the door because she recognizes that the doorbell is ringing,” (23). Byrne goes on to say, “The ‘because’ is intended to mark the kind of reason-giving causal connection that is often discussed under the rubric of ‘the basing relation.’ Mrs. Hudson might recognize that the doorbell is ringing, and believe that there is someone at the door for some other reason; in this case she does not form her belief because she recognizes that the doorbell is ringing,” (23).
Byrne then states that an individual S follows an epistemic rule of the form “If C obtains, then believe that p” if and only if
(i) S believes that that p because she recognizes that conditions C obtain,
which, according to Byrne, entails the following three conditions:
(ii) S recognizes (hence knows) that conditions C obtain,
(iii) Conditions C obtain; and
(iv) S believes that p. (23)
Byrne goes on to claim that one can gain knowledge of one’s own beliefs by following the rule:
BEL If p, believe that you believe that p.
Byrne thinks that we should explain knowledge of one’s own beliefs in terms of a capacity to follow BEL. Doing so seems to allow us to account for the fact that we have peculiar and privileged knowledge of our own beliefs.
Obviously, one cannot follow BEL in the course of forming beliefs about the beliefs (or other mental states) of others. Since following BEL is a way that I (and only I) can come to know my own (and nobody else’s) beliefs, Byrne thinks that BEL accounts for the peculiar access I have to my beliefs.
What about privileged access—the fact that I typically know what I believe in a way that seems largely immune to the sorts of errors that I am prone to make in forming beliefs about others’ minds, or in forming perceptually-based beliefs about things in my environment?
Well, as Byrne points out, to follow BEL is to believe that you believe that p because you recognize that p obtains. But according to Byrne, “recognizing that p obtains” entails knowing—hence believing—that p. So if you form the belief that you believe that p by following BEL, this belief will be true; there’s no way for you to follow BEL without believing that p, hence without making the belief that you believe that p true. Thus, according to Byrne, “BEL is self-verifying in this sense: if it is followed, the resulting second-order belief is true,” (26). Note that many other rules of belief-formation, such as
BEL-3 If p, believe that Fred believes that p
are not self-verifying: “the result of following [BEL-3] may be (indeed, is very likely to be) a false belief about Fred’s beliefs,” (26).
As Justin Fischer emphasized in discussion (and as Byrne himself notes), BEL can only be (successfully, actually) followed in cases where the belief one thereby self-attributes is true. If I falsely believe that p, then I can’t know that p—hence I can’t recognize that p obtains, so my belief that I believe that p cannot be caused by my recognition that p. In such cases, Byrne says that I can know that I believe that p by trying to follow BEL, where trying to follow BEL involves mistakenly concluding that p, and (on the basis of this mistaken belief) coming to know that I believe that p.
I want to raise a series of (often related) questions about Byrne’s account.
First, a purely clarificatory point: on pages 27-28, Byrne sketches three kinds of errors that an individual could make, which Byrne labels “Type I,” “Type II,” and “Type III” errors. I just want to ask about Type III errors. Byrne characterizes a Type III error, when attempting to follow a rule of the form “If C, then believe that p,” as a case in which “not-p, and one believes that p, but not because one knows or believes that conditions C obtain. Perhaps too much coffee might have led one to believe [that p], even if [condition C had not obtained],” (p. 28). I take it that Type III errors can occur when one recognizes that condition C obtains and forms the belief that p, but where the belief that p is not caused (or: not caused in the appropriate way) by one’s recognition that C obtains.
But a few paragraphs later, Byrne states, “If one follows BEL, only Type III errors are a threat to one’s knowledge: perhaps too much coffee would have lead one to believe that one believes that the doorbell is ringing, even if one had not believed that the doorbell is ringing,” (28). My question is this: if one really follows a rule of the form “If C then believe that p,” doesn’t that just entail that one believes that p because one recognizes that C obtains? If it does, then if Type III errors are cases in which one “believes that p, but not because one knows or believes that conditions C obtain,” then doesn’t it follow that it’s impossible both to make a Type III error and follow a rule? If the rule is, “If C, believe that p,” and I recognize that conditions C obtain, and I believe that p but not because I recognize that conditions C obtain, then won’t I have failed to follow the rule? (Perhaps I will have tried but failed? See below…)
Second, it seems a little odd, or awkward—to me, anyway—that following an instance of BEL is not, on Byrne’s account, strictly speaking a necessary condition for authoritative (peculiar and privileged) knowledge of one’s own beliefs. Suppose it is false that p. Presumably, if I believe that p, I can still know, in the peculiar and privileged way, that I believe that p. But of course, if p isn’t true, then I cannot recognize that p—I can’t know that p obtains. Hence, by Byrne’s account of what it is to “follow” BEL, I can’t follow an instance of BEL.
Byrne’s response to this is to say that one can try to follow BEL: “this situation is commonplace: trying to follow BEL, one investigates whether p, mistakenly concludes that p, and thereby comes to know that one believes that p. (In these cases, one will know that one believes that p on the basis of no evidence at all)” (29). So one can have authoritative self-knowledge even when one fails to follow any instance of BEL, as long as one at least tries to follow it.
Presumably, there are many different ways of trying, but failing, to follow BEL. For example, what if I recognize that p, and try to figure out whether or not I believe that p, but I end up coming to believe that I believe that p but not because I recognize that p—i.e., what if, when trying to follow BEL, I make a “Type III” error? Presumably, in such a case, Byrne would want to say that my resulting meta-belief wouldn’t constitute privileged and peculiar self-knowledge. So there needs to be a distinction between different ways of trying but failing to follow BEL, and the epistemically relevant distinction looks like it should be something like this. In the case, where your initial belief that p is false (and everything else works in the usual way), your belief that p causes your belief that you believe that p—you believe that you believe that p because you believe that p (in the relevant sense of “because”). That kind of “trying” to follow BEL still allows self-knowledge, whereas trying to follow BEL doesn’t give self-knowledge if one falls short of actually, successfully following BEL due to a Type III error. Call the case where the only reason you don’t follow BEL is simply because your belief that p is false a “brute error.”
So now, the account of self-knowledge looks something like this: I authoritatively know that I believes that p if (and only if?) I either form the belief that I believe that p as a result of following BEL, or I form the meta-belief by trying to follow BEL but fail due to brute error.
This expanded account of self-knowledge is thus disjunctive and seems a little, well, awkward (at least to me). It seems somewhat odd, I think, to say that one can succeed in following no epistemic rule whatsoever (because the self-attributed belief happens to be false), yet have perfectly good self-knowledge. But I think we could get a more-or-less extensionally equivalent account by doing one of the following:
Option #1: Emend the account of following a rule. Say that S follows a rule of the form “If C, then believe that p” just in case S believes that p because S believes that C obtains. Then, saying that S has authoritative self-knowledge just in case S (actually, successfully) follows BEL wouldn’t rule out cases of brute error.
Another way to go would be to emend the conception of how rules should be formulated. My (extremely rough-and-ready) proposal is something like this:
Option #2: Epistemic rules are always rules of the form “If C, then believe that p,” but the variable ‘C’ is only allowed to range over possible states of the individual to whom the rule applies. (Go ahead and allow factive and other relational states if you want.) Now, replace BEL with BEL*: If you believe that p, believe that you believe that p. To avoid trivializing the account, emend the account of following a rule as follows: S follows a rule of the form “If C, then believe that p” just in case S forms the belief that p because S is in state C (not: because S recognizes that S is in state C). As before, let’s assume that the relevant sense of “because” is a reason-giving one.
Here’s what we might regard as a reason to prefer the second option. Suppose that we want to allow perceptual states to appear as antecedents of our epistemic rules. Suppose, for example, that we want rules roughly like, “If you see an orange, believe that That’s an orange.” Then on the original account, you’d need to recognize that you see an orange in order to follow this rule. That seems rather intellectualist. Roughly put, I think we might want the seeing of the orange (or, at least, one’s perceptual state as of an orange) itself to be treated as the reason or epistemic basis of the belief that That’s an orange, rather than one’s belief or knowledge that one sees an orange. (I think it’s most plausible to say that a little child can know that That’s an orange, without knowing that he or she sees an orange, or that he or she is in an orange-presenting perceptual state, etc.)
Let me address some worries about the second option. First, the fact that the antecedent condition specified in BEL* is a mental state might lead one to conclude that BEL* isn’t “transparent.” One might worry that one wouldn’t follow it by focusing “outward,” on the world as it were, but “inward” on one’s own mental states. But my proposal was to emend the notion of “following” a rule, so that you don’t follow a rule by “looking to see” if the antecedent condition obtains. You follow a rule just whenever you form the relevant belief because you’re in the state specified in the antecedent. And of course, you only get to be in the state specified in the antecedent of BEL*—namely, you only get to be in the state of believing that p—by “looking outward,” by determining what’s true. So, given the emended account of following a rule, BEL* (like any rule with a first order belief or perceptual state in the antecedent) is, I think, transparent in the relevant sense. Note also that BEL* is self-verifying: you can’t follow it without making your belief that you believe that p true. (To follow BEL*, after all, you must actually believe that p, otherwise you can’t believe that you believe that p because you believe that p.)
I think that, with either of these emendations, Byrne could then just say that to have privileged and peculiar self-knowledge, one must simply follow BEL (or, on the second proposal, BEL*). A separate account of trying to follow a rule (together with a distinction between epistemically good and bad instances of trying) wouldn’t be needed. The question is: am I missing something? Am I missing a reason why Byrne doesn’t take one of these other options?
Third, there is a worry that, according to Robert Howell, has been raised by Brie Gertler. As I understand it (I haven’t read Gertler on this; I’m going largely off Robert’s comments in discussion) the worry is that Byrne’s account requires one to reconsider the question of whether p in order to have authoritative knowledge of one’s own beliefs. In order to follow BEL, it seems that one must either revise or at least re-assess one’s beliefs. But doing so will inevitably change what one believes. So the question of “What do you believe (now, at time t1)?” can’t be answered by following BEL. By the time one follows BEL (time t2), it is altogether likely that one will have changed one’s mind. BEL seems better addressed to a request like, “Deliberate for a while, and then, once you’re done, tell me what you believe,” rather than “What is it that you presently believe?” I think the worry, then, is that BEL is limited. A broader account of self-knowledge, one that applies to reporting one’s own standing, settled opinions as well as to one’s re-considered (or newly-considered) judgments, is wanted.
Fourth and finally, I want to come back to this matter of what it is for one to believe that one believes that p “because” one believes that p. This is an extremely hard problem: as Byrne notes, it’s more or less the familiar “basing relation” problem. I think that getting clear on this problem—or, at least, making some headway on the problem as it crops up in specific epistemological contexts—is important. In the present context, we need to be able to say why BEL couldn’t be followed in a Rylean way. Suppose I believe that my boss is trying to ruin my career, but I only come to know that I believe this because I reflect on my behavior (I avoid my boss, etc.) and infer that I must believe that he’s trying to ruin my career. I assume this is a case in which my self-knowledge would be neither privileged nor peculiar; I know my own mind in this case in very much the same way a therapist would know my mind.
Assuming my belief about my boss is actually well-founded, what’s to stop us from saying that the following conditions obtain:
(1) I recognize that my boss is trying to ruin my career,
(2) My boss is trying to ruin my career,
(3) I believe that I believe that my boss is trying to ruin my career,
and thus, that I have followed an instance of BEL?
The answer is supposed to be, I think, that as I described the case, I wouldn’t believe that I believe that my boss is trying to ruin my career because I recognize that he’s trying to ruin my career, but instead because I inferred that I have this belief by observing my own behavior. But why characterize the route to the self-knowledge in this way? After all, don’t I believe that I believe that my boss is trying to ruin my career in some sense precisely because I believe he’s trying to ruin my career? (Let’s suppose that, if I didn’t believe this, then I wouldn’t engage in the behaviors that evince this belief.)
Presumably, Byrne would say that the sense in which I hold the meta-belief because I hold the first-order belief, in the case as I described it, is a “purely causal” or “non-reason-giving” sense of “because.” Now, I think I have a grip on the distinction between causal and reason-giving senses of “because.” Let’s grant that there’s a sufficiently intuitive distinction there. But what is unclear—and, in my view, the philosophical crux of the matter—is what it means to say that my first-order belief that p causes my higher-order belief that p in the reason-giving way. Presumably, the first-order belief doesn’t cause the higher-order one all by itself; various specific conceptual competencies, for instance, must be activated. But then there are two questions: first, what are those competencies?; and second, why doesn’t their activation rob my first-order belief of its reason-giving causal powers, the way that my inferences about my behaviors result in a non-reason-giving relationship between my belief about my boss and my knowledge that I have this belief? I guess my suggestion is that we need to make some progress on these sorts of questions in order to fill out exactly what it means to follow a rule like BEL.
-->
Hi all,
Since the Scwitzgebel piece generated such lively discussion, we’re going to stay on topic reading Alex Byrne’s “Introspection” next week. Those who like to play at home are invited to join us!
Best
RH
We’ve just read Eric Schwitzgebel’s interesting “The Unreliability of Naïve Introspection”, recently published in Phil. Review. In general, I seem to find myself sympathetic to the sort of worry he’s trying to raise. In the philosophy of perception, in particular, one often hears claims about the so-called ‘naïve’ conception of experience (or about the metaphysics [...]
After an unfortunately long hiatus, with some of us preparing for tenure review, Brainpains is back with more brains and more pains. SMU has added two new philosophers of mind to our tenure-track ranks: Justin Fisher, previously a student at Arizona and a postdoc at UBC, and Matthew Lockard, fresh out of the excellent program [...]
Back from the beach at last! We’ve just read Sydney Shoemaker’s “On the Way Things Appear” in the 2006 Gendler/Hawthorne collection [Perceptual Experience, OUP]. Here, I’m interested in a very small point about the ontology of appearance properties (nothing to say about Shoemaker’s nice response to Michael Thau).
[Warning: This post was intended to be simple [...]