Neta and Pryor on the Basing Relation

In the “Hot Topics” post, a new topic of discussion has begun, and it may be useful to call attention to it in a separate post so that it doesn’t get lost in the other one. It is about the basing relation, and it began with Jim’s comment:

Here’s an example: one promising way to draw an internalism/externalism contrast is to say: X is eligible to be an INTERNAL justifier of your belief iff X is the sort of thing upon which you could BASE that belief. This preserves some of the familiar contrasts: e.g., intuitively, the reliability of your belief is not a candidate basis. But it may end up diverging in some ways from the more familiar definitions of internalism. Still, I think it’d be a fruitful contrast—if only we had some informative, independent story about what can count as a basis…

Ram followed up with 2 proposals:

(1) S’s belief that p is based on X iff S can know, by reflection alone, that S’s belief that p is caused by X. (So, for S’s belief that p to be based on X, it has to at least be caused by X. But not just any old cause can be the basis — it has to be a cause that S can recognize as such by reflection alone.)

(2) S’s belief that p is based on X only if S believes — or is at least inclined to believe — that X is a good reason for believing that p. (If S is not even inclined to believe that X is a good reason for believing that p, then, X might cause S to believe that p, but X is not a reason upon which S’s belief is based.)

(1) and (2) both strike me as promising.


To which Jim and I each raised concerns:

Both proposals do have some appeal. But a worry re (2): I’m inclined to think subjects can have based beliefs even if they lack epistemic concepts and epistemic beliefs. They can respond to reasons without yet being able to think about reasons as reasons. If that intuition is worth respecting, then (2) needs to be weakened somehow.

A similar worry re (1): I think subjects can respond to reasons without yet having the concept of a cause. It’s not clear to me, for instance, that Homer’s Greeks had either the concept of a cause or the concept of a reason. But I think they had based beliefs. You have a “can” in your formulation of (1), so that might be able to accommodate this. But a lot will depend on how that “can” gets articulated.

More pressure on that “can”: what if the subjects has background evidence, or mere beliefs, that interfere with their knowing that X caused them to believe P. E.g., they may be skeptics about causation, and so refrain from forming any beliefs to the effect that this caused that. Or, if they do form any such beliefs, the beliefs may fail to count as knowledge due to the incoherence with their skeptical commitments. Yet I’d have thought they could still form based beliefs. Maybe we should say, they’re at least _in a position_ to know by reflection alone that X caused their belief, if only they abandoned their skeptical beliefs. As I said, though, this puts more pressure on how exactly to spell out the “can” in your formulation of (1).

Ram, maybe you can explain idea (1) a bit more. My initial reaction is to think it is not promising. If causation exists, I don’t see how it could ever come to be known by reflection alone what the relata of a given causal relation are. Inferential basing, I assume, would be a good test case, but coming to believe something by inferring it from something else you believe doesn’t strike me as a case where you know, by reflection alone, that the second causes the first. If you reflect, that may be the conclusion you come to, but I don’t see how you could know it by reflection alone. Think of common causes, mere correlations, overdetermination, pseudo-overdetermination, pre-emption, double pre-emption, and all the other landmines that exist whenever we try to ascertain what causes what. I don’t see how reflection alone puts us in a position to know that, e.g., there was no common cause, Freudian or otherwise, at work (nor am I sure that such a common cause would undermine proper basing…). (Note that closure opponents have some wiggle room here that the rest of us don’t…)

We haven’t had much discussion of the basing relation here, so these comments may provide a useful starting point for one.


Neta and Pryor on the Basing Relation — 53 Comments

  1. Ram’s first proposal has to be at least partially correct, it seems to me, though any causal account of course raises the familiar problem of deviant causal chains. Ram’s proposal that the causal relation be accessible by reflection will handle many such problem cases, but presumably not all. My own feeling is that the deviant cases can best be handled by a dispositional account.

  2. This is a question about Pryor’s proposal:

    PB: X is eligible to be an INTERNAL justifier of your belief iff X is the sort of thing upon which you could BASE that belief.

    Is it all that clear that reliabilism comes out as externalist on PB? It seems a reliabilist could say that there is an internal justifier on which our beliefs are based while insisting that the justifier is a justifier because of conditions external to the subject. I thought this was pretty much Alston’s view on the matter. I’m tempted to classify Alston as an externalist because the question as to whether the subject’s belief is justified cannot be settled by focusing on the intrinsic properties of what he regards as the belief’s basis, so I think there is something to PB, but the situation is complicated once the distinction between justifier/that which makes the justifier a justifier is introduced.

    Another interesting test case is found in Williamson’s account of justification. Some of his remarks suggest that one’s beliefs are justified by one’s evidence and one’s evidence is what one knows (I’m not confident that we can ascribe to Williamson the view that one’s justification for believing p when one knows p non-inferentially is the fact that p or something else, such as the fact that p is known, but Brewer (?), Dancy (?), Hyman, and Kenny seem to think that one’s reason or justification for the belief that p when one knows p perceptually is the fact made evident to us through perception, so if I’m misreading Williamson, substitute as you see fit). Assuming we know some facts about our surroundings non-inferentially, there will be some who say that our basis can be something in the world rather than in the head. Do we count these views as externalist if these authors also insist that we have a kind of immediate access to these external facts? If my hand or the fact that I have a hand really can be ‘internal’ to my perspective on the world, I can understand calling it a basis for my belief, but unless I’m holding my hand _really_ close to my face and was talking amongst friends, I wouldn’t unhesitatingly call the hand/the hand-fact an internal justifier.

    Jon, do you think we can have immediate or non-inferential observational knowledge of causal relations? I’m inclined to think not in part for the reasons you offer for thinking we cannot know certain facts about causal relations by reflection or observation alone, but I’ve met stiff opposition from those who insist that when it seems that x and y are causally related and the seeming isn’t accidentally connected to the fact that x and y are causally related, the mere fact that there could have been alternative explanations of the seeming doesn’t undermine the judgment’s status that x and y are causally related as knowledge. Depending upon how things can seem (I’m not sure that causality is part of the phenomenology of perception), I have to admit that there is a real issue here. Is the case of ‘knowledge by reflection’ different from immediate observational knowledge?

  3. Clayton, I don’t want to remark too strongly at this point, since I expect what I’m thinking and what Ram is thinking about knowledge by reflection alone are different, but I will say this much at least. If we suppose that we can have observational knowledge of causation, I’m not sure how that will help with knowledge by reflection of causation. I can’t know by reflection alone that there is a tree in my backyard, though I can by observation. Again, though, this remark is based on a particular construal of ‘knowledge by reflection alone,’ one that may not be what Ram had in mind.

  4. Hey Jon,

    I thought part of the problem for knowledge of causal relations amongst mental states had to do with the issue of immediacy or non-inferentiality. There are those who think that we cannot know immediately or non-inferentially whether the ball broke the window even if we witness Billy throw the ball through the window but admit that we might know Billy’s ball broke the window (me on odd days). I was just curious to see if you thought that knowledge of causal relations could not be immediate or non-inferential and for that reason troubles arose for Ram’s proposal or whether the resistance had a different source. Of course, there will be differences between what you can know by way of immediate observation and what you can know non-inferentially by way of reflection if the judgments of observation and reflection are about different subject matters. I do think there is a very cool (hot? Probably a stretch) issue concerning the epistemology and experience of causation. If someone were to argue that the representation of causal relations between particulars were a part of the content or phenomenology of experience then when our experience is veridical, what grounds have we for denying immediate knowledge of causal relations? If none, it is hard to see what sort of principled objection stood in the way of developing Ram’s first proposal.

  5. Jim, I think we should rephrase your suggestion to include the notion of the basing being proper. I think epistemological usage of the phrase is pretty well-entrenched in terms of the possibility of basing a belief on things that can’t justify (such as a reading of the tea leaves or the entrails of goats). (I ignore for the present those who think anything can justify anything, if put into the right total system of beliefs.)

    Rephrased that way gives us: X is eligible to be an internal justifier of a belief that p iff X is the sort of thing that you could properly base a belief that p on.

    I wonder, though, if you have a reason for leaving out the normative term ‘properly’?

  6. Clayton, re your post 4, do you think that immediate knowledge of causal relations is possible only if such relations are represented phenomenologically? Or do you mean that that sort of account is the only sort that would satisfy someone who thinks that reflective access to the causal relation is a central aspect of basing?

  7. Some miscellaneous additions…

    1. Behind Ram’s proposal (1) is the idea that we can (at least sometimes) know by reflection alone what a belief is based on. That sounds extremely plausible to me. I’m not sure whether we can ever know by reflection alone that a causal relation obtains; but if we can’t, then I’d take that to count against a causal analysis of the basing relation.

    2. Re Jon’s amendment: “X is eligible to be an internal justifier of a belief that p iff X is the sort of thing that you could PROPERLY base a belief that p on.”

    Sure, if what we’re after is an account of the internal JUSTIFIERS, then that amendment is just going to take us closer to our quarry. What I really hanker for, though, is a two-stage story that tells us:
    (a) Such-and-such are the things that are (by internalist lights) internal enough to be ELIGIBLE to be justifiers. These are things it’d be POSSIBLE for you to base a belief that P upon.
    (b) However, some of those things don’t do anything to support P. So beliefs based on them would be improperly based.
    I was just suggesting an account of (a); that’s why I didn’t require the basing to be proper. I’d like to see a two-stage account because I think an informative answer to (a) would be independently interesting.

    3. Clayton aired some worries about how closely this internal/external divide would correspond to more familiar taxonomies. Alston is a useful test-case. I was thinking this:
    (i) Facts about how reliable your belief are aren’t candidate bases for your belief. (I think that’s TRUE, but I also think it NEEDS DEFENDING; on some views, any external fact you’re aware of can be a basis for your belief.)
    (ii) Since they’re not candidate bases, they can’t be justifiers.
    (iii) Something counts among the justifiers of your belief =def. it’s among the facts that make your belief justified.
    Alston might accept the letter of (ii), but I think that’s because he uses “justifier” differently than me. If we read (ii) with “justifier” understood as in (iii), then Alston would reject it. Hence, he’d count as an externalist by the criterion I had in mind.

    I can see how the criterion I proposed might count people like Williamson, McDowell, Brewer etc., as internalists. E.g., if you say that what justifies your belief that P is your genuine perception that P, and you allow the belief to be based on the perception itself, then you’d count as an “internalist” by the criterion I proposed. McDowell seems to regard HIMSELF as an internalist, so maybe that’s OK.

    4. I _suggested_ that this might be a good way to define an internalism/externalism contrast; but that doesn’t mean I’m confident it will pan out. Counting McDowell-style views as internalist doesn’t bother me so much. What does worry me are our inclinations to say that people have based their beliefs on the tea-leaves, or on any other external fact they’re aware of. That threatens to trivialize the proposal. We could say, “Well they’re not really basing the belief on the tea-leaves, they’re basing it on their beliefs/experiences concerning the tea-leaves.” But since I don’t have a good account of the basing relation, I don’t know how to adjudicate that kind of dispute.

  8. Stephen,

    I’m not sure what it would take to rule out or rule in immediate knowledge of causal relations, but the one time I had been in the room when it was seriously defended, it was on the grounds that when one sees Billy break the window, it looked like Billy broke the window. If it turned out that, say, Billy’s rock flew through the window just as something hidden smashed the glass in the right way, one’s experience would mislead one into thinking something false. This was taken as a clue as to the content and perhaps of the phenomenology of the experience. The idea was then that if it looked as if p and it turned out that p, even though there were possible scenarios in which the look of p didn’t get explained in terms of the fact that p, this didn’t seem to threaten the perceptual judgment that p’s status as knowledge unless one held an implausibly restrictive account of what grounds could be adequate for knowledge. Now, this view struck me as somewhat implausible as I had a hard time thinking that when one mistakenly thinks it is Billy’s rock that broke the window, one is dealing with a case of perceptual illusion. However, the situation is complicated further by the possibility that the scope of perceptual knowledge is not limited to what is ‘manifest’ or part of the phenomenology of experience (I’m thinking of Alan Millar’s paper in which he argues that what one can know perceptually goes beyond what is contained in the phenomenology of experience). I guess a really impressive demonstration of the possibility of immediate knowledge of causal relations would inter alia involve some argument to the effect that there is something in the experience that represents causal relations but worry also that the failure to establish this wouldn’t serve as a convincing case against the possibility of such immediate causal knowledge. In other words, the situation seems far too murky to dismiss the possibility of immediate knowledge of causation. I think we have immediate access to all sorts of things external to us so I’d not be immediately disposed to accept certain kinds of undeteceted or undetectable switching arguments as good arguments against such knowledge or a good guide to what can be known via reflective access.


    I’m not sure I’d want to defend the standard taxonomies. Maybe it wouldn’t be so bad if McDowell and McDowell-esque views turned out to be internalist. It does seem that many of the arguments for internalism are treated as if they were arguments against those who seem to have picked up the cause of Oxford Realism (McDowell, Brewer, Williamson?) as well as the reliabilists.

    If I might ask you about (iii), there are a few examples that worry me. On Sutton’s view (most recent Nous), your belief that p is justified iff you know p. Whether you know p depends upon whether you are in a Gettier situation. The absence of the conditions that would Gettierize my belief don’t seem like justifiers. The absence seems to count as the absence of muckers rather than the presence of a justifier. This issue doesn’t seem to go away if we move to more standard accounts of justification. Consider the access internalist view according to which only that to which we have access counts as a justifier. It seems that to properly motivate their view, the AI has to take aim at the McDowellian who thinks that we have access to the external world. Access, says the defender of access internalism, is not provided by perception because we do not have access to what distinguishes perception from hallucination. The problem is that it doesn’t seem that we have access to that in virtue of which introspection provides us with access either–introspection provides me with access to my present sensations but not that which allows me to access those sensations. Appealing to (iii), the access internalist might charge McDowell with externalism but it seems that if there are non-accessible conditions for introspection to work, appealing to (iii) allows us to likewise charge the access internalist with externalist commitments. The general worry–there will always be conditions C such that C is necessary for J to be a justifier where C won’t intuitively seem to be a justifier. At least, the problem arises if ‘justifier’ is being used not as a term of art being stipulatively introduced but as something like ‘normative reason’, a notion we have an excellent grasp of, as illustrated by the pellucidly clear literature on that topic.

  9. Wow: I come back to the blog one day later, and Jim’s prediction about a hot topic has become self-fulfilling. Thanks for all the ideas!

    Let me see if I can sidestep Jon’s and Jim’s worries about causation by slightly rephrasing my proposal and then adding a couple of points:

    So, the proposal:

    S’s belief that p is based on X IFF: S can know, by reflection alone, that it is because of X that S believes that p.

    Jim worries that subjects can lack the concept of causation but still have their beliefs stand in the basing relation to something. But if we consider a subject who lacks the concept BECAUSE — i.e., a subject who can’t understand “why?” (and so can’t give or ask for reasons) — then is this a subject whose beliefs really do stand in the basing relation to anything, or are this subject’s beliefs simply CAUSED by (but not based on) something or other?

    (I guess it would be nice to have a relatively theory-neutral way to answer this last question, but I’m afraid I don’t see how to be theory-neutral here.)

    Jon worries that we can’t know about causal relations by reflection alone. But consider a case. I just remembered — or seemed to remember (to avoid worries about the factivity of “remember”) — that I have to pick up the dry cleaning, and I just remembered it because I saw (or seemed to see) a piece of paper that said “dry cleaning” on it. Now, I can know by reflection alone that I seemed to remember to pick up the dry cleaning. And I can know by reflection alone that I seemed to see the piece of paper. Can’t I know by reflection alone that the latter event reminded me of the dry cleaning — and thereby that the latter event caused the former? I’m inclined to think so, but again, maybe the question can’t be answered without appeal to some substantive account of reflective knowledge.

  10. A couple of thoughts about the basing as causal and the possibility of learning about cause through reflection alone.

    [This is just fiddling around in the space of possibilities. I find Ram’s example of knowing about cause through reflection reasonably compelling, and I also suspect that in the end I’ll want to reject the idea that basing is available to reflection alone. If you tell me that p, and I forget that you told me, then I think my belief that p is based on your testimony (or appropriate internal correlate) even though that’s not reflectively accessible.)]

    Suppose that we want to hold that we can know by reflection alone what a belief is based on, but that we can never know by reflection alone that a causal relation attains. It is still possible for the basing relation to be causal, so long as we can’t know by reflection alone that it’s causal. This, I think, is a cheap shot.

    More interesting, maybe: suppose we’re anomalous monists. We can hold that the basing relation is always reflectively accessible, and that it’s causal, and that we can know by reflection alone that it’s causal, and we don’t have to accept too much knowledge by reflection alone of causal relations. Namely: Maybe all we can come to know by reflection alone is “The physical state that realizes this belief (whatever that state is) is caused by the physical state that realizes this basis (whatever that state is).” But we need not be able to come to know reflectively anything of the form “Physical state A causes physical state B” where the states are described as physical states rather than as realizers of mental states. Maybe that would draw the sting of reflective knowledge of causal relations.

    (Is anyone else finding it virtually impossible to type inthe comment box? The scroll bar keeps hopping to the top of the box, so I can’t see the cursor. I’m using Firefox.)

  11. Ram,

    In general I think we should be suspicious of transparency-of-mind accounts of basing. We might want to count cases of wishful thinking and beliefs motivated by irrational fears to be beliefs that are based on wishful thinking or irrational fears. In that case, transparency of mind doesn’t seem all that plausible. Some of the time I might be able to know by reflection alone that it is because I have a phobia of snakes that I believe that they are vile creatures. But other times, I might not be so able, for reasons that Freudians might point out. It doesn’t seem to me that there are any differences with respect to the bases of my beliefs in cases in which I can know by reflection that my fears explain my beliefs, and cases in which I cannot for Freudian reasons.

    Of course, we might not want to say that beliefs are sometimes based on wishful thinking or irrational fears. But why not? Surely such beliefs are paradigm cases of irrationally held beliefs, and it seems like they are irrational because they’re based on considerations that are not good reasons. Maybe you want to say that they are irrational because they are based on no reasons, but rather, have weird causes. In that case, we should reject causal accounts of basing, for reasons given above.

  12. I just realized that a sentence from my post #10 may be difficult to understand because I used myself as the subject of the example. So:

    If you tell me that p, and I forget that you told me, then I think my belief that p is based on your testimony (or appropriate internal correlate) even though that’s not reflectively accessible.

    should be

    If A tells B that p, and B forgets that A told her, then I think B’s belief that p is based on A’s testimony (or appropriate internal correlate) even though that’s not reflectively accessible [to B].

  13. There may be some cases, even a decent number of them, in which something like that reflection principle can apply successfully. But: what are we to say of the vast array of our ordinary beliefs for which we have no particular causal story at our fingertips, but for which it would be ludicrous to say they were baseless? To take a favored example from my colleague Adam Leite (and really this whole comment should be read as my attempting to apply some of his insights without (I hope) mangling them too badly), consider the belief that I currently reside in the United States. Surely the original causes of that belief for me are lost to the ages, somewhere in the dark recesses of the Ford (or maybe early Carter) administration. And I’d be extraordinarily hard pressed to come up with, upon reflection, some set of beliefs, perceptions, memories, or whatever that are the sustaining causes for that belief. In a very real sense, I’ve just got too darn much evidence for the belief for any reflectively tractable subset of it to be the causes of the belief. Moreover, a huge proportion of our beliefs work just like that; we don’t tend to notice such beliefs, because they are really very boring beliefs, unlikely to be challenged precisely because they are so extremely easy to defend. So any too-causal account of basing risks rendering a large chunk of our beliefs as epistemically unintelligible, where in fact they should come off as our most secure & justified pieces of knowledge.

    (Would it work to read the “because” in the principle as merely being a contributing cause, and not the cause? I suspect not, as our cognition is sufficiently holistic that almost every belief plays at least some contributing role in the sustaining of the others, at a minimum for the continual confirming of our sense that our cognition is well-working. The biconditional on this weak reading of “because” would make the basing relation vacuous in the other direction — instead of most everything being based on nothing, rather we’d have that most everything would be based on most everything.)

  14. I spent 50+ pages of my dissertation on the epistemic basing relation, so I’m loving this!


    Doesn’t your account let far too much in? Suppose I know that it’s (partly) because neurons fired in a certain way that I believe that I must bring the garbage down to the curb tomorrow. But that belief of mine is not, in the relevant sense, based on neurons firing a certain way. And by “relevant sense” I mean that the neurons firing a certain way is not among my reasons for believing that I must bring the garbage down tomorrow.

    Also, I don’t think we need to imagine a case where the subject “lacks the concept BECAUSE.” Rather, we only need a case where the subject cannot, for some reason or other, effectively employ that concept in his reflective deliberations. I’m inclined to think that such a subject can still believe for reasons.

  15. Matt,

    Minor point: If B harbors an “appropriate internal correlate” of A’s testimony that p, can it be that B has really “forgotten” A’s testimony?

    And, yes, I’m having trouble with the comment box, too. I’m also using Firefox. I’ve taken to just watching the preview pane, just visible above “LEAVE A REPLY” as I type, but the text box must be situated near enough the bottom of your monitor for it to work.

  16. Barry,

    Maybe you want to say that they are irrational because they are based on no reasons, but rather, have weird causes. In that case, we should reject causal accounts of basing, for reasons given above.

    That seems a bit too quick. Why can’t this problem be taken care of by restricting the sort of things that can count as reasons in the first place? Wishes could get ruled out at that stage, but we could hold on to a causal theory for all the eligible items.

    In my experience of presenting cases to people, they almost invariably classify a case of wishful thinking as a case where the subject believes for “no reason at all.” When I try to draw them out by asking, “Do you mean no good reason?” most people stick to their original characterization. This has caused me some problems in formulating my own theory of the basing relation, because we obviously want to allow for beliefs to be based on bad reasons, but it’s tricky to draw the line between “a bad X” and “not an X at all.”

    I’ve settled on the above strategy for dealing with this complication. If someone wants to rule wishes out, fine. I’ll apply my theory wherever they draw the line.

  17. Oops, I see that my block-quotation formatting didn’t work in my response to Barry. For those keeping track, that first paragraph should be in quotes.

    “Maybe you want to say that they are irrational because they are based on no reasons, but rather, have weird causes. In that case, we should reject causal accounts of basing, for reasons given above.”

  18. Jim,

    I think that is a good way to draw an important contrast in epistemology, though as Clayton remarked (and I think as you yourself have said in print!), it does seem to cut across the internalism/externalism controversy in many of its guises (though not all, if we observe your distinction between Simple Internalism and Weak Externalism in sec. 3.3 of “Recent Highlights,” reflected in your comment 7 above).

    It’s hard to deny that there’s some sense in which contingent non-mental facts form part of the basis of our beliefs. As you point out, we say things like this all the time. But that is consistent with there being a more restrictive, epistemologically important basis.

    Having recognized an external basis of belief, how might we defend the view that there’s an important sense in which “facts about how reliable your belief is aren’t candidate bases for your belief”? By appealing to new evil demon intuitions. We have a powerful intuition that you and your recently victimized twin, Vic, believe what you do for the very same reasons. But then it can’t be the case that reliability-facts are among your reasons for holding your belief. We might call this the rational basis of belief, to distinguish it from the total basis, which includes the sundry external things.

    Of course, that doesn’t give you everything you want because there’s still the further question of whether reliability-facts, or other contingent non-mental facts dear to externalists, which may form part of a belief’s total basis, can affect the belief’s normative epistemic status. (Though if Alston’s right, there’ll be no single answer to this question- -certain epistemic statuses will be so affected, whereas others won’t.)

  19. John in 15: If B harbors an “appropriate internal correlate” of A’s testimony that p, can it be that B has really “forgotten” A’s testimony?

    The simple answer is that I was thinking of a past internal correlate–B’s past belief that A had told her that p, or her past experiences of seeming to be told that p by A, or something like that. So B need not still harbor the correlate at the time at which the belief is based on the correlate. (And I only meant this as a possibility, to leave open the possibility that the baser is a completely internalstate–it may be that the testimony really is the basis.)

    That said, it’s not obvious to me that the answer is”no” even if B still harbors the internal correlate. It may be that there’s some sort of trace somewhere in B’s brain that still isn’t reflectively accessible.

  20. Matt,

    I think there’s a sense in which A’s testimony is part of the basis of B’s belief, but I don’t think anything that no longer exists, such as a past seeming, could now be among B’s reasons for believing that p.

    I’m inclined to agree that an existing but reflectively inaccessible trace could be B’s reason.

  21. What the data to which an account of the basing relation must answer? What does an account of the basing relation need to explain, or predict?

    I am assuming that an account of the basing relation must explain — or in any case help to explain — the fact that some beliefs are reasonable (or justifiable) and others are not. Now, what makes it reasonable for Jonathan to believe — at this moment — that he lives in the United States? Here’s a possible answer: the fact that he has learned it. What makes it reasonable for Matt to believe — at this moment — something that he has learned by means of some long forgotten episode of testimony? Again, here’s a possible answer: the fact that he has been apprised of it.

    These answers are of course not very theoretical or informative. But they seem to me to be correct. So why can’t these answers state the bases of the specified present beliefs?

  22. Ram, I thought you’d say that! What if I’ve forgotten whether I was told that p, or saw it for myself, or what? Do you want to say that that belief can’t be justified? That’s not necessarily incorrect, just trying it out.

  23. Doesn’t “I learned that p” here just have to mean, more or less, “I have in the past formed the belief that p on the basis of some sort of proper evidence or other”? If that’s right, then it’s very odd to say that my basis for p is my having some-basis-or-other for p — shades of dormitive virtues! (Of course there are some very esoteric cases, like Loeb’s theorem, that maybe do work something like this, but surely they do not provide a general model for basing.)

    If “I learned that p” means something more specific, like, “A teacher or parent taught me that p”, then we’ve got something like Matt’s case on our hands, since I don’t recall even faintly how I first came across the information; maybe I read it in a book? Anyway, I just don’t remember any of that basic instruction that I might have received as a kid, so I suppose that I now base that belief, if on anything at all, on the much more diffuse base of my knowledge about world geography, passport-issuing practices, some general beliefs about people being generally very reliable about what country they are in, and so on.

  24. Maybe seeing that p, being informed that p, figuring out that p, etc. are all species of the genus: learning that p. Does learning that p require having formed the belief that p? Hmmm… Consider the perversely skeptical student who learns everything that she’s told in school, but who pathologically doubts all of it. (I think I knew someone like this in my chem 10 course, back in college.)

    Maybe learning that p is the most generic way of acquiring a basis for believing that p — even if it doesn’t include or require the belief itself…?

    Maybe I can know by reflection alone that I believe that p because I’ve learned that p — which is just to say that my belief has that (maximally generic) basis.

  25. Jim Pryor writes:
    Behind Ram’s proposal (1) is the idea that we can (at least sometimes) know by reflection alone what a belief is based on. That sounds extremely plausible to me. I’m not sure whether we can ever know by reflection alone that a causal relation obtains; but if we can’t, then I’d take that to count against a causal analysis of the basing relation.

    Jim’s diagnosis sounds correct to me, but I’m skeptical where he is optimistic. Viz. I’m skeptical that

    (1) S’s belief that p is based on X iff S can know, by reflection alone, S’s belief that p is caused by X.

    is ever satisfied. And this is because widely accepted and influential data from social psychology suggest that we never identify the causes of our mental states by reflection alone. Rather, as Nisbett and Wilson (1977) argued, we use a posteriori theories about the relationship between our mental states to identify which mental states cause which others. Nisbett and Wilson review a host of evidence about cases where the causes (including the rational causes) of our mental states deviate from our accepted theories about what sorts of mental states cause one another, and where (consequently) people give wrong accounts of the causes of or reasons for their beliefs/actions.
    This data need not give rise to the skeptical conclusion that we never or rarely know the causes of our mental states. It’s compatible with the view that usually, in the real, non laboratory world, we get the relationship right. But it does suggest that we never have introspective access to a mental state qua cause. We just fill in the causal story after the fact, using our general (and a posteriori) knowledge about plausible relationships between mental states. It’s obvious why this is: we might have introspective access to the mental states themselves, but we sure don’t have introspective access to the processes that connect them. (That’s why we don’t just know what the correct descriptive theory of, say, inductive reasoning is). But if that’s right, then we never have simple reflective access that S’s belief that p is caused by X.

  26. This was more or less what I proposed in my paper “Internalism Explained” (PPR 2002). Admittedly, I decided not to use the term ‘basing’ in that paper, but it was what I had in mind.

    When one forms a belief in p, the basis for one’s forming this belief is what I there called the “proximate folk-psychological explanation” of one’s forming that belief. I then argued that the proximate folk-psychological explanation of an instance of belief-formation is always some fact that supervenes on the thinker’s (non-factive) mental states, and so is in a sense an “internal” fact.

    If the justificatory status of a belief depends on what the belief is based on (and on what any antecedent beliefs on which that first belief is based are themselves based on), and the bases of beliefs are always in this way internal, then a sort of “internalism” about justification is true.

    Of course, this is a slightly different conception of the basing relation from Ram’s, since I would actually reject both of his proposals (1) and (2) above, essentially for Jim’s reasons. But the fundamental idea is the same as Ram’s.

  27. Re Ron’s post 26, the problem with Nisbett and Wilson and similar studies in that vein is that the subjects’ beliefs in the studies generally aren’t epistemically justified. So the studies don’t furnish counter-examples to internalist constraints. Moreover, it seems to me that when a person’s belief is properly based on an adequate ground (or whatever formula one prefers) then that person is much more likely to access that ground (and its causal connection to the belief at issue) on reflection. Proper basing enhances reflective power, as it were.

    I do think that skepticism about reflective access to causal connections is motivated by a particular sort of “Cartesian” theory of introspection. This skepticism in turn is used as a basis for objecting to internalism, on the supposition that internalism is committed to this “Cartesian” theory. But it seems to me that on a naturalist/reliabilist conception of introspection, there’s no difficulty in principle (or in practice) with reflective access to causal connections, and I don’t see why an internalist couldn’t adopt such a view.

  28. Stephen, I’m not sure I totally understand your response to Ron. Is the idea that N&R just weren’t looking at the kind of case we’re talking about — i.e., where one has formed a belief on some sort of basis, and then attempts reflectively to determine what that basis is? It’s true that that’s not the sort of case they were looking at, but the basic challenge Ron is making from N&R still seems a good one. (At least this is so if we raise the bar from “we can tell by reflection, at least sometimes, what our beliefs are caused by” to the more appropriate “we can tell by reflection, approximately as often as our beliefs are justified, what our beliefs are caused by.”) The N&R results seem to show that we don’t have any generally reliable capacity for reflective access to the sources of our mental states, so you’d need to offer something of an argument — preferably an empirical one — to back up the claim that some special reflective epistemic yumminess obtains when we’ve got some basing. It’s a fair hypothesis to explore, but what available evidence there is doesn’t seem to point particularly in its favor. (I’d note that the Leite-style cases I mentioned above should be further evidence against the claim; whatever the causal origin may be of my belief that I reside in America, it’s far from reflectively available to me.)

    Also, I don’t think we need to interpret Ron as even trying to argue against internalism in epistemology. After all, one could be an internalist, and think that we don’t have good reflective access to the causes of our beliefs, and therefore just reject a reflective-access-on-causal-origin condition on justified belief. (Fwiw, I’m fairly sure from my knowledge of other aspects of his work that Ron has no Cartesian commitments regarding introspection; and I promise you I’ve got none myself.)

  29. Jonathan has the story right, I think. I (like Stephen, I think) think of introspection as a form of defeasible perception (though I’m not sure Ram does). But the questions are: Do we everdirectly introspect a causal connection? Is there any evidence that we do so reliably? That’s where N&W lead me to be skeptical.

  30. Most people are terrible at reporting, or reproducing in paint, the colors they see in a landscape right in front of them. (Anyone who’s learned to paint can verify this.) That doesn’t imply that those colors aren’t visible.

    Similarly, most people are terrible at reporting facts that can be known by reflection alone. That doesn’t imply that those facts aren’t reflectively accessible. Even that arch Cartesian Descartes thought that most people were extremely bad at reflection: that’s one reason why there need to be six meditations, and not just one.

  31. Ram writes:

    Similarly, most people are terrible at reporting facts that can be known by reflection alone. That doesn’t imply that those facts aren’t reflectively accessible.

    You are of course right that it does not imply those facts aren’t reflectively accessible. But it does provide evidence for the view that they are not. Moreover, N&W report many cases of mistaken reason giving: people give reasons for their judgments that are not the causes of those judgments. So, you need no only that we can fail to access the causes/reasons we have, but also that we can wrongly identify other belief states as the causes/reasons for our judgments. One hypothesis for this is the N&W one: we do not typically have reflective/introspective access to the processes connecting our mental states to one another or to behavior. Another hypothesis is yours: we actually do have reflective/introspective access, only (a) we are lousy at reporting successfully what we introspectively ‘see’ and (b) we sometimes report ‘seeing’ something that’s not there.
    One of these hypotheses sounds mighty ad hoc. So, perhaps I should ask: Why should we believe your hypothesis, other than that it saves your theory?

  32. Hi Ron,

    So two possible explanations of the N&W data: (1) people are lousy at exercising their reflective abilities with respect to the bases of their beliefs, and (2) people don’t generally have reflective access to the bases of their beliefs. The big question is: which explanation of the N&W data is better?

    Here’s my anecdotal (and therefore not highly dispositive) reason for favoring (1) over (2). It seems to me that there are lots of cases in which the following sort of thing happens. Someone who misidentifies the basis for a belief of hers can, under suitable circumstances (usually involving conversation), be brought to recognize that the basis for her belief is actually something quite different from what she thought it was. I think that I’ve had this experience quite often, and I think that others have also.

    Of course, it’s possible that I’m wrong about precisely what was going on in the cases that I’m thinking about. But what we’d need to do to turn this into a testable hypothesis, I suppose, would be to specify the kinds of circumstances under which — as it seems to me — people are being brought to reflect more accurately on the bases of their beliefs. And then we’d need to test whether, under those circumstances so specified, people do in fact reflect more accurately on the bases of their beliefs.

  33. The presumption that people routinely are able to reflectively access the grounds of their beliefs is so firmly embedded in daily life that it could only be rebutted by extraordinary evidence. To my mind, Nisbett et al. do not rise to that level. The circumstances of those studies are too unique and too engineered to draw any general conclusion about the reliability or power of human reflective faculties.

    Moreover, reflective access to the grounds of one’s beliefs is close to incomprehensible unless one also has access to the relations by which those grounds confer justification. It is very difficult to understand how one could cite a ground in support of a belief (something we do, and are expected to be able to do, in most everyday situations) unless one also appreciates that this is indeed the basis for the belief at issue. Why would one be citing that ground unless one appreciated it as such? On pain of regress, this appreciation cannot always be inferential, but must on occasion (and, presumably, on most occasions) be non-inferential.

    If one accepts that we have reliable non-inferential faculties for accessing causal connections in the external world (and it seems very likely that we do have such faculties), then the complete lack of such faculties as to the internal world would be very surprising and extraordinary. What could explain that?

  34. I think the really thorny problem with causal accounts of basing is not access, but ruling out the deviant cases in a non-ad hoc way.

    I think the best way to avoid such cases is to explain basing in terms of being “sensitive” to the ground. This ultimately leads to something like a virtue account of basing.

  35. Stephen,

    You might be interested to learn that I develop a virtue-theoretic account of causal non-deviance for basing in my dissertation (precis here). The basic idea is that a reason non-deviantly causes a belief just in case it does so through a manifestation of one of the subject’s cognitive dispositions. So I think your instincts on this are right!

    Ralph Wedgwood defends a similar view in his excellent paper, “The Normative Force of Reasoning” (forthcoming, Nous; see esp. section 2; link to online version). William Alston also defends a similar approach in chapters 5 and 6 of his Beyond ‘Justification’ (Cornell, 2005). Alston basically applies his earlier insights into solving the generality problem for reliabilism to the basing relation.

  36. I worked out an account along those lines in my dissertation as well. I always just assumed nobody would find it that interesting.

  37. Hi Stephen,

    If the causal chain from X to S’s belief that p is deviant, then is it possible for S to know, by reflection alone, that it is because of X that she believes that p? What’s an example?

  38. Ram,

    Your proposal that the causal relation be accessible on reflection may eliminate a number of deviant cases. I suppose this would depend on whether all such deviant cases involve nonaccessible external factors, and whether the presence of such factors means that the causal relation itself is not accessible.

    For example, Moore’s case (from his Commonplace Book): I believe that someone is in the house. This causes me to open the door, which causes me to believe that no one is in the house. This appears to be a deviant case: my belief that no one is in the house clearly is not based on my prior belief that someone is in the house. The crucial question is whether the causal relation between the prior belief and the present belief is accessible on reflection, given that the sequence involves an external intermediary (my act of opening the door). I’m inclined to say that the fact that the prior belief caused the present belief indeed is accessible on reflection, but I can see the opposing argument as well.

    Plantiga’s case: My belief that I see Sylvia causes me to spill my tea, which causes me to believe that my leg is in pain. The latter belief is not based on the former. Again, there seems to be an external intermediary. Is the causal relation accessible on reflection? As with Moore’s case, I could see people going either way, but my inclination is to say that it is.

  39. Hey Stephen,

    Hmmm… Can you think of a case that’s like the ones above, except that the causal intermediary is not an external (i.e. perceptible) event?

  40. Ram,

    I believe that I have to go to the bank tomorrow. I seem to remember learning that anytime I have a belief with that content, it’s because of neuron assembly N firing in manner M. And in fact I do remember this, so I can come to know, on this occasion, that I believe I have to go to the bank tomorrow because N is firing in manner M.

    If drawing upon memory counts as part of reflection, which seems at least plausible, then your proposal implies that my belief is based on N firing in manner M.

    So one question is: Does drawing upon memory count as part of reflection, or does it take us beyond reflection so that your proposal doesn’t imply a basing relation here?

  41. Ram–how about this, which (in some form) happens all the time to me: I seem to hear a lecturer talk about partnerships in accounting firms. This causes me to think about partnerships in publishing firms, which causes me to think about Marian Keyes’ latest novel, which causes me to think (for the first time) that in it we see a character refusing to let herself by a relationship. Then I think, “Why am I thinking this right now?” I reflect on my train of thought and conclude that my current belief was ultimately caused by (seeming to hear) the lecturer talking about accounting partnerships. But that cause, which is reflectively accessible (we’ll suppose), definitely isn’t the basis for the belief.

    Now, if I thoughtabout what the basis for the belief was–“Why do I think this?”—then I wouldn’t think of my train of thought. But that won’t help us analyze the basing relation in terms of causal chains, because it presupposes a successful analysisof the basing relation.

  42. Ram, couldn’t deviant causal sequences such as those I mentioned above be recharacterized in a way that involves reference only to internal states? It isn’t the act of opening the door that is causally relevant but the experience as if of opening the door?

    Matt, the sort of case you’re talking about is interesting, and I think does illustrate one way in which one internal state may cause another without that causal connection being a manifestation of a cognitive disposition or virtue (one prerequisite of proper basing). But still I think part of the reason why your case isn’t an example of basing is simply that the relata of the basing relation are beliefs and your case involves a sequence thinkings or imaginings (though I confess I wouldn’t be able to explain the precise way in which thinkings differ from believings).

  43. Hm. I would have thought that we might want to leave it open that a belief could be based on an experience, but I confess that I haven’t thought the issue through.

    In this case I intended the end thought to be an occurent belief, and it ought to be possible to change the example so the beginning thought is also an occurrent belief. Maybe that shows that we really don’t want occurent beliefs to be the relata either. Still, I think it should be possible to construct an example where every stage is something I start to believe for the first time as I think it; then if asked “How did you come to believe this?” I could answer “First I came to believe this, then I came to believe this, etc.” But that would still probably fall on the wrong side of the thinking/believeing divide, if we knew where that was.

    [Final random thought; maybe this sort of thing can manifest a cognitive virtue; what if I come upon a lot of interesting and true beliefs because of my free-associative habits? Wouldn’t the free assocation then be cognitively virtuous?)

  44. Hey guys,

    Good questions.

    Matt and Stephen: In the examples that you give, I can know by reflection alone that I’m thinking about X because I was thinking about Y. But can I know by reflection alone that I BELIEVE that p because of something? For instance, in Matt’s example, if I wonder “why am I thinking this right now?” I could retrace my mental transitions. But if I wonder “Why do I believe that Keyes’ character is so stubborn?”, then I can answer that question only by appeal to features of (my experience of) that character: when her brother tries to talk to her, she rebuffs him, and she won’t listen to her parents, etc. etc. Talking about my own mental transitions doesn’t answer the question “why do I believe that Keyes’ character is so stubborn?”

    John: I was assuming that memory would rule out a case like yours, and that the case doesn’t involve reflection alone. Of course, whether or not this assumption is correct depends on what it is to know something by reflection alone, and I don’t know anyone who has a good account of that (despite the large role played by the concept of reflective accessibility in lots of contemporary theory).

  45. Ram, my suspicion is that skepticism about whether causal relations are the sort of thing that could ever be grasped reflectively is driven by a particular view of reflective access, a hint of which I think I can detect in your post. You suggest that one can know something on reflection if one “can answer [the] question only by appeal to features of [one’s] experience . . . .” My sense is that this is the same conception of reflective access that was behind Clayton’s question in post # 4:

    “If someone were to argue that the representation of causal relations between particulars were a part of the content or phenomenology of experience then when our experience is veridical, what grounds have we for denying immediate knowledge of causal relations?”

    The implication here is that reflective access to causal relations is possible only if such relations are represented in the content or phenomenology of experience. But it has always seemed to me that a reasonable alternative view is that causal relations can be accessed reflectively, despite not being represented in experience in this way. I worked out an account along these lines in my dissertation years ago, but never pursued it beyond that.

    Matt, you’re quite right that I spoke too hastily. Presumably beliefs can be based on experiences as well as other beliefs. I was focusing more on the fact that your case seemed to involve a chain of musings or imaginings (rather than believings; though I see now that I had not read your post carefully enough) and I was wondering how that might limit what the case showed about basing.

  46. Ram,

    I think there might be a dilemma of sorts taking shape for your proposal.

    Earlier in response to Jon’s worry that reflection alone could never put you in a position to know a contingent causal claim, you suggested that you could “know by reflection alone that I seemed to remember to pick up the dry cleaning. And I can know by reflection alone that I seemed to see the piece of paper.” But I don’t see how this is any different from my case where the subject seems to remember learning that anytime he has a belief with a certain content, it’s because of neuron assembly N firing in manner M. Yet you’re inclined to say that, in my case, reliance on memory ruins the claim to reflective status.

    So it looks like you’ll need to rely on seeming memories in order to be in a position to know that you believe p because of X. But if you do that, then cases like mine will be possible, and the view will imply that beliefs could be based on things like neuron assemblies firing (and who knows what else). Either way, it’s a problem.

    This dilemma does not appear to assume any particular view about reflective knowledge or reflection. It just assumes that we’ll treat similar cases similarly.

  47. Pingback: Fake Barn Country

  48. Hi Stephen and John,

    I think it’s becoming increasingly clear how badly we need an account of what it is to know something by reflection alone. Without such an account, I can’t feel very certain about anything in this area. Nonetheless, I think I might be able to deal with the dilemma that John raises by appeal to the following principle (which seems to me to be true):

    If S knows, by reflection alone, that p, then S’s knowledge that p is not achieved by inference from a premise that S doesn’t know by reflection alone.

    I can know by reflection alone that I had the experience of seeing (or seeming the see) the piece of paper that says “dry cleaning” on it. And I can know by reflection alone that the former experience caused me to remember (or seem to remember) that I need to pick up the dry cleaning. So I know by reflection alone that I am in the latter mental state because I was in the former mental state. But I don’t need to infer this from any premise that is not known by reflection alone. In John’s case of knowing that I am in a certain mental state because of some neurological cause, presumably I can know that that causal relation obtains only if I infer it from some premise that I know empirically, no?

  49. “The presumption that people routinely are able to reflectively access the grounds of their beliefs is so firmly embedded in daily life that it could only be rebutted by extraordinary evidence.” But what’s at issue here is, exactly, whether our knowledge of our beliefs’ _grounds_ should be cashed out in terms of knowledge of our beliefs’ _causes_. I agree that we generally & as a matter of course (albeit perhaps not infallibly or anything like that) have good first-person access to our grounds. That, plus the Nisbett et al. stuff, is exactly why we should reject reflective-access-to-the-cause accounts of basing like the proposal on the table here. As things stand now, the best interpretation of the evidence is that we just don’t have much of anything like the required degree of access.

    This is all the more so if we endorse Ram’s last comment (as it seems plausible to me that we should); for our sense that we at least _sometimes_ can figure out what our beliefs’ causes were, is going to include a fair number of cases where extra-reflective resources were brought to bear.

  50. Jonathan,

    I’m not sure whether this case was one of Nisbett’s, but there is an example discussed in Adler’s Belief’s Own Ethics in which customers are shown an array of indistinguishable socks, asked to select one, and later justify their choice by reference to distinguishing characteristics that just weren’t there. I guess there was a tendency to select socks based on position but when asked, subjects offered some bogus reason to justify their choice.

    It seemed that if what these examples showed was that we had better access to our reasons than what caused our beliefs, presumably some distinguishing characteristic appeared to the agent to justify their decision, but isn’t the more plausible empirical hypothesis that there were no such characteristics present to the mind regardless of whether they were causally efficacious or not? I don’t quite see how the challenges you raise would count against causal accounts of basing without threatening access accounts of basing, too.

  51. I think that’s a good point, Clayton, but here’s how I’d want to address it: the N&R results are part of the evidence showing that we do not have much sensitivity to the causes of our mental states. Inasmuch as the reflectively-known-causes theory requires of us that we have such a sensitivity, that theory is in trouble.

    (Note that the trouble is not that it cannot handle the laboratory cases considered in themselves. For I presume that the right thing to say about the preferences in the stockings case and others like it is that such preferences are, in fact, baseless. And the reflectively-known-causes theory presumably gets that right.)

    Now, we do have a capacity to confabulate when asked to offer a rationale for something for which there is no existing accessible rationalizer, as with the stockings case. And there are some areas where we take ourselves to have reliable accessible rationalizers, but empirical science shows our reliability to be much poorer than we thought; e.g., the work on false recovered memories by Loftus and others, and the ever-mounting literature on the unreliability of eye-witness testimony. It is an empirical question just which parts of our reflectively-accessible cognition are actually worth accessing. But I don’t yet know of any results showing that our mass of standing beliefs — surely the most standard items to appeal to as rationalizers — are subject to massive systematic distortion. I don’t know of evidence that shows that, for example, when we report having a particular belief, we are systematically wrong about what belief it is or whether we even have it. So an access account of basing may have to watch its empirical step somewhat, but it doesn’t at this point seem to have any empirical commitments that are already falsified as badly as the reflectively-known-causes theory does.

Leave a Reply

Your email address will not be published. Required fields are marked *