draft of paper on epistemic defeat

I’ve just put a draft online of a paper of mine on Two Approaches to Epistemic Defeat. It’s a follow-up to some of my earlier posts on the topic, and as always, comments welcome.

The central idea is that there is a frontdoor and backdoor approach to the concept of defeat. Where the house itself is the noetic system, the backdoor approach conceives of defeat in terms of putting something in the house and seeing what gets kicked out the back door. The front door approach characterizes defeat prior to entering the house, the primary example of which is a relationship between the propositional contents of possible beliefs. Pollock’s account of defeat is an example of the latter; Plantinga’s of the former. The argument is a further defense of the view that propositional justification is more basic than doxastic justification, since I argue that there are irremediable problems for a backdoor approach to defeat.


Comments

draft of paper on epistemic defeat — 9 Comments

  1. Haven’t read your paper yet Jon, but I look forward to doing so beause we need more work on the nature of defeat. I lean towards a pluralistic approach — I think defeaters come in many varieties and I’ll be surprised if any narrow account (one that says they all come in the same kind of package) is correct. This issue crosscuts the question, or so it seems to me, of whether propositional J is more basic than doxastic J. That’s an interesting question too of course.

  2. Fritz, do you have examples that you think Plantinga’s account can’t address? I’ve been worrying about whether I have an adequate stock of examples of types of defeaters in mind, so any ideas would be helpful.

  3. I don’t have a challenge to Al ready to offer. As I recall Al’s approach is better suited to subjective defeaters (defeating propositions that I bear some attitude towards) than to purely objective defeaters (eg, something you should believe but don’t and indeed that you don’t bear any attitude towards). But it’s been awhile and I need to re-reread some relevant material including your new paper. In general, what I’ve thought in the past is that it’s very hard for one non-disjunctive account of defeat (and/or defeasibility) to capture these two very different sorts of beasts under one umbrella account. But perhaps progress has been made that I haven’t tracked properly.

  4. Hi, Jon.

    I enjoyed reading your paper (and the blog in generalâ�� sorry not to have had the opportunity to comment until now). I’m pretty sympathetic to your ultimate verdict on a doxasticist account of defeat, though I wonder if Plantinga could at least successfully modify his account to handle the “defeaters defeaters” worry you give. The result you argue for is that on the most plausible Plantingan treatment of defeater defeaters, it is possible for too many beliefs to get counted as potential defeaters for some belief B. Here is, I take it, how your argument goes:

    Plantinga needs to distinguish actual defeaters- defeaters that actually do defeat- from potential defeaters- defeaters that would defeat if they themselves were not defeated by “defeater defeaters.” The most plausible way to characterize, in doxasticist terms, a potential defeater for a belief B is to say (among other things) that it is a belief D that satisfies the following counterfactual: if any defeater defeater DD of D (and grounds for DD) were removed from the noetic structure of the properly functioning believer in question (call her S), then D would be a defeater of B. Now, the problem is that some defeater defeaters are such that they not only defeat the defeater D for B, but those defeater defeaters also provide S with her grounds for B. Removal, then, of some such defeater defeater DD (and S’s grounds for believing DD) from S’s noetic structure will result in S’s withholding B. Now, at the same time that she withholds B, S holds many other beliefs. According to Plantinga’s account of defeaters, the coincidence between the holding of those beliefs and the withholding of B is what makes Plantinga’s account classify those beliefs as defeaters of B (under the condition that DD and its grounds are removed from S’s noetic structure), even if we would not ordinarily classify those beliefs as defeaters of B. Since the beliefs get improperly counted as defeaters of B under the condition that DD and its grounds are removed from S’s noetic structure, the beliefs get wrongly classified as potential defeaters of B, according to the above characterization of potential defeaters.

    Here is what I have in mind to remedy the defect in Plantinga’s account. If Plantinga modified the final part of his original proposal’s clause (2) to read “…would withhold B on the basis of believing D,” then not just any old belief could count as a potential defeater of B according to your proposed characterization of potential defeaters. Under the modified Plantingan proposal, it’s not enough that some belief D is coincident with S’s withholding B for D to count as a defeater of B; the adoption of D needs to provide the reason that S withholds B. The modification to Plantinga’s proposal insures that only those beliefs bearing the proper defeating relation to B get counted as potential defeaters of B. (The proper functioning condition in Plantinga’s (2a) clause insures that the withholding of B on the basis of D is appropriate.)

    Now, maybe something like what I just mentioned is what you had in mind in offering your final modified Plantingan proposal (p. 9), where (2d) is modified to indicate that S*’s response to D is part of S’s design plan. I don’t think your p. 9 proposal goes far enough, though, to insulate Plantinga against your objection. Take any one of the beliefs that Plantinga’s original account would wrongly classify as a potential defeater of B; call it D. S* could respond to such a belief in whatever way her design plan calls for, thereby satisfying (2d), and still withhold B (since, in the counterfactual cases in question where DD is not present in S’s noetic structure, she lacks the grounds DD for B). Such an accidental connection between holding D and withholding B can result in your modified Plantingan proposal also wrongly classifying D as a potential defeater.

    Regardless of whether Plantinga’s account can be patched up to handle your defeater defeaters worry, I do think that what you say in Section IV of the paper is convincingâ�� there are just too many ways that an improperly functioning agent’s noetic structure can be repaired for the counterfactuals in Plantinga’s account to render an unambiguous verdict about what count as defeaters for that improperly functioning agent.

  5. Hi again, Jon.

    One small revision in my proposal for fixing Plantinga’s original doxasticist account of defeat: the final part of Plantinga’s original proposal’s clause (2) is to be modified to read “…would withhold B AT LEAST PARTLY on the basis of believing D.” In the counterfactual circumstances that are relevant to your “defeater defeaters” worry (the circumstances in which DD and its grounds are absent from S’s noetic structure), S, of course, withholds B on the basis of her lack of grounds for holding B. But we could also think of her as withholding B partly as the result of her holding the belief D, since D, in its capacity as a defeater for believing B, thereby contributes to her lack of grounds for B.

    — Chad

  6. Chad, thanks so much for the comments! Your suggestion about defeater defeaters is a good one that I should take into account. So, as I understand your proposal, the counterfactual sends us to a close world in which DD is absent and D is present, the same as in the account in the text. The difference is that we look for the basis of the withholding of B.

    Here’s a quick question about your proposal. If we suppose that the withholding is not itself a propositional attitude taken by the person toward the proposition in question (since there need not be any such attitude), then we have a non-event being caused. (I’m assuming here that the basing relation is at least partially causal). That is, there is a world in which DD is absent from the noetic system, D is present, and B is absent, where the absence of B is caused by the presence of D. That requires that non-events have causes, which sounds strange. Why wouldn’t it be an adequate explanation of a non-event that nothing caused the event?

  7. You wrote:

    Here’s a quick question about your proposal. If we suppose that the withholding is not itself a propositional attitude taken by the person toward the proposition in question (since there need not be any such attitude), then we have a non-event being caused. (I’m assuming here that the basing relation is at least partially causal). That is, there is a world in which DD is absent from the noetic system, D is present, and B is absent, where the absence of B is caused by the presence of D. That requires that non-events have causes, which sounds strange. Why wouldn’t it be an adequate explanation of a non-event that nothing caused the event?

    Thanks for the good question, Jon. Here are my initial thoughts on it:

    In describing reasons for withholding belief, maybe it’s best to eschew talk of the basing relation, for the reason you give- withholding is a non-event that it sounds odd to say is “caused.” I think, nonetheless, a request for an explanation of the reasons why a person withholds belief, or, at least, the reasons why a person withholds belief rather than believes, can still be in order. Just as we are often curious about why people believe certain things, we are also frequently interested in determining what state they are in that keeps them from believing certain things. That is especially true when it is puzzling to us why an epistemic agent withholds belief when the information that we expect the agent to have suggests why believing rather than withholding is called for. We want to know what additional information the agent has or what information she lacks that leads her to withhold. Regardless of whether the agent’s withholding is puzzling or not, an explanation of it can proceed by citing the person’s lack of evidence and/or possession of counterevidence for the withheld belief in question.

    In light of these thoughts, perhaps it’s best to reformulate my modification to the final part of Plantinga’s (2) clause in the following explanation-based terms: “…would withhold B (or believe B less strongly), and S*’s withholding B rather than believing B is partly explained by her believing D .”

    We can understand this reformulation as my gloss on what the basing relation amounts to when applied to withholding belief- S withholds belief B partly on the basis of believing D just in case S’s believing D partly explains why S withholds B. Perhaps invocation of explanation is even a nice way of generalizing the basing relation for both believing and withholding- S believes/withholds B on the basis of believing R iff S’s believing R explains why S believes/withholds B. I confess that I haven’t read much lately on the basing relation, though, so perhaps there are counterexamples to an explanation-based account of that relation that I’m overlooking? In any case, even if we cannot characterize basing in terms of explanation, I hope the above explanation-based modification to Plantinga’s (2) clause is an improvement on my original modification.

  8. Hi, Jon.
    I like the new part of the paper. The final “back door becomes a front door” reply sounds right to me, and I think the first part concerning explanations is promising. Let me throw an idea at you about that part; see what you think of it.

    When I suggested my “Believing D partly explains S’s withholding B” addition to Plantinga’s proposal, I had a specific idea of “partly explains” in mind. Having read the new part of your essay, I realize now I wasn’t being as precise about “partly explains” as I should have been. What I meant by “X partly explains Y” is not merely that X is one component in a multi-component explanation of Y, but that (1) X, on its own, offers an explanation of Y, even while (2) there are additional salient explanations of Y in the near vicinity. What I was thinking of, for the defeater defeaters case you raise, is that there is a certain sort of “overdetermination” of S’s withholding B- not only does B, not believing DD, lack positive evidence for B, but in D, she also has evidence against B. So in response to the request for explanation, “Why does S withhold B?,” we could say either that S lacks evidence for B or that she has counterevidence to B. Either reply on its own constitutes an explanation of why S withholds B, even if each on its own does not give the full explanatory story of why S withholds B. In the sense of not giving the full story, we can say each only “partly” explains B.

    Before I explain the relevance of this more particular understanding of “partly explains” to what you say in your new reply, let me say a little more about what sorts of explanations count as legitimate explanations of S’s withholding B. The following claim sounds plausible to me: relative to the stipulation (which Plantinga should add to his account) that S’s withholding B is a display of the truth-aiming part of S’s design plan, some fact will explain S’s withholding B only if it refers to S’s lack of evidence for B or S’s possession of evidence against B. That’s because in the case that S’s withholding B is a display of the truth-aiming part of her design plan, that withholding will be a response to her evidential state (lack of evidence for / possession of evidence against B).

    Other facts- about S’s emotional health or prospects for survival, for instance- could explain S’s withholding B, but S’s withholding B for those non-epistemic reasons would not be a display of the truth-aiming part of S’s design plan. Those facts, then, won’t enter the explanations of S’s withholding B that enter my modified version of Plantinga’s proposal. So as long as S’s withholding B is a display of the truth-aiming part of S’s design plan, any explanation of that withholding will make essential reference to facts about what evidence she has or lacks. Any explanation, though, that makes essential reference to S’s lack of evidence for B or S’s evidence against B does legitimately count as a defeater for B, in virtue of describing S’s lack of evidence for B or S’s evidence against B. The stipulation that the explanation make essential reference to the evidential facts is what insures that only appropriate explanations get counted as defeaters according to my modified version of Plantinga’s proposal.

    To highlight a couple salient points about what I’ve suggested so far:
    on the particular understanding of “partly explains” I lay out above, S’s believing D won’t “partly explain” her withholding B unless it offers an explanation, on its own, of her withholding B; the “partly” modifier here only refers to the fact that S’s believing D doesn’t give all the relevant explanations of S’s withholding B
    an explanation of S’s withholding B, assuming it is a display of the truth-aiming part of her design plan, will make essential reference to S’s evidential state. (N.B.: Plantinga need not make a potentially circular reference to confirming or defeating evidence in his characterization of defeat- he need only say that S’s withholding B is a display of the truth-aiming portion of her design plan.)

    Let me apply these points to the potential counterexample you raise in the new part of your paper. As I understand it, here are relevant features of the example:
    Joe knows that Jeff and Jimmy are his only potential (immediately accessible?) sources of information about the color of the leaves on a certain tree (obviously, they aren’t the actual sources of Joe’s info about the color of the leaves, since, as you go on to stipulate…
    Joe does not have any information from Jeff and Jimmy about the color of the leaves; their testimony is absent from his noetic structure.
    The fact that Jeff and Jimmy are Joe’s only sources of information about the leaf color helps explain why Joe withholds the belief that the leaves are red, yet
    the fact that Jeff and Jimmy are Joe’s only sources of information about the leaf color does not count as a potential defeater of the belief that the leaves are red.

    (Is that an accurate understanding of your example?) The account I’ve laid out above agrees that Joe’s knowledge about his information sources does not count as a potential defeater of the leaf color belief. The reason: the fact that Jeff and Jimmy are Joe’s only sources does not, on its own, constitute an explanation of Joe’s withholding belief about the leaf color. What needs to be added to that fact for it to explain Joe’s withholding is the fact that Joe does not have any leaf color information from Jeff and Jimmy. So on the above proposal, the fact about information sources does not, on its own, get counted as a potential defeater of the leaf color belief. Yet if we supplement that fact with information about Joe’s lack of evidence for that belief (as we would when we add that Joe has no testimony from Jeff and Jimmy about the leaf’s color), then we do a get a defeater for the leaf color belief, just as we should: certainly Joe’s lack of evidence counts as a defeater for that belief.

    Well, if I were Plantinga, that’s what I would say in reply to the new part of your paper. Having said that, I still think, as I mentioned before, that the doxasticist account of defeat is ultimately going to be found lacking (or, at the very least, it’s going to be considerably more complex than the propositional account of defeat, as the defense of Plantinga’s approach against counterexamples is proving the doxasticist account to be). Having attempted in my dissertation-writing days a doxasticist characterization of the evidential relation “e is evidence for h relative to b,” I have a hard-earned appreciation of the difficulties of defending doxasticist characterizations of confirming/defeating evidence! And as you say, there are just too many ways that an improperly functioning agent’s noetic structure can be repaired for the counterfactuals in Plantinga’s account to render an unambiguous verdict about what count as defeaters for that improperly functioning agent. So even if what I say here helps Plantinga out a little, my sympathies are with you in characterizing evidence and defeat.

  9. Chad, I really appreciate all the time you’ve put in here! A couple of comments about your revision. First, you’re right to distinguish what we might call partial explanation versus incomplete explanation, and the distinction may affect the example I used. I’ll think about the distinction some more… One point worth mentioning, though: your response on behalf of Plantinga makes essential reference to facts as opposed to beliefs or experiences, and facts are never purely epistemic defeaters. They can undermine knowledge, but not in the way purely epistemic defeaters do. I like Bergmann’s way of characterizing the difference: one kind is purely internal, and relates to the standard third condition for knowledge, and the other is external, relating to the fourth condition.

Leave a Reply

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