Posted by: jadamcarter | 28 October, 2008

Truth, ability and epistemic luck

One thing virtue epistemologists (as well as other epistemologists) try to sort out is how the connection between true belief and abilities should be understood in cases of knowing. One widely-shared intuition is that knowledge requires of a true belief that it ‘depend’ on ability. In a recent paper, Michael Levin (2004) attacks virtue-theoretic analyses of knowledge by arguing that there is no clear sense in which an explanation for how a true belief ‘arises out of ability’ (i.e. depends on ability) will suffice for ensuring that the true belief will not depend on luck (of the sort we think to be incompatible with knowledge). If Levin is right, then the virtue epistemologist has as an important task the elucidation of the dependence relationship a true belief has on ability and luck, respectively. Here, I just want to pose a question: what exactly is it for a true belief to ‘depend on luck.’ If a true belief is lucky just in case it’s false in most nearby worlds in which it is formed in relevantly the same way that it is formed in the actual world, then we might naturally say that a true belief ‘depends on luck’ when it depends on the actual world, given the way the belief was formed, being unlike most nearby worlds in which the conditions giving rise to the belief were held fixed. If this way of thinking of the conditions under which a true belief depends on luck is plausible, then a virtue theoretic approach to analyzing knowledge would seem to have two strategies open for making the case that when the truth of a belief depends on ability (in the sense that knowledge requires) it will not ‘depend on luck.’ One way would be to hold an incompatibilist thesis with respect to what it is that a belief’s being true can depend on. Such a thesis might be:

Dependence incompatibilism: The truth of a belief depends on ability iff it does not depend on luck.

Alternatively, a virtue-theoretic account of knowledge might appeal to a different conception of this dependence relation: call it ‘dependence compatibilism.’

Dependence compatibilism: (i) The truth of a belief can depend at the same time on both ability and luck; (ii) a belief is known only if it depends on ability moreso than luck.

Strictly speaking, I think any attempt to preserve the incompatibilist thesis is doomed; for example, I know immediately after I park my car where it is, and my belief’s being true here seems to depend little on luck–it doesn’t depend much on the world being different from most nearby worlds–which are also ones in which my car will still be there (presumably, it’s a far-off world in which my car disappears moments after I park it). HOWEVER, I still know my car is where I parked it 10 minutes after I parked it… and that my belief is true ten minutes later plausibly depends at least a little more on luck than it did immediately after I parked it. After all, the world had 10 minutes more of an opportunity to be different from what I thought it was. Further… we might say that after a week, I no longer know where my car was; the truth of my belief seems now to be such that it ‘too easily’ could have been false. 

These intuitions would license us to think that the truth of our beliefs depends on ability and luck in a way that is gradient rather than rigid. If this is right, then a theory of knowledge that requires of a knower that her true belief ‘depend’ on ability should be concerned not with explaining how the true belief does not ‘depend on luck’ but rather with explaining how knowledge requires the truth of one’s belief to depend on ability ‘moreso’ than luck. Or at least, this is the idea I want to advance. 

I’m interested in what thoughts anyone might want to share about how either (i) My reasoning here is mistaken; or (ii) how a theory of knowledge could go about preserving this particular intuition–that knowledge requires a true belief depend (comparatively speaking) on ability moreso than luck. What might such a an analysis look like once its conditions are specified? 




  1. Nice post Adam. Can you give us the reference for Levin (2004)? As you know, I defend the same claim that you attribute to him here (albeit not in a fashion that takes into account your new proposal).

  2. I guess that’s the one:

    M Levin, Virtue epistemology: no new cures, PPR 69:2, 2004.

    I guess the motivation for “incompatibilism” arises in part from lottery cases. They motivated Unger’s requirement that a belief was not *at all* accidentally true (Unger 1968) and (if i am right) Duncan’s requirement that a belief be true in all close possible world. (If I remember right the condition was that it be true in “most, if not all” close worlds.)

    So here’s a case. You have a ticket and you know that if you take it to a certain person they might give you a gift. After extensive and cunning enquiry, you gather enough evidence to know that the ticket is actually a lottery ticket, and that the odds of your getting a gift are 1 against a million. You come to believe that it’s not worth bringing the ticket to the person because you won’t get a gift. And as it happens that’s true.

    The belief that you won’t get a gift is true, and it seems that its truth owes a lot to your ability to discover the odds. It actually doesn’t “depend on luck” by your definition (“a true belief ‘depends on luck’ when it depends on the actual world (…) being unlike most nearby worlds”) since the actual world is like *most* nearby worlds in the relevant respect. If you strengthen the definition a bit (“a true belief ‘depends on luck’ when it depends on the actual world (…) being unlike SOME nearby worlds”), you get the conclusion that it does depend on luck, but just a bit: so to speak, it’s not like it’s a big piece of luck to be in a loosing worlds, most of them “in the area” are such.

    So it looks like your ability does most, and luck helped just a bit. If that’s so, your belief depends “moreso” on ability that belief, but (the usual intuition goes) you don’t know.

    A second worry is how to cash out “depend on ability” in the compatibilist view. One way to cash out “the truth of the belief depends on ability” is to say: in all close worlds, using the ability guarantees (implies) truth. By that definition incompatibilism follows (if a true belief depends on luck just if there is a close world in which it’s formed through the same ability but false, then the ability doesn’t guarantees truth in all possible worlds.) So how one is to cash out “the truth of a belief depends on ability” in a compatibilist view? Do you want to say something like the probability of one’s belief being true is greater conditional on one’s having formed it that way?

    Here is one way to go I can think of. Let C be the proposition that obtains just at all close worlds. (If C is the case, you’re in a close world, and C is the case in all close world.) Let A be the proposition that you use the relevant ability. You could say: the truth of a belief depends on ability iff the probability of having a true belief is greater on C&A than on C.

    Two coments about that definition:
    – “dependence on ability” in that sense is a very weak condition. As long as your ability just does a tiny bit towards making your belief more likely to be true, then the true belief counts as “dependent on ability”. It clearly won’t be sufficient for knowledge.
    – if an ability is used in all close worlds, you automatically get the truth of the belief NOT depending on the ability.
    But there’s a problem with that. In order to make “moreso” comparisons we need a measure of the contribution of luck. One natural way to do it is to use the conditional probability of having a true belief on C alone. (The greater the probability, the less it depends on luck; if the belief is true in all close worlds (or would be if formed) then it doesn’t depend on luck at all). So we’d say that a belief depends more on luck than on ability if the conditional probability of true belief on C is greater than C&A. Now here’s the problem: by those two definitions we get incompatibilism: the truth of belief depends on ability iff it doesn’t depend on luck.

    There’s an alternative way of doing things though. The conditional probability of having a true belief on C&A is your chance of having a true belief, given that your belief is formed through A. Now when it’s just .9 probable that forming a belief that way would make it true, your belief depend a bit on luck. If forming the belief that way made its truth .1 probable, and it’s true, then it’s very lucky that it’s true. So you could say that the measure of luck is 1 – (the probability of having a true belief given C&A). In that way you get that dependence of ability and dependence on luck always add up to 1; the more lucky it is that a belief is true, the less its truth depends on ability, and conversely. Maybe that’s the kind of picture you have in mind.

    In that picture you get the lottery problem though. If I just know that the odds are .9 of loosing, then my (suppose) true belief that i loose depends 90% on ability, 10% on luck.

  3. (sorry for the many spelling mistakes!)

  4. NB, after the two comments, read “But there’s a problem with that” as “But there’s a problem with that way of defining dependence on ability.”

  5. Is it entirely “luck” if an individual “A” manages to make it to point “2” even if he had no prior knowledge of “2”, and no help from another person “B”, even though “A” was suppossed to go to point “1”? Keep in mind point “2” is no better than point “1”, but by “A” going to point “1”, he gained an advantage over the efforts of “B”?

  6. Fitswater, I have always contended that knowledge cannot depend on luck, and hence, that knowledge can be undermined by luck. In the past, attempts to define a ‘lucky’ event have been sparse and unhelpful. Luck has either been taken as a basic concept requiring no recursive definition, or defined in such a way that intuitive features of luck—such as subjective significance—have been ignored. Consequently, luck, when defined, has been conflated in the literature with logically distinct concepts such as ‘accident’ and ‘chance.’ Now hook yourself to these thoughts and let me know.

  7. “K” likes carrots, “A” likes peas, “W” likes bananas.

    “K” invites “W” to dinner and serves bananas, however “K” had no prior knowledge that “W” likes bananas.

    Does “K” have knowledge or pure luck?

  8. Let me tell you about the truth about mere true belief. If I had believed what you wrote, I could have never won all seven of those games between 2002-2003.

  9. Be careful not to let these thoughts consume your life. We need you to be focused, don’t go getting soft on me.


%d bloggers like this: