Philosophical Studies

, Volume 144, Issue 1, pp 121–125

Knowledge as aptness


    • Department of PhilosopyArizona State University

DOI: 10.1007/s11098-009-9376-8

Cite this article as:
Cohen, S. Philos Stud (2009) 144: 121. doi:10.1007/s11098-009-9376-8


I raise several objections to Sosa’s account of knowledge as aptness. I argue that aptness is neither necessary nor sufficient for knowledge. I also raise some objection to Sosa’s treatment of dreaming skepticism.


AptnessCompetenceReflective knowledgeDreaming skepticism

I greatly admire Ernie Sosa’s book. It is filled with interesting and provocative arguments, and I learned a great deal by working through them.

The central claim of the book is that knowledge is apt belief. I will argue that aptness is both too weak and too strong for knowledge.

1 Aptness is too weak for knowledge

There are two kinds of cases that show, I think, that aptness is too weak for knowledge:
  1. (1)

    Suppose in the kaleidoscope case, the Jokester provides you with a white surface illuminated by red lights continuously over the period of 1 h, except for a one-second interval when the surface is red. You are staring at the surface for the entire hour believing it is red. During the one-second interval when the surface is red, your belief that it is red is apt—it is accurate because of the exercise of your perceptual competence. But it seems wrong to say that you know the surface is red for that 1 second interval. One could invoke the animal/reflective knowledge distinction and claim that you have animal knowledge for that one-second interval. But of course “animal knowledge” and “reflective knowledge” are technical terms. So we cannot appeal to our intuitions about whether you have animal knowledge in this case. Rather, invoking the animal/reflective knowledge distinction is appropriate only if we feel ourselves pulled in two directions—the subject knows and the subject doesn’t know. But in the case described, I feel no intuitive pull toward saying that you know the surface is red for that one-second interval. So there is no basis for claiming that you have animal knowledge in the case.

  2. (2)

    Suppose in the original Jokester case, a reliable source tells you about the Jokester. Then, even when the surface is red, intuitively, you fail to know the surface is red. But still your belief is accurate because of the exercise of your color vision competence, i.e., your belief is apt. In response to a similar problem for reliabilism, Goldman tried to appeal to there being an available reliable process (the proper use of evidence) which had you used it, you wouldn’t have believed the proposition in question. Perhaps a similar move could be made with respect to the aptness view, viz., there is an available competence (the proper use of evidence) such that had you used it, you would not have believed the surface is red. But there is no reason why we need to suppose that one must be competent at properly using evidence in order to be competent at color vision. One might argue that if one is not competent in the proper use of evidence, then one is not competent at color vision, because color vision depends on your properly using your perceptual evidence. So your belief would not be apt after all.. But if that’s the case, then your incompetence at properly using evidence would prevent your belief from being apt even in the normal case where you do not know about the Jokester, because your belief would not be based on a competence. But intuitively, even if you are not competent at taking defeaters into account, you can still know the surface is red on the basis of your color vision when there are no defeaters.


2 Aptness is too strong for knowledge

It seems that the correctness of one’s belief is attributable to a competence even if it derives from an exercise of that competence in inappropriate conditions. In such a case, it is not clear why the belief should not count as an instance of knowledge. If this is so, then the aptness account is too strong, because it allows that someone knows, only when one’s belief is based on a competence in appropriate conditions for the exercise of that competence.

Consider Sosa’s archer analogy. Suppose the target is very far away. Normally under these conditions, the archer would not be able to hit the target. But let’s suppose that in a particular case, the archer manages to hit the target. Sometimes people manage to perform much better than they are typically able to. Here it seems like the accuracy of the shot is attributable to the competence of the archer. Perhaps you might say that his success on this particular occasion is attributable to more than his mere competence. But surely the success of the shot is creditable to the archer. And I take it that this is the notion that aptness is supposed to capture.

Now consider someone who exercises an epistemic competence in inappropriate conditions but acquires a true belief anyway. Can his success be attributed to his competence? Can he be credited with his success? Suppose I am a competent reasoner, but only in conditions where I have had enough sleep. Nonetheless, on a particular occasion when I have not had enough sleep, I manage to reason soundly. Why isn’t the correctness of my belief attributable to my competence? It is at least true that my success should be creditable to me. Yet, since I was sleep deprived, I did not exercise my competence in appropriate conditions. Thus on the aptness view I would not know the conclusion of my reasoning. But intuitively, I would know the conclusion.

3 Reflective knowledge

Sosa’s alleges that there is an important difference between the between the Jokester case and ordinary perception, viz., that in the Jokester case, we have mere animal knowledge whereas in ordinary perception (typically) we have reflective knowledge. But I do not think this difference between the cases can be sustained. Condition C is supposed to distinguish between the two cases by requiring that the exercise of the competence in appropriate conditions would not too easily have issued a false belief. But as we will see, condition C does not rule out reflective knowledge in the Jokester case.

Competences exist only relative to appropriate conditions. But one is competent at detecting whether or not the conditions are appropriate, i.e., whether the lighting is good, only in conditions where inappropriate lighting displays itself in a certain typical way, e.g. dimness, flickering, visible colored light bulbs, unusually colored objects (e.g. red skin). That is to say, an appropriateness condition for exercising one’s meta-competence in determining whether the conditions are appropriate for color vision, is that bad lighting display itself in one of these typical ways. But the bad lighting when the jokester operates is not displayed in a typical way. Presumably, it is done in a way that makes the red lighting impossible to detect. So it’s not the case that the exercise of your competence for detecting good lighting in appropriate conditions would too easily have issued a false belief. For when the Jokester operates, the conditions are not appropriate for determining the quality of the lighting. Thus you can have apt knowledge that you have apt knowledge, i.e., you can have reflective knowledge that you see a red surface.

There is another way, one might turn out to have reflective knowledge in the Jokester case. Suppose you have an unusual condition whereby red lighting will kill you. If so, then exercising your meta-competence in those appropriate conditions would not too easily have issued a false belief. Were the Jokester to present you with a white surface illuminated by red lights, the red lighting would kill you and thereby prevent you from believing anything. Thus there is no barrier to saying that you have reflective knowledge that the surface is red. But intuitively, the fact that red lighting is deadly for you should not affect whether you have reflective knowledge. This objection is perhaps unfair, because this kind of objection can be made to virtually any analysis that uses subjunctive conditionals.

4 Dreaming and reflective knowledge

One of Sosa’s goals in the book is to disarm the skeptical threat to his account of knowledge posed by dreaming. Because we dream, it looks as if the exercise of our perceptual competence(s) could too easily issue a false belief. If so, condition C will determine that we fail to know. Sosa gives two reasons why C does not prevent us from having reflective knowledge that we are not dreaming. Reason (ii) notes that we would not be using the same perceptual faculties (competences?) when we dream. So the fact that we dream does not entail that the exercise of our perceptual competence would too easily issue in a false belief. But notice that this means that C would allow us to have reflective knowledge that we are not dreaming, even if dream states were qualitatively indistinguishable from waking states. It would still be the case that when we dream we are not exercising the same competence. But that seems to prove too much. Surely the fact that dream states differ from waking states is an essential ingredient for refuting dream skepticism. If our dream states were qualitatively indistinguishable from waking states, then it is hard to see how we could have reflective knowledge that we are not dreaming.

Reason (i) does say that we can avoid dream skepticism because when we are dreaming there are signs that what we seem to perceive is not real. But if this claim is sufficient to refute dream skepticism, then the refutation in no way depends on Sosa’s virtue theory.

In the first chapter of the book, Sosa attempts to refute dreaming skepticism about rational belief without appealing to his virtue theory. An important premise in the argument is that when we are dreaming, we do not have beliefs. We merely dream that we have beliefs. While I do not have space to examine these arguments, I find them convincing. The anti-skeptical argument proceeds by claiming that I have three options with respect to the proposition that I am awake—belief, suspension, and disbelief. But I know that taking the last option is self-defeating. If I believe that I’m not awake, then it follows that I am awake, since I can only have beliefs when I am awake. So I know the last option is defective. Moreover, I know that taking the first option will result in my having a true belief. If I believe that I am awake, it follows that I am awake, because I have beliefs only when awake. So the suspension option is also defective because my taking it would thereby prevent me from having a true belief. So I know that the only non-defective option is to believe I am awake. It follow that the rational thing for me to do is believe I’m awake.

Does this argument succeed? The initial that I have these three options presupposes that I am awake. For I have those three option only if I am awake. So to suppose I have those options is to suppose I am awake. This means that at most, the argument shows that if I am awake, it is rational for me to believe I’m awake.1 This in itself would be a significant anti-skeptical advance. The skeptic makes the unconditional claim that it is not rational for me to believe I am awake, even if I happen to be awake. So the argument would show that the skeptic has not made his case. All the same, I have three objections to this argument:
  1. (1)

    It is not clear why the argument would make it rational to believe one is not dreaming for someone who has not gone through the reasoning. The argument says that because I know that believing I am awake is the only undefective option, that is the rational option to take. So at most Sosa has established that it is rational for him (and anyone who has gone through the reasoning) to believe he is not dreaming. Moreover, the argument does not show that it was rational for Sosa to believe he was not dreaming, prior to his constructing the argument. But presumably, a response to dream skepticism should show that we are all rational to believe we are not dreaming, and that we always have been. (Slightly overstated.)

  2. (2)

    The reasoning of the argument would seem to go through just as well even if we knew we dreamed 99% of the time, and that our dreams were qualitatively indistinguishable from waking states. But surely if we knew that, we would not be rational in believing (when awake) that we are awake.

  3. (3)

    The argument would not work if instead of dreaming every night, we went into a hallucinatory state. For presumably one has beliefs when one is in a hallucinatory state. But it is at least odd to think that a skeptical argument based on hallucinations should have more skeptical force than a skeptical argument based on dreams. Suppose we were unsure whether we actually dream or just go into a hallucinatory state. It’s hard to see how we could think that finding out which is the case would make the difference as to whether or not our normal beliefs are rational.


Nathan Ballantyne and Ian Evans make essentially the same point in "Sosa's Dreams" Philosophical Studies, forthcoming.


Copyright information

© Springer Science+Business Media B.V. 2009