Kantian Autonomy and the Value of Knowledge

Here’s a proposal about the value of justification, and the related issue of the value of knowledge. Suppose we think of justification along Kantian lines, as follows. What is valuable about justification is that it constitutes an expression of our autonomy as cognitive beings (in search of the truth). Autonomy here has nothing to do with libertarian freedom, but instead involves following rules that are self-given rather than imposed by some external authority.

In the epistemic context, the rules in question have to be the evidential theory that determines whether or not one’s beliefs are justified. Some versions of evidentialism will give some objective account of the evidential theory, and thereby eliminate the possibility of an appeal to autonomy to explain the value of holding beliefs in accord with the theory. The best example I know of where a theory preserves autonomy is a version of Bayesianism, where coherence at a time is necessary for justification, and updating by conditionalization completes the picture. What is interesting about conditionalization here is that, to posit it as a requirement on justification presupposes that every person has a theory of evidence implicit in their conditional doxastic attitudes, a theory of evidence that tells them how to update their attitudes in the face of any possible learning experience whatsoever. In the context of my question about the value of justification, this Bayesianism is perfectly suited to an explanation of the value of justification in terms of following self-given epistemic rules as opposed to rules imposed “from the outside.”

The problem, of course, is that this version of Bayesianism is false. It nonetheless provides a model for how to develop the theory of justification so as to embody this Kantian intuition. The trick is to make the theory of evidence somehow elicitable from the attitudes (or dispositions?) of the epistemic agent. I think this is what Lehrer’s theory of knowledge is trying to do, even if it doesn’t quite honor the Kantian requirement in the end. So the proposal is that there is a kind of justification that is valuable in virtue of licensing beliefs that conform to one’s own internal epistemic standards.

Characterized in this way, what we get is Foley’s theory of justification, though there is a complication. For Foley, one’s deepest epistemic standards are defined counterfactually in terms of what conclusion one would come to given a unlimited amount of time to reflect on the matter. I think the theory would be better if the counterfactual were thought of as being good evidence for what one’s deepest standards are (since counterfactual theories can only be philosophically respectable by including an account of how the particular counterfactual theory in question escapes the conditional fallacy explained by Shope in his 1979 piece). Technical emendations aside, however, it looks from here that Foley’s theory can sustain a Kantian explanation of the value of justified belief by appeal to the autonomy of the epistemic standards in question.


Kantian Autonomy and the Value of Knowledge — 12 Comments

  1. Jon,
    I find it intuitively appealing that it is admirable for a person’s beliefs to express standards of acceptance he embraces. But one question I would have about it is whether (or perhaps how far) autonomy is indeed a valuable thing in the sphere of belief. The problem would arise if one wants to think of truth as external to the believer and wants to have truth in the drivers seat epistemically speaking. Truth should be in control and epistemic value should follow the control of truth over a believers beliefs, rather than the control of the believer over them. This seems to me to be a possible disanalogy with the case of autonomy in action…at least if one hews to a Kantian line, viz., valuable actions express control of the agent over his actions. Actions don’t have to conform to an external standard or be evaluated as such; a good actions just is one generated in the right way, from an autonomous will.

  2. So you’re sittin’ on the dock of the bay, wastin’ time…

    Truth does enter into the picture here, but not directly. The standards are the standards appropriate, by one’s own lights, for getting to the truth and avoiding error.

    If truth is in the driver’s seat, we’ll have a deep problem explaining how inhabitants of evil demon worlds can have justified beliefs. I think it is obvious that they can, so truth can’t be in the driver’s seat in too strong a fashion. Naturally, I think the above picture has truth in the picture in exactly the right seat!

  3. All but the ‘wastin’ time’ bit!
    From what you’ve said, by one’s own lights truth is in the driver’s seat, and surely by one’s own lights that is external to oneself as a cognitive agent. So by one’s own lights, the standards are laid down and ought to be controlled by something external to you qua cognitive agent, viz., truth. Perhaps standards can be autonomous yet not be viewed as autonomous, but not entirely. To this extent it seems the standards are not autonomous. Don’t we want believers to view their standards of belief formation as good for achieving something laying quite beyond their control?

  4. Robert, you’re certainly right about the standards as referring beyond themselves, so this account would have to be disanalogous to the case of action where autonomy is directed at expressions of a good will (though I’m not quite sure what to do with Kant’s remarks about the afterlife here). So in that way, the two cases will be disanalogous. But still there is this: in the case of belief, there is still the autonomy of rules given to yourself, of being judged by your own standards and living up to them or not, even if one views the standards as goal-oriented. (OK, this is too murky, but maybe someone will help in cleaning it up…)

  5. One other thought, Robert. Chisholm, in the third edition, changed his account of the epistemic goal. It was getting to the truth and avoiding error, but became holding rational beliefs and avoiding irrational beliefs. I take it you’d see more of Kant in this proposal than in one that had truth as the epistemic goal?

  6. Yes, that seems more like it.
    But, that said, then we should form our beliefs according to standards whose goal is that our beliefs are maximally rational, no reference to truth. And is that as valuable as forming beliefs according to standards whose goal is that our beliefs be true? I’m not sure.
    I *think*, when I’m engaged in epistemic activity, I want standards that to my mind will get me to the truth.

  7. An ideal epistemic agent wants to have as many true and as few false beliefs as possible (on topics that matter). A merely authentic agent wants his/her actions to be determined by his or her own choices. An autonomous agent wants more: that his or her actions be determined by choices that are not merely authentic but that satisfy appropriate standards or principles that are, in turn, endorsed by the agent in light of his or her interests. So, if we have an interest in having true beliefs (on topics that matter to us)–and I take it that at least we have an interest in having true beliefs about the objects we seek and about the best means for securing them–we don’t want truth per se to be in the driver’s seat, though we want the person in the seat to perform actions that are informed by the way things are. We want our actions to be determined (partly) by our beliefs and our beliefs to be determined by our search for the truth in light of standards we endorse for distinguishing what is true from what is false. So the fully autonomous agent needs to have an interest in the truth (a merely authentic agent can have beliefs that satisfy purely aesthetic standards, for example) and to endorse the procedures by which true beliefs are best acquired. Thus, full autonomy requires that one have both (i) reasons for one’s beliefs and (ii) reasons for thinking that one’s belief-forming procedures are reliable. Alas, (i) leads to the epistemic regress problem and (ii) leads to the problem of the criterion (which is a special case of the regress problem (more on that soon)). So an account of how autonomy is possible requires a solution to these two problems. That will not be easy, but, if my ruminations are on the right track, has an important moral dimension.

  8. Andy–I was hoping to flush you out with the post! Excellent ideas here, I especially like the distinction between authentic existence (a little closet existentialism in epistemology!) and autonomy.

    One thought about autonomy. I think it is right to say that such individuals wish to endorse, or actually endorse procedures or principles about belief formation that answer to the idea of using the principles best suited to getting to the truth and avoiding error. But I wouldn’t quite say that an autonomous agent will “endorse the procedures by which true beliefs are best acquired.” Perhaps best acquired by his own lights, but not the objective construal of this phrase.

    You then qualify it in point (ii) above. I think something like this point is right, but I worry as soon as talk of reliability creeps in, on account of the generality problem. There is also, by the way, Foley’s attack on this idea, but I can’t remember where it is (it’s in a paper from the early 80’s; I’ll find it if you’re interested). Maybe there’s a way to put the point without raising these issues: for example, maybe all we need is that the autonomous agent rationally endorses principles which, from his perspective, are the best available in the search for truth. Given a suitable solution to the problem of generality, that may imply your characterization, but it doesn’t raise that problem immediately. (Though a fair criticism is that it doesn’t raise it because it is so dang vague!)

  9. Yes, Jon, my reference to authenticity was deliberately intended to refer to existentialists and other “post-modernists” who want to replace the search for truth which must, barring miracles, need to be conducted in light of principles of evidence which, in order to make genuine autonomy possible, themselves be endorsed in light of reasons. If truth is either unimportant (or not very important), incoherent, or unattainable (as they maintain) then even if autonomy would be valuable, it is futile to pursue it. Of course, a complete response to post-modern alternatives to traditional epistemology will need to tackle directly the attacks on the possibility, coherence, or value of truth. In my opinion, the most sophiticated attacks of this sort are not to be found in thinkers such as Nietzsche, Heidegger, and Rorty, but in the work of “constructivists” such as Nelson Goodman and Catherine Elgin. Like you, Elgin takes understanding to be an important goal of inquiry, but she argues that understanding is often best promoted by propositions that are false (e.g. we understand a lot about kinematics and dynamics by principles that refer to (non-existent) frictionless surfaces). And Goodman seems to hold that understanding can be promoted by adopting systems of statements (constitutive of different “worlds”) that are, when taken together, contradictory, hence false. I’m not as impressed with Stich’s attacks on the value of truth. By my lights, they seem to be that since false beliefs are sometimes more valuable to us than true ones, that true beliefs are not valuable: a non-sequitur. Perhaps more charitably, he is arguing that since we sometimes have no interest in the nature of truth, we have no interest in truth itself. This strikes me as analogous to arguing that since we normally have no interest in the chemical composition of salt, we have no interest in salt itself (not so, especially for a saltaholic such as myself!). But perhaps I am misunderstanding him and should include him in the list of philosophers with deep challenges to the value and/or coherence of truth. Let me think some more about your other comments. It does strike me, however, that I might attain a degree of doxastic autonomy by having reasons for believing that my belief-forming procedures are reliable even if I have no direct control over those procedures. Perhaps our non-inferential beliefs about the physical world that are prompted by sensory experiences are like this. In the near future I plan to add a post arguing that the the problem of the criterion and the problem of induction are special cases of the epistemic regress problem. I’m still working the bugs out of the part of the argument dealing with the problem of induction. But I think all of these problems are deeply important because (i) they rest on premises that seem to be necessary conditions of having the kind of justified belief that is required for understanding and autonomy and (ii) if correct, they show that such autonomy is impossible.

  10. Andy–I agree with your ranking about which objections to take most serously. I need to look at Elgin’s book, but the two examples you give don’t move me much. The second one I don’t understand, but the first I think I do, and I think it mischaracterizes what physicists do when they idealize. The statements are statements within the model that is being used, and the point of the model is that it is an idealization of what is actually the case. The statements are not false, but true statements about the model. Their cash value in terms of understanding is a function of the value of idealization, and that can be accounted for without ever claiming that it is in virtue of maintaining that certain false claims are true that the idealization aids our understanding of the actual phenomena.

    That’s not very clear, but I hope it gives some idea of how to avoid saying that understanding depends crucially on falsehoods.

    There is a way in which understanding can involve minor falsehoods: say you understand Comanche dominance of the southern plains from 1775-1875. You might have some peripheral false beliefs about this subject matter, and yet have understanding. But that is a far cry from saying that understanding depends on falsehoods.

  11. Jon, in your Value book you reject Swinburneâ??s account of the value of objective internalist justification because for him this entails a priori apprehension of the *true* principles of inductive reasoning. Thus his concept of justification employs the concept of truth and he faces the swamping problem (p. 55). Now when you write

    â??What is valuable about justification is that it constitutes an expression of our autonomy as cognitive beings (in search of the truth).â??

    it seems to me that you face the exact same problem you say Swinburne does. Can you help me see it from an angle where it doesnâ??t look like that?

  12. Trent, the primary issue for the swamping problem is explaining what of value is found in justification. If the value in justification is instrumental to truth, then once truth is already in the picture, no additional value is present in virtue of a property understood in terms of instrumentality toward truth. That’s a problem for Swinburne, but more directly for reliabilism.

    An expression of autonomy, in the sense involved in the quote, doesn’t involve any instrumental relationship to truth. You simply take a being engaging in cognition understood in terms of a search for truth, and ask what is valuable, if anything, about the activity. Autonomy in the sense in question involves following rules you give to yourself, so that the theory of evidence to which you are answerable in forming and holding beliefs is subjective in an important sense.

Leave a Reply

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