Reply to a Question about Street

(I originally posted the following on Twitter (hence the less formal writing), in response to someone asking me if I had ever read Sharon Street’s criticisms of Korsgaard’s moral theory, and what my thoughts were if so. My response was long and so I decided to post it here as well. )

I’ve read her paper “Coming to Terms with Contingency” and I have the following thoughts. First, it is unfortunate that she didn’t address Korsgaard’s criticism of the Humean theory of motivation. In “The Normativity of Instrumental Reason” Korsgaard pointed out that although the Humean theory cannot account for the *normativity* of the instrumental principle, which tells us to take the means to our ends. By this she means it can’t account for the fact that the instrumental principle is a rule that, as a result of our consciousness of it, regulates our actions; this is because on the Humean theory such a rule simply cannot be violated, since any action can be interpreted to be in accord with the instrumental principle.

For instance, suppose I have to work on an essay but I also want to watch Netflix instead. It’s impossible for me to make an irrational choice here, since if I decide to watch Netflix that just means, for the moment, that my desire for that was stronger than my desire to write a good essay. When deliberating the instrumental principle doesn’t have any authority over me, since I am not obligated to take the means to any particular end, and my past decisions about what ends I want to obtain don’t have authority over me in the present. “Norms” of instrumental rationality, therefore, become vacuous. This will be important in a sec.

 

So, Street’s critique of Korsgaard is this. In order to get Kantianism, Korsgaard performs a sort of regress, asking how the objects of our inclinations can be valuable. Ultimately she says that, in order to consider the ends we set for ourselves valuable, we have to consider our own humanity valuable, in the sense that we have to regard our will as having the capacity to confer value upon things as a result of its choice. So, Korsgaard is making me put all of my values in a little box, and then asking me how I can consider what I have before me to be valuable, and she claims (in a nontrivial step; I won’t go more into this because it’s not what Street is focusing on) I have to go Kantian.

This is what Street objects to; according to her, it doesn’t make sense to ask about what justifies value-judgement from such an abstracted perspective. Such judgements are always relative to a “web of interlocking desires” had by the agent, and are presupposed in making any practical judgements (I shall continue to refer to one’s complex system of desires as their “web”). Korsgaard, then, is begging the question against the Humean by saying that I have to be able to justify the whole web in the first place on grounds independent of any particular web, since this is just denying the web-relativity of normative judgements.

Before getting at why I think this objection fails, I need to make a minor formal point. As described above, what Street is saying is that, semantically, a value judgement cannot be true simpliciter, rather it is only true relative to a web of values. So to talk in a way that makes this explicit, strictly speaking, it doesn’t make sense to talk about whether an end is valuable full stop, but rather we can only talk about whether it is valuable_W, for some web of values W. When I ask without making any qualification if some option of mine should be taken or whether something is valuable, I’m asking if it’s valuable_X where X is my web of values. Now, just as a matter of notation, I can transform any of these web-relative judgements into a neutral judgement that does not presuppose any such web, namely by replacing the value-judgement with a conditional whose antecedent describes the web. So, my utterance of “E is valuable” is not true or false simpliciter, but only relative to a web of interlocking values; it is, we may suppose, true relative to my web X. But then the “de-relativized” judgement “E is valuable relative to the web X” (or, equivalently, “Given X, E is valuable”; this is the conditional I just mentioned) is completely value-neutral; the utterance is true simpliciter, not true relative to a particular web of values.

As an analogy, consider the sentence “Jane is happy.” It cannot be said that the property of happiness may ascribed to Jane full stop. It can only be ascribed to her relative to a time, and when I say “Jane is happy” I am saying the property applies to Jane now. But I can remove the time-relativity of the sentence just by building it into the sentence itself; the sentence “Jane is happy at time t” is, now, true or false simpliciter, and not merely true or false relative to some time.

 

You can probably see the move I’m going to make now. For the purposes of this reply I hereby grant that statements of the form “E is valuable” or “I ought to do this action” cannot be true simpliciter, but only relative to a given system of contingent values. Korsgaard’s reasoning nevertheless applies to the de-relativized versions of the relevant normative judgements; indeed this is what she originally intended them to refer to. And because these versions indisputably do not presuppose any contingent circumstances, Street’s objections do not apply. Non-Kantians must therefore object to how she answers the question (of what neutral grounds can allow us to justify the de-relativized normative judgments), and not the question itself.

The point is, Kant’s discussions of practical normativity were, from the beginning, set up in such a manner so as to be impervious to Street’s objection. For Kant, the proper object of normative evaluation is not an action simpliciter, but rather a maxim, which is a rule-candidate of the form “I will do A in contexts C in order to achieve E”—it includes both the action and whatever contingent grounds the agent may have for doing that action. One’s web of values can be built into C and E if they are relevant to the agent’s deliberation (which, of course, they are). There is nevertheless a further question as to whether the maxim, taken as a whole, can be permissibly adopted.

It might be objected that, sure, we can de-relativize practical judgements in this manner, but once we do so justifying them on neutral grounds becomes extremely easy. Let’s say I have a web of values X, and I want to do an action A in the context C because it maximizes some of my values. Street points out that whether I should do A cannot be answers except relative to my web X; I reply that, nevertheless, the de-relativized judgement “Given the web X, should I do A in C?” must be justified on neutral grounds, since all the needed contingent stuff is now built into the judgement itself. But this seems trivial; obviously doing A in C is worth doing relative to X, by hypothesis. Why do we need to bring in any Kantian/Korsgaardian machinery to do this?

 

This is where the first point, about the apparent emptiness of instrumental rationality under the Humean conception, comes in. That X is “my web of values” says nothing other than that I am, all things considered, inclined to evaluate options according to X. Now, either such inclinations are sufficient to justify the action in question, or they are not. If they are, then we face the emptiness problem; it will be impossible for me to actually do something irrational. If I find myself inclined to do B instead, thereby acting against X, that B is wrong relative to X will not motivate or obligate me to act differently. Rather, I will simply decide to evaluate my actions according to some other web Y, and that will be permissible. We therefore do not have a non-Kantian theory of practical reason, but rather an error theory, not just about morality, but about all practical normativity. Sure, someone can permissibly call my actions irrational since they’re judging them relative to their web of values, but there’s still the distinctively first-personal issue that no normative consideration can motivate or obligate me to act against my strongest inclinations, since no particular web has any authority over me. So, while we can justify de-relativized judgements from a neutral standpoint for free thereby avoiding the specter of Kantianism, we do so at a pretty great cost.

The other option is that there are some substantive constraints on my actions, that my inclinations are not sufficient to justify my choice of a web-action package. And indeed this is true; the self-conscious nature of human practical reasoning makes it the case that there is a further question beyond my being inclined toward an action of whether I should act as that inclination recommends. There is, then, a non-trivial question as to what the standard of evaluation is of judgements like “Given the web X, I should do A in C,” because, if we are to avoid global practical error theory, there must be at least some cases in which I accept and act on a judgment of that form and I am yet doing something irrational.

 

In summary: Kant and Kantians are frequently criticized, e.g. by Street, on the grounds that they view practical reason as an a priori endeavor which must ultimately be in conformity with a law which is empirically unconditioned, i.e. which has authority independent of any contingent presuppositions (Kant justifies the authority of the Categorical Imperative on the basis that it is the unique empirically unconditioned practical law). Normative judgements, they claim, can only be evaluated from within certain contingent circumstances. But this line of thought is mistaken, for let us suppose it is correct; normative criteria can only be formulated given some contingent facts. Given an arbitrary description of such contingent facts C, let N_C denote the collection of resulting normative considerations which have authority over anyone in C. Then we can make up a more fundamental unconditioned law L which states: for each C, if you find yourself in C, obey N_C. L is empirically unconditioned because, by hypothesis, any contingent circumstances needed to justify the norms are built into the law. The authority of L must therefore be demonstratable a priori and without appeal to any contingent circumstances the agent may find herself in. Thus the fact that Kant starts his investigation by looking for an empirically unconditioned, categorical law is totally unproblematic.