Tuesday, September 25, 2018

Korsgaard 01: On Instrumental Rationality

Assume that you had, as an end or goal, to learn a new language. You judge that a satisfactory way to learn a language is to enroll in an online course. Why should you enroll in the online course? Just because something is a means to an end, why should that be a reason to do it?

A participant in a discussion elsewhere has directed me to The Normativity of Instrumental Reason by Christine M. Korsgaard. This is the reason why I posted How Desires Work in my previous posting. Korsgaard argued extensively against the Humean theory of reasons. Hume’s theory is nearly 300 years old and I sought to present a more modern alternative.

Here, I want to address Korsgaard more directly.

Korsgaard answers this question of why one ought to perform the means to an end by saying that, first, one needs to endorse the end as a legitimate end. If the end is merely something desired or, worse, the result of fear or depression, then it does not provide a reason to pursue the means. One must assign an “ought” to the end. Then the “ought” can be transferred to the means to that end.

This goes against the Humean way of understanding things. On the Humean model, desires identify ends. The desire to learn a foreign language itself is an end and is sufficient to provide a reason to enroll in the course. Note that I said, “a reason”. There may also be reasons not to. Money. Time. One may lack the ability and experience only frustration in the attempt. But the desire provides “a reason”.

According to Korsgaard, this move violates Hume’s own “law” that one cannot derive an “ought” from an “is”. One must first “ought” (normalize - convert into a norm or make into a personal law) the end before one can “ought” (normalize) the means to that end. That is where the endorsement of the end comes from.

In response, it seems to me that this significantly over-intellectualizes action. An animal with an aversion to pain has a reason to avoid the fire without having to “normalize” the aversion to pain. It is simply in virtue of having the aversion that one has a reason to avoid pain. It is merely in virtue of being hungry that one has a reason to eat; in wanting sex that one has a reason to seek sex. Again, I said “a reason” - he may also have reasons not to.

Another way in which this over-intellectualizes action can be demonstrated by looking at more common actions. I am thirsty and decide to get a soda from the refrigerator. First, do I need to normalize thirst? Can it not be the case that my reason to go to the refrigerator is quite like the deer’s reason to walk to the stream?

It is also the case that my going to the refrigerator, getting a soda, and drinking it requires a number of actions. I must walk to the refrigerator, putting one foot in front of another, stepping around the backpack I had set on the floor, open the refrigerator door by applying just enough force on the door to open it without flinging it open, direct my hand and fingers to grasp the can, open it, pour its contents into my mouth, and swallow. Each of these actions can be further broken down.

Two things to note about these actions is that they serve ends other than quenching my thirst. There are interests served by not opening the refrigerator door too quickly, going to the refrigerator for a soda rather than the cupboard (where I keep a spare six-pack) or to the convenience store across the street. There are additional ends involved in getting a soda rather than water or orange juice, walking rather than running or bunny-hopping into the kitchen, and the like. I have a great many ends to normalize to get my drink. And I do all of this while my mind is busy outlining the next few paragraphs of this post.

Specifically, I was able to get myself something to drink while I considered trying to make an argument that each of these actions needed to be normalized – and then figuring out that Korsgaard could respond by saying that the normalization of the end of quenching my thirst would carry through to all of the various means, however small. Then, I thought of the issue whereby I serve a great many ends while I get a drink. A desire-based theory can handle these – there are a number of desires guiding my action as I go for a drink. Korsgaard’s thesis of normalization of ends seems to require an additional unnecessary component to each of those desires.

When the deer I mentioned gets to the stream and discovers that the water level is too low for her to drink from it, she paws the ground in order to dig a small hole. Again, this is a means to an end, but the deer does not seem to need to normalize her thirst to perform this action.

Now, what about the fact that we have ends that, we think, do not give us reason to perform an action. For example, a person who is shy and timid wants to avoid conversation, but sees her shyness as something to be overcome. Here, we have to ask, what reason does she have to overcome her shyness? The only viable answer is that she has some other interest which the shyness comes into conflict with. Without the shyness, she could accomplish these other ends. With it, those other ends are frustrated. In order for her to have a reason to preserve her shyness, it must fulfill some other end – an end that then gives her a reason. If it only thwarts other desires and fulfills none, she has reason to get rid of it and none to keep it.

As for the is/ought problem, I have dealt with this on a number of occasions. Most recently, in a response to Sam Harris' most recent attempt to derive "ought" from "is".

If an “ought” is going to cause physical substances to change its properties – to move in different ways – then there must be something of an “is” to each “ought”. I can draw support for this from Donald Davidson's thesis that reasons are causes - and causes must exist in the realm of "is". (See: Davidson, Donald (1963). Actions, reasons, and causes. _Journal of Philosophy_ 60 (23):685-700.)

I argue that “Agent ought to do X” means “There is a reason for Agent to do X”. This latter is an “is” statement and can be derived from other “is” statements. Please note that this is referring to a pro-tanto “ought” not a moral or practical “ought” – though they can be derived from pro-tanto “oughts”. Anyway, “Agent has a reason to do X” means, “Agent has a desire that would be served by Agent doing X.” Here, I made a shift from “There is a reason for Agent to do X” and “Agent has a reason to do X”. These should follow a distinction between “There is a desire that would be served by Agent doing X” and “Agent has a desire that would be served by doing X.” Of course, reasons for A to do X can exist without Agent having a reason, in the same way that there can be red cars even though Agent does not have a red car.

Effectively, the deer's thirst gives it a reason to walk to the stream, dig a hole, and take a drink, just as my thirst gives me a reason to go to the refrigerator and get a soda. Desires themselves produce reasons for action. There are some desires we have reason to get rid of if we can but, until we do, they provide reasons for action. After all, the only reasons to rid of it are its conflict with other desires.

No comments: