In “Two Distinctions in Goodness” (Philosophical Review 1983), Christine Korsgaard argued that the distinction between intrinsic and extrinsic goodness should not be conflated with the distinction between final and instrumental goods. As it happens, I believe that she is entirely right that these two distinctions are distinct from each other. Still, her account of the distinction between final and instrumental goods does not seem quite correct to me.

As she puts it, a final good is something that “is valued as an end or for its own sake”, while an instrumental good “is valued as a means or for the sake of something else” (p. 170). There are at least three problems with this account.

1. The fact that something “is valued” is surely not enough to show that it is a “good” of any sort. To borrow an example from David Lewis (“Dispositional Theories of Value”), a sick or crazy person might value seasickness; but it surely does not follow that seasickness is a good!

This defect in Korsgaard’s formulation is easily rectified. Instead of saying that for x to be good is for it actually to be valued, we should say that for x to be good is for it to be fitting or appropriate for the relevant agent to value x. But there are still two other problems with this account.

2. Even if xy, is valuing x “for the sake of y” really the same as valuing x “as a means to y”?

To do something “for the sake of y” is to do it because of one’s interest in or concern or regard for y (see the Oxford English Dictionary on ‘sake’, 7a). That is, if you φ for the sake of y, then in some way, you φ because you have this concern or regard for y; your φ-ing is explained or motivated by your concern or regard for y.

However, it is not obvious that this explanatory or motivational connection must always involve a means-end relation. Couldn’t it sometimes be true that your φ-ing is at least explained or motivated by your concern or regard for y, even if your φ-ing is not one of the means that you use in order to achieve y? (Perhaps your concern or regard for y does not take of the form of an intentional attempt to “achieve” y at all?)

In short, the distinction between your valuing x “for the sake of y” (where xy) and your valuing x “for its own sake” concerns the explanation or motivation for your valuing x. It is not at all clear that this is what lies behind the means/ ends distinction. Strictly, the distinction between “instrumental” and “final” goods must have more to do with the distinction between means and ends than with the distinction between “for its own sake” and “for the sake of something else”.

3. What is it to “value” something “as a means”? The term ‘valuing’ covers a wide range of different kinds of attitudes and responses. But it does not obviously make any sense to take many of these attitudes towards something merely “as a means”. Is it really possible for me to admire your achievements merely as a means, or to be proud of my own achievements merely as a means? How could I revere the memory of my late father merely as a means? Could anyone cherish or adore something merely as a means? Is it even possible, really, to rejoice in something merely as a means? (You can rejoice in something because it seems auspicious, or to herald further good things to come; but this is not obviously the same as rejoicing in it merely as a means.)

If means are what we intentionally use in order to achieve an end, then it seems that the attitude that it is most clearly possible to take towards something merely “as a means” is intention: that is, we can certainly intend a course of action merely as a means. Since a choice is just the formation of an intention, one can also choose a course of action merely as a means.

When you intend a course of action merely as a means, that course of action is a subordinate part of a larger plan: your execution of this subordinate part of your plan will be guided or regulated by your monitoring of how well your behaviour is achieving the end or goal of the plan.

If this is right, the distinction between ends and means is a distinction between the different roles that intended events or courses of action can play within a larger plan. The distinction has to do with the structure of plans (i.e. with the way in which different parts of these plans guide or regulate behaviour), not with the explanation or motivation of attitudes.

It seems to me that we cannot in any natural sense “value” a thing “as an end” or “as a means”. We can only choose or intend something as an end or as a means. So we should amend Korsgaard’s account, in the following three ways.

  1. We should replace the idea of what “is valued” with the idea of what it is “fitting or appropriate (for the relevant agent) to value”.
  2. We should drop the distinction between “for their own sake” and “for the sake of something else” and focus exclusively on the distinction between “ends” and “means”.
  3. We should replace the highly generic idea of “valuing” with much the more specific idea of “choosing”.

So, according to this amended account, for x to be “instrumentally good” is for it to be fitting or appropriate for the relevant agent to choose x, not as an end, but only as a means to some further end y (where xy).

A more important point that emerges from this is that it would be a radical mistake to think that all value is either final value or instrumental value. Both final and instrumental value are instances of the rather special kind of value that is essentially connected to choice: there are also many other kinds of value, which are essentially connected to other kinds of “valuing” attitudes.

10 Replies to “What is instrumental goodness?

  1. Interesting stuff. I just had two quick points.
    First point; this might be explained by her Kantian notion of valuing, which really should be rational valuing. It might be that she thinks that when the CI is not satisfied, then you have only inclinations and desires which are not sufficient for valuing. Maybe this can deal with the deviant cases.
    In the second point, to be charitable to Korsgaard, it seems like she is using a disjunction ‘as a means *or* for the sake of something else’. This could mean that she doesn’t think of these as the same thing.
    I don’t know what her reasons for adding the second disjunct are. Here’s one thought though. It could me that the ‘as a means to’ implies some sort of causal relation. However, maybe there are non-causally connected values that are not final, and she thus wants to put these under instrumental category. So, I have an interest in having friends and this interest motivates me to do many actions within my friendships. These actions probably are not means in the causal sense to friendships but in some other sense constitutive of it.
    Finally, I’m not certain about this, but it seems to me that your definition given the word ‘only’ in it does not allow things to have both instrumental and final value. Maybe this is not intended or even implied by the definition, but surely you would want to leave room for that kind of goods too.

  2. Thanks Jussi — By my count, you have three (excellent) points, not two!

    1. You are probably right that by speaking about “valuing”, Korsgaard was really thinking about “rational valuing”. Back in 1983, however, a reader would have had to have amazing powers of mind-reading to figure that out just from reading her paper!
    2. You could well be right that the disjunction is not intended to be “epexegetic” (i.e. the kind of disjunction where the two disjuncts are equivalent in the context, and serve to clarify the content of what is being said). But then it would turn out that there would be two very different ways of being instrumentally good, and the notion of “instrumental goodness” would fail to cut the evaluative realm “at its joints”.
    3. You’re right: I shouldn’t have included “only” in my definition of instrumental goodness, since we certainly do want to allow that some things have both instrumental and final goodness. It will probably be all right simply to omit “only” here, but I’ll try to think about this a bit more.
  3. Doesn’t your third idea – replacing valuing with choosing – wrongly tie instrumental goodness to human action? Suppose that x causes y and y is a final good (say a pleasurable sensation). The goodness of x doesn’t depend on x being chosen. It is instrumentally good simply if it occurs.
    The fact that x is instrumentally good, of course, may typically make it choice-worthy. But the fact that it is choice-worthy doesn’t seem essential to its instrumental goodness. On the contrary, its instrumental goodness seems typically to render x choice-worthy. And that suggests that the choice-worthiness of x depends on x being valuable – and hence being capable of being valued – in some way that is independent of its being chosen or choice-worthy.

  4. Interesting questions.
    First, I bet that Korsgaard’s use of “for the sake of” is drawn from Aristotle and thus I am not sure the OED is the best source for interpreting it. Just a worry.
    Second, I am having trouble sharing your worry about attitudes other than intention. Imagine John is tempted to mistreat Jane but willfully resists the temptation and treats her well. His strength of will is admirable (or worthy of approval) in part because of the end it furthers; I admire it, at least in part, because it enables him to treat her well. By extension, one could say I admire it because it is a means to treating her well or as a means to treating her well.
    None of this precludes, of course, also admiring his strength of will for its own sake (or not) – something you seem to assume in some of the questions you ask (e.g. “How could I revere the memory of my late father *merely* as a means?”).
    It seems there are similar cases for pride and some of the other attitudes you mention.

  5. Just re-read the paper. Very impressive and rich with ideas.
    Any way, you are right to equate instrumental with means-end sorts of cases. My first comment above should be ignored. Interestingly, the stuff about rational willing and the Kantian account of value is actually already in the paper.

  6. Victor and Brad — You both seem to be conflating two separate ideas: (a) the idea of something being good as a means, and (b) the idea of something’s being good because of its causal effects.
    Clearly, the second idea has no necessary connection with choice or intention; but the first idea does seem to, at least if the term ‘means’ is being used in its strictest and most precise sense.
    I might very well admire someone’s achievements at least partly because of their causal effects. (Indeed, this is not just possible but common!) That is, I can certainly take the view that these achievements are admirable at least partly because of the value of these effects. (It can also happen that my admiration for these achievements is explained or motivated by my valuing of these causal effects — in which case, it would be true to say that I admire these achievements for the sake of these effects.)
    But I still don’t think that this would be a case of my admiring someone’s achievements “as a means”!

  7. Thanks for the response, Ralph.
    Korsgaard suggests this test for valuing something merely as a means: you value it, but would be indifferent about swapping it for some equally effective means. We can apply that to the case I mentioned:
    John is tempted to mistreat Jane but willfully resists the temptation and treats her well. Now imagine that this were possible: instead, John could take a “silencing pill” that causes the temptation to disappear and enables him to treat Jane well.
    Korsgaard’s test then supports this conclusion: I admire John’s actual strength of will because it enables him to treat her well, and I count as admiring it merely as a means, because I would be indifferent between his acting as he does and his taking the pill instead.
    What do you think?

  8. Interesting post. A few comments.
    I think Jussi is on the right track here with respect to the first point. It’s kind of quick, Ralph, but right at the beginning she does frame her discussion in terms of judgments of value, and making distinctions in kinds of judgments. This is surely an important project for a Kantian, for whom such attitudes play a central role in determining the goodness of other objects. Even using the language of means and ends, I think, is already to bring us withing the realm of rational valuing, because actions (or objects, or states of affairs) are only means and ends in our judgments. X can be a cause of Y regardless of my purposes, but x is only a means to y given my end of y.
    With respect to the second point, my guess is that the “or” is a re-description. Korsgaard glosses what it is to value something as an end as “to value it for its own sake,” so valuing something as a means is then to value it for the sake of something else (that is, for some other end). So, I suspect it’s just to emphasize the contrast using that particular, not uncommon way of speaking. You may well be right that the “for the sake of” language is not precise enough to fully track the actual means/end distinction.
    Finally, I do think it can make sense to value something as a means. Let me try an example: Joe thinks that going to the store to buy milk is worth doing. Joe thinks this is worth doing because he needs the milk to bake a cake, which he really wants. Joe then values the state of affairs in which he goes to the store and buys milk, but he values it as a means to his ultimate goal of baking the cake. Does that work as a case of “valuing as a means”?

  9. Ralph
    thanks for the post, and for the response. I’m not sure now. Here are some further doubts about your focus on choice.
    1) Here’s a simple story
    x will cause y and y is valuable for its own sake. I value y for its own sake. I value x in virtue of the fact that it causes y. As a result I form an intention to bring x about.
    Isn’t it right to say that I form the intention to bring x about in virtue of the instrumental value that I know that it has – namely its value in causing y?
    2) You focus on the idea that x being an means or instrument to bring y about depends on agency. Perhaps so, but it doesn’t follow that we value the thing in virtue of its connection with agency. Why not say: x is valuable in virtue of the fact that it will cause y. It is instrumentally valuable only on condition that an agent could use it as an instrument to cause y.
    3) We might extend Brad’s comment that we value x as a means if we are indifferent about swapping x for another means as follows. Suppose that it might be the case that I have x or w. If I have x I can use it to bring y about. If I have w it will cause y about without my using it. If x has only instrumental value I ought to be indifferent between these two states of affairs. Don’t I value x and w identically? On the strict reading, only x has instrumental value. I cannot use w as an instrument. But doesn’t this suggest that valuing something instrumentally is identical to valuing it for is causal properties?
    4) The strict idea of using is independent of the idea of choice, and instrumentality is connected to the first rather than the second. Beavers use sticks to make dams. Sticks have instrumental value for them. But they don’t choose sticks.

  10. Ralph, not sure what I think on these issues, but I heartily recommend Julie Tannenbaum’s paper “Categorizing Goods,” Oxford Studies in Metaethics, v. 5 (2008).

Comments are closed.