Let NORID be the thesis that there is no instrumental desire (NID) or instrumental desire reduces (RID) to non-instrumental desire plus belief about what is a means to what (e.g., an instrumental desire for jogging reduces to a non-instrumental belief that jogging promotes health and a non-instrumental desire for health).
One argument for NORID is something I got from Mark Murphy. Desires explain actions. But if M is a means to an end E, we do not need to posit any desire for M to explain the pursuit of M—all we need to posit is a belief that M promotes E and a desire for E.
Here are two more arguments, perhaps unoriginal.
Argument 1: Inertia. Start with this premise:
- If x has an instrumental desire for M, then x believes that M is a means to something that x desires.
Argument 2: Rationality. Suppose I am perfectly rational and I non-instrumentally desire E with strength dE. Suppose I know with certainty that M is a necessary and sufficient means to E. Then, rational as I am, I will be motivated to pursue M at least with strength dE. Now suppose that I instrumentally desire M with strength dM. Surely a desire that does not reduce to non-instrumental desire and belief adds to motivation. So, now, I will be motivated to pursue M with strength greater than dE (maybe dE+dM?). But it is irrational to be motivated to pursue a means with a strength beyond one's perfectly rational desire for the end, and that was dE. So, NORID is true of a perfectly rational being. But it is plausible that NORID is also true of imperfectly rational beings. First, it would be odd if a basic kind of desire—instrumental desire—were never rational. Second, we can say the following. The above argument shows that if I have an irreducible instrumental desire for M, that desire adds to the motivational force of my desire for E combined with my certainty of the necessity and sufficiency of M. But whatever it adds to that motivational force goes over and beyond the merely instrumental motivation for M, since the merely instrumental motivation for M is fully accounted for by the motivation in favor of E combined with my certainty of the necessity and sufficiency of M.
I am grateful to Dan Johnson for showing me that my arguments do not show that there are no non-instrumental desires, but only that either there are none, or they are reducible.