In The Possibility of Altruism, Thomas Nagel advocates substituting reasons for the role in human motivation traditionally played by desires.  He sees the need because what people desire is an empirical matter revealed in behavior, and varies too greatly to provide the kind of inescapability he thinks morality requires:  reasons can be "agent-neutral", whereas desires must always be "agent-relative" (to use terms later introduced by Parfit).  But Nagel specifically rejects the possibility that this approach conflates causal explanation with normative justification:  "a close connection between the two is already embodied in the ordinary concept of a reason" (15).  

He proposes to dethrone desires by drawing attention to a particular problem regarding the role of "future desires" in practical reasoning.  For example, suppose I now purchase a bottle of water for quenching the thirst I anticipate I will experience later in my drive home.  How do we explain this purchase?  Nagel claims that the anticipated thirst cannot explain it:  that is a future desire.  But I am not thirsty now, hence that cause does not yet exist.  So if desire is needed to explain my purchase, it must involve some other desire I have now -- e.g. a desire to be prudent with regard to my expected thirst.  Nagel then argues this raises all sorts of further difficulties, involving the relation between current and future desires as well as other problematic aspects of the desire account.

Instead of venturing into those thickets, I wish to show that Nagel's notion of "future desires" (or "future interests") is itself inherently and fatally flawed, and embodies the very conflation he seeks to skirt.

To begin with, thirst (nor hunger, sleepiness, tumescence, etc.) is not a desire.  This is not to quibble over the fact that thirst is something to which one is typically averse; we can allow aversions as "negative desires".  The point rather is that thirst as such is simply a bodily sensation, a dryness in the throat betokening a state of relative dehydration.  Now typically it is something to which one is averse, something one would prefer to avoid or quickly ameliorate, but this need not be, and is not, always the case.  An ascetic, for example, might celebrate the sensation as validation of self-denial, or a stoic might regard it with utter indifference.  In other words, having thirst is one thing, valuing thirst is quite another.  To desire something is inherently to ascribe value to it, positive or negative:  to see it as something one would prefer to be, or not to be, the case.

So if valuing thirst is distinct from having thirst, then presumably the time that the valuing begins need not be the same time that the thirst begins.  It would be extremely odd to suppose that a person, having had previous experiences of thirst, could not e.g. negatively assess it prior to future occurrences.  Indeed, surely this is generally the case.  Desires are not conditions that only come upon us when their objects obtain.  Rather, they are persistent evaluations that represent enduring assessments of ourselves and the world we are in.

Consequently, it is aberrant to see one’s own desires as subject to significant variation, predictable or otherwise.  And one can take such variations into account now only if doing so comports with the desires one has now -- one can hardly transport oneself completely out of one’s own valuation of things to assess the relative importance of one’s valuations at different stages of time.  But without wading more deeply into such waters, a simple point remains.  I purchase the water now because I desire not-to-be-thirsty-later now:  because right now I regard that pending quenching of thirst as a valuable thing.  I don’t need to wait until I actually experience that thirst to suppose I won’t like it.  So in particular I don’t need any other problematic supplementary desire, distinct from my aversion to thirst, to explain my anticipatory purchase of the water.  

Only if you confound thirst with the desire not to be thirsty will you find “future desires” problematic for practical reasoning.  Nagel falls victim after all to "an illegitimate conflation of explanatory and normative inquiries" (15-16).  He confuses persons’ sensory/bodily states, appropriate to causal accounts of behavior, with their evaluations of such states, appropriate to normative and justificatory accounts.  In short, when a desire exists and persists should not be conflated with when the conditions appropriate to the satisfaction of that desire exist.  Hence, Nagel’s notion of "future desires" is hopelessly compromised, and any dismissal of the appropriateness of desire in accounting for human motivation predicated upon it is premature.  Concern about morality in this regard may also be premature.  The rational inescapability of logic and mathematics does not depend on whether people actually reason and behave accordingly; why should morality be thought any different?