Somebody tells you, "You should do X."
You ask, "Why?"
The only sensible answer to this question is that the speaker must give you a "reason for action" for doing X. If he says anything that does not count as a reason for action, then you have grounds to challenge him with the question, "If you are not giving me a reason for action, then why are you saying that your claim is relevant to what I should do?"
Furthermore, that sensible answer must refer to something real. It must exist. If he gives you a reason for action that does not exist, then . . . well, then his claim that he has a reason that is relevant to a real-world action is false. Imaginary reasons are only suitable for imaginary actions.
So, when you ask, "Why should I do X?" the appropriate answer would be one or more reasons for action that exist.
One can change the meaning of "should" if she wants. She can change the meaning so that it has nothing to do with reasons for action that exist. However, if she does this, then she changes the meaning away from anything that is at all relevant to prescribing a given action. Recommendations for courses of action require reasons for action that exist. Either the speaker's statement refers to reasons for action that exist, or it has nothing to do with recommending an action.
Desires are the only reasons for action that exist.
However, desires ARE reasons for action and they DO exist.
Desires are also real, just as protons, black holes, and subroutines in computer programs. As with these other entities, they yield statements that are true or false in the same sense that any claim in science can be true or false.
Desires can be expressed in the form of propositional attitudes. To say that, "Agent desires that P" is to say that Agent is motivated to act so as to realize states of affairs in which "P" is true. The "desire that P" selects the goal or the end of his action. He then goes about selecting the means that will realize that end.
The selection of means depends on beliefs, but the selection of ends does not. Desires alone select the ends. Desires and facts about the world determine what is the best or most efficient means. However, desires and beliefs will determine what the agent will do.
A person may be thirsty. She sees a glass that she believes contains fresh water on the table. Her thirst (desire-as-end) and her belief that the glass contains good thirst-quenching liquid, drives her to drink the contents of the glass.
However, we are assuming that the contents of the glass had been poisoned.
Our agent's desires select the end (a state in which her thirst is quenched without killing her). Our agent's desires and her beliefs (e.g., that the glass contains thirst-quenching liquid) determines her actual actions.
However, our agent's desires plus facts about the world (including the fact that she has desires that are more likely to be fulfilled if she remains alive, and the fact that glass contains a poison that will kill her) determines what she should do - which may be different from what she will do
She should not drink from the glass.
Meaning, there are reasons for action that exist that recommend against drinking from the glass.
"Why shouldn't I drink from the glass?"
"Because you have desires that can best be fulfilled if you continue to live and those desires will be thwarted if you drink from the glass."
Where this statement is true, it is objectively true. Where this statement is false, it is objectively false.
If somebody wants to claim that desire utilitarianism does not account for an account of what a person should do, ask them this. "What would you use in answering this 'why' question? Other than desires, your only option is to answer with a statement that does not refer to reasons for action at all, or to reasons for action that do not exist. Neither option is relevant to answer the question, 'Why should I do X?'"
This, however, still leaves us with an interesting question. What do we do with desires that are not our own? Other people's desires are 'reasons for action that exist'. However, they are reasons for their action, not for mine. When it comes to my action, other people's desires seem to have no relevance. Others may be harmed by my action. However, how does their aversion to states that would result from my action count as a reason for me not perform that action?