The Paradox of Motivation isn't at all a paradox under meta-ethical subjective relativism. Under MESR, all of our discussion about ethics is a discussion about our motivations; there is no need to supply motivations between the evaluation of good and the performance of actions.
Under MESR, the Paradox of Motivation translates to, "I don't want to do X, but I will do X," which is still puzzling.
I'm always adding a proviso to specific statements about ethics; with the proviso, the problem with the PoM becomes a little more clear: "Some things considered, I don't want to do X, but all things considered, I will." With the proviso, it's clear that the statement is not at all puzzling.
It is a trope in the scientific study of psychology and cognition that the human mind is not a singular, perfectly synchronized entity; the human mind is made up of parts, parts that often compete. You have Freud's id, ego and superego; Berne's child, adult and parent; Wilson and Leary's circuits; Minsky's agents; and the layered evolutionary nature of the brain.
Ethical Philosophy tends to focus on — and thus privilege — only one of the parties to the conversation: that part of our minds concerned with logical, abstract thought. But such arbitrary privilege is prima facie unjustified; worse yet, logical abstract thought requires facts as a foundation (see The Scientific Method).
Furthermore, many of our ethical beliefs are not a result of logical, abstract thought, but are rather socially and culturally constructed: i.e. the product of some sort of social evolution (which, although employing very different specific mechanisms from biological evolution, still retains the fundamental mechanism of natural selection). It is at least a partial justification for some ethical beliefs that, regardless of their origins, we alter them at our peril: they have worked (in some sense) for scores or hundreds of years, and many of our other beliefs depend on those "traditionally justified" beliefs.
It seems obvious that our feelings of empathy and sympathy, of identifying the feelings of others with our own feelings, depends critically on our faculties of logical, abstract reasoning: to identify the "other" with the "self" requires a sense of self, which is a considerably abstract notion. It's likewise seemingly obvious that our concerns about our future self require abstract reasoning; the future is as much an abstract concept as the self.
But the entailment is only one way: a sense of self does not entail any sort of empathy; that we human beings are in truth empathic is a result of our specific evolutionary history, of specifically how our species itself evolved conscious minds. Abstract thought is necessary to conceive of the well-being of others and one's future self, but to desire the well being of others additionally requires the desire itself.
Under MESR, the dilemma of the Happiness rapist is obvious: He has a conflict between two desires: the desire not to hurt other people, and the desire to have sex with young boys. The plural-mind theory renders this simply a conflict rather than a paradox, and MESR obviates the need to discuss anything other than his desires.
There is an irritating tendency in ethical philosophy to label our desires, especially our "primitive" desires (those with the longest evolutionary history) as irrational by definition, because they are not strongly related (as is empathy) to our abstract reasoning. Desires, from our most primitive to our most abstract, are facts, and the facts are as intrinsically a part of rationality as is logical, abstract thought.