I'm going to answer Dr K as an answer because I require extra space and don't want to create an annoying string of comments.
I think there is a levels-of-analysis issue here that reflects your background as an MD and the type of research that drives MD practice and training. Levels-of-analysis is also a pretty standard topic in the philosophy of science, so we're dealing with a classic problem (so humanities types might even find this interesting, not just science geeks).
The food reward concept focuses to some extent on what we might call the behavioral level, which is the level of analysis on which social scientists - including cognitive psychologists like Roberts - typically work. Neurons and other components of biochemical mechanisms and processes are on a much smaller level of analysis, which is the general focus in med school, and of the hard sciences that med students are tested on by the MCAT.
Obviously, the two levels are always present and true: all behaviors co-occur with brain or neural activity. Thus, when we develop and discuss a concept like food reward we can do so on both - and presumably many other - relevant levels of analysis. If people are inclined to respond to foods as rewarding, then this will be evident on the neural level - and indeed obese people show different brain activity patterns than non-obese in fMRI studies - and on the subjective phenomenal levels - i.e., they will experience and describe their experience of food differently than non-obese people.
It is thus odd to begin your response with:
Food reward is just the outflow tract of this system...[of] neurons that drive the dopamine receptors.. (emphasis mine)
I don't think we want to say - or imply - that food reward is just or reduces to the system that you discuss. You may be correct (and I think that you are) that those neurons are important, and that leptins (and probably lots of other stuff) affect them. It is less clear, however, that this serves as evidence that the best treatment of obesity will be one that directly affects this system and the relevant components of the processes you mentioned. That may lead to one set of dietary or medicinal interventions.
However, since the behavioral or phenomenal level or manifestation of the processes you are talking about also co-occur, then focusing on them, at their level of analysis, might lead to another set of interventions that largely overlaps with the former set (especially the dietary ones) but not completely. So, for example, the research you are discussing is probably not likely to result in, say, a new type of behavioral therapy, at least not in a direct manner. Also, on the phenomenal level, when we Paleohackers read Guyenet's series, it gives us some simple conceptual tools that help us make sense of our experience when we make food-related evaluations and decisions. A bio-chemical description can provide this too, but it would require more mental energy and theoretical extrapolation.
One big concern with reducing higher levels of analysis to lower levels is that significant and successful treatments can be discovered and studied on the behavioral level absent any knowledge of the underlying biochemical processes. The best example of this is just about all of Seth Roberts' experimental findings. He has ideas about what is going on beneath the hood, but he doesn't really know, & in some sense he does't care, either. And there's nothing wrong with that.
That said, I find your comment characteristically interesting, helpful, and informative. I just think it commits a theoretical error that could give some readers the wrong idea.