The "obvious thing" is an assumption of the person's preferences. It is entirely possible for a person's happiness/utility to depend on happiness/utility of other people, and it seems likely for such dependence to be evolutionary advantageous -- in the most trivial example, it is clearly advantageous to be helpful to own children -- plus, as someone said, people are not stupid.
It is of course also possible to rationally emulate such behavior, as well as emulate emotions, but it may not be easy, and might come at a substantial cognitive cost.
In addition, I would imagine that being able to feel certain emotions yourself must be of great help in predicting behavior of people who do feel those emotions. For example, I am still not sure what can or cannot trigger jealousy, having never felt it myself -- there is a learned list of things that can do it, but by construction it is never complete; it would have been nice to have some sort of an intuitive understanding. And it extends much further, e.g. I've seen studies demonstrating how people with good emotional intelligence tend to be better stock traders.