I'm a bit like a chameleon. I can change my behaviour to fit in to quite a few places; I am quite social (for a mathematician). While not always a good trait, this actually helps in collaboration. I work well with most people, and this trait has served me very well. I share credit where it is due, taking care to err on the side of giving credit where it maybe isn't (because the other error is much worse), and am aware of the phenomenon of `group discussion': namely that even though you may be the one to articulate an idea, that idea grows through the discussion and all there have taken part in its creation. I'm also aware that in any partnership each partner must be prepared to do 80% of the work (I learned that one from Larry Niven, who was writing about his collaborations with Jerry Pournelle).
Because of this, most people treat me this way in return, and the net result is (usually) smooth collaboration. I have been annoyed at my co-authors, sometimes, but never for very long, and in any case not often. Since my collaborations are smooth, I prefer to do my work this way (though occasionally I do something all on my own, more because I fail to get somebody interested in it than because of selfishness).