We identify the logic behind the recent theory of coherent sets of desirable (sets of) things, which generalise desirable (sets of) gambles and coherent choice functions, and show that this identification allows us to establish various representation results for such coherent models in terms of simpler ones.