What’s interesting is that this sort of division of control is found all over the place, not just for the composition of objects. For example, ‘map’ in functional programming languages takes a function and a list, and returns a list of the function applied to each of the elements of the old list.

map(add1, list(1, 2, 3)) == list(2, 3, 4), where add1 is a function which returns its argument plus 1

In this case the caller handles what behaviour is used inside of map, while the map function itself handles looping and constructing the new list.

It seems like a different concept, but it’s really the same idea. What’s powerful is the fact that behaviour itself, though it may be represented as a function or as an object, can be held in a variable and passed around your program.