Sure, in its simplest form, recursion can be function (input) -> side effect (output) -> function input (side effect (output)) -> you're passing side effects into f and computing f with that data until some state is reached (unless you're infinitely computing...). So, in this case, a result is passed to itself (function), not itself, which is the basis of the question (by definition, a side effect of x that is x is x).
A function, f, that returns f would be a sufficient answer to this question. Or would it? More interestingly, as Erik Meijer answered ( he did the homework, too ), energy, e, could be a correct answer. How so?
PS: On an unrelated note (or is it? Is knowledge a side effect of itself?), did you read the history of 0/1 based indexing? What do you think of the author's take?