Is hyperbolic discounting rational?

In dynamic economic models, future periods are typically discounted geometrically, i.e., at a constant rate. There is, however, mounted evidence from experiments that people discount hyperbolically: there is much discounting the first period, and then it is geometric again. This evidence has been put forward as an argument against the rationality of individuals.

Doyne Farmer and John Geanakoplos argue that hyperbolic discounting is observationally equivalent to geometric discounting with uncertainty about future discount rates. While this may rescue rationality, one has to ask oneself whether this makes any empirical sense. And for that there is little support at the moment.