... as, according to the paper you linked, do rats. Good point. Can this be explained in similar ways? A pigeon or rat's environment could surely provide uncertainty about whether a future reward will be delivered at all. Time and energy spent keeping track of a future reward might still be a significant cost. Could these factors have selected for neural circuits that just apply a hyperbolic discount function as a default in the absence of specific reason to do otherwise? This does seem to be evidence suggesting the answer might be yes.