Recursive Self-prediction in Self-control and Its Failure

Abstract

The combination of human foresight and the discounting of delayed events in a hyperbolic curve is all that is needed to explain the learning of higher mental processes from the bottom up. These processes are selected by delayed re- wards insofar as they counteract the over-valuation of imminent rewards that is also predicted by hyperbolic discounting. For instance, these processes come to interpret repeated, similar choices as moves in an intertemporal bargaining game resembling an iterated prisoner’s dilemma. Perception of current choices as test cases for coop- eration in such a game recruits the extramotivation experienced as willpower. Lines seen as criteria for such tests may be experienced as beliefs rather than resolutions. The chance that shifts of self-prediction may cause radical swings of motivation makes choice unpredictable from just knowing the person’s prior incentives, even by the person herself; the resulting introspective uncertainty is arguably the subjec- tive basis of freedom of will. A similar kind of recursive self-prediction explains how surges of emotion or appetite can be occasioned by symbols that convey no information about the availability of external rewards.