Modeling choices in delay discounting
Abstract
In a recent article, Ericsson and colleagues (2015) compared traditional utility-discounting models with a set of heuristic models of intertemporal choice using a cross-validation approach. Consistent with earlier reports, Ericsson and colleagues concluded that heuristic models (specifically their novel intertemporal choice heuristic or ITCH model) explain intertemporal choices better than discounting models do. More surprisingly, their results showed that all discounting models performed nearly at chance level, and did not outperform even the baseline model. In this article, however, we demonstrate that these conclusions are premature. Specifically, we reanalyzed the Ericson et al. (2015) data under different, arguably more appropriate, conditions and found that models of both classes are rather good at predicting choice. We conclude that the jury is still out on which (type of) model is best and end by making suggestions towards conducting more informative model comparisons.
