Abstract
In patch foraging tasks, animals consistently overharvest — they stay in a patch with diminishing rewards longer than is predicted by optimal foraging theory. While overharvesting appears to be ubiquitous, its cause remains poorly understood. To address this question, we characterized the behavior of rats in a series of patch foraging tasks. Most notably, rats left patches earlier if a delay was imposed between the decision to harvest and receiving reward (pre-reward delay), and rats were sensitive to increasing the delay between receiving reward but before the next decision to harvest or leave (post-reward delay). This behavior was best explained by a temporal discounting model that maximizes all future reward. To test the external validity of this hypothesis, rats were further tested in a standard delay discounting task, in which we separately manipulated pre- vs. post-reward delays. As in the foraging paradigm, rats were less likely to select larger rewards if the pre-reward delay was longer and rats were sensitive to post reward delays. Strikingly, parameters obtained from fitting the temporal discounting model to foraging behavior provided an excellent fit to behavior in the delay discounting task. These results suggest that rats employ a common computational approach to both foraging and delay-discounting decisions, and that overharvesting may reflect operation of the same mechanisms that underlie discounting of future reward.
Significance Statement Two common paradigms to study decision making include foraging tasks, in which animals choose to accept an available reward vs. reject it to search for better opportunities, and intertemporal choice tasks, in which animals choose between a smaller reward after a short delay or larger reward after a longer delay. In both tasks, animals exhibit a preference for the more immediate reward. Interestingly, a common model of intertemporal choice behavior — temporal discounting — has failed to explain the similar bias in foraging tasks. Using carefully designed behavioral experiments and quantitative analysis, we show that rats exhibit identical time preferences in the two tasks, and their behavior is well explained by a temporal discounting model that captures important features of both tasks.