The effect of risk on intertemporal choice and preference reversal

Xu et al. (2022) asked adults to pick between two imaginary money amounts. One choice always came with a chance the reward might not show up.

The team added different levels of risk to one option. They also varied how long each reward would be delayed. Everyone made choices on a computer in a quiet lab.

When the risky side might pay nothing, people jumped to the safe side. The switch happened even if the safe reward was smaller or came later.

Even with risk in the mix, the classic reversal still showed up. Folks picked the big late reward at long delays, then flipped to the small early one when both waits were short.

DeHart et al. (2015) changed how delays were worded. Saying "December 15" instead of "in 30 days" cut discounting. Xu adds risk rather than wording, yet both tweaks steer choice away from the impulsive pick.

Cox et al. (2015) saw shorter gaps between trials push rats and pigeons toward impulsive choices. Xu finds a similar push in humans, but the trigger is reward uncertainty, not trial timing.

Lord et al. (1986) showed pigeons work harder when delay is unpredictable. Xu agrees: uncertainty matters, and it can outweigh pure delay.

If you offer a learner two reinforcers and one might not deliver, expect them to flee to the sure thing. You can use this bias by pairing the target skill with the safer reward. When you must use chance-based tokens, keep the delay short or the safe amount large to keep responding strong.