It will be worth it, in the end: a model of naturalistic intertemporal choice

Intertemporal choices, which involve consequences spread over time, often display a bias toward short-term rewards, which was recognized in ancient Greek philosophy as "akrasia." This bias is nowadays often modelled via hyperbolic discounting models, where intertemporal choice failings are due to an inherently irrational way of trading-off present and future rewards. Hyperbolic discounting models lead to precommitment as a suggested strategy to avoid short-term rewards, as in the Greek story of Odysseus and the sirens. We contrastingly propose a boundedly rational model of naturalistic intertemporal choice, where the complexity of the choice sets lead to people discounting future rewards rationally, but only up to a finite, mentally simulated horizon. This is consistent with only three unique types of error which we link to the established naturalistic biases of addiction, procrastination, and choosing smaller-sooner over larger-later rewards. A second assumption regarding the attentional resources required to simulate outcomes at varying delays then leads to the well-known phenomenon of "preference reversals" - where these three errors tend to recur in the present despite a stated preference to do better in the future. Importantly, extending mental simulation further into the future reduces all these errors, and allows people to act more in accordance with their true preferences. This prediction is consistent with a nascent empirical literature on successful intertemporal choice interventions, which all appear to succeed by helping people to think about the future. This model's psychological mechanism can help inform the design of future intertemporal choice interventions.