A computational exploration of choice variability in interpersonal decision-making

When expressing their preferences, people oftentimes make choices that seem to be inconsistent with one another. To explain this variability two competing cognitive models have been proposed (Moutoussis et al., 2016, PLOS Comp. Biol.): the Preference Temperature model assumes that variability comes from the noisiness of preference implementation, whereas the Preference Uncertainty model assumes that preferences change continuously, gravitating towards a particular state. We test these models in an interpersonal decision-making task, in which participants can decrease their own payoffs to increase (‘share’) or reduce (‘compete’) someone else’s. In accordance with previous findings in intertemporal decisions, the Preference Uncertainty model outperforms the Preference Temperature model for other-regarding choices. We additionally find that participants with a sharing attitude have less choice variability than participants with a competitive attitude, and that the more extreme the preferences are (either sharing or competitive), the greater the choice variability is.