A multiattribute attentional drift diffusion model

Although individuals frequently face decisions over multiattribute outcomes, it is unclear how attentional patterns to the attributes and options of a choice set impact decisions. For example, are unfixated choice options and attributes discounted compared to the currently fixated feature? This paper proposes a model that describes how fluctuations in attention to choice set features impacts decision-making. We test and find evidence for the model's predictions across two laboratory studies where participants made incentivized choices between consuming multiattribute bundles as their eye movements were recorded. The first study finds that an attentional drift diffusion model accurately describes choices, response times, and how these variables are correlated with visual attention to attributes and options. On average, only 80% of an unattended attribute's value and 60% of an unattended option's value were integrated in the evidence accumulation process. The second study exogenously manipulated attention to features of the choice set and found this altered choices, as the model predicts. The attentional bias identified here causally affects decisions and has implications for understanding multiattribute choice.

Automated summary

This paper proposes a model that describes how fluctuations in attention to choice set features impacts decision-making. Two laboratory studies found that only 80% of an unattended attribute's value and 60% of an unattended option's value were integrated in the evidence accumulation process. Additionally, attentional bias identified here causally affects decisions and has implications for understanding multiattribute choice.