Neuroforecasting reveals generalizable components of choice
Year: 2025
Type: article
Abstract: Accurate forecasts of population-level behavior critically inform institutional choices and public policy. While neuroforecasting research suggests that measurements of group brain activity can improve forecasting accuracy relative to behavior, less is known about how and when brain activity can effectively improve out-of-sample forecasts. We analyzed neural and behavioral data collected in two experiments to forecast choice in more vs. less demographically representative aggregate internet markets in order to test when forecasts based on brain activity generalize better than behavior. In both experiments, while the accuracy of market forecasts based on behavior varied as a function of sample representativeness, market forecasts based on brain activity remained significant regardless of sample representativeness. These findings are consistent with the notion that brain activity associated with early affective responses can generalize across individuals to index aggregate choice more broadly than downstream behavior. Thus, brain activity from limited samples may reveal generalizable components of choice that can improve market forecasts. These findings inform theory regarding which components of individual choice generalize to improve market forecasts and provide insights into mechanisms that underlie the effective application of neuroforecasting.
Source: PNAS Nexus
Institutions: Erasmus University Rotterdam, Stanford University
Language: English
Subfield: General Decision Sciences
Field: Decision Sciences
Domain: Social Sciences
Open Access status: gold
Estimated APC paid: $4,202
OpenAlex ID: W4407913098