Neuroforecasting reveals generalizable components of choice

Area of Science:

• Neuroscience
• Behavioral Economics
• Computational Social Science

Background:

• Accurate population-level behavior forecasts are crucial for policy and institutional decisions.
• Neuroforecasting suggests group brain activity can enhance prediction accuracy over behavior alone.
• Understanding when and how brain activity improves out-of-sample forecasts remains an open question.

Purpose of the Study:

• To investigate when and under what conditions brain activity forecasts generalize better than behavioral forecasts.
• To test the generalizability of neuroforecasting across varying levels of sample representativeness in aggregate markets.
• To explore the role of early affective responses in brain activity for predicting aggregate choice.

Main Methods:

• Analysis of neural and behavioral data from two distinct experiments.
• Forecasting aggregate choice in internet markets with differing demographic representativeness.
• Comparison of forecast accuracy derived from brain activity versus behavioral data.

Main Results:

• Behavioral forecasts varied in accuracy depending on market sample representativeness.
• Market forecasts derived from brain activity remained significant irrespective of sample representativeness.
• Brain activity forecasts generalized better than behavioral forecasts in less representative samples.

Conclusions:

• Brain activity, particularly early affective responses, can serve as a generalizable index of aggregate choice.
• Neuroforecasting from limited samples may capture generalizable choice components, improving market predictions.
• Findings inform theories on choice generalization and the mechanisms of effective neuroforecasting.