BehaviorBench: Benchmarking Foundation Models for Behavioral Science Tasks

2026-06-23

AI Summary

Existing foundation models lack systematic evaluation across contexts and populations in behavioral science tasks, particularly with respect to a comprehensive assessment of consistency between individual and group-level behaviors. To address this gap, this work introduces BehaviorBench, a multitask evaluation protocol encompassing behavioral prediction, strategic decision-making, trait inference, and knowledge application. It further proposes a novel two-tier evaluation framework that jointly considers individual-level accuracy and alignment with population-level behavioral distributions. Leveraging large-scale behavioral data for fine-tuning, we develop Be.FM-1.5, a behaviorally aligned foundation model. Experiments demonstrate that Be.FM-1.5 significantly outperforms general-purpose foundation models on population distribution alignment metrics while maintaining competitive performance on individual-level tasks, thereby validating the efficacy of domain-specific adaptation in behavioral modeling.

Abstract

Foundation models have been increasingly applied to behavioral science domains such as psychology, sociology, and economics. While these models show promise in individual tasks such as survey response prediction and human-subject experiment simulation, there remains no systematic understanding of how well they perform across diverse behavioral science tasks, contexts, and populations. We introduce BehaviorBench, a comprehensive benchmark that evaluates foundation models along four core capabilities: (1) behavior prediction and simulation, (2) strategic decision-making, (3) subject-trait inference, and (4) behavioral knowledge application. Crucially, BehaviorBench evaluates model outputs at both the individual and distributional levels, capturing not only per-subject accuracy but also population-level alignment, an essential requirement for behavioral validity. Leveraging the tasks in BehaviorBench, we further develop Be.FM-1.5, extending the Be.FM family of behavioral foundation models fine-tuned on behavioral data. Our results reveal a considerable gap: proprietary general-purpose models excel at individual-level prediction and knowledge-intensive tasks, whereas behavioral foundation models, fine-tuned on behavioral data, achieve substantially stronger distributional alignment. Notably, Be.FM-1.5 leads on distributional metrics and remains competitive on individual-level metrics, suggesting that proper behavioral adaptation can close the gap. Our results highlight the importance of distributional evaluation, establish BehaviorBench as a foundation for developing and assessing behaviorally aligned AI systems, and demonstrate Be.FM-1.5's potential for a broad range of behavioral science studies. Our BehaviorBench and Be.FM-1.5 models can be accessed via https://umich-foreseer.github.io/behaviorbench/.

Problem

• foundation models
• behavioral science
• benchmarking
• distributional alignment
• behavior prediction

Innovation

• BehaviorBench
• behavioral foundation models
• distributional alignment
• individual-level prediction
• behavioral science benchmarking