Decoding the Human Factor: High Fidelity Behavioral Prediction for Strategic Foresight

Ben Yellin; Ehud Ezra; Mark Foreman; Shula Grinapol

cs.AIFeb 19, 2026

Decoding the Human Factor: High Fidelity Behavioral Prediction for Strategic Foresight

Authors: Ben Yellin, Ehud Ezra, Mark Foreman, Shula Grinapol

Organizations: OMGene AI Lab

Abstract

Predicting human decision-making in high-stakes environments remains a central challenge for artificial intelligence. While large language models (LLMs) demonstrate strong general reasoning, they often struggle to generate consistent, individual-specific behavior, particularly when accurate prediction depends on complex interactions between psychological traits and situational constraints. Prompting-based approaches can be brittle in this setting, exhibiting identity drift and limited ability to leverage increasingly detailed persona descriptions. To address these limitations, we introduce the Large Behavioral Model (LBM), a behavioral foundation model fine-tuned to predict individual strategic choices with high fidelity. LBM shifts from transient persona prompting to behavioral embedding by conditioning on a structured, high-dimensional trait profile derived from a comprehensive psychometric battery. Trained on a proprietary dataset linking stable dispositions, motivational states, and situational constraints to observed choices, LBM learns to map rich psychological profiles to discrete actions across diverse strategic dilemmas. In a held-out scenario evaluation, LBM fine-tuning improves behavioral prediction relative to the unadapted Llama-3.1-8B-Instruct backbone and performs comparably to frontier baselines when conditioned on Big Five traits. Moreover, we find that while prompting-based baselines exhibit a complexity ceiling, LBM continues to benefit from increasingly dense trait profiles, with performance improving as additional trait dimensions are provided. Together, these results establish LBM as a scalable approach for high-fidelity behavioral simulation, enabling applications in strategic foresight, negotiation analysis, cognitive security, and decision support.

Figures & tables

Component	Metric / Scope	Notes
Sample Size	2,500 participants	Enrollment (Jan 2024 – Dec 2025)
Recruitment Method	Volunteer / Snowball	Self-registration via OMGene App
Psychometric Data	Comprehensive Battery	Administered to all valid enrollees
Behavioral Data	Scenario-Behavior Dataset	Sourced from a bank of 55 unique scenarios
Data Coverage Strategy	Multi-Scenario Response	Maximizes observed vs. possible interactions
Population A	English Speakers (USA)	Primary cohort

Table 1: Dataset Specifications and Study Design

Figure 1: Comparison of Behavioral Alignment across LLMs. Bars report accuracy, balanced accuracy, and macro-F1 for the base backbone model, the SFT-tuned LBM, and frontier LLM baselines. Error bars indicate 95% confidence intervals computed using participant-level bootstrap resampling. (Rendered from PDF page 11.)

Figure 2: Effect of trait dimensionality on behavioral prediction. Lines show model performance when conditioned on 5, 10, 20, 40 and 74 traits. (a) Accuracy. (b) Macro-F1. LBM improves with additional traits, while the base model and Claude remain approximately constant; Grok shows a modest upward trend but remains below LBM.

Appendix figures & tables3 assets

Supplementary material from the paper’s appendix.

Appendix

Trait	Short description	Instrument	Items	Standardization and binning
Openness	Curiosity, imagination, aesthetic and intellectual openness	NEO-FFI	12	S: Gen pop (Schmitt et al., 2007). B: Very low ( $z<-2\sigma$ ), Low ( $-2\sigma<z<-\sigma$ ), Normal ( $-\sigma<z<\sigma$ ), High ( $\sigma<z<2\sigma$ ), Very high ( $z>2\sigma$ ).
Conscientiousness	Organization, dutifulness, persistence, and self-discipline	NEO-FFI	12	S: Gen pop (Schmitt et al., 2007). B: Same as Openness.
Extraversion	Sociability, assertiveness, energy, and positive emotionality	NEO-FFI	12	S: Gen pop (Schmitt et al., 2007). B: Same as Openness.
Agreeableness	Trust, altruism, compliance, and modesty	NEO-FFI	12	S: Gen pop (Schmitt et al., 2007). B: Same as Openness.
Neuroticism	Tendency to experience negative emotions like anxiety and anger	NEO-FFI	12	S: Gen pop (Schmitt et al., 2007). B: Same as Openness.
Lie / Social Desirability	Social desirability, overly virtuous self-presentation, potential response distortion	EPQ-RS	12	S: None. B: (0–4) Low/realistic; (5–8) Normal; (9–12) High socially desirable reporting.

Appendix

Table 2: Trait table (first 15 traits shown in PDF)

Model	Metric	Mean	Std	2.5%	25%	50%	75%	97.5%
LBM-2 (big5)	Accuracy	0.480	0.018	0.443	0.467	0.479	0.492	0.515
LBM-2 (big5)	Balanced Accuracy	0.310	0.020	0.275	0.296	0.309	0.323	0.353
LBM-2 (big5)	Macro-F1	0.256	0.021	0.217	0.241	0.256	0.270	0.296
Llama 3.1 8B (big5)	Accuracy	0.425	0.018	0.392	0.413	0.425	0.437	0.461
Llama 3.1 8B (big5)	Balanced Accuracy	0.240	0.011	0.218	0.232	0.240	0.247	0.263
Llama 3.1 8B (big5)	Macro-F1	0.163	0.009	0.147	0.158	0.164	0.170	0.181

Appendix

Table 3: Figure 1 results (Big Five setting). Mean performance with 95% participant-level bootstrap confidence intervals.

Model	Traits	Metric	Mean	Std	2.5%	25%	50%	75%	97.5%
LBM-2	74	Accuracy	0.628	0.019	0.592	0.615	0.627	0.640	0.668
LBM-2	74	Macro-F1	0.478	0.026	0.428	0.460	0.478	0.496	0.531
Llama 3.1 8B	74	Accuracy	0.427	0.018	0.394	0.415	0.427	0.439	0.462
Llama 3.1 8B	74	Macro-F1	0.164	0.009	0.147	0.158	0.164	0.170	0.181
Claude 4.5 Sonnet	74	Accuracy	0.468	0.016	0.436	0.457	0.468	0.479	0.500
Claude 4.5 Sonnet	74	Macro-F1	0.245	0.015	0.216	0.235	0.245	0.255	0.273

Appendix

Table 4: Figure 2 results. Performance as a function of trait dimensionality. Mean performance with 95% participant-level bootstrap confidence intervals.

Behavioral Foundation Models

See all 15 papers

Apr 26, 2026

LLMs Reading the Rhythms of Daily Life: Aligned Understanding for Behavior Prediction and Generation

Human daily behavior unfolds as complex sequences shaped by intentions, preferences, and context. Effectively modeling these behaviors is crucial for intelligent systems such as personal assistants and recommendation engines. While recent advances in deep learning and behavior pre-training have improved behavior prediction, key challenges remain--particularly in handling long-tail behaviors, enhancing interpretability, and supporting multiple tasks within a unified framework. Large language models (LLMs) offer a promising direction due to their semantic richness, strong interpretability, and generative capabilities. However, the structural and modal differences between behavioral data and natural language limit the direct applicability of LLMs. To address this gap, we propose Behavior Understanding Alignment (BUA), a novel framework that integrates LLMs into human behavior modeling through a structured curriculum learning process. BUA employs sequence embeddings from pretrained behavior models as alignment anchors and guides the LLM through a three-stage curriculum, while a multi-round dialogue setting introduces prediction and generation capabilities. Experiments on two real-world datasets demonstrate that BUA significantly outperforms existing methods in both tasks, highlighting its effectiveness and flexibility in applying LLMs to complex human behavior modeling.

Fanjin Meng, Jingtao Ding, Nian Li +2

Department of Electronic Engineering, BNRist, Tsinghua University, Beijing, China

LLM Alignment Human Behavior Modeling Human Behavior Simulation Human Behavior Prediction

Ask PDF

May 28, 2026

ActTraitBench: Quantifying the Knowledge-Decision Gap in Large Language Models via Human-Grounded Behavioral Validation

While Large Language Models (LLMs) can convincingly simulate personas in explicit self-reports, they often deviate in implicit behavioral decisions, revealing a substantial Knowledge-Decision Gap (

G_{\mathrm{KD}}

). Existing benchmarks struggle to measure this discrepancy due to limited construct validity, multidimensional entanglement, and distributional biases in LLM-based evaluation. To address these issues, we propose ActTraitBench, a human-grounded evaluation framework for measuring personality consistency in LLMs. Grounded in empirical human data, ActTraitBench establishes one-to-one mappings between psychometric facets and behavioral paradigms and applies Distributional Calibration via Quantile Mapping to reduce distributional mismatch between LLM-judge scores and human responses. Experiments on 14 mainstream LLMs reveal substantial knowledge-decision gaps and show that assigned personas are reflected more consistently in self-reports than in behavioral decisions for most evaluated models. To mitigate this gap, we further introduce the Chain of Cognitive Alignment (CoCA), an inference-time intervention that reduces

G_{\mathrm{KD}}

for 12 of the 13 models with paired results. Code and resources are available at https://github.com/Selina233/ActTraitBench.

Yutong Yang, Chenxi Miao, Weikang Li +1

Peking University · Baidu Inc

LLM Evaluation Personality Modeling in Language Models Inference-Time Intervention

Ask PDF

Jul 13, 2026

Reproducing human biases in route choice using large language models: Toward scalable behavioral modeling

Human choice behavior, including route choice, exhibits systematic behavioral biases that deviate from the assumptions of full rationality. Cumulative prospect theory (CPT) has been widely recognized as an effective framework for characterizing such behavioral patterns. However, its large-scale application, particularly in simulation and agent-based modeling, critically depends on specifying individual-level CPT parameters, which remain a major bottleneck. Conventional approaches typically rely on surveys and controlled experiments to calibrate CPT parameters, yet these methods are difficult to generalize and often fail to capture the full diversity of human decision-making. To address this challenge, this paper investigates whether large language models (LLMs) can reproduce human behavioral biases in choice-making without explicit specification of prospect-theoretic parameters. Using route choice as a representative scenario, we design a behavioral evaluation framework and systematically compare LLM-generated decisions with established human behavioral patterns predicted by CPT. Experimental results demonstrate that LLMs are capable of reproducing non-rational human choice biases and can exhibit decision behaviors consistent with prospect-theoretic effects under uncertainty. These findings suggest that generative AI models may provide a scalable alternative for modeling human decision processes and offer a promising foundation for next-generation large-scale agent-based simulation and AI-driven behavioral research.

Jiangtao Han, Shoufeng Ma, Shuxian Xu +4

College of Management and Economics, Tianjin University, China · Faculty of Science and Engineering, University of Nottingham Ningbo China

Cognitive Modeling Human Behavior Simulation Decision-Making under Uncertainty

Ask PDF

Abstract

Behavioral Foundation Models

Explore similar work

LLMs Reading the Rhythms of Daily Life: Aligned Understanding for Behavior Prediction and Generation

ActTraitBench: Quantifying the Knowledge-Decision Gap in Large Language Models via Human-Grounded Behavioral Validation

Reproducing human biases in route choice using large language models: Toward scalable behavioral modeling