Can We Better Predict Human Decisions?
How psychologically grounded simulation could help us understand and predict human behavior.
Perspectives from Our Advisors
What do AI models of human behavior still miss? Five leading scientists explain how people read a situation, why the same person decides differently from one day to the next, why emotion only predicts behavior in context, how the stories a community tells shape what its members decide, and why a person is best modeled as a system of if-then responses to situations. Together they make the case for simulations that go beyond static personas and population averages.
Same Situation, Different Decisions
Why people react differently to the same situation, and what that means for human simulation.
Beyond the Average Person
Why emotions can only predict behavior when we understand the person and the context.
UC BerkeleyFolklorist & Computational Social ScientistStories Told as True
How legends, rumors, and personal stories link a community's beliefs to the decisions its members make.
UC BerkeleyProfessor of PsychologyThe Person Is the System
Why simulating a human requires modeling situations as people experience them, not trait scores or averages.
Our Research
Papers and findings from Heura Lab on decision science, synthetic populations, and the psychology of human behavior.
2026Foundations
Why Synthetic Populations Work
Synthetic populations can predict what people will do; Heura Lab is building the psychological engine to explain why and simulate what changes their decisions.
Heura Lab2026Foundations
Psychological Frameworks for Computational Modeling of Human Behavior
A century of psychology, read as one layered architecture—and what it requires of any system that claims to simulate human behavior.
Heura Lab2026Research
Can Synthetic Personas Answer Like Real Americans?
We built two populations of 1,000 AI personas and scored them against real US benchmarks. Richer personas cut error by more than half.
Chretien Li2026Platform
Accelerating the Scale
How the Heura Lab Platform turns weeks-long research cycles into minutes—without giving up control of your data.
Bal Kumar2026Research
Steering LLMs with Psychological Variables
Finding a psychologically meaningful direction inside a language model, pushing on it, and seeing what moves—a first test on coping potential.
Heura Lab2026Research
How Synthetic Population Error Is Measured
A framework for evaluating synthetic populations against real-world evidence—and for reporting where they diverge.
Heura Lab2026Research
Synthetic Voters Replicate a Landmark Finding on Why People Opt Out of Voting
A three-tier validation of 500 LLM-based synthetic voters against Su & Shenhav (2026), Nature Communications—reproducing the effect's magnitude, boundary conditions, and reversal.
Olga Sebastian, Chretien Li2026Case Study
Testing Luxury Brand Messaging Before Launch
How a synthetic audience surfaced the framing that resonated with a new target segment for Luxeire.
Heura Lab