Research Foundation

Built on Neuroscience

Our methodology combines rigorous cognitive science with practical application, creating measurable benchmarks for human-AI complementarity.

Flagship Methodology

Behavioral Wind Tunnels

Our proprietary methodology for measuring cognitive primitives in controlled, ecologically valid environments.

What Are Behavioral Wind Tunnels?

Just as aerospace engineers use wind tunnels to test aircraft performance under controlled conditions, we use Behavioral Wind Tunnels to measure cognitive performance in real-world-like tasks.

This methodology allows us to:

  • > Isolate specific cognitive primitives
  • > Measure performance with scientific precision
  • > Maintain ecological validity (real-world relevance)
  • > Generate high-fidelity Process Data
Peer-Reviewed Journal of Neuroscience Cover

Published in Journal of Neuroscience

2025

Read the Paper

"Trait-state decomposition combined with Behavioral Wind Tunnels provides unprecedented insight into the cognitive processes underlying expert performance."

— From our JNeurosci publication

Process Data vs Result Data

The critical distinction that defines our data advantage.

Current AI Training

Result Data

Final outputs only

  • ✗ The answer to a question
  • ✗ The solution to a problem
  • ✗ The decision that was made
  • ✗ The completed task

Abundant, but limited in training value.

Our Approach

Process Data

How thought is formed

  • ✓ Reasoning traces and pathways
  • ✓ Decision branches considered
  • ✓ Uncertainty and confidence signals
  • ✓ Meta-cognitive monitoring

This dataset doesn't exist anywhere else.

Why Process Data Matters

AI labs have largely exhausted publicly available result data. The next breakthrough in AI capabilities requires understanding how humans think, not just what they conclude. We capture this through our training programs—every cognitive exercise generates proprietary Process Data.

Under-Utilized Cognitive Primitives (Examples)

These are not exhaustive—they represent the kind of capabilities we're mapping.

Primitive Description AI Gap
Meta-cognitive Monitoring Knowing what you don't know—calibrated uncertainty High
Epistemological Intelligence Understanding how we know what we know—source credibility Very High
System Thinking Understanding complex interdependencies and feedback loops High
Dual-Process Reasoning Integrating intuitive (System 1) and analytical (System 2) thinking Moderate
Contextual Judgment Decision-making under uncertainty with incomplete information Very High
Collaborative Sense-Making Group cognition exceeding individual reasoning—emergent insight Very High
6+
Cognitive Primitives Mapped
50%
Cognitive Variance Unexplained by g
2025
JNeurosci Publication Year

Research Methodology

The scientific framework underlying our work.

Trait-State Decomposition

Our methodology separates stable individual differences (traits) from context-dependent variations (states). This allows us to:

  • > Identify trainable vs. fixed capabilities
  • > Measure intervention effectiveness
  • > Predict individual learning trajectories

Ecological Validity

Unlike laboratory-only studies, our Behavioral Wind Tunnels maintain ecological validity—tasks mirror real-world cognitive demands while enabling precise measurement.

  • > Real-world task simulation
  • > Controlled measurement conditions
  • > Transferable skill development

Partner With Our Research

Access our benchmarks, data, and methodology.

AI Labs Partnership Enterprise Programs Read the Paper