Renwei Meng
Skip to content
CORTEX.HUMAN → INTELLIGENCE

HUMAN COGNITION × NEXT INTELLIGENCE

The next lesson
for AI is
within us.

We taught AI to learn from our answers. Now, we move upstream—to the attention, reasoning and decisions that come before them.

From human output to verifiable cognitive structure.
FIG. 01A DIFFERENT VIEW OF COMPUTATION
HUMAN COGNITIONattention / relations / decisions
AI LATENT STATESalignable intermediate computation
CONCEPT VISUALIZATIONDRAG TO ROTATE · PROCEDURAL POINT CLOUD
THE CORTEX PRINCIPLE

Beyond learning the answer. Learn how it comes to be.

01 / DEPTH OF SUPERVISION

Supervision goes deeper.

Move through five layers of learning signals. Greater depth brings richer information, higher costs and new responsibilities.

03 / PROCESS

Process

Follow attention, evidence and conflict through time to explore the intermediate computation before an answer.

Information
high
Cost
high
Noise
high
Real-time
mid
Privacy sensitivity
high

Qualitative concept scales, not empirical comparisons.

HUMAN
RELAYRELIABLE TEACHER
AI

02 / FROM STATIC TO ACTIVE

A reliable starting point.
A future that asks questions.

RELAY first turns fixed neural trajectories into risk-calibrated teachers for visual reasoning. CORTEX-Loop asks what happens when AI actively selects what is worth learning from humans.

03 / BEYOND A SINGLE TASK

One paradigm. New frontiers.

Cognitive structure reaches beyond a single reasoning problem. Step into three distinct interactive experiments.

CORTEX distills the computation.

What is worth learning next?

Distill the computation. Discover what is worth learning next.

© 2026 RENWEI MENGPortfolio