AI That Thinks Like a Human Brain

A neuroscience-grounded cognitive architecture with 17 modules simulating dual-process thinking, emotion, memory consolidation, curiosity, and metacognition.

Try Simulator Read Research →
0
Cognitive Modules
0
Simulation Scenarios
0
Peer-Reviewed Bases
MIT
Open Source

The Problem

Current AI systems (LLMs, RL agents) lack fundamental cognitive mechanisms that humans use daily: emotional regulation, memory consolidation during sleep, curiosity-driven exploration, and metacognitive self-monitoring.

  • ✕ No emotional processing or somatic markers
  • ✕ No memory decay, priming, or consolidation
  • ✕ No intrinsic curiosity or metacognition
  • ✕ No theory of mind or social reasoning

Our Solution

Human Cognition AI implements a complete neuroscience-based cognitive architecture. Each module maps directly to brain research, creating a system that processes information the way humans actually do.

  • ✓ Dual-process thinking (System 1 + System 2)
  • ✓ Emotion system with somatic markers
  • ✓ 4-store memory with sleep consolidation
  • ✓ Curiosity, metacognition, theory of mind

The 17 Cognitive Modules

Each module maps to neuroscience research. Color-coded by architectural layer. Click any tile to explore.

Foundation
Processing
Motivation
Interface
Advanced
View all 17 modules in detail →

Cognitive Architecture

Five concentric phases from foundational processing to advanced cognition. Click any layer to expand.

Phase 5: Advanced Creativity, Sleep Consolidation, Time Perception
Creativity Engine
Conceptual blending, analogical reasoning, and divergent thought generation
Sleep Consolidation
Offline memory replay, synaptic homeostasis, and memory integration
Time Perception
Temporal reasoning, event sequencing, and duration estimation
Phase 4: Interface Language, Social Cognition, Embodied Cognition
Language Processing
Semantic parsing, pragmatic reasoning, and natural language generation
Social Cognition
Theory of mind, empathy modeling, and social norm understanding
Embodied Cognition
Sensorimotor grounding, body schema, and environmental coupling
Phase 3: Motivation Curiosity, Emotion System, Self-Awareness
Curiosity Drive
Information-gap detection, intrinsic motivation, and novelty seeking
Emotion System
Somatic markers, valence/arousal modeling, and affective forecasting
Self-Awareness
Recursive self-modeling, metacognition, and confidence calibration
Phase 2: Processing Dual-Process Thinking, Executive Function, Decision Making
Dual-Process Thinking
System 1 fast intuition and System 2 deliberate reasoning
Executive Function
Task switching, inhibitory control, and working memory management
Decision Making
Multi-criteria evaluation, expected utility, and satisficing
Phase 1: Foundation Predictive Coding, Memory Systems, Perception
Predictive Coding
Hierarchical prediction error minimization (free-energy principle)
Memory Systems
Episodic, semantic, procedural, and working memory
Perception
Multi-modal sensory processing with top-down prediction

See It in Action

Watch how information flows through the cognitive architecture in real time. The interactive simulator shows each module activating as the system processes stimuli, makes decisions, and generates responses.

  • 1. Choose from 5 cognitive scenarios
  • 2. Watch modules light up as information flows
  • 3. Inspect parameters and adjust in real-time
  • 4. Compare processing times to human cognition
Launch Full Simulator
S
Sensory Input
Visual: large shape approaching fast
A
Amygdala Fast Path
THREAT DETECTED: fight-or-flight
M
Motor Output
IMMEDIATE: jump back, raise arms
C
Cortical Analysis
Object identified: ball, not threat
E
Executive Control
False alarm, downregulate response
Danger Response Scenario - 150ms human equivalent

Built With Leading Technology

Research-grade cognitive science infrastructure

Python Python
NumPy NumPy
SciPy SciPy
pytest pytest
Jupyter Jupyter
GitHub GitHub

How It Compares

Benchmarking against established cognitive architectures.

Capability Human Cognition AI ACT-R SOAR Global Workspace
Cognitive Modules 17+ ✓ ~12 ~8 Theory only
Emotional Integration Full ✓ Limited None Theoretical
Memory Types 4 stores ✓ 2 1 Theoretical
Self-Awareness Recursive ✓ No No Theoretical
Sleep Consolidation Yes ✓ No No No
Open Source MIT ✓ Academic Academic N/A

Research Foundations

  1. [1] Baars, B.J. (1988). "A Cognitive Theory of Consciousness." Cambridge University Press.
  2. [2] Friston, K. (2010). "The Free-Energy Principle: A Unified Brain Theory?" Nature Reviews Neuroscience, 11(2), 127–138.
  3. [3] Kahneman, D. (2011). "Thinking, Fast and Slow." Farrar, Straus and Giroux.
  4. [4] Damasio, A.R. (1994). "Descartes' Error: Emotion, Reason, and the Human Brain." G.P. Putnam's Sons.
  5. [5] Dehaene, S. (2014). "Consciousness and the Brain." Viking Press.
  6. [6] Tononi, G. & Cirelli, C. (2006). "Sleep function and synaptic homeostasis." Sleep Medicine Reviews, 10(1), 49–62.
  7. [7] Premack, D. & Woodruff, G. (1978). "Does the chimpanzee have a theory of mind?" Behavioral and Brain Sciences, 1(4), 515–526.

Support This Research

An open-source framework for modeling human cognition. Join researchers and institutions worldwide in advancing cognitive AI.

All prices in EUR.

Researcher
€15/mo

Name in README, early access to new modules

Learn More
Most Popular
Lab
€79/mo
  • ✓ Logo on site
  • ✓ Priority support
  • ✓ Research collaboration
  • ✓ Module customization
Get Started
Institution
€299/mo

Co-research, custom modules, dedicated support

Learn More

Frequently Asked Questions

LLMs are statistical pattern matchers trained on text. Human Cognition AI implements actual cognitive mechanisms from neuroscience: dual-process thinking, emotional processing with somatic markers, multi-store memory with decay and consolidation, curiosity-driven exploration, and metacognitive self-monitoring. These are architectural features, not emergent behaviors.

Python 3.8+ with NumPy as the primary dependency. Optional Numba support provides 10-100x speedup through JIT compilation. The architecture achieves 2,000+ cognitive operations per second with sub-millisecond core operations.

Yes. The project is MIT licensed and free for commercial use. For enterprise-level support, custom module development, or co-research partnerships, see our sponsorship tiers on the pricing page.

Yes. A live cognitive simulation API is deployed at zedigital-human-cognition.fly.dev. The demo page connects to it automatically when available. You can also run the full architecture locally with a single Python command.

Click the "Cite This Work" button in the footer for a ready-to-use BibTeX entry. The citation includes the full project title, author, and repository URL.

BibTeX Citation

@software{agielo2026,
  title     = {Human Cognition AI (AGiELO):
               A Neuroscience-Based Cognitive Architecture},
  author    = {Zekaj, Elvi},
  year      = {2026},
  url       = {https://github.com/ezekaj/agielo},
  license   = {MIT},
  note      = {17 cognitive modules implementing
               brain-realistic cognition}
}