CortexLab

CortexLab is a desktop-first, open-source cognitive training engine built in Go and GIO.

It provides research-oriented implementations of validated cognitive tasks with precise timing, reproducible training protocols, and full raw-data export.

CortexLab is not a game.
It is an instrument.

Philosophy

CortexLab is designed around the following principles:

• Evidence-informed task design
• Analytics-first architecture
• Reproducible experimental protocols
• Raw trial-level data export
• Privacy by default (local-first)
• No gamification, no dark patterns

The goal is to provide a transparent and extensible cognitive training and measurement platform suitable for serious users, researchers, and engineers.

---

Scientific Scope

CortexLab focuses on experimentally validated cognitive domains:

• Working memory
• Short-term memory
• Executive control
• Attention
• Processing speed
• Learning and retention

Included tasks (MVP roadmap):

• N-Back (visual & auditory)
• Digit Span (forward & backward)
• Stroop Task
• Reaction Time (simple & choice)

Each task logs:

• Trial-level accuracy
• Reaction time (ms precision)
• Difficulty level
• Session metadata
• Timestamped events

All raw data can be exported as CSV or JSON.

Design Principles

1. Reproducibility

All training sessions are defined by explicit protocol configurations.

Example:

{
  "exercise": "n-back",
  "n": 3,
  "trials": 200,
  "stimulus_interval_ms": 2000,
  "seed": 12345
}

The same protocol produces identical stimulus sequences.

2. Analytics from Day One

CortexLab provides:

• Accuracy metrics
• Reaction time distributions
• Learning curves
• Session comparisons
• Longitudinal tracking

Raw data is always preserved.

3. Transparency

CortexLab makes no claims about increasing IQ or general intelligence.

Improvements observed in tasks may reflect task-specific learning. Transfer effects are not assumed and must be empirically evaluated.

Architecture

CortexLab is structured as a modular engine:

/cmd/app
/internal/core
/internal/exercises
/internal/analytics
/internal/storage
/ui

Exercises implement a shared interface:

• Deterministic protocol definition
• Stimulus generation
• Response capture
• Metric reporting

The UI layer (GIO) is strictly separated from task logic.

Technology Stack

• Go
• GIO (cross-platform desktop GUI)
• Local storage (SQLite or BoltDB)
• AGPLv3 license

License

CortexLab is licensed under the GNU Affero General Public License v3 (AGPLv3).

This ensures that modified versions used over a network must also provide source code to users.

See LICENSE for details.

Target Audience

• Researchers interested in reproducible cognitive experiments
• Engineers building cognitive measurement tools
• Serious individuals tracking cognitive performance
• Open-source contributors interested in cognitive science

Non-Goals

CortexLab does not aim to:

• Provide medical diagnosis
• Replace clinical neuropsychological assessment
• Guarantee cognitive enhancement
• Compete with gamified brain-training apps

Contributing

Contributions are welcome.

Before submitting a new exercise:

• Provide references to the cognitive paradigm
• Describe expected metrics
• Ensure deterministic protocol support
• Include documentation

Why CortexLab?

Most cognitive training tools prioritize engagement (and your money of course).

CortexLab prioritizes:

• Measurement
• Transparency
• Reproducibility
• Scientific clarity

It is designed as a foundation for long-term cognitive tracking and research-grade experimentation.

Status

Early development. API and architecture may evolve.