Open-source research systems at the intersection of computational neuroscience, closed-loop neural engineering, and neuroadaptive educational technology.

Noggin Labs develops lightweight, local-first computational systems for studying and optimizing closed-loop cognitive interactions.

Our work focuses on two complementary areas:

1. Neural Engineering — Quantifying end-to-end feedback latency and temporal constraints in closed-loop neural systems.
2. Neuroadaptive Education — Building deterministic, rule-based Socratic tutoring systems with configurable accommodations for neurodivergent learning profiles, including ADHD, autism, and dyslexia.

Our goal is to make experimental cognitive-computing systems measurable, reproducible, inspectable, and openly accessible.

The Write-Back Gap as a Trial-Level Temporal Variable in Closed-Loop Learning Systems.

NFOT investigates the Write-Back Gap,

$$ L \equiv t_{\mathrm{fb}} - t_0 $$

as an empirical, trial-level temporal variable rather than treating feedback latency solely as an engineering overhead.

Core components:

• NFOT Preprint v3
• Pre-registered test protocols (PTSP)
• Trial-level data extraction and validation workflows
• Cross-sectional audits of public EEG/BCI datasets, including BCI-FIT and data from Chandravadia et al.
• Computational models for evaluating feedback-timing relationships

Stack: Python 3 · MNE-Python · NumPy · pandas · SciPy

Local Socratic AI Tutoring Engine & Neuroadaptive Scaffolding Loop.

Noggimigo is a local-first tutoring engine designed to guide learners through mathematical and logical problems in the Noggin and PenPal platforms using structured Socratic micro-questions rather than directly generating completed solutions.

Core components:

• Deterministic Socratic tutoring loops
• Structured Misconception Error Taxonomy
  • INVERSION_ERROR
  • HALVING_ERROR
  • INVERSE_ERROR
• Algorithmic fallback and recovery logic
• Response-latency logging through $L_{\text{edu}}$
• Configurable scaffolding and pacing mechanisms
• Based on the paper "Your LLM is an Incompetent AI Tutoring System".

Accessibility: Feature-flagged accommodation adapters can modify text density, reading level, and pacing for different learning profiles.

Noggin Labs' software systems provide computational and runtime components for the following research:

Kassim, F. A. (2026).
Neural Feedback Optimization Theory (NFOT): The Write-Back Gap as a Trial-Level Temporal Variable in Closed-Loop Learning Systems. Zenodo Preprint, v3.

DOI: 10.5281/zenodo.21378901

Kassim, F. A. (2026).
Your LLM is an Incompetent AI Tutoring System: A Survey of AI Tutoring Paradigms, Neural Solvers, and Student Simulation Frameworks. Zenodo Preprint.

DOI: 10.5281/zenodo.22975179

Noggin Labs emphasizes:

• Local-first computation
• Reproducible experiments
• Explicit measurement over implicit assumptions
• Deterministic system behavior where appropriate
• Open-source tooling
• Trial-level and event-level data analysis
• Transparent experimental protocols
• Accessibility-aware system design
• Safety-first AI

Noggin Labs is an independent research workspace open to technical collaboration and critical feedback.

We welcome:

• Data-sharing collaborations
• Open-source contributions and pull requests
• Research discussion and reading groups
• Reproducibility reviews
• Technical feedback from computational neuroscience, neural engineering, and educational technology researchers

Principal Lead: Folarera Ayobami Kassim
Contact: folarera.kassim@gmail.com