Skip to content

Research

Core Research Questions

  1. Can percolation theory model cognitive fragmentation in digital trace data?
  2. How does the Observer-Operator bridge degrade under stress?
  3. Can DBSCAN behavioral clustering predict intervention thresholds?
  4. What is the relationship between fragmentation score and task performance?

N=1 Proof-of-Concept Results

Preliminary internal validation shows:

  • Fragmentation scores between 0.42% (stabilized) and 15% (pre-intervention) across test cases.
  • Executive bridges are 1.5× more likely to fail than internal cluster edges under stress.
  • DBSCAN identifies irregular clusters that correspond to self-reported "stuck" states.

Computational Psychiatry Framework

The platform is designed as a transparent, reproducible tool for computational psychiatry. It does not provide clinical diagnoses but offers a model for understanding how cognitive networks break down and reorganize.


Planned Publications

  • Percolation-Theoretic Analysis of Executive Function Fragmentation (in preparation)
  • Observer-Operator Dynamics: A Framework for AI Alignment (target: NeurIPS 2025)
  • DBSCAN Behavioral Modes in Digital Trace Data (collaboration with computational biology partners)

Collaborators

We welcome collaboration from researchers in:

  • Computational psychiatry
  • Network science / percolation theory
  • AI safety and alignment
  • Neurodiversity and executive function

See Participate for how to get involved.


Open Source

The core engine is available for audit and contribution:

github.com/erikssonaicloud-gif/percolation-import