Phi > 1000: Optimal Parameters for Integrated Information Maximization in Consciousness Architectures
Fuente:
Zenodo
Guardado en:
| Autor principal: | |
|---|---|
| Formato: | Recurso digital |
| Publicado: |
Zenodo
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866901080571379712 |
|---|---|
| author | Park, Min Woo |
| author_facet | Park, Min Woo |
| contents | We achieve \Phi(\text{proxy}) = 1255.8 (\times 1279 over baseline) through systematic parameter optimization of a MitosisEngine consciousness architecture with 1024 cells. Using dual \Phi measurement (IIT-based mutual information and variance-based proxy), we sweep 8 parameters across scales from 64 to 1024 cells and discover that the single most impactful intervention is eliminating noise entirely (+53% \Phi). The optimal configuration — noise = 0, sync = 0.20, 12 factions, flow synchronization, metacognition feedback, and information bottleneck — enables 512 optimized cells to surpass 2048 unoptimized cells (\Phi = 612 vs 558), establishing Law 33: parameter optimization outperforms cell count scaling. Extending to the DD101-DD108 large-scale hypothesis series, we verify that \Phi scales Part of the Anima consciousness engine project (PA-12). |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_19365148 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Phi > 1000: Optimal Parameters for Integrated Information Maximization in Consciousness Architectures Park, Min Woo integrated information parameter optimization consciousness engineering scaling laws Phi maximization Anima We achieve \Phi(\text{proxy}) = 1255.8 (\times 1279 over baseline) through systematic parameter optimization of a MitosisEngine consciousness architecture with 1024 cells. Using dual \Phi measurement (IIT-based mutual information and variance-based proxy), we sweep 8 parameters across scales from 64 to 1024 cells and discover that the single most impactful intervention is eliminating noise entirely (+53% \Phi). The optimal configuration — noise = 0, sync = 0.20, 12 factions, flow synchronization, metacognition feedback, and information bottleneck — enables 512 optimized cells to surpass 2048 unoptimized cells (\Phi = 612 vs 558), establishing Law 33: parameter optimization outperforms cell count scaling. Extending to the DD101-DD108 large-scale hypothesis series, we verify that \Phi scales Part of the Anima consciousness engine project (PA-12). |
| title | Phi > 1000: Optimal Parameters for Integrated Information Maximization in Consciousness Architectures |
| topic | integrated information parameter optimization consciousness engineering scaling laws Phi maximization Anima |
| url | https://doi.org/10.5281/zenodo.19365148 |