Neosis: A Computational Framework for Biologically-Inspired Self-Modifying Organisms

Author: Tooraj Helmi

Date: 2025

Abstract

Artificial neural networks have achieved remarkable success, but they remain fundamentally static—their architectures are fixed at design time and only weights are adjusted during training. This structural rigidity stands in contrast to biological cognition, where neural circuits develop, specialize, and reorganize throughout an organism's lifetime.

Neosis addresses this limitation by proposing a foundation in which cognitive systems are alive in a computational sense. The basic organism, the Neo, is a self-modifying graph dynamical system whose nodes carry local computational rules, whose edges define information flow, and whose structure can reorganize over time through mutation and energy-driven selection.

This document develops the formal model of Neosis, analyzes its computational and learning capabilities, and demonstrates how populations of Neos can evolve increasingly sophisticated adaptive behaviors. The framework bridges micro-level node dynamics with macro-level mathematical analysis, providing both mechanistic understanding and predictive power for large-scale evolutionary systems.

---

Copyright © 2025 Tooraj Helmi. All rights reserved.

This work is submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy.