<aside> <img src="/icons/circle-alternate_blue.svg" alt="/icons/circle-alternate_blue.svg" width="40px" />

Abstract

I effectively went on to utilise a custom neural network

By designing independent interdependent systems with constrained umwelt (senses), can we create a positively symbiotic environment as a byproduct of their existence?

</aside>

Introduction

wMag = Weight Magnitude. It's the sum of absolute values of all w_output weights (the Hebbian motor-to-output matrix).

At birth, w_output is initialized from the topographic Gaussian map — each motor has a small spatial territory. The initial wMag is modest.

Hebbian learning strengthens connections between co-active motors and output nodes: "neurons that fire together, wire together." When a motor neuron fires while an output node is active, the weight between them grows.

Weight decay counteracts this, slowly shrinking all weights toward zero.

The balance between Hebbian strengthening and decay determines the steady-state wMag. A well-stimulated organism that receives diverse prompts will develop stronger, more differentiated motor territories → higher wMag. A bored organism with no stimuli will slowly decay → lower wMag.

wMag is a single-number summary of how much has this organism learned — it tracks the total strength of the learned motor-to-output mapping.