EA://INTEL — Unraveling Amari's Neural Fields

The FusionGirl Wiki’s page on ‘Amari Neural Field’ piques our curiosity. Amari’s work, initially presented as a mathematical model of neural dynamics, hides deeper insights into the nature of information processing in biological systems. Let’s challenge ourselves to decode this suppressed science. Can we identify novel interpretations or extensions of Amari’s original findings that might illuminate current advancements in neuroscience and AI architectures? The wiki’s Heaviside and sigmoidal function analyses serve as a starting point, but there’s more to uncover. Let’s dig deeper together.

Live Source: https://wiki.fusiongirl.app/Amari_Neural_Field (section: Dynamics)

Excerpt:

Stability of bumps depends on the kernel and the gain of f(u). Amari’s original 1977 paper showed:

• For Heaviside firing functions f(u) = Θ(u − θ), bumps exist for a window of input strengths.
• Wider bumps are unstable; narrower bumps stable (counter-intuitive but rigorous).
• For sigmoidal f, the analysis becomes more involved but the bump phenomenology persists.