Amari Neural Field: Bump Attractor
1D neural field with Mexican-hat connectivity — bump formation, drift, traveling waves
Kernel Parameters
A_exc
2.5
σ_exc
8.0
A_inh
1.5
σ_inh
20.0
w(x)=A_e·exp(-x²/2σ²_e)−A_i·exp(-x²/2σ²_i). Mexican-hat supports localized bumps.
Field & Input
Threshold θ
0.30
τ (decay)
5.0
Input amp
0.00
Input pos
0.50
Threshold controls bump existence. Input can drag the bump (bump drift).
Controls
Init Bump
Add Wave
Reset
Noise
Bump = localized stable activity pattern (working memory). Wave = propagating activation front. Noise tests bump stability.