Wilson–Cowan — S(u)=1/(1+exp(−u)), dx/dt=(−x+(1−x)·S(cee·x−cei·y−a·z+P))/τe, dy/dt=(−y+(1−y)·S(cie·x−cii·y+Q))/τi, dz/dt=(x−z)/τz. Fixed cee=16 cei=12 cie=15 cii=3 a=5 Q=0 τe=1 τi=2. Classic P=2 τz=8 (a nudged 1.5→5, P 1.25→2 for clear limit-cycle; brief a=1.5 P=1.25 near-fixed). Dual P+τz sliders. Display embed (x,y,z). Distinct from morris-lecar.html, fitzhugh-nagumo.html, hindmarsh-rose.html, arneodo.html, lotka-volterra.html, brusselator.html. Bulk centroid ≈ (0.2272, 0.4423, 0.2273). RK4 neural fold. Gold low-alpha trail, cap 650. RM = settled silhouette, no animate-on.