All simulation is structural

Simulation is widely invoked in cognitive science, yet the criterion that distinguishes it from modeling or inference remains unclear. The dominant replication assumption holds that simulation requires concrete resemblance: the simulating system works by the same principles or undergoes the same kinds of states as its target. I argue that this assumption can neither demarcate simulation nor fix simulation content. Neural reuse shows why: the same experiential resources participate in multiple task dependent mappings, so resemblance alone underdetermines what is being simulated. Paradigmatic cases point the same way. Flight simulators, wind tunnel models, and neural mirroring succeed not by sharing materials, forces, or state types with their targets, but by preserving dependency structure. I therefore propose a structural account. Generic simulation operationally realizes target relevant dependencies in a way that supports counterfactual surrogative reasoning. Cognitive simulation is a species of this genus: it realizes such dependencies in experiential systems, including perceptual, motor, proprioceptive, vestibular, or affective systems, rather than encoding them propositionally. Because these systems are multifunctional, cognitive simulation also requires contextual selection: current task, neural configuration, and downstream use fix which correspondence is operative. This account preserves the simulation and theory contrast without appealing to replication.