Computationally Constrained Renormalisation in Virtual Brain

This note is conjectural and should be read accordingly. For example, function induction may be vulnerable to minute approximation errors, and the resolution required to ensure accurate behavioural simulation may be ruinously costly. Indeed, the author conjectures that even if in-sample reconstructions are faithful, minute errors in coordinates and interactions will compound during rollout, producing cumulative drift and a progressive loss of behavioural fidelity.Author’s Note (Informal). The author imagines a continued tension between reinforcement updates that tether the reconstruction to its in-vivo target and the accumulated degradation induced by approximation error. In sufficiently complex systems, infinitesimal errors may produce arbitrary finite-time deviations. Thus, the author visualises the reconstruction as gradually “derezzing” beyond the data horizon.