Abstract

Many medical and regulatory domains adopt Linear No-Threshold (LNT) assumptions: any non-zero exposure produces proportional harm. This paper argues that LNT is conceptually inadmissible as a default for adaptive biological systems. Because means have conditional value—value dependent on the ends they serve—they necessarily have bounded effective ranges. This structure, captured by the Delphic maxim mēden agan (nothing too much), is a formal entailment of instrumental reasoning. In adaptive systems, monotonic linear harm should be the exception requiring explanation, not the null hypothesis. The paper's contribution is twofold: to derive bounded dose-response as a conceptual entailment that shifts the burden of proof to defenders of LNT; and to rehabilitate natural philosophy—systems theory applied to living systems—as capable of generating the general form correct medical hypotheses must take, while leaving to empirical science the determination of where specific bounds fall.