This paper (Part I of a two-part study) proposes a novel demarcation criterion for distinguishing scientific from non‑scientific knowledge. Building on the Reflexive-Constructive Methodology (RCM), the criterion focuses not on the static properties of statements (such as verifiability or falsifiability) but on the nature of the cognitive act that generates knowledge. A cognitive act is defined as reflexive if it changes the agent’s subjective models and evaluation criteria and cannot be reduced to Bayesian updating. Knowledge produced by reflexive acts is considered scientific, regardless of whether it reduces uncertainty (δ0, e.g., discovery of new phenomena). The paper shows that Popperian falsifiability is a special case of this broader criterion. Importantly, the criterion does not require empirical testability; logical falsifiability (e.g., by thought experiment, logical contradiction, or counterexample) suffices. Thus, mathematics, theoretical physics, humanities, and natural sciences all satisfy the criterion on equal terms. The criterion is self-applicable and falsifiable, and it resolves classical counterexamples (logarithmic tables, accidental discoveries) without reducing science to empiricism. The formal apparatus of RCM is introduced, and the relationship with Kuhn’s paradigm shifts and Lakatos’ research programmes is discussed.
Alexander Romannikov (Sat,) studied this question.