The limits of falsifiability: Dimensionality, measurement thresholds, and the sub-Landauer domain in biological systems

Abstract

Karl Popper’s falsifiability criterion assumes that scientific hypotheses can be reduced to binary tests. We show this assumption is scale-dependent and can saturate in high-dimensional biological systems operating near physical measurement limits, especially near criticality. In neural networks, much relevant information exists as patterns below the Landauer threshold for irreversible bit recording—signals too weak for individual neurons to detect but detectable when pooled across populations. These sub-threshold patterns cannot be projected into binary outcomes without destroying their causal structure. We develop a framework connecting dimensionality, thermodynamic measurement limits, and biological epistemology, showing that Popperian logic represents a special case applicable only to low-dimensional systems with strong signals. Our analysis has implications for neuroscience, where aspects of conscious processing may in part depend on sub-threshold coherence patterns that resist binary measurement, motivating a shift from single-case hypothesis tests to multi-scale, ensemble-based inference. The framework extends to other complex biological systems including ecological networks, protein folding dynamics, and evolutionary processes where causal relationships exist as irreducible multi-dimensional structures operating below classical measurement thresholds.