Excursus 091 – On Reconstruction as a Form of Scientific Inference

Hossein Jorjani

First publication: 2026 – 07 – 19

Scientific inference is commonly discussed in terms of interpolation and extrapolation. Yet many important forms of modern scientific reasoning belong to neither category. They are better understood as instances of reconstruction.

Interpolation estimates unknown values that lie within an observed domain. Extrapolation extends established relationships beyond the domain in which they have been validated. Reconstruction differs fundamentally from both. It seeks to recover unknown components within an already existing structure by exploiting the constraints that relate its constituent parts.

This distinction can be summarized succinctly:

• Interpolation: the unknown lies within the observed domain.
• Extrapolation: the unknown lies outside the observed domain.
• Reconstruction: the unknown lies within the observed structure.

Reconstruction derives its inferential power not primarily from statistical regularities but from structural organization. Missing DNA variants may be inferred from linkage disequilibrium, damaged manuscripts from linguistic and textual regularities, incomplete pedigrees from genealogical relationships, and missing observations from causal or statistical models. In each case, the unknown is recoverable because the observed elements constrain the range of admissible solutions.

Unlike extrapolation, reconstruction does not extend a theory into an untested domain. Instead, it exploits relationships already present within the system itself. The principal source of information is therefore structural coherence rather than spatial, temporal, or conceptual extension.

This distinction has broader implications for the philosophy of science. Discussions of scientific inference have traditionally focused on induction, deduction, prediction, and extrapolation. Modern scientific practice increasingly relies upon reconstruction, where missing information is recovered through structural constraints rather than by extending empirical regularities.

Whether reconstruction deserves recognition as a methodological category alongside interpolation and extrapolation remains an open question. Nevertheless, its growing importance in genetics, statistics, computational science, historical scholarship, and many other disciplines suggests that it constitutes a distinctive form of scientific reasoning worthy of separate philosophical analysis.