Is Reproducibility Always Important or Even Possible for a Scientific Experiment?

Abstract

This article provides an extended commentary on three books by R. Laymon and A. Franklin about the methodology and epistemology of the scientific experi­ment, as well as their article on the issue of reproducibility of experiments. The reproducibility of scientific results has historically been considered one of the methodological standards of science, and it is associated with ideas about the truth and intersubjective nature of scientific knowledge. The problem of re­producibility has received particular attention in recent decades because special­ized studies have revealed that more than half of the results from the social sci­entific studies cannot be reproduced; many cases of fraud in biomedical sciences have been uncovered; and the collective nature of subjectivity in elementary par­ticle physics has accentuated the instability of the knowledge obtained by large collaborations. In reconstructing discussions about reproducibility in the philo­sophical literature, we distinguish between replicating an experiment by repeat­ing it in a way that is as close as possible to the original and actually reproducing it by re-obtaining a previously observed phenomenon in a significantly modified instrumental-theoretical setting. We also introduce the concept of replication-2 as an intermediate form between replication and reproducing. These kinds of re­search repetitions perform different functions in experimental practice. We show that a variety of kinds of replication and reproduction are at the heart of a set of epistemic strategies: experimental methodological standards identified by Franklin based on decades of research in scientific practice. We analyze a num­ber of experiments in which a single measurement, in the absence of epistemic strategies, was sufficient for the community to accept a new theory. In these cases, we argue, a theory based on high-value symmetry principles turned out to be the dominant lens of the community, while the experiment played a role only as a demonstration. Such examples, in our opinion, indicate that the experiment’s role in a situation of shifting scientific paradigms is different from its role in nor­mal science: the requirements for reproducibility and epistemic strategies are significantly alleviated in the former in comparison to the latter.