The Failed Ego-Depletion Replication

Popper would see the replication as a potential falsification. The original theory predicted a medium-to-large effect; the replication, with 200 participants and a pre-registered protocol, found a small effect with a confidence interval that includes zero. If the replication is a genuine test of the same hypothesis, then the theory has been refuted and should be replaced or revised. But Popper would also inspect the original author's response. The claim that the effect appears only under 'high depletion' conditions may be an ad hoc auxiliary hypothesis that saves the theory from refutation. Popper would demand an independent, risky test of that condition before accepting the revision. Kuhn would see the situation differently. For Kuhn, normal science works within a paradigm; a single failed replication is an anomaly, not a paradigm shift. The community will decide whether the anomaly is tolerable or whether it accumulates into a crisis. Kuhn would expect the original author to defend the paradigm and the replicators to press the anomaly. The editor's decision is therefore not just about one paper; it is about how the community will treat an anomaly. Popper gives a sharper critical standard; Kuhn gives a more realistic sociology of science.

Value judgements are unavoidable in editorial decisions, but they should not be confused with epistemic ones. The editor faces inductive risk: accepting a paper that overstates the replication could harm the field, while rejecting it could suppress useful evidence and harm the student's career. The student's need for a CV line is a real interest, but it is not evidence about the truth of the replication. The editor should separate the two. Longino's account of objectivity is useful here: objectivity is not value-freedom but critical dialogue under publicly shared standards. The editor should ask whether the paper makes its methods transparent, whether the conflicting reviews are addressed, and whether the conclusion is proportionate to the evidence. The 10-day deadline is a constraint, not a reason to lower standards. A defensible decision might be to send the paper for a third review focused specifically on whether the measure is a fair test, and to ask the authors to soften the 'failed replication' claim if the evidence is inconclusive. The editor's value judgement should be about the health of the field, not about the student's graduation.

On Duhem-Quine grounds, the replication is not a decisive falsification. A hypothesis is never tested in isolation; it is tested as part of a web that includes the measure of self-control, the manipulation of depletion, the population, the statistical model, and the theory itself. The replication's negative result could be due to the different measure, to the absence of a high-depletion condition, to the sample, or to the theory being false. The original author's auxiliary hypothesis is part of that web. This does not make the replication worthless. It shifts the burden: the original author must specify in advance what counts as a fair test, and the replicators must show that their measure is valid. The editor should publish the paper with a clear discussion of these ambiguities, not as a decisive refutation. A decisive falsification would require a test that both sides agree is fair, or a series of replications that rule out the auxiliary explanations. The replication is best read as evidence that the effect is fragile, not as a knockout blow. That distinction matters for how the field updates.