The Rally Paradox: Compensatory Theta Coherence Dominates Sustained Cognitive Load, Reframing Vigilance Decrement as Compensation Failure

Corrigendum The Rally Paradox: Compensatory Theta Coherence Dominates Sustained Cognitive Load, Reframing Vigilance Decrement as Compensation Failure Zenodo DOI 10.5281/zenodo.18303204 — corrigendum, July 2026 Three empirical claims in this preprint require correction. The theoretical framework is retained. 1. Data source and pre-processing attribution.** The paper does not credit the EEG pre-processing. All filtering, artifact handling and reduction to 30 EEG channels are those of Cao, Chuang, King & Lin (2019), *Scientific Data* 6:19, which should be cited. The Figshare citation is corrected from the raw deposit (10.6084/m9.figshare.6427334) to the **pre-processed** deposit actually used (10.6084/m9.figshare.7666055). The two are not interchangeable: applying the same coherence pipeline to the raw deposit yields Δρ values essentially uncorrelated with those reported (r = −0.03 across the same 17 sessions). 2. Headline rally count.** The abstract reports 88% rally and 12% fade (15 and 2 of 17 sessions). Under the paper's own threshold rule (Δρ ≥ 0 rally, Δρ < 0 fade), the underlying session values yield **14 rally and 3 fade — 82.4% and 17.6%**. The figures 15 and 2 are not recoverable from those values, and the abstract should read 82.4% / 17.6%. This does not affect the framework claim. Reanalysis of the complete deposit — 61 unique sessions from 26 participants — gives a rally rate of **86.9% (95% CI 75.8–94.2%)** under the same rule, an interval that encompasses the originally claimed 88%. Rally dominance is confirmed on a sample three times larger, including 17 participants not previously analysed. 3. Phenotype counts.** The reported split of Adaptive Rally n = 5 (29%) and Drowsy Rally n = 10 (59%) is not reproducible from the underlying session values under any documented threshold rule. Applying the sign of Δα to rally sessions gives 7 Adaptive and 7 Drowsy under Δρ ≥ 0, and 6 and 4 under the ±0.03 rule. These specific counts are withdrawn; readers are referred to the corresponding sections of *Compensatory Fronto-Parietal Theta Coherence Under Sustained Cognitive Load* (Dobbin, 2026) for phenotype breakdowns on the full sample. Retained. The theoretical framework stands: rally as compensatory recruitment, fade as compensation failure, and the alpha-based Adaptive/Drowsy distinction. The expanded evidence reinforces it. Rally is the majority response at both session and participant level, and the Adaptive/Drowsy behavioural contrast — non-significant at the original sample size, where achieved power was 0.27 — reaches significance at 26 participants (p = 0.015, d = −1.18). The vigilance decrement literature has long assumed that sustained cognitive load leads to progressive degradation of neural coordination and performance. We present preliminary evidence challenging this assumption. In an analysis of 17 subjects performing a sustained-attention driving task (43-118 minutes), 88% showed increased fronto-parietal theta coherence over time (the rally pattern), while only 12% showed the predicted decrease (the fade pattern). This rally paradox suggests that compensatory neural mechanisms may be the dominant response to sustained load, with fade representing a failure of compensation rather than normal fatigue. We propose a theoretical framework distinguishing three response phenotypes: Adaptive Rally (coherence up, alpha down; n=5, 29%), Drowsy Rally (coherence up, alpha up; n=10, 59%), and Fade (coherence down; n=2, 12%). If this pattern generalises, it would reframe vigilance research from preventing universal decline to identifying the minority who fail to compensate. These findings, while preliminary, warrant replication in larger, multi-task cohorts to test generalisability.