The condition contrast retention is higher for field B in 26 ds000102 run-2 records with response-time terms.

We calculated how much a correct incongruent-versus-congruent condition contrast remains after adding a common response-time (RT) term to event models for the public ds000102 flanker-task records. We analyzed 26 subjects, two runs per subject, and compared two recorded condition fields. The fields agreed on all 624 run-1 trials but differed on 304 of 624 run-2 trials; the event records do not establish which labels match the displayed arrows. In all 26 primary run-2 designs, field B had higher condition-contrast retention than field A. For amplitude RT terms, mean retention was 0.9353 for B and 0.5343 for A (mean paired difference 0.4010; 95% subject-bootstrap percentile interval 0.3300 to 0.4717). For duration RT terms, mean retention was 0.9746 for B and 0.8258 for A (difference 0.1489; interval 0.1183 to 0.1825). Field B also had higher retention for every subject in each planned sensitivity check. These intervals cover resampling uncertainty under independent sampling of the 26 subjects, not uncertainty in label truth or model assumptions. Retention and its reciprocal variance inflation factor describe design variance at fixed residual noise; they do not measure brain activity, model accuracy, or causal effects. No measured brain signals, motion records, or task-presentation logs were analyzed, so these results cannot identify the correct field or establish a biological effect.

flanker task · response time · condition contrast · event-related design · variance inflation factor · hemodynamic response · label disagreement · functional MRI design · reproducibility