The condition time summaries do not identify the labels of two ds000102 snapshot 00001 trials.

A finite sensitivity calculation on 1,248 trials from 26 participants in OpenfMRI ds000102 snapshot 00001 shows that participant-level condition time summaries do not identify trial labels. Two allowed run-2 label exchanges preserve the exact condition counts, time sums, sums of squares, means, sample variances, coefficients of variation, and condition correlations. The constructed examples use equal recorded reaction times (420 ms for sub-18 trials 7 and 21; 701 ms for sub-25 trials 14 and 18). Of 890 permitted participant-level pair changes, 406 meet both reported condition-correlation comparison rules and 484 do not; two alternatives also preserve exact moments. The source fields disagree on 304 of 624 run-2 trials. The correct-positive-response overall correlation is 0.4093874792819571, which does not match the reported 0.38; including all positive responses gives 0.38422366989519363. These are constructed assignments on observed times, not corrected labels or new observations. No original task-presentation record was obtained, so the actual arrow directions remain unresolved. The result is conditional on the stated model and does not establish label correctness, population uncertainty, neural effects, biological causes, or a general proof.

trial labels · reaction time · coefficient of variation · label ambiguity · sensitivity analysis · ds000102 · Flanker task