The ds000102 snapshot 00001 run-2 mean contrast space overlap with three response terms is in all three calculated 95% label reference intervals.

We reproduced all 156 primary contrast space overlap values from the target publication using the recorded ds000102 snapshot 00001 event files and a newly written QR-based implementation. The maximum absolute difference was 2.1094237467877974e-15 (tolerance 1e-9). For the three-term model, the run-2 mean overlap was 0.05981983229381024. Under each of three specified synthetic label generators, this observed mean fell within the calculated 95% label reference interval; the reference fractions at or above it were 0.18601245, 0.18088612, and 0.18430367. The run-1 condition fields agreed at all 624 rows; the two run-2 fields differed at 304 of 624 rows, including 296 correct trials. These references describe the chosen generators on fixed schedules. They are not population confidence intervals or experimental p-values, and they do not establish which recorded condition field matches the displayed stimuli. The analysis used no brain images, measured hemodynamic response, movement, or measured noise covariance. Its Euclidean metric, response model, clock, drift cutoff, and label generators limit interpretation.

ds000102 · contrast space overlap · label reference interval · hemodynamic response function · QR factorization · replication · synthetic labels · flanker task