The same digit order in 334 ds003823 SART tables allowed perfect no-go prediction from trial number with correct trial alignment

All 334 SART trial tables in OpenNeuro ds003823 version 1.3.5 contain the same 225-digit order. The tables cover 189 people and 75,150 trials. We checked all files against the pinned public source and checked that the source selection was complete. A lookup from one person's pre visit used trial number to predict every digit and no-go label in 74,700 trials from 188 other people. Both visits of the training person were removed from the test. Correct trial alignment was required. A model that used five preceding digits also classified all go and no-go trials correctly. Different-order controls gave mean balanced accuracy 0.499933 for the fixed lookup. The analysis is exploratory, and the full corpus was inspected before the local plan. The results describe stored stimulus labels. They do not show that participants learned the order or that a brain signal predicted a stimulus. The package preserves source files, hashes, all model predictions, controls, independent checks, code, and repeat instructions.

Sustained Attention to Response Task · SART · stimulus schedule · trial order · no-go prediction · reproducibility · OpenNeuro