Removing adjacent direction repeats changed the random trial response time contrast in ds008017 v1.0.0 runs 3 and 4 under shared context effects

This study tested a response time contrast in 15 people from OpenNeuro ds008017 v1.0.0. The model assumed shared physical triplet effects and gave each person total weight one. The planned late-run contrast was 14.08 ms. Its 95 percent person bootstrap interval was 9.14 to 19.13 ms. A planned check removed adjacent direction repeats. Its contrast was -7.30 ms, with interval -15.03 to -0.12 ms. The exploratory paired change was 21.38 ms, with interval 13.69 to 28.82 ms. The small negative estimate included zero under a cluster interval. An exact class identity showed that unrestricted personal context effects absorb the class. A second cycle relation showed how repeat selection changes the overlap between two matches. These results do not prove a neural cause or growth in learning. The package preserves 125 source files, a trial ledger, code, bootstrap draws, proofs, and instructions. A fresh retrieval and full replay reproduced all 27 result files byte for byte.

sequence learning · response time · trial selection · repetition · alternating serial reaction time · bootstrap