With the first error fixed, mean recall matches after that error exceeded the uniform random order mean in the first nine positions of 9- and 13-digit lists in 86 ds003838 v1.0.6 participants

This study tested digit order after the first recall error in OpenNeuro ds003838 v1.0.6. It used 6,163 digit transcripts from the 9-digit and 13-digit conditions in 86 participants. The main test scored positions 1 through 9 in both conditions. It fixed the correct prefix, first error digit, and empty positions. It calculated the exact mean for uniform orders of the remaining recorded digits. The equal-condition, equal-participant recorded count exceeded this mean by 0.270 matches per trial. The 95% participant bootstrap interval was 0.234 to 0.306. The change from 9-digit to 13-digit lists was -0.032, with an interval of -0.081 to 0.020. This does not establish equal condition effects. All planned checks retained a positive excess. A separate exact calculation checked all 9,253 digit transcripts, including the descriptive 5-digit condition. Complete small permutation sets checked both reference rules in 5,310 cases. A fresh download matched all 183 source files. All 11 result files were byte equal in two complete runs. The package includes source files, hashes, an exact score tool, a proof, results, bootstrap arrays, figures, and reproduction steps. The local plan was fixed before the new scores but was not externally registered. The result measures recorded order relative to a stated reference. It does not identify a neural cause or stored item count.

serial recall · digit span · working memory · recall order · conditional permutation · random order reference · participant bootstrap · behavioral data