Under uniform target-order control, alignment gave a larger 9-to-13-digit recall score decrease for positions 1–9 in 86 ds003838 v1.0.6 participants

We compared position scoring with sequence alignment in OpenNeuro ds003838 version 1.0.6. We used 6,163 digit responses from 86 participants. We scored target positions 1 through 9 in the 9-digit and 13-digit conditions. Each scored target had distinct digits. We held each response fixed and calculated exact mean scores under uniform target orders. The adjusted position score decreased by 0.696 points. The adjusted alignment score decreased by 0.880 points. The paired difference was -0.184 points, with a participant bootstrap 95% interval of [-0.273, -0.096]. All planned checks with nine target positions gave negative differences. The first-five-position check gave +0.065 points, with a descriptive interval of [0.010, 0.120]. The score comparison thus depended on the positions that were scored. These scores do not identify memory capacity or error causes. The conditions also differed in task time. The package preserves raw source files, fixed source hashes, code, exact control histograms, resamples, figures, and independent checks. A repeat run on fresh source copies reproduced all checked result and figure files byte for byte.

digit recall · working memory · sequence alignment · longest common subsequence · position scoring · exact conditional control · participant bootstrap