Other nine-digit targets retained 91% of the mean unordered overlap score in 86 ds003838 version 1.0.6 participants

We tested how much a digit recall score depends on its target list. We used 3,085 nine-digit responses from 86 participants in OpenNeuro ds003838 version 1.0.6. The unordered overlap score counted matching digit tokens and ignored order. For each response, the control used every other retained target from the same participant and list length. Other targets retained 91.1% of the mean unordered score and 18.3% of the mean correct-position score. The paired difference was 72.8 percentage points. Its 95% participant bootstrap interval was 71.8 to 73.7 points, from 20,000 samples. Every nine-digit target used nine different digits from an alphabet of ten digits. An exact proof showed that the unordered score for a fixed response can change by at most one digit across such targets. Planned checks gave similar results. The control measures dependence on the target; it does not estimate the number of remembered digits. The package provides a score audit for later neuroscience work. It includes unchanged source inputs, code, trial scores, pair scores, bootstrap draws, figures, and retrieval instructions. A separate calculation checked all 331,961 primary target-response pairs. A full offline run reproduced all 14 result and figure files exactly. All 183 source files matched a fresh download at the fixed source commit. No physiological signal was analyzed.

digit span · serial recall · unordered overlap · target replacement · behavioral measurement · bootstrap