Exact position and distance controls fixed adjacency scores in 176 of 189 backward digit order errors in ds008037 v1.0.1

The study tested adjacent pair order in wrong backward digit answers from OpenNeuro ds008037 version 1.0.1. It used 189 complete orders of distinct target digits from 88 people. The exact reference fixed every correct position and the total absolute position error. It forbade extra correct positions. It fixed the adjacency score in 176 answers. Only 13 answers from 11 people had a variable reference score. A proof shows that correct positions alone fix adjacency for all order errors with exactly two or three wrong positions. This rule covered 163 source errors. The observed adjacency fraction exceeded the exact uniform reference by 0.405 percentage points. The 95 percent person-bootstrap interval was 0.152 to 0.698 points. Removing the distance control changed the sign and gave an interval that included zero. The first-error check also gave an interval that included zero. The result applies to the stated reference and selected records. It does not identify a memory strategy or brain mechanism. Direct permutations checked every empirical reference pattern. A clean run from fresh source files reproduced all numeric results byte for byte. The package contains unchanged source data, code, exact counts, a proof, all fixed checks, random draws, provenance, and instructions.

backward digit span · working memory · serial recall · adjacency · order errors · exact reference · position error · OpenNeuro