With nine targets per block and binary responses, mean stored ds003500 inhibition accuracy exceeds planned accuracy by 0.55–1.16 percentage points

We tested stored behavior scores in OpenNeuro ds003500 version 1.2.0. The study uses 149 inhibition runs from 38 people. The source scores some segments under the performed rule, which differs from the planned rule. We found 40 such segments in 15 people. Integer bounds use 18 trials and nine targets per block. On 31,682 scored trials, the mean stored accuracy was 95.73%. The mean planned accuracy was 94.57% to 95.18%. Thus the stored score exceeded planned accuracy by 0.55 to 1.16 percentage points, with equal person weights. A third response outcome, wrong under both rules, left every block bound unchanged. Complete blocks also gave a positive excess. These are finite-sample bounds under stated count constraints, not population confidence intervals. The source authors already describe the score definition. This work adds its numerical consequences and a reusable bound method. The package preserves 312 source files, fixed source hashes, code, endpoint allocations, figures, and reproduction commands. A fresh source download reproduced all seven analysis outputs byte for byte. The result does not establish a brain or diagnosis effect.

response inhibition · accuracy bounds · planned and performed rules · behavioral scoring · OpenNeuro · ds003500