In 22 people from ds004117, an additive model gives a 34 ms previous memory versus ignored letter contrast with a 95% interval that includes zero

This study tests correct negative probe responses in OpenNeuro ds004117 version 1.0.1. It compares letters from the previous memory set with letters from the previous ignored set. Both groups contain letters shown in the previous sequence. The fixed additive model gives equal total weight to each person. The primary sample has 830 trials from 22 people. The adjusted contrast is 34.3 ms. The 95% person bootstrap interval is -105.4 to 162.6 ms, from 20,000 samples. The interval includes zero and substantial differences in both directions. All nine planned checks also have intervals that include zero. Including three responses above five seconds raises the estimate to 141.1 ms. One person has only recorded no responses. The planned estimate with that person omitted is 39.2 ms. The package preserves all 176 fixed source files and hashes. It also preserves the local plan, trial history labels, code, random samples, results, and instructions. Independent parsers and weighted fits agree. A clean run reproduced all 69 analysis files by exact byte comparison. A full fresh-source run with a separate numerical build passed numerical checks for all files. The result assumes independent person records and a common additive model. It is conditional on correct current responses. The observed histories do not identify a causal effect or a brain process.

verbal working memory · negative probes · task relevance · response time · trial history · OpenNeuro · person bootstrap