A 95% person interval included zero for the 20-degree previous-minus-next target coefficient in 121 ds008037 v1.0.1 adults

We tested whether orientation reports had a larger association with the previous target than with the next target. The source was OpenNeuro ds008037 version 1.0.1. We used 121 adults and 7,018 trials. The local plan was saved before the first study score. It was not registered with an external service. Each person had a separate model with a fixed 20-degree curve. The outcome was a bounded sine error score in degree units. It was not the raw angle error. The mean previous-minus-next coefficient was 0.933. Its paired 95% interval from 20,000 person draws was -0.207 to 2.092. The interval did not meet the saved rule for a positive difference. It did not establish equal coefficients or no history effect. Wider curve checks had small positive lower limits. The interval result thus depended on curve choice. The next target was a timing control, not a causal control. The load difference interval also included zero. All 250 source files matched a fresh public download. A separate implementation checked all coefficients and intervals. Reversing the sequences exchanged the previous and next coefficients. A repeat analysis gave 26 byte-identical output files. The package preserves source data, version and hashes, trial links, exclusions, coefficients, complete person draws, figures, and reproduction code.

working memory · serial dependence · orientation report · trial history · circular statistics · person bootstrap · OpenNeuro ds008037