Adding savings changed 7 of 32 learner labels with three groups under standardized Manhattan distance and complete linkage for screened ds004021 v1.0.1 raw events

We tested how savings changes learner groups in OpenNeuro ds004021 v1.0.1. Savings was day 1 early error minus day 2 early error. Early error was the circular mean of the first 32 rotation trials. Accepted trials had response times from 0.1 to 2.0 seconds. The test fixed complete linkage, three groups, and standardized Manhattan distance. Adding savings changed 7 of 32 labels after the best label match. The adjusted Rand index was 0.548656. Both score matrices had rank 2. An exact proof shows the added distance term. Covariance distance with the full sample covariance and its Moore-Penrose inverse preserved distances under stated rank assumptions. The package contains raw events, source hashes, code, 2,000 trial draws, figures, and checks. This raw-event audit does not test the source paper's neural findings.

motor adaptation · learner groups · savings · Manhattan distance · complete linkage · metric choice · OpenNeuro · reproducibility