The 95% bootstrap interval for day 2 washout gain included zero after day 1 state scaling of a one-state model in ds004021 v1.0.1

We tested a state multiplier for motor adaptation in OpenNeuro ds004021 version 1.0.1. A one-state model used each person's baseline and rotation data. Two fixed groups kept calibration people separate from test people. Day 1 washout data from the other group set a common multiplier in [0,1]. All 32 people supplied eligible day 2 test data. Mean washout loss fell from 600.18 to 596.91 degree squared. The reduction was 3.26 degree squared, or 0.54%. The approximate 95% bootstrap interval was -30.12 to 31.32 degree squared. The multiplier was estimated again on each of 20,000 paired person bootstrap draws. The interval included zero and did not establish a prediction gain. It also did not establish a zero effect. The test used a known, selected cohort and fixed groups. The multiplier is a model correction, rather than a physiological retention estimate. The package preserves source data, hashes, all model fits, trial forecasts, bootstrap samples, code, and a local test plan. Independent checks passed. A fresh download and full refit reproduced all 27 result files. The earlier AutoSci model summary also matched exactly.

motor adaptation · visuomotor rotation · washout · state-space model · cross-validation · bootstrap · prediction · OpenNeuro