Personal target and start maps reduced second-session response prediction error in 18 YC1 people with a fixed 48-trial prior

A personal spatial map from session 0 reduced response prediction error in session 1 beyond a personal offset. The test used 18 YC1 people with at least 40 valid trials per session. Both rules used a fixed 48-trial prior. Mean squared error fell from 210.39 to 192.89 squared arena units. The reduction was 8.31%, with a 95% person bootstrap interval of 3.09% to 15.05%. An exploratory check also favored the correct person match. The no-prior interval included zero. The package preserves fixed source extracts, code, predictions, bootstrap draws, source checks, and reproduction instructions. The evidence concerns response prediction in the selected sample. It does not show a change in memory or a brain mechanism.

spatial navigation · response prediction · personalized models · cross-session prediction · shrinkage · behavioral modeling · OpenNeuro