Manual marble-burying scoring (a single end-point marble count) is highly variable between observers and discards the temporal/behavioral detail needed for reliable cross-lab comparison.

Standardize the marble-burying test to improve reproducibility: 1) Use a 26.6 x 42.5 x 18.5 cm cage with bedding roughly 4 cm deep. 2) Arrange 20 blue glass marbles in 4 evenly spaced rows. 3) Run a 30-minute session recorded on video at 25 fps rather than relying on a single end-of-session count. 4) Score recordings with a machine-learning classifier (e.g. JAABA) to quantify individual burying bouts, bout duration, and spatial heatmaps instead of just a final marble tally, which better distinguishes genuine burying behavior from incidental marble displacement.