A fixed test did not confirm lower high-load response spread in 12 SEEG patients and 16 fMRI adults

We tested whether higher memory load reduces trial-to-trial response spread in separate human SEEG and fMRI studies. All six development people had lower high-load spread. A fixed test used 12 other SEEG patients and 16 other fMRI adults. The score was the high-minus-low Gini mean difference after a common, label-blind feature scale. Mean scores were -0.05224 in SEEG (95% person t interval -0.10010 to -0.00439) and +0.00211 in fMRI (-0.09700 to +0.10122). Both shared directional claims failed Holm correction, with p=1.00. A second fMRI study check in 11 complete IDs outside Study 1 was uncertain. SEEG history adjustment and discharge masking left intervals through zero. A negative fMRI encoding check did not supply maintenance evidence. A synthetic unequal-count check changed the inference plan before reserved scores: raw sign calibration did not match a mean null. People remained the statistical units. The study does not establish equal spread, a modality difference, a common neural cause, or causality. Clinical sampling, source region selection, adaptive task history, beta error, and distinct response scales limit scope. The package preserves code, permitted inputs, all results, plans, figures, source versions and rights, an arithmetic audit, and repeat instructions. Exact author BOLD trial rows remain external. This is a new-endpoint holdout in reused cohorts, not new-study replication.

working memory · SEEG · fMRI · response variability · Gini mean difference · negative results · reproducibility · subject-level inference