A fixed word recency test did not confirm shared suppression in six iEEG patients and 28 fMRI adults

We tested whether recent occurrence of the same word predicts a smaller neural response during story listening. Separate public studies supplied iEEG high-gamma amplitude and auditory BOLD fMRI. A local plan and code lock preceded new scores in six reserved patients and 28 reserved MRI adults. The same MRI adults heard Merlin and Sherlock. Controls included common-word identities, sound, timing, cumulative occurrence and next occurrence. Mean past partial correlations were -0.00820, +0.00902 and +0.00761. Their 95% person Student limits were [-0.01897,0.00258], [-0.03010,0.04814] and [-0.02513,0.04035]. The required past-minus-future scores also had limits across zero. The fixed shared suppression rule failed, p=0.85584. All negative people and controls remain. An exploratory exact identity expresses complete-link past-minus-future features as changes in weighted intervals between word occurrences. A no-memory model with identical responses at every word gave directional scores after changes to clock time or the response filter. These are model checks, not fitted human causes. The contribution is a numerical recurrence dataset with 10062 rows, a negative benchmark and tools to test filter errors. Source selection, sparse clinical coverage, broad auditory MRI means, different cohorts, missing alignments and fixed response filters limit scope. No positive cross-modal neural generalization or causal effect was established. The package gives source pins and terms, code, all results, figures and repeat instructions.

human neuroscience · word repetition · natural language · iEEG · fMRI · recency · response filter · negative result · public data · reproducibility