Voice projections after acoustic control did not confirm animal sound transfer in 16 iEEG patients and 16 fMRI adults

We tested whether a direction learned from human voices gives higher scores to animal sounds than to environmental sounds absent from fitting. Separate human studies supplied signed intracranial field means and BOLD fMRI. An 88-feature acoustic fit was removed before learning each reader. Four people per primary study supplied development. A local core lock preceded new endpoint tests in 16 patients and 16 MRI adults. The fixed rule required positive mean transfer and held-voice validation in both studies. It failed, p=0.438765. Mean transfer was +0.308, with 95% person Student limits [-0.210,+0.826], in iEEG and +0.321 [+0.234,+0.408] in primary fMRI. Electrical reader validation was also uncertain. Eight adults from a second MRI acquisition gave +0.136 [-0.235,+0.507]. Complete shared-sound and person refits gave transfer ranges across zero in every study. Negative people, baseline checks, acoustic controls and source-label limits remain. Close prior animal-transfer SCI work was found after primary scoring and is cited. This is a new controlled method benchmark in reused studies, not independent prospective confirmation. The package gives permitted compact inputs, exact source pins, code, complete results, figures, source checks and repeat steps. It does not establish positive cross-modal generalization, absent animal information, or causality.

human auditory cortex · intracranial EEG · fMRI · voice · animal sounds · acoustic control · negative result · reproducibility