One iEEG panel and eight fMRI adults showed small pattern transfer across labels for 49 sounds, but centered BOLD error exceeded baseline
We tested whether maps learned from eight sound labels transfer within-label response patterns to four excluded speech and music labels. Separate public studies supplied one fixed panel of 32 distinct iEEG profiles and eight fMRI adults. No test-label sound entered map fitting or scale estimation. Category and scaled spatial means were removed. Mean centered pattern scores for 49 test sounds were 0.16359 for electrical targets and 0.07834 for BOLD targets. Joint test-sound and fMRI-person 95 percent ranges were 0.09985 to 0.22178 and 0.04116 to 0.11589. Both directed label tests passed; shared p was 0.0001. All eight fMRI records had positive means in both directions, but they reused the same electrical panel. The acoustic model scored higher on average in both targets. Mean centered error scores were +0.01051 for electrical targets and -0.32947 for BOLD. All 32 BOLD person-category cases had worse squared error than a zero residual prediction. Fixed checks, negative outcomes, and a later full-refit uncertainty check remain in the package. This is a new endpoint in reused studies, not new-patient electrical confirmation. Missing electrical patient labels, source site selection, related acoustic features across labels, and measurement differences limit scope. The result supports small pattern transfer under the stated label and panel conditions. It does not establish a common cause or useful BOLD calibration. The package gives code, source pins and terms, plans, results, figures, audits, and repeat instructions.
human auditory cortex · intracranial EEG · fMRI · sound categories · cross-modal prediction · calibration · reproducible research
References
4- usesSound relationships after category removal agreed between a fixed iEEG panel and eight fMRI adults sci:sha256:952a3d810a1e0d48021b88ea47f485459d86862c83bacd97cf3f4568e9c5526b
Its residual auditory geometry motivated the new excluded-label transfer test. We reuse its verified source pins and attributed MIT fetch script, but fit new neural maps.
- usesCombined sound features gave positive mean gains over both single sets in 11 iEEG patients and eight fMRI adults sci:sha256:d3f382a918d33ee6d84d34c4e83b6ac1b02dc9350f35af96200eb63c734ddb9a
Its model comparison and source documentation informed the acoustic comparator and measured-site scope; its endpoint and iEEG score table do not enter this analysis.
- usesdoi:10.1016/j.neuron.2015.11.035
Public author response matrix at commit b41b99ff620d7997dbf278fc770f58566707726a supplies separate fMRI training and third-session test responses.
- usesdoi:10.1016/j.cub.2022.01.069
Public author release at commit 03bb41e95d04c191acd3b76b1c11d7185096f1f3 supplies selected single-electrode responses, source labels and acoustic features. Prior work established component prediction across modalities.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- Fixed acoustic metrics did not pass a shared transfer test in 20 iEEG patients and 16 fMRI adults
Related cross-label transfer benchmark; the present test transfers acoustic metrics between modalities with person holdouts.
- usesClipped acoustic fits did not pass a shared 1% voice pattern error test in 20 iEEG patients and 16 fMRI adults
Its distinction between pattern alignment and calibrated error motivated the new primary endpoint.