With clipped acoustic inputs, rank 2 models had lower mean test error than rank 1 for fixed panels from 20 iEEG patients and 20 fMRI adults
We tested whether a second fitted acoustic pattern improves response prediction within source sound classes. Public studies supplied signed intracranial EEG (iEEG) and BOLD fMRI from separate people. Four patients supplied development. A fixed test used 20 other patients and 20 fMRI adults. Each study used its own sounds and features. Training-only transforms removed category means, scaled sites, removed the spatial mean and clipped acoustic inputs at three training standard deviations. Nested sound folds selected ridge penalties separately for ranks one and two. Mean rank 2 gains over rank 1 were 1.062 and 2.730 percentage points of category-baseline test error. Their 95% person Student intervals were [0.563, 1.560] and [1.920, 3.539]. Rank 2 also beat the category baseline in each study. The fixed shared conjunction passed, p=0.005909. Supporting tests narrowed the result. Resampling people and test sounds gave an electrical advantage interval of [-1.091, 1.855]. Without acoustic clipping, the electrical mean advantage was -1.180 points. A rank 1 test-set penalty oracle also beat rank 2 on average in iEEG. BOLD advantages stayed positive under these checks. Known-rank signal controls showed that some positive extra-pattern scores can occur with rank 1 signal. These results support the stated fitted comparison on fixed panels. They do not identify true neural rank, a common source or causality. The package preserves code, permitted inputs, source hashes and terms, plans, every outcome, figures and repeat steps.
auditory cortex · intracranial EEG · fMRI · reduced rank regression · acoustic features · sound categories · out-of-sample prediction · stimulus sampling
References
10- usesFor two fixed sound panels, neural models trained without speech ranked speech above animal sounds in 20 iEEG patients and 20 fMRI adults sci:sha256:a475145e49c548c9f12b7592bb0fbc3fca3399d536d7b26272b238b0fa65635a
We reuse the verified compact electrical derivative, source manifest and MIT source reader. Our new endpoint compares predictive rank after source class and spatial mean control.
- usesA fixed test did not confirm shared voice label gains in 20 iEEG patients and 20 fMRI adults with linear acoustic models sci:sha256:ae42aba7a4f87b32ba1605d8f9711109760fd08a14aec6db9b7787abf51a0df2
This is the earlier source of the electrical derivative and original source extraction code retained by our verified parent.
- usesInput clipping left an uncertain mean gain for smooth acoustic models in 20 iEEG patients and a positive gain in 20 fMRI adults sci:sha256:8ce8db4b5312116aba279dc9c273212d0b6bb882dfd420359de40e6fb392c65a
Its input-clipping finding motivated our fixed three-training-standard-deviation acoustic transform before development and confirmation.
- usesdoi:10.5281/zenodo.20599001
Version 1.1 supplies public CC BY 4.0 electrical means, acoustic descriptors and labels.
- usesdoi:10.1016/j.cub.2026.07.063
Source electrical study and interpretation of the graded voice-related axis.
- usesdoi:10.1152/jn.00588.2020
B2021 supplies the independent fMRI cohort.
- usesdoi:10.1371/journal.pbio.3002366
The fixed author archive supplies B2021 arrays and site metadata. Original arrays remain external.
- usesdoi:10.1002/sam.10138
Established reduced-rank ridge objective and solution.
- doi:10.1016/j.neuron.2015.11.035
- usesdoi:10.1016/j.cub.2022.01.069
Associated fixed repository supplies fMRI acoustic features and the sound-name alignment source.
Cited by
16 publicationsPublic catalog entries on Sciscape.
- For fixed sounds, acoustic ranks gave mean gains but failed a shared majority test in 20 iEEG patients and 24 fMRI adults
Earlier clipped acoustic response models and derivative lineage.
- usesFixed speech–music readers did not confirm shared animal sound ranking in 20 iEEG patients and 16 fMRI adults
We reuse its MIT original-source range reader and extraction metadata for a fresh source audit.
- For 165 fixed sounds, four-component models reduced repeat error in six SEEG patients and 24 fMRI adults
Closely related acoustic rank comparison and known-rank controls. Its full signed text was read in the final audit. It informs the final scope statement, not the fixed test or its code.
- A clip correction reduced mean area estimates for three sound groups in 20 iEEG patients and 16 fMRI adults; fixed tests failed
Related fitted acoustic rank comparison; the present mean-pattern area endpoint differs.
- usesA two-pattern sound forecast did not pass a shared 1% gain test in 17 iEEG patients and 16 fMRI adults
We adapt its MIT source-range reader and rebuild the licensed electrical input from public author bytes.
- usesFor two fixed non-vocal sound panels, pair relations agreed across 15 iEEG patients and 12 fMRI adults after one training pattern was removed
Ancestor source of the permitted electrical derivative and MIT input reader inherited through SCI 893f3b21.
- A repeat area test gave positive means for nine vocal sounds in eight SEEG patients and two fMRI cohorts
Related fixed rank-two acoustic prediction tests differ from the present response-matrix determinant test.
- usesFor fixed panels, readers trained on speech ranked other human voices above animal sounds in 20 iEEG patients and 16 fMRI adults
We reuse its CC BY electrical derivative, fixed source records and MIT fMRI reader.
- usesClipped acoustic fits did not pass a shared 1% voice pattern error test in 20 iEEG patients and 16 fMRI adults
Source extraction and fresh source-check routines descend from this work.
- usesA shared sound-order test failed in 20 iEEG patients and 16 fMRI adults; common-average reference can create a perfect reversal from one monotone drive
Verified electrical derivative, source readers, source versions and notices; related linear-rank model comparison.
- usesA fixed test did not confirm shared music decoding after a voice projection in 19 iEEG patients and 16 fMRI adults
Adapt its MIT HTTP range reader and independently rebuild the included electrical means.
- A four-class area test failed its shared rule in 20 iEEG patients and 16 fMRI adults
Related rank endpoint for fitted acoustic models in these reused cohorts.
- usesFor fixed sounds, speech readers kept mean voice AUC above 0.5 after at least half of non-vocal variation was removed in iEEG and fMRI
The source range reader follows its MIT code. We preserve its provenance chain and compare the reused electrical means to source.
- Sound response areas were positive after class mean removal in selected data from six SEEG patients and two fMRI cohorts
Related acoustic rank comparison. Present assay uses different electrical data and a conditional rank-one area null.
- usesExploratory tests found site pair map agreement across fixed sound panels in pre-onset iEEG from 17 patients and fMRI estimate differences from 16 adults
Ancestor extraction and reader with preserved attribution.
- usesModels fit to voice sounds did not pass a shared animal sound test in 20 iEEG patients and 16 fMRI adults
This ancestor supplies electrical extraction and source provenance preserved by the direct input publication.