{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,11,7]],"date-time":"2026-11-07T02:31:09Z","timestamp":1794018669056,"version":"4.8.0"},"reference-count":31,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,7,20]],"date-time":"2026-07-20T00:00:00Z","timestamp":1784505600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Anima Institute","award":["12\/2025"],"award-info":[{"award-number":["12\/2025"]}]}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Neurosci."],"abstract":"<jats:p>\n                    Modern deep learning has broadened the tools available for non-invasive neural decoding, but its advantage over well-engineered classical pipelines remains unclear at clinical neural-engineering sample sizes. We compared four classical decoders, three multilayer perceptron (MLP) variants, and three deep-learning architectures (an EEGNet-style compact convolutional network, a four-layer Transformer encoder trained from scratch, and a graph attention network) on the public WAY-EEG-GAL grasp-and-lift dataset (12 participants, 3,528 trials). Models were evaluated using leave-one-subject-out (LOSO) cross-validation to decode object weight (165, 330, and 660\u202fg) and grasp-surface friction (sandpaper, suede, and silk). After Benjamini\u2013Hochberg false discovery rate (BH-FDR) correction within the primary\/robustness family, none of the deep-\n                    <jats:italic>versus<\/jats:italic>\n                    -best-classical comparisons reached significance. The graph attention network led nominally on weight (0.643), and the compact convolutional network (CNN) led nominally on surface (0.565), but neither exceeded the best classical baselines (HGBM\u202f=\u202f0.617 for weight; logistic regression\u202f=\u202f0.562 for surface). Two-direction bandwidth controls showed that the nominal graph neural network (GNN) advantage on weight reflected access to higher-frequency electromyography (EMG) content rather than a robust architectural gain. The GNN weight signal was concentrated in the first 500\u202fms of sustained hold (0.639 early\n                    <jats:italic>vs<\/jats:italic>\n                    . 0.540 late,\n                    <jats:italic>q<\/jats:italic>\n                    \u202f=\u202f0.007), whereas the compact CNN was comparatively robust to bandwidth and phase. A four-layer Transformer trained from scratch underperformed (\n                    <jats:italic>q<\/jats:italic>\n                    \u202f=\u202f0.003 for both tasks), consistent with a parameter\u2013data mismatch at\n                    <jats:italic>n<\/jats:italic>\n                    \u202f=\u202f12. Modality ablation showed that both tasks were dominated by peripheral EMG features: electroencephalography (EEG)-only decoding was near chance for weight (0.34\u20130.37) and surface (0.36\u20130.38; all\n                    <jats:italic>q<\/jats:italic>\n                    \u202f&amp;lt;\u202f0.0015\n                    <jats:italic>vs<\/jats:italic>\n                    . EMG-only). EMG-only HGBM achieved the highest accuracy in the study (0.712 for weight and 0.584 for surface), and adding EEG channels reduced HGBM weight decoding (\n                    <jats:italic>q<\/jats:italic>\n                    \u202f=\u202f0.007) and CNN surface decoding (\n                    <jats:italic>q<\/jats:italic>\n                    \u202f=\u202f0.012). In contrast, the graph attention network was robust to EEG-induced fusion dilution (\n                    <jats:italic>q<\/jats:italic>\n                    \u202f&amp;gt;\u202f0.5), consistent with attention-based down-weighting of low-information channels. Together, bandwidth, modality, phase, and conditioning controls indicate that, at clinical sample sizes, the critical design choice is not architectural complexity but modality-relevant channel selection and the ability to suppress low-information channels.\n                  <\/jats:p>","DOI":"10.3389\/fnins.2026.1874302","type":"journal-article","created":{"date-parts":[[2026,7,20]],"date-time":"2026-07-20T05:35:09Z","timestamp":1784525709000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":2,"title":["Architecture-data matching for EEG-EMG decoding: compact deep models match classical spectral decoders on the WAY-EEG-GAL grasp-and-lift dataset"],"prefix":"10.3389","volume":"20","author":[{"given":"Osmar","family":"Pinto Neto","sequence":"first","affiliation":[{"name":"Biomedical Engineering Postgraduate Program, Anhembi Morumbi University","place":["S\u00e3o Jos\u00e9 dos Campos, Brazil"]},{"name":"Department of Kinesiology, California State University San Marcos (CSUSM)","place":["San Marcos, CA, United States"]},{"name":"Neurometra","place":["Carlsbad, CA, United States"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,7,20]]},"reference":[{"key":"ref1","article-title":"EEG-MFTNet: an enhanced EEGNet architecture with multi-scale temporal convolutions and transformer fusion for cross-session motor imagery decoding","volume-title":"Arxiv (Preprint)","author":"Andrikopoulos","year":"2026"},{"key":"ref2","doi-asserted-by":"publisher","first-page":"21002","DOI":"10.3389\/FNINS.2012.00039","article-title":"Filter bank common spatial pattern algorithm on BCI competition IV datasets 2a and 2b","volume":"6","author":"Ang","year":"2012","journal-title":"Front. Neurosci."},{"key":"ref3","doi-asserted-by":"publisher","first-page":"920","DOI":"10.1109\/TBME.2011.2172210","article-title":"Multiclass brain-computer interface classification by Riemannian geometry","volume":"59","author":"Barachant","year":"2012","journal-title":"IEEE Trans. Biomed. Eng."},{"key":"ref4","doi-asserted-by":"publisher","first-page":"133","DOI":"10.3390\/NEUROLINT17090133","article-title":"High-accuracy classification of Parkinson\u2019s disease using ensemble machine learning and stabilometric biomarkers","volume":"17","author":"Carolina Brisola Brizzi","year":"2025","journal-title":"Neurol. Int."},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1016\/j.jvoice.2024.04.020","article-title":"A temporally augmented graph attention network for affordance classification","author":"Chopra","year":"2026","journal-title":"Arxiv (Preprint)"},{"key":"ref6","doi-asserted-by":"publisher","first-page":"3548","DOI":"10.1152\/JN.00325.2010","article-title":"Identification of oscillations in muscle activity from surface EMG: reply to Halliday and Farmer","volume":"103","author":"Christou","year":"2010","journal-title":"J. Neurophysiol."},{"key":"ref7","doi-asserted-by":"publisher","first-page":"917","DOI":"10.1113\/JPHYSIOL.1995.SP021104","article-title":"Synchronization between motor cortex and spinal motoneuronal pool during the performance of a maintained motor task in man","volume":"489","author":"Conway","year":"1995","journal-title":"J. Physiol."},{"key":"ref8","first-page":"1061","author":"Demir","year":"2021"},{"key":"ref9","doi-asserted-by":"publisher","first-page":"3547","DOI":"10.1152\/jn.00222.2010","article-title":"On the need for rectification of surface EMG","volume":"103","author":"Halliday","year":"2010","journal-title":"J. Neurophysiol."},{"key":"ref10","doi-asserted-by":"publisher","first-page":"345","DOI":"10.1038\/NRN2621","article-title":"Coding and use of tactile signals from the fingertips in object manipulation tasks","volume":"10","author":"Johansson","year":"2009","journal-title":"Nat. Rev. Neurosci."},{"key":"ref11","doi-asserted-by":"publisher","first-page":"17701","DOI":"10.1109\/TNNLS.2023.3307470","article-title":"Graph neural networks on SPD manifolds for motor imagery classification: a perspective from the time-frequency analysis","volume":"35","author":"Ju","year":"2023","journal-title":"IEEE Trans. Neural Networks Learn. Syst."},{"key":"ref12","doi-asserted-by":"publisher","first-page":"8838","DOI":"10.1523\/JNEUROSCI.20-23-08838.2000","article-title":"Human cortical muscle coherence is directly related to specific motor parameters","volume":"20","author":"Kilner","year":"2000","journal-title":"J. Neurosci."},{"key":"ref13","doi-asserted-by":"publisher","first-page":"493","DOI":"10.1109\/TNSRE.2024.3355750","article-title":"Graph neural network-based EEG classification: a survey","volume":"32","author":"Klepl","year":"2024","journal-title":"IEEE Trans. Neural Syst. Rehabil. Eng."},{"key":"ref14","doi-asserted-by":"publisher","first-page":"056013","DOI":"10.1088\/1741-2552\/AACE8C","article-title":"EEGNet: a compact convolutional neural network for EEG-based brain-computer interfaces","volume":"15","author":"Lawhern","year":"2018","journal-title":"J. Neural Eng."},{"key":"ref15","doi-asserted-by":"publisher","first-page":"100","DOI":"10.3389\/FNHUM.2019.00100","article-title":"Corticomuscular coherence and its applications: a review","volume":"13","author":"Liu","year":"2019","journal-title":"Front. Hum. Neurosci."},{"key":"ref16","doi-asserted-by":"publisher","first-page":"031005","DOI":"10.1088\/1741-2552\/AAB2F2","article-title":"A review of classification algorithms for EEG-based brain-computer interfaces: a 10 year update","volume":"15","author":"Lotte","year":"2018","journal-title":"J. Neural Eng."},{"key":"ref17","doi-asserted-by":"publisher","first-page":"140047","DOI":"10.1038\/SDATA.2014.47","article-title":"Multi-channel EEG recordings during 3,936 grasp and lift trials with varying weight and friction","volume":"1","author":"Luciw","year":"2014","journal-title":"Sci. Data"},{"key":"ref18","doi-asserted-by":"publisher","first-page":"190","DOI":"10.1016\/j.jneumeth.2011.11.001","article-title":"Rectification of the EMG is an unnecessary and inappropriate step in the calculation of Corticomuscular coherence","volume":"205","author":"McClelland","year":"2012","journal-title":"J. Neurosci. Methods"},{"key":"ref19","doi-asserted-by":"publisher","first-page":"036008","DOI":"10.1088\/1741-2560\/12\/3\/036008","article-title":"Power spectrum of the rectified EMG: when and why is rectification beneficial for identifying neural connectivity?","volume":"12","author":"Negro","year":"2015","journal-title":"J. Neural Eng."},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1016\/J.JVOICE.2024.04.020","article-title":"Harnessing voice analysis and machine learning for early diagnosis of Parkinson\u2019s disease: a comparative study across three datasets","author":"Neto","year":"2024","journal-title":"J. Voice"},{"key":"ref21","doi-asserted-by":"publisher","first-page":"1093","DOI":"10.1152\/JN.00792.2009","article-title":"Rectification of the EMG signal impairs the identification of oscillatory input to the muscle","volume":"103","author":"Neto","year":"2010","journal-title":"J. Neurophysiol."},{"key":"ref22","doi-asserted-by":"publisher","first-page":"255","DOI":"10.1007\/S00421-010-1498-3","article-title":"Task failure during standing heel raises is associated with increased power from 13 to 50 Hz in the activation of triceps surae","volume":"110","author":"Pereira","year":"2010","journal-title":"Eur. J. Appl. Physiol."},{"key":"ref23","doi-asserted-by":"publisher","first-page":"051001","DOI":"10.1088\/1741-2552\/AB260C","article-title":"Deep learning-based electroencephalography analysis: a systematic review","volume":"16","author":"Roy","year":"2019","journal-title":"J. Neural Eng."},{"key":"ref24","doi-asserted-by":"publisher","first-page":"5391","DOI":"10.1002\/HBM.23730","article-title":"Deep learning with convolutional neural networks for EEG decoding and visualization","volume":"38","author":"Schirrmeister","year":"2017","journal-title":"Hum. Brain Mapp."},{"key":"ref25","doi-asserted-by":"publisher","first-page":"214223","DOI":"10.1109\/ACCESS.2025.3645762","article-title":"Investigating the impact of rational dilated wavelet transform on motor imagery EEG decoding with deep learning models","volume":"13","author":"Siino","year":"2025","journal-title":"IEEE Access"},{"key":"ref26","doi-asserted-by":"publisher","first-page":"100961","DOI":"10.1016\/J.ARRAY.2026.100961","article-title":"Ratiowavenet: a learnable wavelet-based framework for robust and interpretable electroencephalogram motor imagery classification","volume":"30","author":"Siino","year":"2026","journal-title":"Array"},{"key":"ref27","doi-asserted-by":"publisher","first-page":"710","DOI":"10.1109\/TNSRE.2022.3230250","article-title":"EEG conformer: convolutional transformer for EEG decoding and visualization","volume":"31","author":"Song","year":"2023","journal-title":"IEEE Trans. Neural Syst. Rehabil. Eng."},{"key":"ref28","first-page":"5998","volume":"30","author":"Vaswani","year":"2017","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"ref29","author":"Veli\u010dkovi\u0107","year":"2018"},{"key":"ref30","doi-asserted-by":"publisher","first-page":"331","DOI":"10.1007\/S13534-023-00340-5","article-title":"High and low pitch sound stimuli effects on heart-brain coupling","volume":"14","author":"von Jakitsch","year":"2023","journal-title":"Biomed. Eng. Lett."},{"key":"ref31","doi-asserted-by":"publisher","first-page":"4134","DOI":"10.1109\/JBHI.2025.3622725","article-title":"DBConformer: dual-branch convolutional transformer for EEG decoding","volume":"30","author":"Wang","year":"2026","journal-title":"IEEE J. Biomed. Health Inform."}],"container-title":["Frontiers in Neuroscience"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fnins.2026.1874302\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,20]],"date-time":"2026-07-20T05:35:11Z","timestamp":1784525711000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fnins.2026.1874302\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,20]]},"references-count":31,"alternative-id":["10.3389\/fnins.2026.1874302"],"URL":"https:\/\/doi.org\/10.3389\/fnins.2026.1874302","relation":{},"ISSN":["1662-453X"],"issn-type":[{"value":"1662-453X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,20]]},"article-number":"1874302"}}