{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,21]],"date-time":"2026-09-21T06:02:46Z","timestamp":1789970566127,"version":"4.0.1"},"reference-count":69,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,9,21]],"date-time":"2026-09-21T00:00:00Z","timestamp":1789948800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Comput. Neurosci."],"abstract":"<jats:sec>\n                    <jats:title>Introduction<\/jats:title>\n                    <jats:p>Biologically motivated artificial visual learning requires local plasticity rules to be evaluated within architectures whose connectivity assumptions remain explicit. Convolutional Hebbian networks demonstrate that competitive local learning can support visual classification, but their performance is coupled to exact weight sharing, a prior that is not anatomically literal. This study asks whether Hebbian principal component analysis (HPCA) can support representations when spatial locality is preserved but exact translational weight sharing is removed.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>HPCA was evaluated on CIFAR-10 as the primary benchmark, with MNIST and STL-10 used as controls. We compared fully connected HPCA, position-specific locally connected HPCA with fixed receptive fields, non-shared synaptic weights, and population-wise divisive response normalization, and HPCA substitutions in two shared-kernel convolutional Hebbian benchmarks. Representations were learned without supervised feature-layer targets or BP-based feature-layer fine-tuning, frozen, and evaluated through supervised readouts.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>Position-specific locally connected HPCA outperformed the fully connected reference. On CIFAR-10, the best locally connected model reached 61.42 \u00b1 0.36%. Shared-kernel HPCA reached 74.51 \u00b1 0.38% in the modular benchmark with common HPCA preprocessing and 76.57 \u00b1 0.55% in the SoftHebb architecture with reference preprocessing.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion<\/jats:title>\n                    <jats:p>The results show that HPCA remains effective across position-specific locally connected and shared-kernel convolutional architectures, but that performance depends strongly on the surrounding architecture and benchmark protocol. The tested shared-kernel configurations achieved higher accuracy, yet useful frozen visual representations still emerged when exact parameter sharing was removed. Locally connected HPCA therefore provides a controlled model for examining how local plasticity interacts with receptive-field structure and convolutional inductive bias in biologically constrained visual learning.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fncom.2026.1943064","type":"journal-article","created":{"date-parts":[[2026,9,21]],"date-time":"2026-09-21T05:35:01Z","timestamp":1789968901000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Locality and weight sharing shape Hebbian principal-component learning in biologically constrained visual models"],"prefix":"10.3389","volume":"20","author":[{"given":"Patrick","family":"Inoue","sequence":"first","affiliation":[{"name":"KEIM Institute (Knowledge Engineering and Information Management), Albstadt-Sigmaringen University","place":["Albstadt, Germany"]},{"name":"Department of Computer Science, Chemnitz University of Technology","place":["Chemnitz, Germany"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Florian","family":"R\u00f6hrbein","sequence":"additional","affiliation":[{"name":"Department of Computer Science, Chemnitz University of Technology","place":["Chemnitz, Germany"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Andreas","family":"Knoblauch","sequence":"additional","affiliation":[{"name":"KEIM Institute (Knowledge Engineering and Information Management), Albstadt-Sigmaringen University","place":["Albstadt, Germany"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,9,21]]},"reference":[{"key":"B1","unstructured":"Abadi\n              M.\n            \n            \n              Agarwal\n              A.\n            \n            \n              Barham\n              P.\n            \n            \n              Brevdo\n              E.\n            \n            \n              Chen\n              Z.\n            \n            \n              Citro\n              C.\n            \n          \n          TensorFlow: Large-scale machine learning on heterogeneous systems\n          \n          2016"},{"key":"B2","doi-asserted-by":"publisher","first-page":"63","DOI":"10.1523\/JNEUROSCI.3874-05.2006","article-title":"A resilient, low-frequency, small-world human brain functional network with highly connected association cortical hubs","volume":"26","author":"Achard","year":"2006","journal-title":"J. Neurosci"},{"key":"B3","doi-asserted-by":"publisher","first-page":"284","DOI":"10.1364\/JOSAA.2.000284","article-title":"Spatiotemporal energy models for the perception of motion","volume":"2","author":"Adelson","year":"1985","journal-title":"J. Opt. Soc. Am. A"},{"key":"B4","doi-asserted-by":"publisher","first-page":"324","DOI":"10.1007\/978-3-030-30642-7_29","article-title":"\u201cHebbian learning meets deep convolutional neural networks?,\u201d","author":"Amato","year":"2019","journal-title":"Proceedings of ICIAP 2019, volume 11751 of Lecture Notes in Computer Science"},{"key":"B5","doi-asserted-by":"publisher","first-page":"196","DOI":"10.1162\/neco.1992.4.2.196","article-title":"What does the retina know about natural scenes?","volume":"4","author":"Atick","year":"1992","journal-title":"Neural Comput"},{"key":"B6","doi-asserted-by":"publisher","first-page":"51","DOI":"10.1016\/j.neunet.2016.07.006","article-title":"A theory of local learning, the learning channel, and the optimality of backpropagation","volume":"83","author":"Baldi","year":"2016","journal-title":"Neural Netw"},{"key":"B7","doi-asserted-by":"publisher","first-page":"e1007692","DOI":"10.1371\/journal.pcbi.1007692","article-title":"Learning to represent signals spike by spike","volume":"16","author":"Brendel","year":"2020","journal-title":"PLoS Comput. Biol"},{"key":"B8","doi-asserted-by":"publisher","first-page":"51","DOI":"10.1038\/nrn3136","article-title":"Normalization as a canonical neural computation","volume":"13","author":"Carandini","year":"2012","journal-title":"Nat. Rev. Neurosci"},{"key":"B9","doi-asserted-by":"publisher","first-page":"25","DOI":"10.3389\/fnsyn.2010.00025","article-title":"Voltage and spike timing interact in STDP \u2013 a unified model","volume":"2","author":"Clopath","year":"2010","journal-title":"Front. Synaptic Neurosci"},{"key":"B10","first-page":"215","article-title":"\u201cAn analysis of single-layer networks in unsupervised feature learning,\u201d","volume-title":"Proceedings of the fourteenth international conference on artificial intelligence and statistics","author":"Coates","year":"2011"},{"key":"B11","doi-asserted-by":"publisher","first-page":"1240348","DOI":"10.3389\/fncom.2024.1240348","article-title":"Information bottleneck-based hebbian learning rule naturally ties working memory and synaptic updates","volume":"18","author":"Daruwalla","year":"2024","journal-title":"Front. Comput. Neurosci"},{"key":"B12","doi-asserted-by":"publisher","first-page":"263","DOI":"10.1146\/annurev-vision-082114-035334","article-title":"Functional circuitry of the retina","volume":"1","author":"Demb","year":"2015","journal-title":"Ann. Rev. Vision Sci"},{"key":"B13","doi-asserted-by":"publisher","first-page":"60403","DOI":"10.52202\/085713-2020","article-title":"\u201cRethinking hebbian principle: Low-dimensional structural projection for unsupervised learning,\u201d","author":"Deng","year":"2025","journal-title":"Advances in neural information processing systems"},{"key":"B14","doi-asserted-by":"publisher","first-page":"415","DOI":"10.1016\/j.neuron.2012.01.010","article-title":"How does the brain solve visual object recognition?","volume":"73","author":"DiCarlo","year":"2012","journal-title":"Neuron"},{"key":"B15","doi-asserted-by":"publisher","first-page":"474","DOI":"10.1007\/s42979-026-04932-7","article-title":"Neuroplastic neural networks: adaptive learning through structural plasticity and Hebbian updates","volume":"7","author":"Fahyaz","year":"2026","journal-title":"SN Comput. Sci"},{"key":"B16","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1093\/cercor\/1.1.1","article-title":"Distributed hierarchical processing in the primate cerebral cortex","volume":"1","author":"Felleman","year":"1991","journal-title":"Cereb. Cortex"},{"key":"B17","doi-asserted-by":"publisher","first-page":"2038","DOI":"10.1152\/jn.01311.2006","article-title":"Model of birdsong learning based on gradient estimation by dynamic perturbation of neural conductances","volume":"98","author":"Fiete","year":"2007","journal-title":"J. Neurophysiol"},{"key":"B18","doi-asserted-by":"publisher","first-page":"194","DOI":"10.1162\/neco.1991.3.2.194","article-title":"Learning invariance from transformation sequences","volume":"3","author":"F\u00f6ldi\u00e1k","year":"1991","journal-title":"Neural Comput"},{"key":"B19","doi-asserted-by":"publisher","first-page":"47","DOI":"10.1098\/rspb.1993.0125","article-title":"Principal component analysis learning algorithms: a neurobiological analysis","volume":"254","author":"Friston","year":"1993","journal-title":"Proc. Biol. Sci"},{"key":"B20","doi-asserted-by":"publisher","first-page":"193","DOI":"10.1007\/BF00344251","article-title":"Neocognitron: a self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position","volume":"36","author":"Fukushima","year":"1980","journal-title":"Biol. Cybern"},{"key":"B21","doi-asserted-by":"publisher","first-page":"2901","DOI":"10.1016\/j.visres.2006.03.008","article-title":"Can the theory of \u201cwhitening\u201d explain the center-surround properties of retinal ganglion cell receptive fields?","volume":"46","author":"Graham","year":"2006","journal-title":"Vision Res"},{"key":"B22","article-title":"Local unsupervised learning for image analysis","author":"Grinberg","year":"2019","journal-title":"arXiv preprint arXiv:1908.08993"},{"key":"B23","volume-title":"The Organization of Behavior: A Neuropsychological Theory","author":"Hebb","year":"1949"},{"key":"B24","doi-asserted-by":"publisher","first-page":"181","DOI":"10.1017\/S0952523800009640","article-title":"Normalization of cell responses in cat striate cortex","volume":"9","author":"Heeger","year":"1992","journal-title":"Vis. Neurosci"},{"key":"B25","article-title":"The forward-forward algorithm: some preliminary investigations","author":"Hinton","year":"2022","journal-title":"arXiv preprint arXiv:2212.13345"},{"key":"B26","doi-asserted-by":"publisher","first-page":"19542","DOI":"10.1109\/ICASSP55912.2026.11464818","article-title":"\u201cHebbian learning with global direction,\u201d","author":"Hua","year":"2026","journal-title":"ICASSP 2026\u20132026 IEEE international conference on acoustics, speech and signal processing"},{"key":"B27","doi-asserted-by":"publisher","first-page":"106","DOI":"10.1113\/jphysiol.1962.sp006837","article-title":"Receptive fields, binocular interaction and functional architecture in the cat's visual cortex","volume":"160","author":"Hubel","year":"1962","journal-title":"J. Physiol"},{"key":"B28","doi-asserted-by":"publisher","first-page":"1705","DOI":"10.1162\/089976600300015312","article-title":"Emergence of phase- and shift-invariant features by decomposition of natural images into independent feature subspaces","volume":"12","author":"Hyv\u00e4rinen","year":"2000","journal-title":"Neural Comput"},{"key":"B29","article-title":"Constrained hebbian learning supports efficient representational allocation under structural constraints","author":"Inoue","year":"","journal-title":"arXiv preprint. arXiv:2607.16027"},{"key":"B30","article-title":"Guiding sparse neural networks with neurobiological principles to elicit biologically plausible representations","author":"Inoue","year":"","journal-title":"arXiv preprint arXiv:2603.03234"},{"key":"B31","article-title":"\u201cHebbian deep learning without feedback,\u201d","volume-title":"The eleventh international conference on learning representations","author":"Journ\u00e9","year":"2023"},{"key":"B32","doi-asserted-by":"publisher","first-page":"63","DOI":"10.3389\/fnana.2016.00063","article-title":"Structural plasticity, effectual connectivity and memory in cortex","volume":"10","author":"Knoblauch","year":"2016","journal-title":"Front. Neuroanat"},{"key":"B33","author":"Krizhevsky","year":"2009","journal-title":"Learning multiple layers of features from tiny images"},{"key":"B34","first-page":"1097","article-title":"\u201cImageNet classification with deep convolutional neural networks,\u201d","volume-title":"Advances in neural information processing systems","author":"Krizhevsky","year":"2012"},{"key":"B35","doi-asserted-by":"publisher","first-page":"633674","DOI":"10.3389\/fnins.2021.633674","article-title":"Scaling equilibrium propagation to deep ConvNets by drastically reducing its gradient estimator bias","volume":"15","author":"Laborieux","year":"2021","journal-title":"Front. Neurosci"},{"key":"B36","doi-asserted-by":"publisher","first-page":"6503","DOI":"10.1007\/s00521-021-06701-4","article-title":"Comparing the performance of hebbian against backpropagation learning using convolutional neural networks","volume":"34","author":"Lagani","year":"2022","journal-title":"Neur. Comput. Applic"},{"key":"B37","doi-asserted-by":"publisher","first-page":"127867","DOI":"10.1016\/j.neucom.2024.127867","article-title":"Scalable bio-inspired training of deep neural networks with FastHebb","volume":"595","author":"Lagani","year":"2024","journal-title":"Neurocomputing"},{"key":"B38","first-page":"26826","article-title":"\u201cRole of locality and weight sharing in image-based tasks: a sample complexity separation between CNNs, LCNs, and FCNs,\u201d","volume-title":"International conference on learning representations","author":"Lahoti","year":"2024"},{"key":"B39","doi-asserted-by":"publisher","first-page":"1332","DOI":"10.1126\/science.aab3050","article-title":"Human-level concept learning through probabilistic program induction","volume":"350","author":"Lake","year":"2015","journal-title":"Science"},{"key":"B40","doi-asserted-by":"publisher","first-page":"36","DOI":"10.1038\/236","article-title":"The metabolic cost of neural information","volume":"1","author":"Laughlin","year":"1998","journal-title":"Nat. Neurosci"},{"key":"B41","doi-asserted-by":"publisher","first-page":"2278","DOI":"10.1109\/5.726791","article-title":"Gradient-based learning applied to document recognition","volume":"86","author":"LeCun","year":"1998","journal-title":"Proc. IEEE"},{"key":"B42","doi-asserted-by":"publisher","first-page":"1502","DOI":"10.1126\/science.1160028","article-title":"Unsupervised natural experience rapidly alters invariant object representation in visual cortex","volume":"321","author":"Li","year":"2008","journal-title":"Science"},{"key":"B43","doi-asserted-by":"publisher","first-page":"335","DOI":"10.1038\/s41583-020-0277-3","article-title":"Backpropagation and the brain","volume":"21","author":"Lillicrap","year":"2020","journal-title":"Nat. Rev. Neurosci"},{"key":"B44","doi-asserted-by":"publisher","first-page":"143","DOI":"10.3390\/make7040143","article-title":"Delving into unsupervised hebbian learning from artificial intelligence perspectives","volume":"7","author":"Lin","year":"2025","journal-title":"Mach. Learn. Knowl. Extr"},{"key":"B45","doi-asserted-by":"publisher","first-page":"33","DOI":"10.1002\/syn.890050103","article-title":"Associative synaptic potentiation and depression: quantification of dissociable modifications in the hippocampal dentate gyrus favors a particular class of synaptic modification equations","volume":"5","author":"Lopez","year":"1990","journal-title":"Synapse"},{"key":"B46","doi-asserted-by":"publisher","first-page":"213","DOI":"10.1126\/science.275.5297.213","article-title":"Regulation of synaptic efficacy by coincidence of postsynaptic APs and EPSPs","volume":"275","author":"Markram","year":"1997","journal-title":"Science"},{"key":"B47","article-title":"Hebbian learning with gradients: Hebbian convolutional neural networks with modern deep learning frameworks","author":"Miconi","year":"2021","journal-title":"arXiv preprint arXiv:2107.01729"},{"key":"B48","doi-asserted-by":"publisher","first-page":"e2021925118","DOI":"10.1073\/pnas.2021925118","article-title":"Local dendritic balance enables learning of efficient representations in networks of spiking neurons","volume":"118","author":"Mikulasch","year":"2021","journal-title":"Proc. Natl. Acad. Sci. U.S.A"},{"key":"B49","article-title":"NM-Hebb: coupling local hebbian plasticity with metric learning for more accurate and interpretable CNNs","author":"Mili\u010devi\u0107","year":"2025","journal-title":"arXiv preprint arXiv:2508.19896"},{"key":"B50","doi-asserted-by":"publisher","first-page":"044017","DOI":"10.1088\/2634-4386\/aca710","article-title":"SoftHebb: Bayesian inference in unsupervised hebbian soft winner-take-all networks","volume":"2","author":"Moraitis","year":"2022","journal-title":"Neuromor. Comput. Eng"},{"key":"B51","doi-asserted-by":"publisher","first-page":"107628","DOI":"10.1016\/j.neunet.2025.107628","article-title":"Advancing the biological plausibility and efficacy of hebbian convolutional neural networks","volume":"190","author":"Nimmo","year":"2025","journal-title":"Neural Netw"},{"key":"B52","doi-asserted-by":"publisher","first-page":"267","DOI":"10.1007\/BF00275687","article-title":"A simplified neuron model as a principal component analyzer","volume":"15","author":"Oja","year":"1982","journal-title":"J. Math. Biol"},{"key":"B53","doi-asserted-by":"publisher","first-page":"61","DOI":"10.1142\/S0129065789000475","article-title":"Neural networks, principal components and subspaces","volume":"1","author":"Oja","year":"1989","journal-title":"Int. J. Neural Syst"},{"key":"B54","doi-asserted-by":"publisher","first-page":"235","DOI":"10.1016\/j.neunet.2020.03.016","article-title":"Learning in the machine: to share or not to share?","volume":"126","author":"Ott","year":"2020","journal-title":"Neural Netw"},{"key":"B55","first-page":"13924","article-title":"\u201cTowards biologically plausible convolutional networks,\u201d","volume-title":"Advances in neural information processing systems","author":"Pogodin","year":"2021"},{"key":"B56","doi-asserted-by":"publisher","first-page":"129440","DOI":"10.1016\/j.neucom.2025.129440","article-title":"Unsupervised representation learning with hebbian synaptic and structural plasticity in brain-like feedforward neural networks","volume":"626","author":"Ravichandran","year":"2025","journal-title":"Neurocomputing"},{"key":"B57","doi-asserted-by":"publisher","first-page":"533","DOI":"10.1038\/323533a0","article-title":"Learning representations by back-propagating errors","volume":"323","author":"Rumelhart","year":"1986","journal-title":"Nature"},{"key":"B58","doi-asserted-by":"publisher","first-page":"12978","DOI":"10.1523\/JNEUROSCI.0179-10.2010","article-title":"Selectivity and tolerance \u201cinvariance\u201d both increase as visual information propagates from cortical area V4 to IT","volume":"30","author":"Rust","year":"2010","journal-title":"J. Neurosci"},{"key":"B59","doi-asserted-by":"publisher","first-page":"459","DOI":"10.1016\/0893-6080(89)90044-0","article-title":"Optimal unsupervised learning in a single-layer linear feedforward neural network","volume":"2","author":"Sanger","year":"1989","journal-title":"Neural Netw"},{"key":"B60","doi-asserted-by":"publisher","first-page":"24","DOI":"10.3389\/fncom.2017.00024","article-title":"Equilibrium propagation: Bridging the gap between energy-based models and backpropagation","volume":"11","author":"Scellier","year":"2017","journal-title":"Front. Comput. Neurosci"},{"key":"B61","doi-asserted-by":"publisher","first-page":"252","DOI":"10.3390\/e26030252","article-title":"To compress or not to compress\u2013self-supervised learning and information theory: a review","volume":"26","author":"Shwartz-Ziv","year":"2024","journal-title":"Entropy"},{"key":"B62","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1109\/IJCNN60899.2024.10650478","article-title":"\u201cWeight perturbation and competitive hebbian plasticity for training sparse excitatory neural networks,\u201d","author":"Stricker","year":"2024","journal-title":"2024 international joint conference on neural networks (IJCNN)"},{"key":"B63","first-page":"59","article-title":"\u201cBayesian modeling of human concept learning,\u201d","volume-title":"Advances in neural information processing systems","author":"Tenenbaum","year":"1998"},{"key":"B64","doi-asserted-by":"publisher","first-page":"892","DOI":"10.1038\/36103","article-title":"Activity-dependent scaling of quantal amplitude in neocortical neurons","volume":"391","author":"Turrigiano","year":"1998","journal-title":"Nature"},{"key":"B65","author":"Werbos","year":"1974","journal-title":"Beyond Regression: New Tools for Prediction and Analysis in the Behavioral Sciences"},{"key":"B66","doi-asserted-by":"publisher","first-page":"2699","DOI":"10.1162\/089976605774320539","article-title":"Learning curves for stochastic gradient descent in linear feedforward networks","volume":"17","author":"Werfel","year":"2005","journal-title":"Neural Comput"},{"key":"B67","doi-asserted-by":"publisher","first-page":"96","DOI":"10.21236\/AD0241531","article-title":"\u201cAdaptive switching circuits,\u201d","author":"Widrow","year":"1960","journal-title":"1960 IRE WESCON convention record"},{"key":"B68","article-title":"Distance-forward learning: enhancing the forward-forward algorithm towards high-performance on-chip learning","author":"Wu","year":"2024","journal-title":"arXiv preprint arXiv:2408.14925"},{"key":"B69","doi-asserted-by":"publisher","first-page":"021006","DOI":"10.1103\/PhysRevX.13.021006","article-title":"Weight versus node perturbation learning in temporally extended tasks: weight perturbation often performs similarly or better","volume":"13","author":"Z\u00fcge","year":"2023","journal-title":"Phys. Rev. X"}],"container-title":["Frontiers in Computational Neuroscience"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fncom.2026.1943064\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,21]],"date-time":"2026-09-21T05:35:08Z","timestamp":1789968908000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fncom.2026.1943064\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,21]]},"references-count":69,"alternative-id":["10.3389\/fncom.2026.1943064"],"URL":"https:\/\/doi.org\/10.3389\/fncom.2026.1943064","relation":{},"ISSN":["1662-5188"],"issn-type":[{"value":"1662-5188","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,21]]},"article-number":"1943064"}}