EUVIMEDEUVIMED entdecken
Crossref

Inhibitory Feedback Enables Predictive Learning of Multiple Sequences in Neural Networks

Matteo Saponati, Martin Vinck

Neural Computation · 17.03.2026 · 38 (4) · 471-498

DOI: 10.1162/neco.a.1504 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Neural Computation, online erschienen am 17.03.2026. Verfasst von Matteo Saponati und 1 weiteren Autor:innen. Beteiligte Einrichtung: Ernst-Strüngmann Institute for Neuroscience in Cooperation with Max Planck Society, 60528, Frankfurt Am Main, Germany. Hauptthemen laut MEDLINE: Neurons, Neural Networks, Computer, Neural Inhibition, Models, Neurological. Die Arbeit nennt 78 Literaturangaben. Gefördert von: ERC, NWO VIDI, BMF.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Anticipating future events is a key computational task for neuronal networks. Experimental evidence suggests that reliable temporal sequences in neural activity play a functional role in the association and anticipation of events in time. However, how neurons can differentiate and anticipate multiple spike sequences remains largely unknown. We implement a learning rule based on predictive processing, where neurons exclusively fire for the initial, unpredictable inputs in a spiking sequence, leading to an efficient representation with reduced postsynaptic firing. Combining this mechanism with inhibitory feedback leads to sparse firing in the network, enabling neurons to selectively anticipate different sequences in the input. We demonstrate that intermediate levels of inhibition are optimal to decorrelate neuronal activity and to enable the prediction of future inputs. Notably, each sequence is independently encoded in the sparse, anticipatory firing of the network. Overall, our results demonstrate that the interplay of self-supervised predictive learning rules and inhibitory feedback enables fast and efficient classification of different input sequences.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

NeuronsNeural Networks, ComputerNeural InhibitionModels, NeurologicalLearningAction PotentialsAnimalsPredictive Learning ModelsPrediction AlgorithmsHumansFeedback

Autor:innen und Institutionen

  • Ernst-Strüngmann Institute for Neuroscience in Cooperation with Max Planck Society, 60528, Frankfurt Am Main, Germany.
  • IMPRS for Neural Circuits, Max-Planck Institute for Brain Research, 60438, Frankfurt Am Main, Germany.
  • Donders Centre for Neuroscience, Department of Neurophysics, Radboud University Nijmegen, 6525 Nijmegen, the Netherlands matteosaponati@gmail.com.
  • Donders Centre for Neuroscience, Department of Neurophysics, Radboud University, Nijmegen, 6525 Nijmegen, the Netherlands martin.vinck@donders.ru.nl.

Förderung

  • ERC
  • NWO VIDI
  • BMF

Zitieren

APA

Saponati, M., & Vinck, M. (2026). Inhibitory Feedback Enables Predictive Learning of Multiple Sequences in Neural Networks. Neural Computation, 38(4), 471-498. https://doi.org/10.1162/neco.a.1504

Vancouver

Saponati M, Vinck M. Inhibitory Feedback Enables Predictive Learning of Multiple Sequences in Neural Networks. Neural Computation. 2026;38(4):471-498. doi:10.1162/neco.a.1504

BibTeX
@article{Saponati2026,
  author = {Saponati, Matteo and Vinck, Martin},
  title = {Inhibitory Feedback Enables Predictive Learning of Multiple Sequences in Neural Networks},
  journal = {Neural Computation},
  year = {2026},
  volume = {38},
  number = {4},
  pages = {471--498},
  doi = {10.1162/neco.a.1504}
}
Kostenlos starten →