EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A feedforward mechanism for human-like contour integration

Fenil R Doshi iD, Talia Konkle, George A Alvarez

PLOS Computational Biology · 2025

DOI: 10.1371/journal.pcbi.1013391 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS Computational Biology, 2025. Verfasst von Fenil R Doshi und 2 weiteren Autor:innen. Beteiligte Einrichtung: Department of Psychology, Harvard University, Cambridge, Massachusetts, United States of America. Hauptthemen laut MEDLINE: Neural Networks, Computer, Form Perception, Visual Perception. Die Arbeit nennt 90 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Science Foundation, Kempner Institute.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep neural network models provide a powerful experimental platform for exploring core mechanisms underlying human visual perception, such as perceptual grouping and contour integration—the process of linking local edge elements to arrive at a unified perceptual representation of a complete contour. Here, we demonstrate that feedforward convolutional neural networks (CNNs) fine-tuned on contour detection show this human-like capacity, but without relying on mechanisms proposed in prior work, such as lateral connections, recurrence, or top-down feedback. We identified two key properties needed for ImageNet pre-trained, feed-forward models to yield human-like contour integration: first, progressively increasing receptive field structure served as a critical architectural motif to support this capacity; and second, biased fine-tuning for contour-detection specifically for gradual curves (~20 degrees) resulted in human-like sensitivity to curvature. We further demonstrate that fine-tuning ImageNet pretrained models uncovers other hidden human-like capacities in feed-forward networks, including uncrowding (reduced interference from distractors as the number of distractors increases), which is considered a signature of human perceptual grouping. Thus, taken together these results provide a computational existence proof that purely feedforward hierarchical computations are capable of implementing gestalt “good continuation” and perceptual organization needed for human-like contour-integration and uncrowding. More broadly, these results raise the possibility that in human vision, later stages of processing play a more prominent role in perceptual-organization than implied by theories focused on recurrence and early lateral connections.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Neural Networks, ComputerForm PerceptionVisual PerceptionHumansComputational BiologyModels, Neurological

Autor:innen und Institutionen

  • Department of Psychology, Harvard University, Cambridge, Massachusetts, United States of America.
  • Kempner Institute for the Study of Natural and Artificial Intelligence, Harvard University, Cambridge, Massachusetts, United States of America.

ORCID-Profile: Fenil R Doshi

Förderung

  • National Science Foundation
  • Kempner Institute

Zitieren

APA

Doshi, F. R., Konkle, T., & Alvarez, G. A. (2025). A feedforward mechanism for human-like contour integration. PLOS Computational Biology. https://doi.org/10.1371/journal.pcbi.1013391

Vancouver

Doshi FR, Konkle T, Alvarez GA. A feedforward mechanism for human-like contour integration. PLOS Computational Biology. 2025. doi:10.1371/journal.pcbi.1013391

BibTeX
@article{Doshi2025,
  author = {Doshi, Fenil R and Konkle, Talia and Alvarez, George A},
  title = {A feedforward mechanism for human-like contour integration},
  journal = {PLOS Computational Biology},
  year = {2025},
  doi = {10.1371/journal.pcbi.1013391}
}
Kostenlos starten →