Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A feedforward architecture accounts for rapid categorization.

Thomas Serre1, Aude Oliva, Tomaso Poggio

  • 1Center for Biological and Computational Learning, Massachusetts Institute of Technology, Cambridge, MA 02139, USA. serre@mit.edu

Proceedings of the National Academy of Sciences of the United States of America
|April 4, 2007
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Perilesional neuromodulation replaces lost sensorimotor function in persons with spinal cord injury.

Nature biomedical engineering·2026
Same author

MOSAIC: A scalable framework for fMRI dataset aggregation and modeling of human vision.

bioRxiv : the preprint server for biology·2026
Same author

Better artificial intelligence does not mean better models of biology.

Trends in cognitive sciences·2025
Same author

What if eye...? Computationally recreating vision evolution.

Science advances·2025
Same author

From prediction to understanding: Will AI foundation models transform brain science?

Neuron·2025
Same author

Feature binding in biological and artificial vision.

Trends in cognitive sciences·2025
Same journal

The TaMYB55-TaSnRK1α1-TabZIP9 module confers heat stress tolerance in wheat.

Proceedings of the National Academy of Sciences of the United States of America·2026
Same journal

Superstatistics approach to turbulent circulation fluctuations.

Proceedings of the National Academy of Sciences of the United States of America·2026
Same journal

A molecular timescale for evolution of cobamide biosynthesis.

Proceedings of the National Academy of Sciences of the United States of America·2026
Same journal

Pierre Chambon, a pioneer of molecular biology and gene regulation in eukaryotes.

Proceedings of the National Academy of Sciences of the United States of America·2026
Same journal

Granulosa cell glycogen fuels the avascular corpus luteum.

Proceedings of the National Academy of Sciences of the United States of America·2026
Same journal

Synthetic essentiality of TRAIL/TNFSF10 in VHL-deficient renal cell carcinoma.

Proceedings of the National Academy of Sciences of the United States of America·2026
See all related articles

Human visual object recognition, particularly rapid categorization, is highly accurate. A feedforward model, inspired by neural processing, successfully predicts human performance in identifying animals versus non-animals.

Area of Science:

  • Neuroscience
  • Computer Vision
  • Cognitive Science

Background:

  • Primate visual object recognition excels, surpassing current computer vision.
  • Rapid object categorization in primates is highly accurate and likely feedforward.
  • Existing models struggle to fully explain primate visual processing efficiency.

Purpose of the Study:

  • To test a feedforward model of object recognition.
  • To predict human performance on rapid visual categorization tasks.
  • To account for anatomical and physiological constraints in visual processing.

Main Methods:

  • Implemented a feedforward model extending the Hubel and Wiesel hierarchy.
  • Simulated rapid masked animal vs. non-animal categorization.
  • Compared model predictions to human performance data.

Related Experiment Videos

Main Results:

  • The feedforward model accurately predicted human performance levels.
  • The model captured the pattern of human errors and successes.
  • Model performance aligned with known neural processing constraints.

Conclusions:

  • Feedforward processing is a viable mechanism for rapid object recognition.
  • Computational models can successfully replicate key aspects of primate vision.
  • This work bridges computational modeling and biological visual systems.