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

Human visual object categorization can be described by models with low memory capacity.

Robert J Peters1, Fabrizio Gabbiani, Christof Koch

  • 1Computation and Neural Systems, Division of Biology, Caltech, 139-74, Pasadena, CA 91125, USA. rjpeters@klab.caltech.edu

Vision Research
|July 30, 2003
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

Neuropixels Opto: combining high-resolution electrophysiology and optogenetics.

Nature methods·2026
Same author

The functional organization of retinal input to the mouse superior colliculus.

bioRxiv : the preprint server for biology·2026
Same author

Protocol for testing global neuronal workspace and integrated information theories of consciousness in non-human primates and mice.

PloS one·2026
Same author

Consciousness and Controlled Donation After Circulatory Determination of Death.

JAMA·2026
Same author

Predicting neural responses to intra- and extra-cranial electric brain stimulation by means of the reciprocity theorem.

PLoS computational biology·2025
Same author

To jump or not to jump: comparing effects of phenotypic plasticity on the visual responses and escape behavior of locusts and grasshoppers.

Journal of neurophysiology·2025
Same journal

Computational and mathematical models in vision: Quantitative approaches to understanding visual perception.

Vision research·2026
Same journal

Complex interactions between lightness, chroma, and hue in color ensemble perception.

Vision research·2026
Same journal

Driving with autism spectrum disorder: Exploring the impact of tactile hazard warnings on gaze behavior and hazard responses.

Vision research·2026
Same journal

Early visual processing in adults with ADHD: evidence from contrast sensitivity, spatial integration, and external noise.

Vision research·2026
Same journal

Pupil reflexes generate the peripheral drift illusion due to ON/OFF motion responses.

Vision research·2026
Same journal

Perceived direction of glass patterns can flip by 90°: A neural model.

Vision research·2026
See all related articles

Human visual object categorization may rely on sparse representations, not large memory capacity. This suggests abstracted category properties can be extracted directly from images using early-vision models.

Area of Science:

  • Cognitive Science
  • Neuroscience
  • Computer Vision

Background:

  • High-level models of visual object categorization face challenges in neurobiological relevance.
  • Key unresolved issues include feature extraction and the impact of memory capacity on performance.

Purpose of the Study:

  • To compare various models' ability to match human categorization performance.
  • To explicitly account for the number of free parameters in each model.
  • To investigate the role of memory capacity and feature representation in categorization.

Main Methods:

  • Evaluated a comprehensive set of computational models against human observer data.
  • Accounted for the number of free parameters in each model to assess complexity.
  • Analyzed the relationship between model performance, memory capacity, and representation abstraction.

Related Experiment Videos

Main Results:

  • The most successful models demonstrated that large memory capacity is not required for high categorization performance.
  • Findings suggest that categorization performance may be underpinned by a sparse, abstracted representation of category properties.
  • This abstracted representation differs from classical prototype abstraction.

Conclusions:

  • Sparse, abstracted representations, rather than large memory stores, are crucial for effective visual object categorization.
  • Biologically plausible early-vision models can extract these representations directly from 2D images.
  • This approach bypasses the need for experimenter-defined features, offering a more neurobiologically relevant explanation.