Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: May 14, 2026

Creating Objects and Object Categories for Studying Perception and Perceptual Learning

Creating Objects and Object Categories for Studying Perception and Perceptual Learning

Published on: November 2, 2012

A three-layer model of natural image statistics.

Michael U Gutmann¹, Aapo Hyvärinen

¹Dept. of Mathematics and Statistics, P.O. Box 68, FIN-00014 University of Helsinki, Finland; Dept. of Computer Science and HIIT, P.O. Box 68, FIN-00014 University of Helsinki, Finland.

Journal of Physiology, Paris

|February 2, 2013

Summary

This study explores how visual systems achieve pattern selectivity and invariance by learning computations from natural images. Findings reveal simple, complex cell-like processing in early layers and contour/texture selectivity in later layers.

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

Risk-averse optimization of genetic circuits under uncertainty.

Cell systems·2026

Same author

Second-order instantaneous causal analysis of spontaneous MEG.

Imaging neuroscience (Cambridge, Mass.)·2025

Same author

Improving source estimation of retinotopic MEG responses by combining data from multiple subjects.

Imaging neuroscience (Cambridge, Mass.)·2025

Same author

Towards interpretable Cryo-EM: disentangling latent spaces of molecular conformations.

Frontiers in molecular biosciences·2024

Same author

Towards Interpretable Cryo-EM: Disentangling Latent Spaces of Molecular Conformations.

bioRxiv : the preprint server for biology·2024

Same author

Penalized canonical correlation analysis reveals a relationship between temperament traits and brain oscillations during mind wandering.

Brain and behavior·2024

Area of Science:

Computational neuroscience
Computer vision
Image processing

Background:

Visual systems require both selectivity to patterns and invariance to variations, which are opposing requirements.
Previous theories proposed iterative computations to achieve selectivity and invariance, but the specific computations at each level remained unclear.

Purpose of the Study:

To investigate and learn the computations underlying visual selectivity and invariance from natural images.
To propose and estimate a probabilistic model of natural images to understand hierarchical visual processing.

Main Methods:

Developed a three-layer probabilistic model for natural image processing.
Trained and evaluated the model on two datasets: image patches and downsampled visual scenes.
Analyzed the computations performed at each layer of the model.

Keywords:

Deep learning Invariance Natural images Probabilistic modeling Selectivity Sparse coding Visual processing

Related Experiment Videos

Last Updated: May 14, 2026

Creating Objects and Object Categories for Studying Perception and Perceptual Learning

Creating Objects and Object Categories for Studying Perception and Perceptual Learning

Published on: November 2, 2012

Main Results:

The first two layers of the model exhibited simple and complex cell-like computations, consistent with early visual processing.
The third layer showed selectivity for longer contours.
Specific findings for the third layer included texture selectivity for image patches and curvature selectivity for downsampled scenes.

Conclusions:

The learned computations from natural images provide insights into how visual systems balance selectivity and invariance.
The hierarchical model successfully replicates key aspects of biological visual processing.
Findings suggest that different types of selectivity (contour, texture, curvature) emerge at different levels of visual processing.