Related Experiment Video
Updated: Mar 28, 2026
07:08
Investigating Object Representations in the Macaque Dorsal Visual Stream Using Single-unit Recordings
Published on: August 1, 2018
8.8K
The Development of Hand-Centered Visual Representations in the Primate Brain: A Computer Modeling Study Using Natural
Juan M Galeazzi1, Loredana Minini1, Simon M Stringer1
1Department of Experimental Psychology, Oxford Centre for Theoretical Neuroscience and Artificial Intelligence, University of Oxford Oxford, UK.
Frontiers in Computational Neuroscience
|December 24, 2015
Summary
This study shows how the primate visual system, using a neural network model called VisNet, can develop hand-centered visual representations. These representations emerge through unsupervised learning, even with complex visual scenes containing multiple objects.
Area of Science:
- Computational neuroscience
- Primate visual system modeling
- Artificial neural networks
Background:
- Neurons in primate brains process visual information using hand-centered reference frames.
- Understanding the developmental mechanisms of these representations is crucial for neuroscience.
Purpose of the Study:
- To investigate the development of hand-centered visual representations in the VisNet neural network model.
- To explore how unsupervised competitive learning and self-organization contribute to this process.
- To simulate biologically plausible mechanisms for neuronal receptive field formation.
Main Methods:
- Training the VisNet neural network model with computerized images of hands against natural visual scenes.
- Simulating scenarios with single and multiple simultaneously presented targets relative to the hand.
- Analyzing output cell receptive fields for hand-centered representations and localization.
Main Results:
- VisNet successfully developed hand-centered representations with single, localized receptive fields.
- Statistical decoupling principles explained the emergence of localized receptive fields despite multi-object training.
- Increased object overlap during training led to decreased shape selectivity and broader spatial tuning in some cells.
- Training with real-world natural visual scenes further supported the emergence of these receptive fields.
Conclusions:
- Unsupervised competitive learning and self-organization in VisNet can generate hand-centered visual representations.
- The model demonstrates biologically plausible mechanisms for receptive field development under ecologically realistic conditions.
- These findings advance our understanding of how the primate visual system processes spatial information relative to the body.
