Related Experiment Videos
Encoding of binocular disparity by complex cells in the cat's visual cortex
I Ohzawa1, G C DeAngelis, R D Freeman
1Group in Vision Science, School of Optometry, University of California, Berkeley 94720-2020, USA.
Journal of Neurophysiology
|June 1, 1997
Summary
Complex cells in the visual cortex are crucial for stereopsis, showing high disparity selectivity and position invariance. They offer a refined mechanism for depth perception and solving the stereo correspondence problem.
Area of Science:
- Neuroscience
- Computational Vision
- Sensory Systems
Background:
- Stereopsis, the perception of depth from binocular vision, relies on processing visual information from both eyes.
- Complex cells in the striate cortex are known to respond to visual stimuli but their specific role in stereopsis requires further elucidation.
Purpose of the Study:
- To investigate the function of complex cells in stereopsis by analyzing their binocular response properties.
- To determine the selectivity of complex cells for binocular disparity and stimulus position.
Main Methods:
- Extracellular recordings from single neurons in the striate cortex of anesthetized cats.
- Measurement of binocular responses using a comprehensive stimulus set varying positions and contrasts for both eyes.
- Characterization of binocular receptive field (RF) profiles.
Main Results:
- Complex cells exhibit strong binocular disparity selectivity, with selectivity for disparity being significantly narrower than for position.
- Binocular RFs are largely consistent for stimuli of the same contrast sign but inverted for opposite contrast signs.
- A disparity energy model successfully accounts for most complex cell behaviors, though some deviations were observed.
Conclusions:
- Complex cells provide a higher level of abstraction for stereopsis by integrating inputs from simple-type subunits.
- Their narrow disparity tuning and position invariance contribute to solving the stereo correspondence problem in complex scenes.
- These cells significantly reduce the complexity of depth perception processing.