Sensors for impossible stimuli may solve the stereo correspondence problem.

Area of Science:

• Neuroscience
• Computational Vision
• Sensory Processing

Background:

• Binocular vision faces the challenge of objects projecting to different retinal positions (binocular disparity).
• Visual cortex neurons exhibit tuning to both position disparity and phase disparity, arising from receptive field differences.
• Phase disparity is notably absent in natural images, raising questions about its neural encoding.

Purpose of the Study:

• To investigate the functional role of phase-disparity detectors in the visual cortex.
• To propose a mechanism by which phase-disparity neurons address the correspondence problem in binocular vision.

Main Methods:

• Theoretical analysis of receptive field properties and disparity tuning.
• Modeling of neural responses to natural and artificial image stimuli.
• Examination of phase-disparity neuron activation patterns under conditions of true and false feature matches.

Main Results:

• Phase-disparity neurons are not activated by phase disparity present in natural images.
• These neurons show heightened activation in response to false matches during the correspondence problem.
• Phase-disparity neurons exhibit a preference for stimuli that do not represent true object correspondence.

Conclusions:

• Phase-disparity detectors may serve as 'lie detectors' in the visual system.
• By identifying false matches, these neurons facilitate the deduction of correct feature correspondence.
• This mechanism aids in resolving the ambiguity inherent in binocular vision and object recognition.