Transparent motion perception as detection of unbalanced motion signals. III. Modeling

Area of Science:

• Neuroscience
• Computational Vision
• Perception

Background:

• Previous studies explored conditions for transparent motion perception via psychophysics and neural mechanisms through physiological recordings.
• Perceptual experiments revealed that balanced motion signals lead to non-transparent perception, while transparent displays exhibit unbalanced signals in direction, disparity, or spatial frequency.

Purpose of the Study:

• To explicitly test a proposed two-stage model of motion perception.
• To compare the model's response to perceptual and physiological results using balanced and unbalanced motion stimuli.

Main Methods:

• Computational modeling and simulations were employed to analyze the two-stage motion perception model.
• The model's first stage extracts motion energies across spatial frequency and disparity ranges, analogous to area V1.
• The second stage implements opponent-direction suppression, mimicking area MT's function.

Main Results:

• The model's first stage, like V1, did not differentiate between balanced (non-transparent) and unbalanced (transparent) stimuli.
• The second stage, employing subtractive or divisive inhibition, showed significantly stronger responses to transparent stimuli compared to non-transparent ones, mirroring MT cell activity.
• The computational model's performance aligned with both perceptual observations and physiological data.

Conclusions:

• The findings support a two-stage model of motion perception.
• Area V1 acts as a motion measurement stage, while area MT functions as an opponent-direction suppression stage.
• This model effectively explains the neural basis of transparent motion perception.