A neural model of motion processing and visual navigation by cortical area MST

Area of Science:

• Neuroscience
• Computational Neuroscience
• Visual Perception

Background:

• Dorsal medial superior temporal cortex (MSTd) cells are crucial for processing optic flow during visually guided navigation.
• Understanding how MSTd cells contribute to self-motion perception is key to deciphering navigation mechanisms.

Purpose of the Study:

• To develop a neural model simulating MSTd cell properties and human navigation behaviors.
• To investigate the emergent properties arising from known neural mechanisms in optic flow processing.

Main Methods:

• A computational model incorporating log polar cortical magnification, Gaussian receptive fields, spatial pooling, and eye movement signals was developed.
• The model's simulated responses were quantitatively compared against neurophysiological data from MSTd cells and psychophysical data from human navigation tasks.

Main Results:

• Model cells accurately replicated MSTd neuron responses to optic flow stimuli across the visual field, including position invariance and tuning properties.
• The model successfully predicted preferred spiral directions, direction reversals, and response characteristics of MSTd cells.
• The model demonstrated how MSTd cell activity explains human heading judgments without complex templates and clarifies the role of retinal and extraretinal signals.

Conclusions:

• The proposed neural model provides a quantitative account of MSTd cell function in optic flow processing and self-motion perception.
• The model elucidates how basic neural mechanisms interact to produce complex navigational behaviors and accurate heading perception.
• The findings highlight the interplay between scene layout, rotation rates, and visual input in determining heading judgment accuracy.