A neural circuit for robust time-to-contact estimation based on primate MST

Area of Science:

• Neuroscience
• Computational Neuroscience
• Computer Vision

Background:

• Time-to-contact (TTC) is crucial for visual navigation and can be derived from an object's expansion rate (E), where TTC = 1/E.
• Primate dorsal medial superior temporal (MST) cells possess receptive field structures suitable for expansion and TTC estimation, but their role has been debated due to large receptive fields and lack of divergence decomposition evidence.

Purpose of the Study:

• To mathematically demonstrate that template models of dorsal MST cells can accurately and robustly estimate TTC.
• To update the ViSTARS model of primate visual navigation with modified templates for TTC estimation.

Main Methods:

• Developed template models for dorsal MST cells to extract and combine motion field components for expansion estimation.
• Integrated modified templates into the ViSTARS model, accounting for population coding of speed in V1 and MT.
• Evaluated the model's ability to code TTC across a population of simulated MST cells.

Main Results:

• The template match model accurately estimates TTC by extracting and scaling motion field components to represent expansion.
• The updated ViSTARS model demonstrates that a population of model MST cells can accurately code TTC.
• The model successfully codes TTC and heading simultaneously, irrespective of receptive field size, object size, or motion representation.

Conclusions:

• Template models of dorsal MST cells provide an accurate and robust method for TTC estimation in visual navigation.
• The updated dorsal MST cell model simultaneously codes TTC and heading, addressing previous limitations.
• A subpopulation of primate MST cells may represent expansion in this manner for navigation.