Distributed encoding of curvilinear self-motion across spiral optic flow patterns.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Vision Science

Background:

• Optic flow from self-motion provides heading cues.
• Primate brain area MSTd neurons are involved in linear heading perception.
• The neural basis for curvilinear self-motion perception remains unclear.

Purpose of the Study:

• To investigate how optic flow signals in MSTd predict curvilinear self-motion perception.
• To model the contribution of different optic flow patterns (radial, spiral, concentric) in "spiral space" to self-motion estimation.

Main Methods:

• Developed a neural model of primate brain area MSTd.
• Simulated optic flow patterns corresponding to radial, spiral, and concentric motion.
• Decoded self-motion estimates from MSTd-like units tuned to various spiral space patterns.
• Compared decoding accuracy and precision with human judgments under varying path curvature and gaze conditions.

Main Results:

• Estimates decoded from the full set of spiral space patterns were significantly more accurate and precise than those from radial expansion alone.
• Decoding using only spiral subtypes closely matched the performance of the full model.
• Only the comprehensive decoding model accurately reflected human judgments when path curvature and gaze direction covaried.
• Predictive units showed peripheral bias in center-of-motion tuning, aligning with existing neurophysiological data.

Conclusions:

• Curvilinear self-motion perception is likely encoded in a distributed manner across the "spiral space" of optic flow patterns.
• Specific spiral subtypes within MSTd contribute significantly to accurate self-motion estimation.
• The findings support a neural mechanism for complex self-motion perception that integrates various optic flow features.