Optimized but Not Maximized Cue Integration for 3D Visual Perception.

Area of Science:

• Neuroscience
• Vision Science
• Computational Neuroscience

Background:

• Reconstructing 3D scenes from 2D retinal images is inherently challenging.
• Human 3D perception relies on integrating multiple visual cues, but this is less understood in non-human primates (NHPs).

Purpose of the Study:

• To assess 3D perception capabilities in macaque monkeys.
• To investigate how NHPs integrate stereoscopic and perspective cues for 3D scene reconstruction.
• To explore the neural mechanisms underlying 3D perception in NHPs.

Main Methods:

• A planar surface orientation discrimination task was used with macaque monkeys.
• Behavioral data were analyzed to understand cue integration strategies.
• Neural network simulations based on recorded neuronal activity were employed.

Main Results:

• Monkeys demonstrated accurate 3D perception across various orientations and distances.
• Perceptual precision varied with the planar surface's pose.
• Monkeys dynamically reweighted stereoscopic and perspective cues based on their reliability.
• Neural network models suggested optimal, but not maximal, cue integration due to canonical computations.

Conclusions:

• NHPs exhibit robust 3D perception through adaptive cue integration.
• The findings provide insights into the neural constraints of 3D visual processing in primates.
• Cue integration for 3D perception in monkeys is optimized but not maximized, differing from theoretical maximum precision.