Related Experiment Videos
Disparity-based coding of three-dimensional surface orientation by macaque middle temporal neurons.
Jerry D Nguyenkim1, Gregory C DeAngelis
1Department of Anatomy and Neurobiology, Washington University School of Medicine, St. Louis, Missouri 63110, USA.
Summary
Middle temporal area (MT) neurons are tuned to three-dimensional (3-D) surface orientation using binocular disparity gradients. This selectivity for surface tilt and slant is independent and arises from complex receptive field mechanisms, not motion or mean disparity.
Area of Science:
- Neuroscience
- Computational Vision
- Visual Perception
Background:
- Gradients of binocular disparity are crucial for perceiving 3-D surface orientation.
- Neurons in parietal cortex show selectivity for 3-D orientation, but its origin is unclear.
- The middle temporal (MT) area is known for its role in depth perception.
Purpose of the Study:
- Investigate if MT neurons signal 3-D surface orientation (tilt and slant) using disparity gradients.
- Determine the neural mechanisms underlying this selectivity in MT.
- Clarify the contribution of MT to high-level 3-D surface structure processing.
Main Methods:
- Recorded responses of MT neurons to random-dot stereograms with linear disparity gradients.
- Manipulated stimuli to isolate tilt and slant tuning, and assess dependence on motion and mean disparity.
- Analyzed receptive field properties to understand the basis of orientation selectivity.
Main Results:
- Many MT neurons exhibit tuning for 3-D surface orientation (tilt and slant).
- Tilt and slant generally affect MT responses independently.
- Tilt tuning is robust to removal of coherent motion and insensitive to mean disparity, suggesting it's not a byproduct of velocity or frontoparallel disparity coding.
- Tilt tuning arises from heterogeneous disparity tuning within MT receptive fields.
Conclusions:
- MT neurons encode high-level signals of 3-D surface structure beyond retinal image velocities.
- MT plays a significant role in processing 3-D surface orientation cues.
- Separable coding of tilt and slant in MT provides insights into visual pathway computations.