Is there opponent-orientation coding in the second-order channels of pattern vision?

Area of Science:

• Visual perception
• Computational neuroscience
• Image processing

Background:

• Investigates the existence of opponency between orientation-selective processes in pattern perception, similar to color opponency.
• Focuses on potential opponency within second-order visual channels.
• Compares SIGN-opponent-only, ORIENTATION-opponent, and BOTH-opponent channel structures.

Purpose of the Study:

• To determine if orientation-selective processes exhibit opponency analogous to color mechanisms.
• To compare the predictive power of different second-order visual channel models.
• To understand the role of contrast gain control in texture perception.

Main Methods:

• Measured human observers' ability to segregate textures made of Gabor grating patches.
• Compared experimental performance against predictions from SIGN-opponent-only, ORIENTATION-opponent, and BOTH-opponent models.
• Incorporated contrast gain control into models to explain contrast-dependent effects.

Main Results:

• Experimental results strongly support the SIGN-opponent-only channel model.
• ORIENTATION-opponent and BOTH-opponent models were ruled out by the data.
• The SIGN-opponent-only model, with contrast gain control, successfully explained observed contrast-dependent effects.

Conclusions:

• Human texture perception relies on SIGN-opponent-only channels, indicating no opponency between orientations in these channels.
• The findings exclude ORIENTATION-opponent and BOTH-opponent channel structures for texture segregation.
• Contrast gain control, a form of normalization, is a crucial component of human texture perception.