Binocular summation and efficient coding.

Area of Science:

• Vision Science
• Neuroscience
• Computational Neuroscience

Background:

• Binocular summation enhances pattern detection, with traditional models suggesting direct input summation.
• An efficient coding framework and prior models of binocular interaction provide a basis for alternative explanations.

Purpose of the Study:

• To propose and test the MAX(S+S-) model for binocular summation.
• To compare the MAX(S+S-) model's performance against conventional models of binocular summation.

Main Methods:

• Developed the MAX(S+S-) model, utilizing sum (S+) and difference (S-) of monocular signals, assuming independent noise and gain control.
• Measured binocular summation for Gabor patches across spatial frequencies and bandwidths at detection and increment thresholds.
• Compared MAX(S+S-) model performance against MAX(LR) and B+ models, both incorporating interocular inhibition.

Main Results:

• The MAX(S+S-) model demonstrated performance comparable to conventional MAX(LR) and B+ models.
• Experimental data supported the MAX(S+S-) model's efficacy in explaining binocular summation.

Conclusions:

• The MAX(S+S-) model presents a viable alternative for explaining binocular summation.
• Evidence suggests separately gain-controlled sum (S+) and difference (S-) signals underlie various binocular behaviors.