Nonlinear analysis of spatial vision using first-and-second-order Volterra transfer functions measurement

Area of Science:

• Visual neuroscience
• System identification
• Psychophysics

Background:

• Linear systems are characterized by one modulation transfer function.
• Nonlinear systems require multiple multidimensional transfer functions for identification.
• Human pattern vision in the spatial frequency domain is complex.

Purpose of the Study:

• To adapt the harmonic input method for nonlinear system identification using Volterra series.
• To investigate human pattern vision psychophysically.
• To empirically estimate the second-order modulation transfer function of the human visual system.

Main Methods:

• Modified harmonic input method.
• Applied Volterra series approach.
• Measured contrast sensitivity to difference harmonics.

Main Results:

• Contrast sensitivity to difference harmonics empirically estimates the second-order modulation transfer function.
• Difference harmonics, perceived as beatlike patterns, can be observed even with high spatial frequencies.
• Contrast sensitivity for difference harmonics depends on frequency difference, not individual frequencies, indicating local nonlinearities.

Conclusions:

• The human visual system exhibits nonlinearities.
• A cascade model (linear-nonlinear-linear) can represent these local nonlinearities.
• The study provides a method to empirically estimate nonlinear characteristics of visual perception.