Contrast discrimination cannot explain spatial frequency, orientation or temporal frequency discrimination

Abstract

Current models of spatial frequency (SF) and orientation discrimination are based on contrast discrimination data. In these “error propagation” models, the precision of all discrimination tasks is limited by “peripheral” noise in contrast-sensitive channels. Therefore, all discrimination thresholds should be proportional to the contrast Weber fraction Δc/c. To test this prediction, increment thresholds were measured for contrast, SF, orientation and temporal frequency (TF) for contrasts ranging from 2 to 50%. All measurements used the same stimuli, procedures and observers. For contrasts of 2% and higher, the contrast discrimination threshold Δc rises with approximately the 0.6 power of contrast, while SF and TF discrimination are independent of contrast. Furthermore, orientation discrimination is nearly independent of contrast at a SF of 4 cpd. No error-propagation model can explain these results. Therefore, SF and TF discrimination, and orientation discrimination at 4 cpd are limited by contrast-independent central noise.