Detection of spatial beats: non-linearity or contrast increment detection?

We measured sensitivity of human observers to the 1 c/deg beat between sinusoidal gratings of 9 and 10 c/deg, at different contrasts of the 2 components. Raising the contrast of one component increases the contrast required in the other component to detect the beat. This is consistent with the hypothesis that the beat is detected because of the local increments in contrast which it produces, but not with the hypothesis that it is detected when a difference-frequency distortion product generated by non-linear transduction of luminance reaches threshold.