When two sinusoidal gratings of the same orientation and similar spatial frequency are summed, the resulting pattern has a periodic spatial variation or beat in contrast. Although the pattern contains no luminance modulation component at the beat frequency, it behaves in some respects as if it did: human observers for example are very good at detecting spatial displacements of the beat. We wished to test the possibility that a non-linearity in the visual system generates a component (a "distortion product") at the beat frequency, and that it is displacement of the distortion product that observers detect. Attempting to "null" the distortion product by adding to the beat pattern a sinusoidal component of the same spatial frequency as the distortion product but 180 deg out of phase with it does not impair performance in detecting motion of the beat; there is no nulling at any amplitude of the added component. Reducing the phase shift of the hypothetical distortion product by adding a static sinusoid to the moving beat pattern fails to produce the predicted fall in performance. These results suggest that distortion products do not contribute to our sensitivity to the displacement of beat patterns. Reversing the contrast of a beat pattern when it is displaced, slightly increases sensitivity to displacement, the same manipulation impairs performance with luminance patterns. This is consistent with the notion that the beat is detected as an unsigned local contrast signal.