Moore-Koch's multiplicative motion detector chip is just a variation
of the Tanner-Mead's motion detector [Moore and Koch 91]. In the motion
detector designed by Tanner the division operation required for
dividing the spatial gradients, 
and 
, by the temporal gradient, 
has
been implemented by a Gilbert multiplier in a feedback loop. In
Moore-Koch's work the feedback loop has been opened and a
multiplication of the temporal and spatial components of intensity is
obtained instead. As the work doesn't contain more conceptual and
practical information, details are referenced to [Moore and Koch 91].
Nonetheless, the design illustrates that even with completely
different functions than what the optic flow theory suggests, still
some motion information can be obtained.