An interesting question is what will happen if we use two patterns moving in opposite directions to define the equiluminous kinetic edge. Can we still observe the perceived positional displacement just as what happens in Demonstration 1. Furthermore, if the perceived displacement does occur, which direction it follows? Surprisingly, in this Demonstration, if you keep staring at the red fixation point, what you will see is completely opposite to Demonstration 1, i.e. now the upper edge (defined by contracting translations) seems more close to the central fixation compared with the lower edge (defined by expanding translations). The direction inversion between Demonstrations 1 & 2 suggests that a new mechanism needs to be introduced in order to account for the new type of perceived spatial displacement. Here, we suggest that this direction inversion effect is due to the anisotropy of sensitivities to motion directions, namely, a greater sensitivity to centripetal rather than to centrifugal motion in peripheral viewing of human observers, as shown by Raymond (1994).