In a [previous experiment][1], we found results which suggested the existence of two, independent filters at work during sustained inattentional blindness. One filter seems to enhance the attended objects and similar objects, while the other suppresses the objects to be ignored (and any similar objects). To try to separate the roles of similarity to the attended and ignored sets, we used white squares and black-and-white checkerboards as the objects in the display. The unexpected objects, which could be either white or black, were designed to vary in similarity to the white squares, but remain constantly similar to the checkerboards. We obtained results consisted with those predicted by two separate filters. However, there are imbalances and potential confounds in the design which necessitate a control experiment. First of all, it may be that the unexpected objects actually are varying with respect to both sets of objects, in that they vary in "proportion of black." In that case, the black object would be more similar to the checkerboards than the white object, and not actually invariant. Secondly, the black unexpected object is unique to the display; no other object is solid black. This could introduce uniqueness effects. To address these possibilities, we designed a new control experiment. We once again pit checkerboards against solid squares, this time medium-gray instead of white. This controls for average luminance between the two sets; one has black and white, and one has the average of these two. The colors are balanced. The unexpected objects now vary in similarity to the _checkerboards_, rather than the squares. The unexpected object can be either a lower spatial-frequency checkerboard (4 x 4 to the display checkerboard's 5 x 5) or a square tessellation made of 16 triangles, each of which is the same surface area as the squares in the checkerboard. The internal pattern of these squares is irrelevant to their similarity to the gray square, and they remain the same average luminance. This removes the "proportion black" problem, as the actual color composition never changes. This design also removes the unique color problem. Both of the unexpected patterns are different from the display checkerboard, and no colors are present in one which are not present in the other. An additional feature of this design is that it predicts the same pattern of results as Experiment 1. If there are two filters, one that enhances the attended objects and one which suppresses the ignored ones, then we expect to see the following: * When subjects attend checkerboards and ignore gray squares, they should notice the more similar unexpected checkerboard more often than the less-similar tessellation, independent of similarity to the ignored set. * When subjects ignore checkerboards and attend gray squares, the more-similar checkerboards should be suppressed and noticed less frequently than the dissimilar tessellations, holding similarity to the attended set constant. See Predictions for more details. [1]: https://osf.io/34z6t/