A large portion of the shapes that were tested with variation of TT had also been tested with various degrees of sparseness, i.e. , display of only some of the full inventory of boundary dots. Hit rates decline when fewer dots are displayed, and the slope of the regression on a given shape reflects the degree to which the dots are providing redundant information for purposes of recognition. Here the slopes of the sparseness regressions are plotted against the slopes of the TT regressions. There is a clear relationship, which supports the proposition that the variation of slope that can be seen in Figures3 and4 are due to differences in the degree to which neighboring dots provide redundant information.