ABSTRACT
Results of Exp. 11.1 are shown in Fig. 6.7. The magnitude estimations give a power function with a curvature (or a knee) toward the darker (lower level) samples. The exponent at the straight segment is close to unity. Almost identical results were reported by Mattiello and Guirao (1974). The geometric means of the numerical values-1.5 for the low end and 80 for the high end of the range-become the two endpoints in the new scale used in Exp. 11.2. A few weeks elapsed before Exp. 11.2 was run , and the same Os participated . As Fig. 6. 7 shows (Xs), the result was a sinusoidal bowing of the points in a curve typical of category judgments . Since the differences between some of the individual ranges in Exp. 11.1 were quite large , it is likely that the Os felt constrained by the averaged fixed points . An alternative was then tried with the same group of Os (Exp. 11.3) . The Os were now given the two extreme points of the numerical
'r !!l
LIGHTNESS c
6. SCALE BASED ON RATIO AND PARTITION ESTIMATES 71
METHOD OF CONVERGING LIMITS (CL)
Results The function depicted in Fig. 6.10 is practically the same as those obtained with magnitude estimation in Exp. 11.1 and with the derived individual ranges in Exp. Il.3 . Most of the Os took advantage of the flexible endpoints. All but one made some modifications in the range they had originally selected.
