The late Professor Morinaga found in 1954 that spatial displacement of direction differed from that of distance even when the measurement was made in the same figure. This finding was named as “paradox of displacement” in geometrical illusion. The present study was designed to prove the existence of the paradox of displacement in more detailed experimentation. Two types of displacements of direction and distance were measured in each of variant forms of Zöllner's, Hering's and Müller-Lyer's figures. In the figures of Zöllner and Hering types the number of dots filling each of parallels was varied from two or three to several (and including solid lines). 1. Paradox of displacement was seen in both Zöllner and Hering types when the number of dots filling the parallels was adequate. This appeared more markedly in a variant form of the Müller-Lyer figure. 2. Displacement of both direction and distance increased gradually with the increase of number of dots filling the parallels. The rate of increase of displacement was, however, more prominent for direction than for distance. These results indicate that the displacement seen in geometrical illusion differs from the objective displacement and is related to the shift in respective dimension, distinguished from each other, of spatial perception. They also show an interaction between two dimensions, direction and distance, in some figural structure. Thus the problem of dimensions seems to be important for the construction of the so-called field-theory.
The variables which are expected to influence the amount of conflict are (1) the absolute strength of competing tendencies, (2) the relative strength of competing tendencies, (3) the number of competing tendencies and (4) the degree of incompatibility. The purpose of the present study is to test the third variable using Berlyne's information theory of conflict. Random-shaped figures devised by Vanderplas and Garvin were used for inducing experimental conflict. Among the figures, 20 high (association) content value cards (H cards) and 20 low (association) content value cards (L cards) were chosen through a preliminary experiment. (See Fig. 1.) Both H and L cards were of almost equal association values. Two hypotheses have been proposed in this experiment. The first hypothesis is that H cards would induce high degree of conflict and high arousal level when the subject is instructed to give only one association to each of them, while L cards would induce low degree of conflict and low arousal level under the same instruction. Following Berlyne's theory, H cards are considered to have high response uncertainty and would induce several competing response tendencies at the same time in the subject's mind, so that the subject would be late in making his decision. Moreover, G. S. R. as a measure of arousal level of conflict is expected to increase as a function of amount of response uncertainty. In this experiment, two kinds of instructions were used. Instruction I was such that the subject was to give only one association to each stimulus as mentioned above. Under Instruction II the subject was to give as many associations as possible to each stimulus. The second hypothesis is that H cards would induce higher conflict and higher arousal level under Instruction I than would under Instruction II. The four experimental groups are designated: H1 (H card×Inst. I), H2 (H card×Inst. II), L1 (L card×Inst. I), and L2 (L card×Inst. II). Twenty subjects were assigned to each group. The reaction time (1/RT×1000) and the G. S. R. score were used as measures of conflict. The results are shown in Tables 1 to 3. No significant difference in response frequency was found between H and L cards under the two different instruction (Table 1). This finding verifies that our selection of H and L cards was adequate as far as the association value is concerned. In addition to this finding, Table 1 indicates that the total frequency of response under Instruction II was greater than was under Instruction I. This might suggest that Instruction II may have a motivational effect of the subjects in the present experimental situation. The mean. G. S. R. score was greater for Group H1 than was for Group L1. This supports the first hypothesis. The G. S. R. score in Group H1 was significantly higher than in Group H2. This validates the second hypothesis. The G. S. R. result is in accordance with the findings of the mean. reaction time (Table 3). In conclusion, this experiment verifies the proposition that the number of competing tendencies is an important variable in determing the amount of conflict.