In the previous articles (1, 2), highly unexpected results were reported which the authors encountered in a series of experiments employing a particular set of pre-illuminating stimuli (Fig. 1 in 2) and dealing with the indirect induction. Namely, the authors made discoveries strongly suggesting the interpretation that, under certain circumstances, the configuration of the indirect induction can not be understood in terms of the pattern of the retinal stimulation. In short, if, I, II, III and IV of Fig. 1 in (2) are presented in succession and fixation is shifted from the left mark to the right one at an appropriate moment prior to the presentation of IV, existence or non-existence of the induction at each point follows the pattern designated P in Fig. 2 in (2). The P-result seems to be accounted for only by the assumption that the configuration of the induction takes its form in accordance with what a subject perceives under the given conditions. On the contrary, if I, II and IV are presented in succession and fixation is shifted in the same way, the results turn to what designated R in Fig. 2 in (2). The R-result is easy to understand in terms of the retinal stimulation pattern. In one of the present experiments, the configuration of the direct induction was examined with the same pre-illuminating stimuli (Fig. 1 in 2). As shown in Fig. 1, the results were exactly parallel to those obtained in the case of the indirect induction. That means, when III was included, the result doubtlessly supported the P-hypothesis, whereas, when III was omitted, the R-hypothesis became tenable. In order to obtain further informations concerning these findings, a series of experiments were designed with the pre-illuminating stimuli as shown in Fig. 2 which involve so-called refraction of a beam of the induction initiated from B in II. When III was included, the arrival of the induction from B to the point 2 was demonstrated by the neutralization of induction of Y at this point (Fig. 5). This is what we called the P-result accompanying the refraction in addition. It is astonishing that the refraction takes place in the region of the retina which isnot stimulated at all by the prism-shaped yellow figure in I. Besides, the refraction index was discovered to coincide with that obtained by Motokawa in the retinal region actually pre-illuminated by the prism shaped figure. On the other hand, when III was omitted, the neutralization was observed at the point 6 and the R-result was obtained (Fig. 7). Hence, it may be said that the results are highly consistent throughout all the experiments reported so far.
This paper is the first report on the experimental studies which were designed to testify to the inhibitory effects of an additional stimulus preceding two stimuli to be compared with each other, in the successive comparison of brightness. Three light-spots of 1cm in diameter were successively presented at an identical location in a perfectly dark room. The additional stimulus (e) always preceded the standard stimulus (N) and the latter preceded the comparative stimulus (V). The durations of three stimuli were 2sec. in length. The interval between e and N was always constant at 1sec., while that between N and V was varied. The light-value of e was made physically equivalent to that of N (6m.a.) to eliminate the so-called assimilation-contrast phenomenon. The effect of additional stimulus was measured according as the PSE, calculated by the method of complete series, was larger or smaller than that obtained without the presence of additional stimulus in a control series. The main results in this study were as follows: (1) When N and V werepresented without an additional stimulus, the normal p-function was obtained, i.e., the time-error gradually changed from positive to negative as the interval increased from 1 to 9sec., except that the error was a little more positive than normal at every interval (Exp. I, see Fig. 1). (2) When the additional stimulus was presented, the time-error shifted towards a negative direction, i.e., underestimation of N occurred independently of whether the interval between N and V was short (1sec.) or long (6sec.) (Exp. II, & III. see Table 3, 5). (3) Nevertheless the amount of the effect of additional stimulus was not constant throughout every interval. Although the time-error also shifted towards a negative direction at the 1-sec. interval, this shift decreased its amount at the 3-sec. interval. And at longer intervals than 3sec., the time-error gradually shifted towards a negative direction once more (Exp. IV, see Fig. 2). (4) The above-mentioned tendency was also recognized when the light-value of e was so controlled as it was phenomenally equivalent to that of N perfectly to eliminate the assimilation-contrast mechanism between them. In this condition, however, the shift towards a negative direction was more remarkable throughout all the intervals (Exp. V, see Fig. 3). (5) According to the observer's introspection, N was seen darker than e, especially at shorter intervals. But e had not a disturbing effect upon the comparative judgement of brightness between N and V at 1-or 3-sec. interval. On the contrary, the jud gement itself was disturbed at longer intervals, images of e and N having fused themselves together with the lapse of time (Exp. II-V). (6) In addition, when two light-spots, placed side by side with each other in a horizontal line at the distance of 1cm, were presented twice successively, the magnitude of time-error did not differ significantly from that in the control series at the 1-sec. interval. On the other hand, the shift towards a negative direction was clearly noticed at the 6-sec. interval (Exp. II, & III, see Table 3, 5).
It was reported in the previous paper that pressure on the skin lowers the sensitivity of pressure sensation in another area. In this paper, the author reports a peculiar fact revealed by successive measurements of pressure sensitivity with and without skin pressure alternately. The sensitivity was measured by the number of pressure spots within a given area of the dorsal side of the right forearm, and the conditioning pressure was given on the forearm, proximally to the test area. A measurement group consists of a pair of sensitivity measurements: such as M1·M2, pM1·pM2, M1·pM2 and pM1·M2 (see Table 1: Abbreviations). Each pair of measurements is made at the interval of 5 min. in series (Table 2). The results were as follows:- 1) M1·M2 series showed almost steady sensitivity (Table 3 & Fig. 1). 2) pM1·pM2 series showed the decrease of sensitivity to 20-30% of the normal value. After the release of pressure, the sensitivity recovered gradually to the normal value within about 90min. (Table 4 & Fig. 2). 3) In M1·pM2 series (conditioned series), the sensitivity was nearly normal in M1, but in pM2 it lowered to 20-30% of M1. Consequently, in this experiment, each group showed a pair of high and low sensitivities. When M1·M2 series (as a test series) was given after this, the decreasing effect remained only in M2, and this after-effect gradually disappeared within about 90min. (Table 5 & Fig. 3). So in this case, the sensitivity pattern formed in M1·M2 group after M1·pM2 series is a pair of high and low ones. It may be said that the after-effect takes place as a group-pattern of high-low. 4) It seemed that the duration of the after-effect on M1·M2 was proportional to the number of conditioned group of M1·pM2 series (Fig. in note). 5) If no test series was given after M1·pM2 series, the after-effect as group-pattern disappeared earlier than in the case in which M1·M2 series followed. 6) When the single sensitivity measurement (M) was made after M1·pM2 series, the after-effect as group-pattern could not be recognized (Fig. 4). And when the test group and the conditioned group took the same interval, the after-effect had the maximum value (Fig. 5). 7) The after-effect as group-pattern could be extinguished by the reversal combination (pM1·M2) of conditioned pattern, and various stimuli such as touch, sound, light, etc. (Fig. 6). 8) If the pM1·M2 series followed M1·M2 series, the sensitivity decreased to 30% in pM1, and 60-80% in M2. Then the after-effect was found, not as the group-pattern of low-high but as the simple decrease (Fig. 7). 9) To form the after-effect as group-pattern, the pattern in the preceding conditioned series must have clear contrast of high and low. The phenomenon described as “the after-effect as group-pattern” would be considered as a conditioned response, in which the unconditioned stimulus is a pressure on the skin, and the conditioned stimulus is the situation of the pressure in the conditioned group. For there are reinforcement and extinction of the response, and generalization of unconditioned stimulus and quasi-secondary reinforcement are also found. It may be considered that this conditioned response is formed in some level of the central nervous system with spatial extension.
Thorndike and his collaborators have reported their highly suggestive studies on the effects of verbal reward and punishment upon multiple seletive learning. But, they observed these effects only in short trial sequences. The writer has tried to examine the effects in relatively longer trial sequences (22-29 trials). Furthermore, some modifications of their old experimental procedurehave been attempted in the present study. These attempts may solve some issues concerning the effect of punishment in the experimental situation of this type. Experiment I. The Thorndikian multiple selective situation was used. The main findings were as follows: While several defects in the procedure of computing the ‘net effect’ of verbal reward and punishment from the ‘empirical base line’ had already been pointed out, another defect was demonstrated in the present experiment. Namely, the repetition scores of choices upon which their appropriateness had never been informed (NC) indicated significant individual difference, whereas the repetition scores of both rewarded and punished choices (RC and PC) indicated no such significant individual differences (Table 1, 2). These results may fail to justify the concept of ‘base line’, the computational base of which is NC. Experiments II and III. In order to isolate the effect of verbal punishment in a Thorndikian experiment, the alternative selective situation may be more effective than the multiple one. Because, in the former situation, the probabilities of occurrences of reward and punishment are equal in the first trial, and therefore the effect of the relative frequency of verbal reinforcements can be eliminated to some extent. The control-group technique was used and it had also never been used in this type of experiment. The application of this technique may make the results sensitive enough to isolate the effect of punishment from that of reward. The main findings were as follows: 1) RC might be said to correlate with the frequency of correct choices throughout the trial series, whereas PC indicated no such simple relation to either RC or the frequency of correct choices (Figs. 2-5, Tables 6, 9). This may suggest the type of learning under these situations. 2) The rewarding-punishing condition (method of contrasted reinforcements) was more favorable for the acquisition rate than the rewarding (unpunishing) condition (Tables 4, 5.) 3) In the long series of trials, the rewarding condition was more favorable for the acquisition rate than the punishing condition in the first half of the trial series, but the reversal of this relation occurred thereafter (Tables 7, 8).