The present study is a genetic approach to the analysis-synthesis process that characterizes structural cognition of humans. It is hypothesized that interaction in the group situation would systematize the complex analysis-synthesis process and create internal structural systematic cognition in each individual. The subjects solved the task of combinating random figures. (FIG.1-1). The systematic solution of this task needs to differentiate following two levels of analysis-synthesis process and to systematize them; (i)analysis-synthesis on each figure, (ii)analysis-synthesis on the relation of all six figures. In the first experiment, the subjects were divided into three groups: (1)10 individuals who were instructed to verbalize about each figure(V group), 10 individuals without such an instruction(NV group), and(3)9 pairs who were instructed to work together but were not asked to verbalize(G group). V group committed less errors and got the solution suddenly and "by insight". In the other hand, NV group made more errors and reached the criterion gradually and "by trial and error"(TABLE 1-1, FIG. 1-3). It took much time for V group to do every trial at the early stages of the task, but it rapidly decreased. But NV group performed rapidly and monotonously through all the stage of the task. (FIG. 1-4). In G group there were more verbalization and more "momentary inhibition of directly reinforced response"(making responses time larger)through all the stage of the task than in NV group. It is suggested that the origins of V's behaviors can be found in group interactions. In the second experiment, what interrelations in the group would create such interactions were examined. One group that consisted of individuals was compared with three groups that had different interrelations. I group(individual group)were given the same instructions as NV in the first experiment. Gc(control)were instructed only to work together. Gd(decision group)were told to present cards after a decision by the pair. Gv(verbalization group)were told to present cards and name each of them together. All three pair groups committed less errors than I group(TABLE 2-1). But Gc's behavior patterns were similar to the I's. They were apt to present cards monotonously and hastily without any agreements and inhibitions from their partners and without verbalizations. Gd took much time at all stages of the task(TABLE 2-3)and had more inhibitions. They often uttered "wait a moment" to their or to themselves. Though the rate of their verbalization of figures was about a half of Gv's, but more than Gc and I. Gv's regulatory response time per every trial showed the same decreasing pattern as in V(FIG 2-1). Thus different inderactions led to different performances and it is showed momentary inhibition has a relation to verbalization. But how the elements of interactions had relations to systematization of the analysis-synthesis process was not directly revealed. This should be examined in the future by corrected procedures.