Überprüfung von Steuerungssystemen zur Strophenanwahl der Amsel durch digitale Simulierung

摘要

Output analysis and input-output experiments had given evidence for five sequential effects occuring independently of each other in the singing behavior of blackbirds (Fig.1). Recently these investigations resulted in a model which summarized the interaction of five functional components evaluated from the found effects; thereby the model explained the data processing controlling the sequential patterning of the behavior (Fig. 2) — During the investigations reported here, this model and alternative models were tested by means of computer programs. Main question: Would the models generate sequences of patterns equivalent to the sequence of vocal patterns produced by a blackbrid? Results:1)Only the original one of the tested models generated sequences of patterns showing the specific regularities and their characteristic variations corresponding to natural blackbird songs. This brought evidence for amultiplicativemode of interaction between the “component of throttling-back” (Drosselungskomponente) and the other four components; furthermore for a terminal functioning of the “extreme-value-passage” (Extremwertdurchlaß) (Figs, 2–4 and 10).2)Concerning the “component of periodicity” (Periodikkomponente) the mechanism of a “reset of periods” was falsified while the mechanism of a “continued periodicity” could be made evident. This implied: The periodicity of the single pattern (phrase) runs over the course of the whole song: the proceeding of the oscillation takes place with each output realisation (Fig. 5).3)Variations of values of the particular parameters given to the different components during the testing programs showed that a) extreme values resulted in “unnatural” sequences of patterns; b) only values lying within distinct intervals generated sequences containing the whole set of the postulated effects; c) exactly with those parameter values, with whichoneof the effects could get produced in a natural form, also the other effects were generated correspondingly optimal (Figs. 6–8).4)The differential variation of the values of the particular parameters given to the components (during the testing programs) allowed a lot of predictions concerning the consequences of increasings or decreasings in the strength of the different components. These predictions have to be tested by experiments with natural systems (bl