User-controlled analysis of behavioural time series by digital computer (author's transl)

Abstract

1. Both the organization of behaviour and communicative interactions can be established by analyzing behavioural time series. By means of an analysis of this kind, conclusions about the control of behaviour and the principles of interindividual communication can be reached. An analysis may be based on simultaneous and successive behavioural events. In these multi-channel time series the temporal arrangement of patterns (“Strukturierung”), the organization and the conditions determining the occurrence of behavioural events can be specified. 2. Data-analysis by digital computer permits rapid processing of a large quantity of material in regard to several respects. “Sets of data” can be stored, thus enabling the user to correct, alter, combine and finally analyze them with respect to various questions. For this purpose, a program system for the digital processing of behavioural time series (title “PROVED”) was developed. 3. The course of behavioural events is composed of temporal patterns which are ordered in a definite hierarchy (Fig. 1). Several quantitative and qualitative characteristics (“Muster-Merkmale”) can be attributed to each pattern (Figs. 2,4). With the PROVED-System, the temporal arrangement of behavioural patterns is reestablished by the computer according to the control orders of the user. These orders must take into consideration the physiological relevance that is to be confirmed by an analysis of time serics (Fig. 3). The orders controlling the temporal arragement as well as the asignment of pattern characteristics to sets of data are available for further evaluation. 4. The PROVED-System includes several procedures involving input, output, storage, administration, correction, assignment of characteristics, temporal arrangement and analysis of data. The examiner can select and combine these procedures at will using statements of a particular control language. 5. During the first step of data analysis, a fixed number of parameters for each pattern is determined, the number of parameters varying with the complexity of the problem. Any measured value which is related to the pattern under investigation can be regarded as a parameter. During the second step, multidimensional frequency distributions of crossclassified parameters are established which can be examined by means of statistical tests (third step) (Figs. 5,6). 6. Determination of the parameters by the examiner decisively influences the further course of the analysis. The control statements for one parameter consist of two parts: 1. quality of the parameter (type of pattern, temporal distance, etc.), 2. position of the parameter in the behavioural time series (pattern under investigation, immediately subsequent pattern, next pattern with the same characteristics, etc.). The latter is determined by a series of shifts of indices controlled by the user (Fig. 7,8). 7. The PROVED-System can be applied to the analysis of various behavioural time series (succession of sounds, recordings of movements, observations of behavioural acts, etc.). Depending upon the aim of the analysis (temporal arrangement, organization, control of behaviour or communication), spontaneous behavioural events, input-output experiments (stimulus-reaction) and social interaction can be evaluated.