Search for a command to run...
The progress of the quantitative study of behaviour depends on the use of concepts which are qualitatively precise. Such concepts as have so far been introduced are deficient in this respect, and hence confusion has arisen in the literature. An attempt is made to meet the problem by establishing simple units from which higher level concepts can be constructed. It is proposed to classify physical variables in several ways, and so define sets and sub-sets of them as to constitute precise concepts. Considerations from physiology and zoology make it desirable to use simple units of overt behaviour. General considerations suggest that we can usefully attempt the cross-classification of messages in the central nervous system in terms of their sources, channels and destinations. Two main kinds of behaviour mechanisms are defined as 'Determination' and 'Adjustment' respectively. The former concerns the decisions between basic patterns of behaviour, the latter the efficient modulation of the chosen patterns to the current environment. Here we are concerned only with Determination mechanisms. Attention is directed to the variables influencing a single 'reaction' or unit of overt behaviour. It is shown that, to avoid circular reasoning, it must be possible to measure the quantitative variation associated with such a unit without reference to causal factors. This dictates the necessity of choosing recognisable units, and of studying the kinds of measurement that can be made on them. An attempt is made, based on the criterion of recognisability, to define a simple unit of overt behaviour, called an 'Act'. Such a unit is to be made up of activity in several effectors, and implies a unit mechanism of co-ordination in the nervous system, to be called an 'Act Centre'. No anatomical assumptions are implied. For simplicity, where a reaction consists of several components, not always occurring together, each component is defined as a separate Act. Consideration of the various kinds of quantitative information obtainable about such an Act leads to the simplifying hypothesis that there are only two behaviourally measurable variables at the Act Centre, to be called 'Tendency' and 'Intensity'. It is supposed that variation in the former can be detected by measurement of frequency of occurrence of the Act or by a number of types of measurement based on time periods great relative to the duration of the Act, while variations in the latter can be detected by differences in the performance of the Act each time it occurs. Both variables are supposed to be continuous. It is recommended that behaviour phenomena be discussed in terms of communication Here, in particular, it is proposed to discuss the distribution of different sets of messages to different Act Centres in the same animal, as an approach to the problem of reaction-specificity. It is assumed, for simplicity, that change in physical variables will affect an Act's Tendency or Intensity only if messages signalling their state can reach the corresponding Act Centre. Energy couplings are regarded as of minor interest in the present study. A given Act Tendency can be expressed as an unique function of a number of physical variables. These are divisible, for a given Act, into what are called 'Effective' and 'Ineffective Variables. Variation in the former results in variation in the measures of Act Tendency or Intensity; variation in the latter does not. The operational criteria are discussed. It is shown how groups of variables can be progressively analysed from this point of view, and it is suggested that the theory of path coefficients be applied in behaviour. It is now possible to develop the classification of variables by the destination of the messages corresponding to combinations of their states. After brief reference to the principles of set algebra, various sets of variables and messages are considered, in particular the set of all Effective Variables for a given Act and the set, which can be generated from this, of all messages which can reach the corresponding Act Centre, to be called 'Act-Available Message Set'. Further classification of variables in terms of their status for the different Acts of an animal leads to the concepts of 'Generic' and 'Specific' Effective Variables. Other possible sets are discussed. A consideration of reactions with several component Acts all having identical Available Message Sets, leads to the concept of a 'Superact Centre'. The possible variables at such a Centre are discussed. In general, the question is asked whether Available Message Sets are identical for the Tendency and Intensity of the same Act or Superact. Consideration is also given to other forms of act association under the headings of Compound Act and Reaction Chain. It is pointed out that Effective Variables for a given Act (Tendency or Intensity) may or may not be Ineffective given certain states of other Effective Variables for the same Act. Variables are divided into two sets-'Independent' and 'Interactive'-on this basis. It is suggested that these sets correspond to what have been called Motivational and Releasing Factors. This approach, and another more fully detailed elsewhere, lead to a precise characterisation of the term 'Instinctive'. A classification by sources and channels leads to the two main sets of 'S' and 'N' variables, corresponding roughly but not in detail to TINBERGEN'S External and Internal Factors. Messages about the state of S variables are supposed to reach any Act Centre they may reach only through the exteroceptors, and to refer to immediately previous states, The set N is defined by exclusion; it is pointed out that many further (intersecting) sub-sets of N can be defined. Special attention is paid to the set of 'Instructions' (which may but need not be constant), the set of N variables associated with previous activity, and the set associated with the state of the Milieu Interieur: sub-sets of the latter are considered. The present approach is based operationally on trials and tests made in principle independently of time. The combination of this approach with the study of time sequences of behaviour can make possible an equally precise discussion of the 'higher levels' of behaviour, of 'centres' controlling patterns of behaviour less simple and specific than Acts or Superacts, of the interactions and channels between Centres and of such notions as 'conflict between drives'. The lines of such progress are briefly sketched. Finally it is pointed out that by means of the cross-classification of variables discussed here it is possible to define a very large number of different sets, each of which can constitute a precise concept, which by reference to Tendency or Intensity can be used in quantitative study. It is hoped that this sort of approach will obviate mis-understandings and facilitate empirical progress.