Perhaps it is to some extent, but the many present-day developments are clearly continuations of the ways of thinking about animal behaviour that were pioneered by Lorenz, and within the diversity one can see a unity of approach or attitude that derives from a common background of general zoology: the animals are still regarded from the standpoint of natural selection, as creatures that can be understood only if seen against the background of their natural setting.

This attitude is most clearly apparent in the students who try to explain behaviour in terms of its function, or use behavioural data to help in working out the adaptive value of some feature of form or function. Tinbergen's attention has been devoted to this sort of problem in recent years. One of his students has worked on the question of why so many sea birds have white plumage: from a study of the feeding habits of the birds and the visual responses of fish to differently coloured models he showed that the whiteness of the birds is very likely an adaptation to a certain kind of fish-catching way of life (Unpublished work by Phillips cited to me in a personal communication by Dr. Tinbergen). Tinbergen and a team of co-workers have examined the question of why Black-headed Gulls carry the broken egg-shells from their nests after hatching (Tinbergen et al, 1962). In a preliminary study (unpublished) I had found that carrying of shells from the nest could be released by placing shells in the nests at any time during the breeding season (not just at hatching), and that the reaction was shown not only to egg shells but to a range of conspicuous objects. Tinbergen and his students systematically tested the responsiveness of the gulls to a wide range of objects and found two peaks in this range: the more conspicuous an object was against the background of the nest the greater its releasing value, and the more like a real egg shell it was the greater its releasing value. Real egg shells were better releasers than any of the other objects even though they were considerably less conspicuous than some of them. This suggested that two selection pressures have acted in the evolution of the response; one favouring responsiveness to conspicuous objects in the nest, the other favouring responsiveness specifically to egg shells. The first suggested a visual predator and there were several obvious candidates for this role in the region such as carrion crows and herring gulls. To test the possibility that removal of conspicuous objects from the nest is advantageous because such objects might catch the eye of an egg-predator, the following experiment was set up: a number of artificial nests were made outside the gullery and eggs were placed in these nests, some containing a conspicuous object and others not. After a time these nests were examined and it was found that significantly more of those containing a conspicuous object had been robbed of their eggs. This result confirms that gulls which tend to remove bright objects from their nests will, on the whole, raise more young than will gulls that do not. The eggs of gulls are also taken by mammalian predators such as hedgehogs and foxes which locate their food mainly by smell. This suggested that the feature of the real egg shell, in addition to its conspicuousness, which favours its quick removal from the nest, is its odour. Another set of artificial nests with eggs was set up but this time half the nests contained a freshly broken and unwashed egg shell, and the other half contained a freshly broken and washed egg shell, and this arrangement was left for one night. The result was that the eggs were found and eaten in a significantly higher proportion of the nests with unwashed (and hence smelly) egg shells. Again this confirmed prediction on the basis of the hypothesis that olfactory predation acts as a selection pressure.

This study illustrates how a question of the form ‘Why does the animal do so-and-so?’ can be put in different ways: it can be directed to the causes acting directly, here and now, to produce the behaviour and these can be divided into those external to the animal and those acting within the animal; it can be directed to the function served by the behaviour — in the natural life of the animal and this in turn leads to consideration of the evolutionary history of the behaviour; it can be directed to the individual history of the animal — how has this behaviour developed in ontogeny? In the egg shell study, investigation of proximate causes (external stimuli) led Tinbergen and his colleagues to reach conclusions bearing on the function of the behaviour and hence on its ultimate causes in the history of the species.

Other workers have directed their attention mainly to unravelling the organisation of proximate factors underlying behaviour and here we can make a division between those who are concerned to make correlations between external factors, physiological states or events and behaviour, and those who are concerned to work out the formal properties that any physiological mechanism must have if it is to account for a particular behaviour pattern. This division is not a hard and fast one; there are various combinations of method and interpretation. In the first group we should include, as ground floor members, the many descriptive studies that establish the quantitative and qualitative regularities of animal behaviour (e.g. Lind, 1959; Manley, 1960; Wickler, 1962). Then there are numerous studies of the effects of natural or experimentally induced differences in such things as the available external stimuli (e.g. Beer, 1961, 1962 a & b, 1963; Hinde, 1958; Lehrman et al, 1961), hormone levels in the blood (e.g. Warren & Hinde, 1959; Lehrman & Brody, 1960), tissue needs (e.g. Tugendhat, 1960a), previous experience (e.g. Tugendhat, 1960b). Notable combinations of observational and experimental techniques are to be found in the studies of ‘biological clocks’ (see Cold Spr. Harb. Symp. quant. Bio. 25, 1960) and studies of homing and navigation in birds (e.g. Sauer, 1961). A particularly important recent contribution that we might also include in this group is the work of von Holst and Saint Paul on electrical stimulation of the mid-brains of domestic chickens (von Holst & Saint Paul, 1958, 1960). By implanting electrodes in the hypothalamus of roosters and hens these workers were able to make the animals perform elements of various behaviour patterns by stimulating the C.N.S. electrically in different localities and with voltages of different strengths. By stimulating two places at the same time they were able to produce interaction effects between two different behaviour patterns, and by stimulating in the presence and absence of certain external stimuli they were able to show interaction between internal conditions and external stimuli. The results were elegantly interpreted in neo-Lorenzian terms although it is doubtful if this interpretation is as logically compelling as the authors would have liked it to be. Some features of the experimental technique were too crude for an accurate judgment to be made about the relation between the data and natural functioning. More precise investigations with electrical stimulation of the brain are being undertaken on herring gulls and ring doves by ethologists at Oxford.

In the second group of workers are first of all the students who have been influenced by Cybernetics and information theory — the models of animal functions, particularly the functions of the brain, that have been drawn up by making analogies with the operations of telephone systems and electronic computers (e.g. Wiener, 1948). By tracing the patterns of qualitative and quantitative relations between dependent and independent variables that describe a behaviour pattern, students such at Mittelsteadt (e.g. 1958) and Hassenstein (e.g. 1960) attempt to work out the range of formal models that would produce such relationships. If these models are genuinely different they will lead to different predictions, and testing of these predictions should eventually eliminate all but the only adequate model. The best known and most accessible example of this type of thinking in ethology is still the Reafference Theory (von Holst, 1954) I mentioned in the last section. Such formal or mathematical models for behaviour tend to remain untranslated into terms of flesh and blood. Until they can be their heuristic value is probably limited.

A second approach to a formal model of the underlying organisation of a behaviour pattern has been by way of statistical methods. The increasing availability of electronic computers and data recording devices has allowed the analysis of a great deal more of the detail of behaviour patterns than was possible in the past, and also more comprehensive statistical treatment of such data (see Beer, 1962c). By application of the techniques of multiple correlation analysis and factor analysis it has been possible, in a few studies, to reduce the variations in a large number of variables to correlation with variation in a much smaller number of variables (see Wiepkema, 1961). Reduction to such independent variables offers the possibility of refinement in the definition of tendency with respect to a particular behaviour pattern: if a tendency is identified with one of these variables it remains an operational concept but, at the same time, makes more profound conditions about the properties that a physiological model for the underlying mechanism of the behaviour should possess.

Although the use of electronic recording devices and computers increases the accuracy and detail of description immensely it has not removed the element of arbitrariness that enters into any analysis of behaviour. Measurement involves the division of behaviour into units that can be counted or timed. Behaviour is a continuum that can be broken up in an infinite number of ways; every classification, of necessity, ignores some ranges of variation — the notion of a complete description is really a contradiction in terms. The selection of units for measurement will not be made at random; there will be some prior idea of what is useful or relevant to the problem on hand. But here there enters a danger if one is not self-critical of one's criteria of selection; it is possible to ensure that you get the answer you expect or hope for by unwittingly sorting the information to that end. If, for example, you want to know how often two postures are associated and you find that one of these postures occurs in a number of situations that seem to have nothing to do with the other posture you might want to leave the latter occurrences out of account. Unless you can find an objective way of making the separation into two classes of occurrences — one relevant and the other not — your result is likely to reflect no more than the association that you think ought to obtain between the two postures. Before units are selected for measurement it should be insisted that the student spend as much time as possible in familiarising himself with the whole behaviour of the animal, preferably in its natural habitat or under conditions which reproduce the essentials (a question-begging word!) of the natural situation. If this is well done the student will be in no doubt about the complexity of the problem with which he has to deal and he will regard his efforts at measurement with a healthy scepticism.

The study of the ontogeny of behaviour has not received the attention that one might have expected from the emphasis that was placed on the innate elements of behaviour in the earlier ethological writings. Tinbergen (e.g. 1955: 102) now admits the term ‘innate’ to be useful only when applied to differences in behaviour between animals (the differences in behaviour between identical twins are clearly not due to differences in genetic factors) or between species (the differences between a man and an ape are clearly not due only to differences in up-bringing) and not when applied to processes of development which give rise to behaviour patterns in the life of an individual. However Lorenz has recently published a rather polemical paper (Lorenz, 1961) in which he defends his dichotomy between innate and learned behaviour and ridicules ‘many modern ethologists, mainly those publishing in English’ who have not had the courage of their convictions. There is little that is new in his arguments except perhaps his introduction of the concept of ‘information’. This is not defined with any precision but seems to mean something like ‘that which determines how a behaviour pattern develops’. According to Lorenz there are only two sources of such information as far as the individual is concerned: its genotype and its environment. Adaptedness between behaviour and the environment or way of life of the animal is obviously the product of natural selection and hence must be secured in the germ plasm. But this argument seems to be as far away as ever from helping in the understanding of the proximate causes or processes acting during the development of any piece of behaviour in an individual animal. The fact that adaptation must be secured in the germ plasm does not answer the question how adaptation is secured in the germ plasm in any instance. Perhaps Lorenz is not really concerned to ask such a question. But when he points out that for certain behaviour patterns to be adaptive they must be ‘environment resistant’, or for others that they must be performed perfectly at the first time of asking without prior experience of trial and error learning, and hence that the information for the developments of such patterns must reside solely in the genotype, he perhaps invites yet another confusion between ultimate and proximate causation — between the problem that was set by Nature and the nature of the solution in terms of the developmental processes in the individual. Further, the notion of information, as used in his paper, is so vague that in saying that the basis of adaptive responses must be genetically carried information Lorenz seems frequently to be saying no more than that the behaviour is the product of evolution, which surely is true of all behaviour in some sense. In spite of the suggestion of technical precision that it imports from the context of information theory, Lorenz's genetically carried ‘information’ seems to be little more than a blank cheque into which he can write any regularity which is species specific or species characteristic irrespective of the precise ontogenetic basis of such regularity. The translation of the ‘information rsquo; (whatever it is) into structure and function is a problem that cannot be decided in advance but can be worked out only by study of the development of individual behaviour patterns through ontogeny.

In a recent review Hess (1962) differs from me in being much more impressed by this paper of Lorenz's. We apparently differ also in our assessments of Hinde's examinations of unitary drive concepts; for Hess makes no reference to Hinde's papers on the subject, and uses the term drive throughout his review without any attempt to qualify its vagueness and ambiguity. I also object to his classification of Hebb, Lehrman, and Schnierla as ‘behaviourists’. This is either capricious or reflects profound misunderstanding of the work of these people.