F.E. Emery and E.L. Trist
Towards A Social Ecology:
Contextual Appreciation of the Future in the Present
The Early Detection of Emergent Processes
The second major difficulty in predicting the future states of large complex social systems, that of early identification of emergent processes, poses far more perplexing methodological problems. However, if social life is properly characterized in terms of overlapping temporal gestalten, then many of those processes that will be critical in the future are already in existence in the present. If this were not the case, it would be difficult to see how such processes could quickly enough muster the potency to be critical in the next thirty years. Thus, for instance, the conditions for World War I were laid before the end of the 19th Century, and correctly perceived around 1900 by such oddly gifted men as Frederick Engels and the Polish banker, Bloch (Liddell Hart, 1944, p. 26).
An obvious question must be asked at this point: ‘Is this not the same class of evidence that is the basis for extrapolative prediction?’ Such evidence does include some evidence of this class, but its most important additional inclusion is of processes that are not recognized for what they are. The early stages of a sycamore or a cancer are not obviously very different from a host of other things whose potential spatio-temporal span is
very much less; likewise with many in social life.
One suspects that the important social processes typically emerge like this. They start small, they grow and only then do people realize that their world has changed and that this process exists with characteristics of its own. Granted that there are genuine emergent processes (otherwise why worry about the next thirty years), then we must accept real limitations upon what we can predict and also accept that we have to live for some time with the future before we recognize it as such.
Yet it is not simply foolhardy to think that we may enable ourselves more readily to recognize the future in its embryonic form. There are almost certainly some regularities about these emergent phases. Social processes which, in their maturity, are going to consume significant portions of men’s energies are almost bound to have a lusty growth. They do not, by definition, command human resources at this stage, and hence their energy requirements must be met parasitically. i.e. they must in this phase appear to be something else. This is the major reason why the key emergents are typically unrecognized for what they are while other less demanding novel processes are quickly seen. A social process which passes for what it is not should theoretically be distinguishable both in its energy and informational aspects. Because it is a growing process, its energy requirements will be substantially greater (relative to what it appears to do) than the energy requirements of the maturer process which it apes. Because it is not what it appears to be, the process will stretch or distort the meanings and usage of the vocabulary which it has appropriated. The energy requirements may be difficult to detect not only because we lack scales for many of the forms of psychic and social energy, but also because a new process may in fact be able to do as much as it claims (e g TV to amuse) but do it so much more easily as to be able also to meet its own special growth requirements. The aberrations of linguistic usage are, on the other hand, there to see.
In trying to go further along these lines, we will first try to explain why there are probably significant although undetectable processes operating in the present. The explanation we will give itself suggests some methodologies that might aid early detection. For reasons of continuity we discuss these before tackling the logically prior question of whether there is any particular reason for trying to achieve early detection.
Complex social systems like the human body rely a great deal on the sharing of parts. Just as the mouth is shared by the sub-systems for breathing, eating, speaking, etc., so individuals and organizations act as parts for a multiplicity of social systems. Just as there are physiological switching mechanisms to prevent us choking too often over our food, so there are social mechanisms to prevent us having too many Charlie Chaplins dashing out of factories to tighten up buttons on women’s dresses (in Modern Times). I think that it is this sharing of parts that enables social processes to grow for quite long periods without detection. If they could grow only by subordinating parts entirely to themselves then they would be readily detectable. If, however, their parts continue to play traditional roles in the existing familiar systems, then detection becomes difficult indeed. The examples that most readily come to mind are the pathological ones of cancer and incipient psychoses. Perhaps this is because we strive so hard to detect them. In any case, healthy changes in physical maturation, personality growth or social growth typically follows the same course. Once we are confronted with a new fully-fledged system, we find that we can usually trace its roots well back into a past where it was unrecognized for what it was.
If this is, in fact, the reason for most or even some important social processes being undetected, then it suggests methodological approaches. Despite the redundancy of functions that the parts tend to have with respect to the role they play in any one sub-system, one must expect some interference in the existing systems as a new one grows. Angyal, from his analysis of competing psychological systems within an individual, has suggested a general classification that could serve as a basis for analysing social systems (Angyal, 1966). This is as follows:
1. When the emerging system is relatively very weak, it will tend to manifest itself only in the parasitical effects it has on the energies of the host system—in symptoms of debility. The host system will find it increasingly difficult mobilize energy (people) for their functions and there will be a slowing
down of their responsiveness to new demands. The balance of forces may oscillate so that these symptoms occur in waves and make the functioning of the existing social systems less predictable.
At any time a social system experiences a fair amount of uncontrolled variance (error) in its operations. The reasons for an increase in this variance, of the kind we are discussing now, will typically be sought for inside the system itself, and may be taken to tighten up its integration. The unpredictable oscillatory effects are likely to encourage a wave of experimentation with new modes of system functioning. All these symptoms have behavioural manifestations and are hence open to study. The methodological strategy of operational research is that of proceeding via analysis of the variance of systems and this would seem particularly appropriate here.
2. When the emerging system is stronger but still not
strong enough to displace the existing system, we can expect to see symptoms
of intrusion. What breaks
through are social phenomena, like the swarming of adolescents at the English
seaside resort of
3. When the emerging system has grown to be roughly in balance with the existing systems there may be mutual invasion. At this stage it should be obvious that there is a newly emerging system but mutual retardation and the general ambivalence and lack of decisiveness may still lead the new system to be seen simply as a negation of the existing system. The methodological task is to identify, in the chaotic intermingling of the systems, characteristics of the new system which are not simply an opposition to the old. Once again we find that this is not an entirely new methodological problem for the social scientist.
The Lewinians gave considerable attention to this in their studies of ‘overlapping situations’, for instance adolescence, when new and old psychological situations are frequently invading each other. Baker, Wright and Gonick (1946) specified five dimensions that they found helpful to sort out what was being done to what, by what. These dimensions are consonance, potency, valence, barriers and extent of sharing of parts.
(a) Relative consonance. Two or more overlapping situations requiring behaviour from the system that is more or less congruent. The degree of consonance ranges from identity, where the same behaviour meets both situations, through consonant where different behaviours are required but they are non-interfering, interfering to antagonistic..
(b) Potency. The influence of one situation relative to all simultaneously acting situations.
(d) Barriers. The relative difficulties confronting the system as it tries to make progress in the different situations.
(e) Extent of common parts.
With the aid of these dimensions, they were able to spell out many of the behavioural properties of invading systems. These conceptual dimensions have been sufficiently well defined to permit ready translation into other theoretical schemes.
The fact that early detection may be possible does not in itself make it worthwhile pursuing. The fact that early detection increases the range of responses and hence the decree of control a system has over its development does make us interested. There are facts about the growth of social change that suggest that each unit step in the lowering of the detection level increase in the time available for response. Put another way, it would yield a disproportionately richer projection of the future from any given time.
The next points I wish to make by referring to Figure 6. Let lines A and B represent two courses of growth over time. If social processes typically grew in the way represented by curve A then we might well feel that early detection was not a pressing problem. At this steady rate of growth, we might expect that when the scale got to the level of ready detection (D on the vertical axis) we would still have the time c - a (horizontal axis) in which to aid, prepare for or prevent the new system getting to critical size (Con vertical axis). All of this is simple enough, and the assumptions do not seem unreasonable because so many of the changes in the physical world and in our physical resources do grow in something like this manner.
In fact, a great many ot the growth processes in social systems appear to be more like that represented by curve B. These growth curves are common enough in all living populations (and some physical systems) where each part has
powers of multiple replication, but in this case we are primarily concerned with recruitment of existing parts to a new social system. What appears to contribute most to the prevalence of type B growth curves in social systems is the fact that these possess the property of highly developed symbolical communication. What is absent (because it is past, distant, or as yet only anticipated) can be represented by one part to the other parts. Their mutual co-ordination and regulation is vastly extended, and so is, as a result, the contagion of changes. One important implication of this is that a new system may, long period of slow and undetectable growth in the interstices of the society, suddenly burgeon forth at a rate which produces a numbing effect on the society, or at least drastically reduces the range of responses to it. The general notion may be explicated by again referring to Figure 6. If the point of critical size is somewhere near where I have marked in C, then it is in the nature of the type B curve that there will be less time between detection and critical size than would occur with a type A growth curve: i.e. T(c — b) <T(c — a).
Although, in this section, I have concentrated on the early detection of emerging systems, the present line of argument has implications for the fate of rapidly growing systems. The sort of growth that occurs between detection at point D and point E on Figure 6 can only too easily be seen as a type A growth. Even if the growth up to point D is reconstructed, the curve 0 to E may be seen as a pure exponential growth curve which will continue on at an increasing rate of growth towards point F. This has been well illustrated by Price (1961). Bringing together statistical evidence on the growth of science he shows that it has the characteristic of the curve 0 to E. This characteristic seems to underly the recent scientific ethos that the sky is the limit for scientific growth. However, he argues that the next stage of growth will be like the curve F to B, not a continuation of the curve from the intersect with D to F.
It is indeed apparent that the process to which we have become accustomed during the past few centuries is not a permanent feature of our world.... The normal expansion of science that we have grown up with is such that it demands each year a large place in our lives and a larger of share of our resources. Eventually that demand must reach a state where it cannot be satisfied, a state where the civilization is saturated with science (p. 113).
For science in the
only about thirty years must elapse between the period when some few per cent of difficulty is felt and the time when the trouble has become so acute that it cannot possibly be satisfied... we are currently in a period in which the onset of a manpower shortage is beginning to be felt… (115-6).
To this I can only add the obvious point that the method of study proposed by Price should include our preceding proposals. The decline in growth rate may occur not only because there is a limited supply of recruitable parts, but also because new systems are competing for existing parts.
Once again we find that elucidating the general nature of social changes is a fruitful way of identifying methodologies for furthering our ability to predict change in individual social systems or processes. The sigmoid type of growth curve (i.e. our B curve) has been a potent tool in the study of all types of living systems.
There remains a further general class of methodologies for early detection. These derive in the first place from the fact that man is not just a symbol-user in the way we have been discussing. His fundamental relation to his environment is a symbolical relation.
… the function of the so-called mental processes is essentially a semantic one. By this we mean that ‘psychological contents’ function as symbols and the psychological processes are operations with these symbols (p. 56). In the psychological realm life takes place, not through the interaction of the concrete individual with a concrete environment - which is only tangential - but by the interaction of symbols representing the individual and the environment (Angyal, 1941, p. 77).
As Tomkins has argued, our present knowledge of man suggests that if our perception mirrors nature, our consciousness is a mirroring of this mirror by the conceptual ordering of our memories.
… afferent sensory information is not directly transformed into a conscious report. What is consciously perceived is imagery created by the organism itself. The world we perceive is a dream we learn to have from a script we have not written... Instead of putting the mirror to nature we are... putting the mirror to the mirror (Tomkins, 1962, Vol. 1, p. 13).
The essential adaptive advantage is that the error inherent in this process makes learning possible. For our purposes the
relevance is that man’s responses are to the world as he symbolizes it and not directly to the world as it presents itself to his eyes, ears, etc.
In the second place, while this mechanism of consciousness (awareness of awareness) is a condition for learning, the learning itself is not conscious (certainly not necessarily conscious). Thus man’s symbolical representation of the world may change to represent changes in that world without his being conscious of the change. In so far as he is unaware of these changes they may remain unrecognized, or, if manifested in his behaviour, be puzzling, trivialized, or segregated parts of his projected world picture.
I have dwelt on these properties of the individual human being because they are basic to any joint human activity whatever the scale or complexity. On available evidence, it would seem that men live and have always lived in a cultural world which is created and maintained by the symbolic transformation of the actual world and the imputation or projection thereon of the meaning and values by and for which we live. My second point about individuals seems also to hold social systems, namely that the social symbols, the myths, beliefs, values, language, fads and fashions change without any necessary awareness of what the change means or to what they correspond. More concisely, there can be awareness of world changes without awareness of that awareness and this awareness can be manifested in man’s communicative behaviour as well as in his other behaviours. When these manifestations are recorded in oral traditions, in art forms or writing, it is theoretically possible that analysis of the records will reveal the existence of social processes which existed at the time, were sensed and lived with but not consciously grasped. At least three methodologies of different levels of generality have begun to emerge here. For convenience we label them as follows: (a) symbol analysis; (b) value analysis (c) analysis of linguistic usage.
We use the term symbol analysis to refer primarily to the methods of Jung and his followers. On the same assumption that basic changes in the life conditions of large groups may he -
detected in symbolic changes, Bion has speculated that we might be able to develop a method of inferring such basic changes from statistical fluctuations in psychosomatic symptoms (as unconscious individual symbolizations) and in the value of money (as in part reflecting aggregate psychological valuation) (Bion, 1961, pp. 105-113). This approach cannot be ruled out. The ethnologists and ecologists have together shown the nearly ubiquitous nature of symbols in living populations and their contribution to the natural selection of populations. Since this, it has been difficult to write off the possibility that the human species might have evolved through the use and selection of some similar innate cognitive programmes involving ‘perceptual concepts’ (Arnheim, 1954).
Less tentatively, we can accept the possibility that cracks and repairs in man’s umbrella of symbols might well presage the obvious emergence of major social processes by a long period of time. Neumann (1966), Marcuse (1956) and McLuhan (1964) made much of the notion that signs of our present condition were present in the painters, poets and writers of fifty years ago. As might be expected, McLuhan is particularly outspoken on this. He quotes Wyndham Lewis as writing: ‘The artist is always engaged in writing a detailed history of the future because he is the only person aware of the nature of the present.’ To this he adds his own judgement, that ‘the artist is the man in any field, scientific or humanistic, who grasps the implications of his actions and of new knowledge in his own time (McLuhan, 1964, p. 65). For these reasons, McLuhan sees his own method of detecting the future in the present as an application of the analytical techniques of modern art criticism. Just because these methods are esoteric, we cannot afford to ignore them.
The analysis of values has already been touched upon because this, like the analysis of symbols and linguistic usage, offers a radical reduction in the complexities with which would have to deal. In each of these we would be using men themselves as a filter of what is important.
The analysis of linguistic usage is at one level a commonsense way of sensing the way a person is developing or the way a
people are tending to go. The very way in which people are speaking about things is often a valid indication of changes in the way they are looking at the world, even though they insist in all honesty that they have in no way changed their views. This method is a basic ingredient of psychiatric practice. At the social level, it has been applied to the content analyses of films, women’s magazines, etc., and, more intuitively, to tracing out the subtle shifts in the meanings of key concepts like ‘work’, ‘leisure’ and ‘justice’ (Arendt, 1958). Marcuse has given us a profound analysis of the relation between experience and linguistic usage. He sets the methodological goal of linguistic analysis as that of ‘analysing ordinary language in really controversial areas, recognizing muddled thinking where it seems to be least muddled, uncovering the falsehood in so much in normal and clear usage. Then linguistic analysis would attain the level on which the specific societal processes which shape and limit the universe of discourse become visible and understandable.’ (Marcuse, 1964, p. 195). Drawing upon the empirical studies of Karl Kraus, he specifies some of the features of the method:
For such an analysis, the meaning of a term or form demands its development in a multi-dimensional universe; where any expressed meaning partakes of several interrelated, overlapping and antagonistic ‘systems’. For example, it belongs:
(a) to an individual project, i.e. the specific communication (a newspaper article, a speech) made at a specific occasion for a specific purpose;
(b) to an established supra-individual system of ideas, values and objectives of which the individual project partakes;
(c) to a particular society which itself integrates different and even conflicting individual and supra-individual projects’ (Marcuse, 1964, pp. 196-7).
It will be noted that these are methods of gathering information about the different levels of system competition which we presented as the general model for early detection.
I mentioned earlier that these methods offered a reduction in the complexity which had to be coped with, because men will, if unwittingly, tend to symbolize the relevant change and filter out for themselves the relevant changes. If acting consciously, they will typically see things through the ideologies of their times. This is, however, only a relative reduction. A profound reduction may occur with a Blake or Joyce. However,
this may be of little use. How do we recognize a Blake or Joyce in our midst or understand what they are saying when they probably don’t understand themselves? If these methods of analysis are to be effective, we shall still have to deal with samples of data that are very complex relative to our current analytical tools. It has been recognized that modern computers bring us within reach of the point where the predictions of such highly perceptive individuals as McLuhan, Marcuse and Neumann can be converted to testable hypotheses. Stone’s (1966) General Inquirer programme is a step in this direction, but it would still be necessary to identify the kind of system which one suspects is emerging. In other words, these methods complement the perceptive intuitive minds.
An example may illustrate and draw together some of the methods I have discussed under the heading of ‘early detection’. It is desirable, of course, that we concentrate upon the general principles, not the concrete features of the example. Assume, for instance, that a resurgence of Nazism is thought to be likely in a given country. Early detection is desired in order to allow counteraction and yet it is expected that any such embryonic movement would actively seek to avoid detection until it had recruited enough strength to challenge existing social systems and overcome the conceivable counteraction. The recruitment of any particular individual can be hidden because recruitment does not entail total subordination to the party. The recruit can still continue to function as civil servant, waiter. Husband, etc., although there may well be some falling off in the enthusiasm with which he now carries out his duties or even some change in the way he conducts them. However, even if each recruit in turn recruits several others each year, the growth rate, while sigmoidal, would put off the achievement of critical mass in a large nation for a long time (and of course increase the probability of detection). Therefore, in a large nation, a resurgent Nazi party would need to use the mass media. (Clandestine leaflets, papers and wail slogans would intensify efforts at detection.) They would have to penetrate and use the media in a covert way in order to avoid detection. However, to use it all they would have to shape the media content and style so that it propounded their Weltanschauung. It is not impossible to do this and at the same time avoid detection and counteraction. The aim would be to reach and to nurture the
thinking of like-minded persons and these will tend to be more sensitive to low intensity messages than all but perhaps the most anti-Nazis. Secondly, people can learn from a large number of trivial cues without being aware of just what led to the learning. This latter counts heavily against the obsessed anti-Nazi. He may well come to the firm conclusion that a particular medium has Nazi flavour and yet be unable to put his finger on anything that constitutes evidence for demanding counteraction.
In this case, how would the methods of symbolical analysis help to test hypotheses about the emergence of such a concealed symptom? Briefly, they would involve some sampling of media content because of the sheer mass of material going through them. The sample, if it were to be at all sensitive, would have to be handled by computers. The computer programme would need to be so designed that it could detect metaphors of the sort that Jung thought central to Nazi thinking, values of the sort that McGranahan (1946) found to distinguish the Nazi Youth from the U.S. Boy Scouts, and the more complex problems of syntax, grammar, vocabulary and even typography which Kraus found so revealing. For practical purposes the last would have to be restricted to the controversial political universe of discourse where in any case the effects are more significant. By repeating the study over time it should, theoretically, be possible to determine whether there is an embryonic growing process, more than one centre of growth or simply unrepentant, unburied remnants. It should not be impossible to go beyond mere detection to inferring structural properties and system orientations that differ from assumptions based on past experience.
As an example this is not entirely satisfactory. The hypothetical social process is conscious of its ends, consciously striving to use the symbolic processes of the society and consciously seeking to avoid detection. The latter does not simply cancel out the first two features to make it equivalent to a non-self-conscious social growth. Hence, although in this case symbolical analysis can only be usefully employed when the weaker system is strong enough to start intruding, it does not argue against symbolical analysis at the earlier stage when all that is present are symptoms of pressure.
Summary of Methodologies Discussed in Chapters 2 and 3
In this section 1 have outlined the following:
1. Two aspects of the general methodological problem:
(a) to identify the system in terms of its members and the dimensions in which they are arranged
(b) to identify the characteristic generating function of the system.
2. Special methodological difficulties that arise with predicting the future of large complex social systems:
(b) early detection.
3. Methods that have been developed or proposed for overcoming these difficulties:
1. Ashby’s model for studying conditions for survival;
2. Models for studying subordinate goals (values), e.g. Ackoff-Churchman, Cantril;
3. Models for studying the starting conditions for change;
4. Method of identifying ‘the leading part’.
(b) Early detection
1. Model derived from the properties of weakly competing systems;
2. Sigmoid growth model;
3. Models based on analysis of symbols, values and linguistic usage.