006 on Communication Models in the Social Sciences Deutsch

American Association for Public Opinion Research

On Communication Models in the Social SciencesAuthor(s): Karl W. DeutschSource: The Public Opinion Quarterly, Vol. 16, No. 3 (Autumn, 1952), pp. 356-380Published by: Oxford University Press on behalf of the American Association for Public Opinion ResearchStable URL: http://www.jstor.org/stable/2745780 .Accessed: 12/06/2011 04:59

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .http://www.jstor.org/action/showPublisher?publisherCode=aapor. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

American Association for Public Opinion Research and Oxford University Press are collaborating with JSTORto digitize, preserve and extend access to The Public Opinion Quarterly.

http://www.jstor.org

On Communication Models in the Social Sciences BY KARL W. DEUTSCH

Models defined broadly by the author as munications models for the evaluation of cer- structures of symbols and rules designed to tain critical aspects of organizational behavior. correspond to the relevant points of an exist- An earlier version of this paper was pre- ing structure or process, are indispensable to sented at the Conference on Organizational the study and understanding of social organ- Behavior held under the auspices of the Or- izations. In this article, Dr. Deutsch lays ganizational Behavior Project at Princeton special emphasis on the implications of cor- University in March 1952.

IN RECENT YEARS, increasing attention has been paid to both the use of symbols in the process of thinking, and to the problems that arise when symbols are combined into larger configurations or models -particularly when these models are then used as an aid in investi- gating or forecasting events that occur in the world outside the thinking system. One important use of such models is in describing the behavior of social organizations.

The organizations to be described may be informal groups, they may be political units or agencies of government, or they may be industrial or business organizations. Each of these organizations is composed of parts which communicate with each other by means of messages; it receives further messages from the outside world; it stores information derived from messages in certain facilities of memory; and all these functions together may involve a configuration of processes, and perhaps of message flow, that goes clearly beyond any single element within the system. Whenever we are discussing the past or future behavior of such an organization, we must use a model for it, and much of the effectiveness of our discussion may depend upon the degree of similarity or dissimilarity between the model and the thing supposedly modeled.

Investigation of such models, therefore, is more than a mere play upon some fine points in the theory of knowledge. We are using models, willingly or not, whenever we are trying to think systematically about anything at all. The results of our thinking in each case will depend upon what elements we put into our model, what rules and

COMMUNICATION MODELS IN THE SOCIAL SCIENCES 357 structure we imposed on those elements, and upon what actual use we made of the ensemble of possibilities which this particular model offered.

In one sense the study of models, and the theory of organizations that could be derived from it, cuts across many of the traditional divisions between the natural and social sciences, as well as between the particular social sciences themselves. In all these fields, symbols are used to describe the accumulation and preservation of patterns from the past and their arrangement into more or less self-maintaining, self- destroying, or self-transforming systems. The resulting models are then used to describe further the impact of outside events upon such systems and the responses which each system makes to them. In this manner we use models in describing the behavior of a social group, or of a state, or of a nation, or of the memories and preferences that make up an individual personality. In a similar way, we use models in describing a system of logic, or in suggesting a theory of games, or in describing the behavior of an array of communications machinery.

The present paper will begin with a very brief discussion of models in the abstract and a brief summary of some work published previously on the subject. Its next section will consider some general functions of models and some yardsticks by which their performance may be evaluated. The following section will discuss some differences between genuine and pseudo-models in the social sciences and some possible misuses of mathematical language in the social sciences. The last and longest section will deal with some implications of communications models for the study of organizations. In particular, it will consider the implications of such models for the evaluation of organizations in terms of the degree of autonomy, self-consciousness, initiative, and capacity for growth and learning which may be found in the observable structure and behavior of each organization.

SOME EARLIER WORK ON MODELS

By a model is meant a structure of symbols and operating rules which is supposed to match a set of relevant points in an existing structure or process. Models of this kind are indispensable for the understanding of more complex processes. The only alternative to their use would be an attempt to "grasp directly" the structure or process to be understood; that is to say, to match it completely point for point.

PUBLIC OPINION QUARTERLY, FALL 1952 This is manifestly impossible. We use maps or anatomical atlases pre- cisely because we cannot carry complete countries or complete human bodies in our heads.

Each model implies a theory asserting a structural correspondence between the model and certain aspects of the thing supposed to be modeled. It also implies judgments of relevance; it suggests that the particular aspects to which it corresponds are in fact the important aspects of the thing for the purposes of the model makers or users. Furthermore, a model, if it is operational, implies predictions which can be verified by physical tests. A rough survey of major models used in human thinking in the course of history suggests that there has been a change in the character of the models that predominated in each period, and that it has been a gradual change from pictures to full- fledged models in the modern sense.

A more extended discussion of the general relationship of models to knowledge has been given in two papers by the present writer- "Mechanism, Organism and Society,"' and "Mechanism, Teleology and Mind."2 I do not propose to repeat the descriptions given in them, but I shall refer to them for the use of those readers who may wish to go more thoroughly into the background of the discussion that follows.

Among other things, these two papers contain discussions of a number of primitive models,3 and of the classical concepts of mechanism, organism, and of historical process.4 All three classical concepts were found to have major shortcomings. Mechanism cannot represent evolution; organisms are incapable of extensive internal rearrangement; models of historical processes have thus far lacked inner structure and quantitative predictability.

During the last two generations there has been a revival in a type of model related to mechanism. It has, however, been derived not from the classical image of clockwork, but rather from the image of circu- lating systems of liquids, pumps, and valves. An early classic applying the analogy of pumps, pipes, and valves to the circulation of the blood had been Harvey's On the Motion of the Heart. More recent applica-

1 Deutsch, Karl W., "Mechanism, Organism and Society," Philosophy of Science, vol. I8, no. 3 (July I95I), pp. 230-252; henceforth referred to as MOS.

2 Deutsch, Karl W., "Mechanism, Teleology and Mind," Philosophical and Phenomenological Research, vol. 12, no. 2 (December 1951), pp. 185-222; henceforth referred to as MTM.

3 MOS, pp. 232-233. 4 MTM, pp. I87-192; MOS, pp. 234-239.

358

COMMUNICATION MODELS IN THE SOCIAL SCIENCES 359 tions of the flow idea have been the concepts of circulation of money and goods in economics, and of ambitious would-be politicians in Vilfredo Pareto's theory of the circulation of the elite. Sigmund Freud's image of the libido of an individual, which when repressed in some manner will seek with increased force an outlet in some other aspect of his personality, has been accused by K. S. Lashley of representing a system of "psychohydraulics." In any case, the analogy between the movement of psychological moods and of bodily liquids-as in the ambiguous use of the word "humour"-has been old in medicine. While many of these flow models have been very crude, some flow models have reached a considerable degree of quantitative precision. Here we may think of the flow charts of industrial engineering and of traffic engineering, and, on a far more sophisticated level, of the output- input economics of Vassily Leontieff, and the economics of the school of John Maynard Keynes.

All these flow models, however, take their channel configurations for granted. They try to predict, so to speak, how much water will flow through a river bed, rather than when and where the river will dig itself a new one. Differently put, flow models cannot represent learning. Those aspects of reality which are connected with learning processes they usually must take for granted (thus, the technological "production function" or the popular "propensity to consume" in Keynesian economics). They cannot deal easily with changes in technology or tastes and have often tried to pass these unwanted babies to the other social sciences under the title of "non-economic factors."

Finally, another group of models has been derived from developments in neuro-physiology and communications engineering, as indicated by Walter B. Cannon's concept of "homeostasis"5 and Norbert Wiener's "cybernetics." Some of the properties and problems of cyber- netic models have been extensively discussed in Wiener's Cybernetics6 and The Human Use of Human Beings: Cybernetics and Society.7 Starting from these communications models I have tried to outline certain specific-and, it is hoped, operational-concepts of social com-

5 Cannon, Walter B., The Wisdom of the Body, New York: Norton, 1932. 6 Wiener, Norbert, Cybernetics, Cambridge, New York: Massachusetts Institute of Technology

Press and John Wiley, 1948. 7 Wiener, Norbert, The Human Use of Human Beings: Cybernetics and Society, Boston:

Houghton Mifflin, 1950.

PUBLIC OPINION QUARTERLY, FALL 1952 munication for the description of such problems or processes as recog- nition, novelty, values, learning capacity, consciousness, will, autonomy, integrity, openness, and growth.8

The present paper will consist of a few comments on general standards according to which models in the social sciences might be evaluated, of discussion of a number of specific pictures of models suggested by the theory of communication and control, and of the specific research approaches which they might suggest.

THE EVALUATION OF MODELS

We may think of models as serving, more or less imperfectly, four distinct functions: the organizing, the heuristic, the predictive, and the measuring (or mensurative).

By the organizing function is meant the ability of a model to order and relate disjointed data, and to show similarities or connections between them which had previously remained unperceived. To make isolated pieces of information fall suddenly into a meaningful pattern is to furnish an esthetic experience; Professor Paul Lazarsfeld once described it as the "Aha!-experience" familiar to psychologists.9 Such organization may facilitate its storage in memory, and perhaps even more its recall.

If the new model organizes information about unfamiliar processes in terms of images borrowed from familiar events, we call it an explanation. The operational function of an explanation is that of a training or teaching device which facilitates the transfer of learned habits from a familiar to an unfamiliar environment. If it actually does help us to transfer some familiar behavior pattern to a new problem, we may feel that the explanation is "satisfactory," or even that it "satisfies our curiosity," at least for a time. Such an explanation might be subjectively satisfying without being predictive; it would satisfy some persons but not others, depending on each person's memories and habits, and since it yields no predictions that can be tested by physical operations, it would be rejected by some scientists as a "mere explanation" which would be operationally meaningless.10

8MTM, pp. 192-216; MOS, pp. 243-252. 9 Paul Lazarsfeld at a meeting of the Columbia University Seminar on Methods in the Social Sciences, March 12, I95I.

10 Conant, James B., On Understanding Science, New Haven: Yale University Press, 1947; cf. also Bridgman, P. W., The Logic of Modern Physics, New York: Macmillan, 1927.

360

COMMUNICATION MODELS IN THE SOCIAL SCIENCES 361

Certainly, such "mere explanations" are models of a very low order. It seems, however, that explanations almost invariably imply some predictions; even if these predictions cannot be verified by techniques practicable at the present time, they may yet serve as heuristic devices leading to the discovery of new facts and new methods.11

The heuristic function of a model may be independent to a considerable degree from its orderliness or organizing power, as well as from its predictive and mensurative performance.

Little has to be said about the predictive function of a model, beyond the well known requirement of verifiability by physical operations. There are different kinds of prediction, however, which form something of a spectrum. At one extreme we find simple yes-or-no predictions; at higher degrees of specificity we get qualitative predictions of similarity or matching, where the result is predicted to be of this kind or of that kind, or of this particular delicate shade; and at the other extreme we find completely quantitative predictions which may give us elaborate time series which may answer the questions of when and how much.12

At this extreme, models become related to measurement. If the model is related to the thing modeled by laws which are not clearly understood, the data it yields may serve as indicants. If it is connected to the thing modeled by processes clearly understood, we may call the data obtained with its help a measure-and measures again may range all the way from simple rank orderings, to full-fledged ratio scales.13

11 For the concept of heuristics, see Polya, George, How to Solve It, Princeton: Princeton University Press, I944.

12 For the relationship of prediction to time series, cf. Wiener, Norbert, Extrapolation, Interpo- lation, and Smoothing of Stationary Time Series, Cambridge: Massachusetts Institute of Tech- nology Press, 1949.

In the natural sciences a yes-or-no prediction might answer a question like this: Will this paper burn or not? A qualitative prediction might answer the question: Will it burn with a bright yellow flame? A quantitative prediction might answer the question: In how many seconds will it heat the contents of a test tube to 400 Fahrenheit?

In economics or politics, yes-or-no questions might be: Will the Jones Corporation build a new plant? Will the Blank party put on a political drive? Qualitative questions might be: Will the Jones Corporation build a large and modern plant? Will the Blank party put on a drive for clean government? Quantitative questions might be: How large a plant will they have built by what date? How many meetings, poster, radio appeals will the Blank party use before next November, and when will the drive reach its climax? It should be remembered that the spectrum formed by these different kinds of questions might well be continuous.

13 Cf. Stevens, S. S., "Mathematics, Measurement and Psychophysics" in Stevens, ed., Hand- book of Experimental Psychology, New York: John Wiley, I95I, pp. 1-48.

PUBLIC OPINION QUARTERLY, FALL 1952 A dimension of evaluation corresponds to each of these four func-

tions of a model. How great is a model's generality or organizing power? What is its fruitfulness or heuristic value? How important or strategic are the verifiable predictions which it yields? And how ac- curate are the operations of measurement that can be developed with its aid? If we collect the answers to these four questions under the heading of the "performance" of a model, we may then evaluate the model still further in terms of the three additional considerations of originality, simplicity and realism.

By the originality of a model, or of any other intellectual contribution, we mean its improbability. Any idea, scheme or model may be thought of as the product of the recombination of previously existing elements, and perhaps of a subsequent process of abstraction omitting some of the traces of its combinatorial origin. The greater the probability, or obviousness or triteness, of a model, the more frequent is this particular recombination in the ensemble of combinatorial possibilities at the immediately preceding stage. Originality or improbability is the re- verse of this value.

A structure of symbols may be highly original but useless. Or a model may be original and perform well but require such a large share of the available means and efforts as to impair the pursuit of other work. Models are therefore evaluated for their simplicity or economy of means. But it turns out that the concept of simplicity is not completely simple. Francis Bacon declared in the controversy between Ptolemaic and Copernican Astronomy that, in the absence of conclusive data from observation, he would choose the simpler of the two hypotheses; he then duly chose the Ptolemaic system on the grounds that it required fewer readjustments of his everyday experience.14 Clearly, all notions of simplicity involve some sort of minimization problem, but what is to be minimized? Is it the number of unverified assumptions or dis- tinctions, as William of Occam seems to have taught? Or is it a number of calculating steps, as Copernicus suggested in praise of his system? Or is it the number of readjustments of acquired habits, as in Lord Chancellor Bacon's reasoning? If we could succeed in reducing the number of logical or calculating steps required in a model by introduc-

14 Frank, Phillip, Modern Science and Its Philosophy, Cambridge: Harvard University Press, I949, pp. 209-10.

362


ing a large number of suitable fictions, have we simplified the model, or have we increased the complexity of its assumptions? Would we not have simplified it according to Copernicus, but made it more complex according to Occam?

Perhaps the concept of simplicity itself is operational, and could be considered to resemble the concept of efficiency in engineering and in economics. Efficiency in economics denotes the attainment of a given result with the greatest economy in the employment of those means which are shortest in supply at each particular time, place, or situation. Since such supply conditions are historical, simplicity, like efficiency, would then be a historical concept. (If there is merit in this approach, we might wonder about the effect of the availability of cheap calculating aids and electronic calculators on the traditional stress on ele- gance in mathematics.)

If simplicity is measured by the economy of means in critical supply, then claims to simplicity on behalf of rival models or theories can be evaluated more objectively. We might also be able to predict cross-cultural disagreements about standards of simplicity, as well as changes in accepted standards of simplicity over time. Some of these considerations of simplicity could also be applied to the evaluation of research programs as well as to the measurement of organizational behavior.

The last consideration for evaluating a model or a conceptual scheme is its realism: that is, the degree of reliance which we may place on its representing some approximation to physical reality. Ac- cording to P. W. Bridgman, we may impute "physical reality" to a construct or model if it leads to predictions which are verified by at least two different, mutually independent physical operations. If we put this somewhat more formally, we may say that the statement "X is real" implies the prediction that "Predictions based on the assumption of X will be confirmed by (2 + N) mutually independent physical operations, where N is any number larger than one." The larger N- the number of independent confirmatory operations-actually turns out to be, the greater the degree of reality, or content of reality, we may impute to X. If N approaches infinity, we may be justified in treating X as real, though by no means necessarily as exhaustive. This approach implies the assumption that every real object or process is in principle knowable but may be inexhaustible. It may seem farfetched to define

PUBLIC OPINION QUARTERLY, FALL 1952 the concept of reality as a prediction about a series of other predictions, but it is a definition that can be tested, and I believe, applied to the evaluation of models, or of statements about the inferred inner structure of organizations.

GENUINE VERSUS PSEUDO-MODELS

Mathematical models in the social sciences may lose much of their usefulness through starting from too naive assumptions, or through the introduction of pseudo-constants: that is, magnitudes represented as constants in the mathematical equations, but incapable of being checked by independent and impersonal operations.

An example of sophisticated mathematical techniques prevented from becoming useful by regrettably naive assumptions is found in Professor Nicholas Rashevsky's discussion of changing levels of activity in social groups and of the "interaction of nations," in his Mathe- matical Theory of Human Relations."5 Professor Rashevsky assumes that members of the politically and economically "active population" differ from the "passive population" by hereditary constitution, and that the relative proportions of "active" and "passive" population then develop according to certain patterns of genetics and natural selection, depending largely on the numbers and density of total population. To what extent Professor Rashevsky's mathematical techniques could be applied to more realistic social and economic assumptions, and particularly to processes of social learning, in contrast to mere heredity, only the future can show.

A far more striking combination of relatively sophisticated mathematics with utter naivete in social science can be found in the work of the late George Kingsley Zipf.'6 According to Zipf the size of com- munities in terms of their number of inhabitants should approximate a harmonic series for each country, if its cities were ranked in the decreasing order of size of population. The closeness of the actual dis- tribution found to the theoretical harmonic series was then naively taken as an indicator of social stability. Thus, Zipf found that Austria

15 Rashevsky, N., Mathematical Theory of Human Relations: An Approach to a Mathematical Biology of Social Phenomena, Bloomington, Indiana: Principia Press, 1947, pp. 127-I48 and esp. pp. I48-49.

16 Zipf, George Kingsley, National Unity and Disunity: The Nation as a Biosocial Organism, Bloomington, Indiana: Principia Press, 1941; and Human Behavior and the Principle of Least Effort, Cambridge: Addison-Wesley, 1949.

364

COMMUNICATION MODELS IN THE SOCIAL SCIENCES 365 between the two world wars had too large a capital city and too few cities of middle size, and that the aggregate series of cities in Germany and Austria after Austria's annexation by the Nazis approximated a harmonic series more closely than before. From this he concluded that the German annexations of Austria and the Sudetenland in 1938 had increased the stability of Germany and the social and economic balance of her "Lebensraum."'7 This "mathematical" conclusion completely overlooked the fact that before 1938 Germany had already been a food- deficit area, dependent on exports for part of her living, and that Aus- tria as well as the Sudetenland had similarly been areas of food deficits, export dependence, and unemployment. What the Nazi annexations of 1938 had produced had been a merger of three deficits. The "greater Germany" of 1939 was more dependent on food imports and on export drives to pay for them than its component parts; the pooled threats of unemployment in all three territories were met by an armament drive, and food supplies and exports were sought by imperial expansion. What Professor Zipf has described as a harmonic series on paper, was in reality a situation of extreme unbalance and disharmony, which led within a year to a violent explosion in the German invasion of Poland and the unfolding of the Second World War.

Perhaps it is too much to expect at this stage that individuals should undergo the highly specialized training of the advanced professional mathematician and at the same time, the at least equally intense training of the experienced social scientist. The difference in the intellectual techniques in these two fields should not obscure the fact that both approaches represent full-time intellectual jobs. The main task of the mathematician is perhaps to concentrate on the single-minded pursuit of long trains of symbolic operations. He may start out on these from any set of given initial conditions, without caring overmuch, as a rule, why just these conditions or assumptions and no others were selected.

Much of the training of the historian and social scientist is just the opposite. He must become familiar with a very wide range of social and economic situations at different places and times. The outcome of this part of his training is at best a sense of relevance, an experience in judging which factors in a situation must be taken into account and which ones may be neglected without much risk of error. To be sure,

17 National Unity and Disunity, p. 196- 97 and figure i8.

PUBLIC OPINION QUARTERLY, FALL 1952 the social scientist can only benefit from analytic training. He does and should study economic, political, and psychological theory, and to an increasing extent mathematics and symbolic logic. Yet all analytic work in the social sciences is primarily tied to judgments of relevance, to evaluating the realism of assumptions and the appropriateness of models. This ability is not easily acquired by mathematicians in their periods of rest between or after their more arduous professional labors. And the advice to younger social scientists to study more mathematics should be tempered with the insistence that they will have to judge the relevance of their models against their fund of factual knowledge as social scientists; no amount of mathematical knowledge or advice can take this task from their shoulders.

The most hopeful answer to this problem at the present time lies perhaps in the development of teamwork between men who are primarily social scientists but who have had enough analytical training to put their problems into a form where mathematicians can go to work on them, and mathematicians who have had enough of a solid training in the social sciences to understand what the social scientists need from them, and how to select lines of mathematical treatment which will lead more closely toward reality rather than away from it.

Another source of trouble with mathematical models in the social sciences stems from the tendency to put arbitrary constants or coefficients into equations so as to make their results fit a known series of numbers or their extrapolations. Thus, Lewis F. Richardson's "Gen- eralized Foreign Politics" attempts to predict the armaments expendi- tures of two rival countries by equations which contain numerical coefficients for the "grievances" and the "submissiveness" of each country vis-a-vis the other.18

It is well known that any finite series of numbers can be fitted by more than one equation, and, on the other hand, that any result can be attained in an equation by introducing a sufficiently large number of arbitrary constants or coefficients. There is all the difference in .the world between such arbitrary coefficients and a constant in physics,

18 Richardson, Lewis F., "Generalized Foreign Politics; A Study in Group Psychology," British Journal of Psychology, Monograph Supplement No. 23, London: Cambridge University Press, 1939; cf. also the summaries in Quincy Wright, A Study of War, Vol. II, appendix 42, pp. 1482-83; Kenneth J. Arrow, "Mathematical Models in the Social Sciences," in Daniel Lerner and Harold D. Lasswell, eds., The Policy Sciences: Recent Developments in Scope and Method, Palo Alto: Stanford University Press, I951, p. 137.

366

COMMUNICATION MODELS IN THE SOCIAL SCIENCES 367 such as Planck's quantum constant h. Genuine constants in physics can be verified by impersonal physical operations of measurement, or by impersonally verifiable inferences from measurement. Such constants are the same for all physicists regardless of their sympathies or political beliefs, and they would be confirmed, in principle, by impersonal recording and measuring devices. The use of such operationally independent and verifiable concepts in models, such as in Bohr's model of the atom, is therefore quite legitimate. As long as social scientists cannot specify an impersonal set of operations for producing a numerical measure of "grievance" or "submissiveness," there will remain a grave suspicion that coefficients based on arbitrary estimates in such matters are somewhat akin to the "variable constants" familiar from the folklore of undergraduate humor.

To be sure, there may be cases where such mathematical pseudo- models may describe, however inadequately, some genuine intuitive insight of their author. It would be folly to suggest that only that is real which is measurable by present-day methods; the perception of Gestalt or the structural vision of a previously unrecognized configuration of phenomena all have their places among our sources of knowledge. In all such cases, however, it is the qualitative insights that are relevant, and not the mathematical disguises which they have prematurely donned.

SOME IMPLICATIONS OF MODELS OF COMMUNICATION AND CONTROL FOR RESEARCH ON ORGANIZATIONS

Communication and control are the decisive processes in organizations. Communication is what makes organizations cohere; control is what regulates their behavior. If we can map the pathways by which information is communicated between different parts of an organization and by which it is applied to the behavior of the organization in relation to the outside world, we will have gone far toward understanding that organization. This will be true of an organization composed of cells in an organism, or of machines in an automatic communications network, or of human beings in a social organization. The fundamental processes of communication and control in all these types of organization follow at least some of the same fundamental regularities. Some of these have been best understood thus far in the field of communication engineering as applied to machines. Others

PUBLIC OPINION QUARTERLY, FALL 1952 may be most familiar to the neuro-physiologist, or to the student of society. In whatever field they are first studied, they may be used to suggest questions for research in the others.

The general viewpoint of communications theory or cybernetics is thus not a substitute for concrete research, but it suggests a strategy for it. It suggests that certain questions are more relevant than others and that certain classes of data are worth obtaining and measuring, even at the cost of considerable effort. Particularly, it suggests that certain areas of experience in social life (such as the phenomena of novelty, creativity, autonomy, and consciousness) are worth examining anew-despite the fact that these questions have been long neglected by empiricists on the grounds that they are meaningless, and by literary social scientists on the grounds that they are ineffable. Some of the lines suggested for research and some of the kinds of data that might be worth collecting will be indicated in the rest of this paper.

Generally speaking, the communications approach suggests lines of attack in the study of organizations. First, instead of concentrating on the ostensible purpose of the organization, it will concentrate on two questions: how are the formal and informal communications channels of the organization connected, and how are they maintained?

This approach would agree with the viewpoint of Chester Barnard that organizations have laws of behavior different from those of their individual members :19

"The primary efforts of leaders need to be directed to the maintenance and guidance of organizations as whole systems of activities. I believe this to be the most distinctive and characteristic sector of leadership behavior, but it is the least obvious and least understood. Since most of the acts which constitute organizations have a specific function which superficially is independent of the maintenance of organization-for example, the accomplish- ment of specific tasks of the organization-it may not be observed that such acts at the same time also constitute organization and that this, not the technical and instrumental, is the primary aspect of such acts from the 19 Organization and Management, Cambridge: Harvard University Press, I949, pp. II4-II5.

"Take, for instance, a man who is always changing, or a corporation of by-laws, or to get farther toward the earth, take a whirlpool. This is a realistic thing to one who gets into it, and it seems real enough to anyone who watches it. ... A whirlpool is a situation in a body of water in whi6h there are comparatively stable uniformities of relations between streams of molecules of water, moving with increasing rapidity, spirally towards a center called a vortex, . . . If the molecules stop moving in this way, there is no whirlpool, because all there is to a whirlpool are streams of molecules of water moving in certain ways." (Italics in original.)

368

COMMUNICATION MODELS IN THE SOCIAL SCIENCES 369 viewpoint of leadership. Probably most leaders are not ordinarily conscious of this, though intuitively they are governed by it."20

Closely connected with our study of channel maintenance and channel layout would be our special interest in whatever feedback channels there are in the organization. What inner disequilibrium drives the organization at all to behave, that is, to change its relations vis-a-vis the outside world? What data concerning its own behavior are fed back into the organization, and how are they applied to the guidance of its actions? At what goal situations in relation to the outside world is the inner disequilibrium of the organization at a minimum? Are there one or several such minima, and is the information available to the organization adequate to reach them?

The second priority in the study of organizations deals with the problem of memory. What information is stored in the organization and at what points? How is it selected, broken down into parts, and recombined? Above all, where and how is it applied to the guidance of behavior? An organization is autonomous insofar as it remembers and is thus guided by its past, provided that this information recalled from memory is confronted or balanced with incoming information from the present state of the outside world and from the organization's own position within it. Where are the points where memories are stored, where are they fed back into behavior, and where are they balanced against current intake from the present? These are critical points for the control of the behavior of the organization as a whole, and if we understand organizational structure and behavior at these points we will understand much about the structure of the decision system by which the organization is guided.

A third line of inquiry would deal with the attachment of secondary symbols to the primary information used by the organization. What and how many secondary messages does the organization use about the primary messages moving through its channels ?21 To what extent does it monitor its own internal processes, and what use is made of the information thus obtained? In short, how conscious is the organization of its own knowledge and behavior and that of its parts? In other words, what internal data about the knowledge stored in the organiza-

20 Barnard, op. cit., p. 89. 21 For definitions of primary and secondary messages and the general concept of consciousness,

see MTM, pp. 205-208.

PUBLIC OPINION QUARTERLY, FALL 1952 tion and the behavior of its elements is abstracted and transmitted to steering committees, policy committees, executives, and governing boards of various kinds, and what use is made of it? What difference, that is to say, does it make in the probability of their subsequent decisions? What are the costs of this consciousness or internal monitor- ing-in resources, labor, and time of delay?

How realistic is this consciousness? How different is the ensemble of secondary symbols, as it comes to the policy boards or the executives, from the ensemble of primary messages that make up the life of the organization? To what extent do whole categories of internal information remain unmonitored or untransmitted, and thus ineffective at the policy level? How effective, for that matter, is the feedback of ostensible policy decisions in changing the organizations' actual behavior? In short, to what extent does the steering system of the organization remain blind to its surroundings or to the consequences of its own behavior? And, on the other hand, how often are leaders of an organization as vividly aware of its behavior as Cassandra was of the impending fall of Troy, but are equally powerless to change it?

To what extent are failures in the steering of an organization due to the absence of some crucial communication link not to the presence of some evil elements? Too often the theory of the improvement of organizations through "shake ups" or "purges" may turn out to be a lineal descendant of the medieval theories of demonic possession and of exorcism-and hardly more efficient than its ancestors. Firing em- ployees may seem less expensive than a thorough survey of the work- ings of an organization, but traffic jams are rarely improved by firing drivers, or telephone switchboards by firing operators.

Perhaps it is worth repeating here that none of the questions listed in the preceding paragraphs are rhetorical. All of them should be capable of being answered by procedures of surveying, mapping, and often measuring and counting.

In mapping the structure of preferences and priorities which govern the transmission of information within an organization (and thus indirectly its behavior toward the outside world), we should obtain a rank order scale of its operating preferences or values, and thus a map of its internal value system. This internal value system may be different from the collection of value images as they may appear among the collection of secondary symbols on the level of consciousness. Con-

370

COMMUNICATION MODELS IN THE SOCIAL SCIENCES 37I sciousness in regard to values, as to other things, may be false consciousness. The organization-that is, its leaders, or even all of its members and all of its permanent records-may "think" that the organization prefers one thing while in its actual operations it might pre- fer another most of the time.

Similarly, of course, this internal value system may be different from the presumable value system which an outside observer might deduce from its gross external behavior over a limited period of time, Such external behavior has been the favorite subject of study for the behavioristic school of psychology, but it does not predict the internal preferences governing the internal storage of information, or the internal rearrangements of the organization's physical resources or communication channels (and thus the changes in the probabilities of its eventual future behavior). During the Second World War, trains carrying valua- ble freight were protected against land mines by a preceding string of empty cars. Designers of land mines countered by constructing mines which did not explode under the first five or six impacts, but did respond to each impact by a change in an internal relay that insured their explosion under the seventh. Stepping repeatedly on such land mines would have surprised behavioral psychologists-and so would the problem of how to disarm time bombs.

What is suggested by such considerations are studies of the relationship between outward behavior and inner structural change, as well as the relationships between goal images and actual goal situations, and between value images and operating value systems. In particular, such operating value systems should be investigated for the extent of circularity, inconsistency, disequilibrium, and goal seeking that pre- vails in them. This might have applications to the personality of individuals as well as to the behavior of large social groups and particular social, economic, or political organizations. Too often, it has been customary to concentrate attention almost entirely on the consistent aspects of such systems. Ruth Benedict has described the culture of an Indian tribe as a single vessel, having one consistent shape, and then broken by contact with the white men.22 Similar emphasis on harmony and unity is customary in descriptions of western medieval culture,23

22 Benedict, Ruth, Patterns of Culture, New York: Penguin Books, 1947, pp. 19-21. 23 E.g. Randall, John Herman, Jr., The Making of the Modern Mind, revised edition, Boston:

Houghton Mifflin, 1940, pp. I7-106.

PUBLIC OPINION QUARTERLY, FALL 1952 in corporation reports to stockholders, and in election platforms of the major political parties. All these embody at best only limited aspects of reality. In studying more realistically the present structure and probable future behavior of organizations, societies, or cultures, we might rather ask: What are their strategic unbalances and inconsist- encies? What are the points where, and what are the rates at which, memories, resources, or strains are accumulated in such a manner as to lead to changes in probable future behavior?

LEARNING AND ORGANIZATIONS

When we deal with changes in inner structure which imply changes in behavior, we are not far from studying the processes of learning. Learning is here used to signify those inner changes in an organization that occur in response to some repeated outside stimulus and that change the system's subsequent response to it. However, the connection with the repeated outside stimulus need not be close. Items of information may be dissociated and abstracted in memory and then recombined to new patterns; simpler patterns may be abstracted from these, thus obliterating the combinatorial origin, and completing the production of novelty. These new internally created patterns may be fed back into the determination of new behavior and thus constitute initiative.

All these processes can be studied in structural or quantitative terms. The probability of novelty and the capacity for learning depend to some extent on the size of the ensemble of possible recombinations of separate items of information and material internal resources. In this sense, learning capacity in organizations depends on the range of internally available recombinations of knowledge, manpower, and facilities. These ranges should increase with the effectiveness of dissociation, that is, the extent to which knowledge or facilities could be subdivided into ever smaller independent pieces.

Unqualified, this statement would be a dangerous half-truth. For the smaller we make the items of information and the vaster we make the ensemble of their possible new combinations, the longer it would take to scan these ensembles, and the less would be the probability of extracting from them relevant or usable combinations within a limited time. A million monkeys typing on a million typewriters offer a range of possible recombinations that includes the collected works of William

372


Shakespeare, but the probability of obtaining them by this method within a limited time is infinitesimal. Learning through the recombination of large numbers of unrelated small items is therefore likely to be slow and may have to be guided heavily by additional selective criteria from outside the ensemble of recombinable items.

This would correspond to the description of "infant learning" given by D. O. Hebb, and to the same author's description of the learning of visual orientation by persons who acquired vision for the first time in adult life as the result of a corneal operation.24 Hebb contrasts this with what he calls "adult learning." This, in his opinion, consists of the recombination of a smaller number of larger subassemblies of memories or habits. The principle would apply mutatis mutandis to recombinations of subassemblies of material facilities. Infant learning, in this view, resembles building a house from bricks; adult learning resembles assembling it from prefabricated panels. Infant learning thus is slower but richer in possibilities; adult learning is more rapid within the limits of combinations of the subassemblies which are given.25

In the search for an increase in the capacity to learn, then, the problems are to find some optimum range between infant type and adult type learning; or to alternate between infant type and adult type learning at various stages within the same organization; or, finally, to establish strategic criteria of interest for the selection of promising configurations from the large ensemble of "infant" type learning for the purpose of developing the selected configurations more intensively by more nearly "adult" type learning methods.

From a structural point of view, the learning capacity of an organization is thus indicated by the amount of its uncommitted inner resources, by the extent of their possible dissociation into discrete items, and by the extent and probable relevance of its fixed subassemblies available for new recombinations. Somewhere between the extreme subdivision of items in infant learning and the relative rigidity of a small ensemble of large fixed subassemblies, there exists probably a region of optimum solutions, combining a high degree of richness and originality-that is, improbability-of new patterns with a high degree of speed in their selection, and with a high probability of their

24 Hebb, D. 0., The Organization of Behavior, New York: John Wiley, 1949, pp. 109-I34, esp. pp. 111-120.

25 Ibid.

PUBLIC OPINION QUARTERLY, FALL 1952 relevance to the challenges offered to the organization by its environment.

Structural data about learning capacity may be checked against observed data of learning performance. Here, research might deal with adaptations of the concept of the learning curve to the behavior of organizations, industries or countries. For a quantitative approach to the problems of learning and innovation in economic history, I should like to repeat an outline of concrete research possibilities which I listed in another paper:26

"We might measure the imitative innovation rate, that is the rate at which selected, standardized, technical innovations were accepted in given countries. We could select those cases for which sufficient records are available. Numerous examples suggest themselves: the linotype machine, the shoe machines of the United Shoe Machinery Corporation, the cash register, typewriter, ball-bearing, telephone, the Westinghouse railroad air-brake, and many like instances. We might then try, let us say in the case of the linotype machine, to find out when and where these machines were first introduced on a significant scale, how long a time elapsed between the first introduction and the stage when more than one-third the total circulation of a country was printed with their aid, and how much longer until the one-half and two-thirds marks were reached. Figures on the linotype machines might be obtainable from the corporation handling the license rights; data on newspaper circulation have been collected for some time now by the Editors' and Publish- ers' Year Book. A sample graph of the speed with which the printing of newspapers was modernized, in this respect, could be correlated with such 'economic' factors as the total numbers of newspapers among which the entire circulation was divided; the capital investment in the newspaper industry; the volume of paid advertising; the profits of the business; and numerous other variables. After some of these correlations have been made, it might be possible to see if there still remained any significant differences in the speed with which typesetting was being modernized in the United States, for example, as compared with, perhaps, France. This comparison could tell us for the limited subject of typesetting, not only whether French entrepreneurs were as quick (or slower) than their American counterparts in modernizing 26 Deutsch, Karl W., "Innovation, Entrepreneurship, and the Learning Process," in Cole, A.

H., ed., Change and the Entrepreneur: Postulates and Patterns for Entrepreneurial History, Cambridge: Harvard University Press, 1949, pp. 24-29; the citations are from pp. 25-27, slightly modified.

374

COMMUNICATION MODELS IN THE SOCIAL SCIENCES 375 their plants, but also, if they had been slower, by just how much they had lagged behind. Such figures as those for the linotype machine alone would mean little, but if the investigation is repeated for, let us say, the cash register (utilizing the records of the National Cash Register Company)-comparing the number of cash registers installed in a given country with the volume of retail sales, or with the number and turnover of retail stores above a certain minimum sales volume per store-the aggregate data would begin to take on significance. (The available figures in each case would, of course, vary with the obtainable statistics for each country.) The outcome of a comparison of these several innovation rates in different countries with those found to have prevailed in the United States might show that the French retail grocers or department store executives had a different rate of adoption of these techniques, that perhaps they were more quick to adopt one innovation than another; and, in the end, we might emerge with a quantitative measure of the differential rates of the acceptance of innovation. This sort of data would be essential for any significant statements about the performance of French entrepreneurship as compared to the American variety, or about the innovating performance of small owner-managed firms as compared to large corporate enterprises; or, to look at the matter from another side, about the innovating performances of all types of business in the classical decades of free trade, say, I846-I873, as compared with the classical period of capital concentration and protection, say, I890- I929.

Complementary to these investigations of imitative innovation would be an investigation of initiative innovation, that is, the frequency with which significant innovations originate and are first significantly applied in a particular country, or in special types of economic institutions. A study of imitative innovation might tell us how quickly American improvements in the technology of coal-mining were introduced in Britain during the era of private enterprise in the Igth century as compared with the rate of such introductions in the 20th century before I945, and with the rate of such innovations since that time under public management. A study of initiative innovation could tell us, at the same time, what significant innovations in coal-mining, if any, originated in Britain during each of these periods. Findings for both types of innovations could then be compared to similar data for coal-mining in France and Germany, with due allowances, of course, for the different geological and geographical conditions in each case.

PUBLIC OPINION QUARTERLY, FALL 1952 It would be against this international background of comparison of the performance of both businessmen and officials in other countries that the contribution of American businessmen in various periods of economic history could be properly appraised, or the chances better estimated for stimu- lating further innovations in a given country by means of economic aid from the United States."

Records of learning performance at successive times may themselves be made the basis for a second-order measurement of what Gregory Bateson has called the "deutero learning" of an organization.27 Deutero learning is second-order learning. Its measurement would measure the speed at which an organization learns to learn; that is, the rate of improvement in its performance when confronted with a suc- cession of different learning tasks.

From a structural investigation of the learning facilities and the learning capacity of an organization, and from measurements of its learning performances as well as of its second-order learning, we may derive a test for evaluating major over-all learning and behavior patterns of the organization. Has the learning of the organization been creative; that is to say, has it increased its ranges of possible intake of information from the outside world and its ranges of possible inner recombinations? Or has the learning of the organization been merely viable; that is, has it neither added nor detracted from the subsequent capacities of the organization for learning and self steering? Or finally, has the learning performance of the organization been pathological; that is, has the organization learned something that has reduced its subsequent capacity to learn or its subsequent capacity to control its own behavior? Such self-destructive learning resembles what moralists call sin, and perhaps what Socrates had in mind when he taught that no man would "err willingly."

Any shift from infant type to adult type learning has (on this showing, at least) a pathological aspect. As large subassemblies of information or resources are frozen and as major pathways of habit and routine become fixed, the speed and probability of the responses of the organization will increase in relation to a limited range of currently probable or frequent stimuli. But this observable improvement in ob-

27 Bateson, Gregory, "Social Planning and the Concept of Deutero-Learning," in Newcomb, T. E., and E. L. Hartley, eds., Readings in Social Psychology, New York: Henry Holt & Co., 1947, PP. 121-I28.

376

COMMUNICATION MODELS IN THE SOCIAL SCIENCES 377 vious competence in routine matters will have been bought at the price of reducing the range of available new recombinations within the organization; that is, at the price of reducing its inner resources of originality and creativity.

The same process may, of course, have involved the hardening of routines in the selection of kinds of information for intake, as well as in the allocation of priorities in its treatment, and in the attachment of secondary symbols to it for treatment in the feedback channels that carry consciousness. All this may result in a narrowing in the range of information that is permitted to enter the organization, or that is likely to become effective in it. Thus, the organization may come to run in blinkers of its own making, and cumulative losses of sensi- tivity may lead to partial blindness behind a facade of seemingly ever more mature performance.

There is evidence that this pathological aspect of adult learning was known intuitively to the early Christians. Their injunction to men to "become like little children" must have shocked disciples of Platonic philosophy, which extolled perfection and maturity, but it should be recognized as a legitimate and significant insight by modern theorists of learning and organization.

Considerations of this kind have a direct bearing on the valuation of the evolution and learning performance of countries and societies. Several times in history we find a conspicuous decline and partial dis- integration in the established routines and fixed subassemblies of formalized learning and established patterns of custom and civilization, coupled at the same time with a broad diffusion of some fundamental items of knowledge and technology and with the emergence of a larger number of smaller units or subassemblies of knowledge or of economic or political activity, offering a wider range of possible new combinations. The so-called "dark ages" in western Europe between A.D. 500 and iooo are an example of this process. The civilization of the Amer- ican colonies, and later the United States, between i750 and I850 is perhaps another. In both these cases we find numerous comments on the loss of many fixed traditions, institutions, or patterns of civilization, summed up in eloquent complaints about a supposed new barbarism or cultural chaos. On closer inspection, these periods turn out to have been periods of great fundamental growth and of the enrichment of the ensemble of learning resources and possibilities, which then in turn

PUBLIC OPINION QUARTERLY, FALL 1952 lead to the emergence of novel and temporarily more relevant patterns.

In the subsequent phase, these new patterns turn into temporarily fixed subassemblies. The subsequent age thus may impress observers with its apparent conservatism and stability, while at the same time embodying continued and important processes of change. These continued changes, however, are then largely changes within the limits of adult-type learning, exploiting a limited range of recombinations or largely ready-made routines. Such adult-type learning may still be original and creative, as these concepts were defined above, but it has strict limits and it may contain serious internal contradictions and incompatibilities. These contradictions and incompatibilities may end in deadlock or in a partial return to a seemingly barbarous or infant- type learning stage.

The demobilization of fixed subassemblies, pathways, or routines may thus itself be creative or pathological. It is creative when it is ac- companied by a diffusion of basic resources, and thus by an increase in the possible ranges of new connections, new intakes, and new recombinations. In organizations or societies the breaking of the cake of custom is creative if individuals are not merely set free from old re- straints but if they are at the same time rendered more capable of com- municating and cooperating with each other and with the world in which they live. In the absence of these conditions, there may be genuine regression; "barbarism" would then mean not merely the loss of prized traditions or routines, but rather the mutual babbling to which the Greek word first referred.

LEARNING AND DECISION SYSTEMS

The problem of learning and learning capacity seems clearly related to the problem of decision systems and the problem of will; that is, the freezing of certain patterns of decision after a point in time by allocating priority to pre-decision messages over post-decision ones.28 Problems of will, or of the hardening of decisions in organizations, could be measured in part by the twin approaches of mapping the structure of the relevant communication channels and memory facilities, and by charting the input-output behavior of the system. The results of these two approaches could then be checked against each

28 For an extended discussion of this point, see MTM, pp. 208-212.

378

COMMUNICATION MODELS IN THE SOCIAL SCIENCES 379 other. Will and the ability to make it prevail against processes, individuals, or organizations in the outside world-that is to say power- are thus related to the ability not to learn, and they have significant possibilities for the potential loss of self-steering capacity and the possible self-destruction of the organization. These possibilities could again be in part charted and measured.9

Just as it should be possible to chart or measure the probability of certain pathological developments in the self-steering or learning of organizations, so should we be able to attach operational meaning to such intuitively understood concepts as integrity and dignity-concepts which we begin to understand painfully when what they describe is being violated and can no longer be taken for granted. As a first super- ficial approximation, we might call the integrity of an organization or person their continued possession of undisrupted inner communication facilities and learning equipment, and we might call the dignity of a person or organization their use of their autonomous learning or steering equipment at nondisruptive speeds. This approach would permit yes-or-no judgments as to whether the integrity or dignity of a person or organization had been impaired to a significant extent or not.

We could go further. We could undertake structural and quantitative studies of the steering process which itself is at the heart of autonomy or self-determination. In addition to mapping its relevant channels and measuring its over-all performance, we could identify particularly relevant magnitudes such as the lag and the gain of the steering system in response to variations in input rates of information and in the loads on its communication channels, and in the performance of its effectors.30 From this approach, the minimum requirements for the actual size and performance characteristics of balancing or controlling facilities in an organization could be developed. To take an example, instead of merely stating in the discussion of the "cobweb theorem" in market fluctuations that advance buying by speculators would "tend to" counterbalance the tendency to increase the fluctuations, it should be possible to state how large a proportion of the total demand present in the market

29 For a discussion of some of them, see Karl W. Deutsch, "Communication in Self-Govern- ing Organizations" in Freedom and Authority Twelfth Symposium of the Conference on Science, Philosophy and Religion, L. Bryson et al., Eds., New York: Harper Brothers, publication scheduled I952-53.

30 For a discussion of the concepts of feedback versus equilibrium analysis and the concepts of lag and gain, see MTM, pp. 197-I99.

PUBLIC OPINION QUARTERLY, FALL 1952 would have to be in the hands of such speculators (this might correspond to the rate of gain in a feedback system), and how early and how reliable the advance information of these speculators would have to be (this would correspond to the lead in the series of data used in certain instances of control engineering). Measurements of these magnitudes are as much subject to quantitative treatment in principle, and as accessible to potential measurement, as any others used in current economic theory. What is true of the proportion of resources devoted to counter-speculation in markets should similarly apply in principle to the proportions between committed troops and operational reserves in certain situations of warfare, and perhaps to proportions of line work and staff work in organizations.31

THE CONCEPT OF GROWTH

I should like to close this paper with the suggestion relating to the measurement of growth in organizations. Such growth could be measured in four dimensions. First, an increase in openness-that is an increase in the range of the organization's channels of intake from the outside world; second, an increase in its inner complementarity or coherence-that is, in the efficiency with which information is transmitted and responded to from one part of the organization to another; third, an increase in power-that is, an increase in the ability to change the environment of the organization in accordance with its projected inner patterns, policies, and needs; fourth, an increase in learning capacity-that is, in the ability of the organization to learn rapidly and yet originally and creatively, and to change its own goals rather than to remain the prisoner of some temporary goal or ideal and to fall victim to what A. J. Toynbee called the "worship of ephemeral institutions."32

This approach to growth is philosophical. It is hoped that it will stem from the philosophy of science rather than of intuition, and that in the course of time every dimension of the growth and performance of organizations which was outlined above will prove capable of mapping and of measurement.

31 Cf. also the discussion of the ratio of total association cortex to total sensory cortex- called the A/S ratio-in Hcbb, op. cit., pp. 124-125.

32 Toynbee, A. J., A Study of History, London: Oxford University Press, I939, vol. 4, pp. 303-422; see also the whole section "The Nemesis of Creativity," pp. 245-584.

380

Article Contentsp. [356]p. 357p. 358p. 359p. 360p. 361p. 362p. 363p. 364p. 365p. 366p. 367p. 368p. 369p. 370p. 371p. 372p. 373p. 374p. 375p. 376p. 377p. 378p. 379p. 380

Issue Table of ContentsThe Public Opinion Quarterly, Vol. 16, No. 3 (Autumn, 1952), pp. 313-446Front MatterDemocratic Theory and Public Opinion [pp. 313-330]The Appeal of Communism to American Middle Class Intellectuals and Trade Unionists [pp. 331-355]On Communication Models in the Social Sciences [pp. 356-380]The Waukegan Study of Voter Turnout Prediction [pp. 381-398]Political Independence in Washington State [pp. 399-409]Some Limitations on the Arbitrary Classification of Non-Scale Response Patterns in a Guttman Scale [pp. 410-416]Linear Segments: A Technique for Scalogram Analysis [pp. 417-431]Living ResearchField Coding Versus Office Coding [pp. 432-436]Socio-Psychological Factors Affecting Predictions of Elections [pp. 436-438]Classifying Communications [pp. 438-439]A Study of Attitudes Toward American Policy in Germany [pp. 440-442]

From POQ Readers [pp. 443-445]News Notes [p. 446]Back Matter

006 on Communication Models in the Social Sciences Deutsch

Documents

Transcript of 006 on Communication Models in the Social Sciences Deutsch