Interaction: Some Ideas

Herbert Blumer, the University of Chicago sociology professor who taught generations of researchers who worked in "the Chicago tradition," was, I think, the first person to speak of "symbolic interactionism" as a point of view,  a "theory." He described "interactionism" as the alternative to three other approaches to the study of human conduct, which he identified as instinct theory, stimulus-response (S-R) theory, and cultural theory. This framework dominated his lectures to students, but was published only in a now unobtainable book (Blumer 1937).

We students in the late 1940s and early 1950s always thought his approach was bizarre, because we couldn’t believe that anyone was so naive as to actually believe any of these clearly (to us) outdated ideas. Surely, no one thought any more that there were instincts that led people to be aggressive, or cooperative, or altruistic. We didn’t realize that psychoanalytic thought was a very sophisticated version of such a theory and we could not have imagined that sociobiology and a gene-based behaviorism would appear in our lifetimes. Nor did we see that the conventional gathering of data through questionaires and similar instruments really took for granted a sort of S-R image of human action. And we thought that culture so obviously existed that there was nothing to question about its use as an explanation of why people acted as they did.

What was wrong with these approaches was that they took as given that behavior was in one way or another an automatic response to something in the person’s internal or external environment. They assumed the following, as contrasted with an approach centered on the idea of interaction, to be described shortly:

Human beings are passive. They sit around waiting for something to stir them into action. If nothing happens to impel them to action, they don’t act. So an understanding of human behavior comes from identifying the links between the pushes, whether internal or external, and the responses. All activity is a response to a push from somewhere.

Instinct theory assumed that people are pushed into action by imperious needs which are built into human nature ("hardwired," as we say these days)

S-R theory assumed that people are pushed into action by external stimuli to which they have learned responses that are rewarding.

Culture theories assumed that people’s actions consist simply of doing what the culture asserts as the right way to do things on the occasions when it is right to do them.

Responses are hardwired into human nature, determined by previous experience, or simply taken, readymade, from the "culture," and thus are not subject to change once activated.

Behavior can be understood as a matter of individuals, not of people in contact with one another. These approaches envision a world of isolated individuals, evidenced for instance by the use of terms like "social stimulus" to refer to people other than the actor being analyzed.

Because interaction by definition is not a solitary activity, a vision of human conduct centered on that idea focuses not on the isolated acts of individuals but on the development of collective action, on how people act together to create an activity that is something they all have contributed to.

David Mamet, the American plauwright, says, somewhere, in a perfect expression of this idea, that in a play every character in a scene is there for a reason, to get something they want, to achieve something they want to achieve; if they didn’t have a reason to be there they wouldn’t be there. The scene consists of each of them pursuing what they are trying to get, but having to deal with the other people present, who are all doing the same thing. The outcome, the result, of the scene is something, very likely, that none of them wanted; it is what comes out of each pursuing his own goal and reacting to others doing the same thing. You could call that outcome a collective act.

An interactionist approach asks how the development of such a collective act is possible. What has to be true of human behavior so that collective action can occur? This leads to the following ideas, opposed to those characteristic of the approaches mentioned earlier:

Human beings are active, not passive. They do not sit around waiting for something to stir them into action. Instead, they are always acting, doing something, trying to do something, and looking to the environment, searching it, for ways to accomplish whatever they are trying to do.

Human conduct is never automatic, but always involves the possibility of a pause, during which the actor can reflect on the action in progress and think about alternative possibilities for responding to what is happening, to what others are doing.

During such pauses, the actor thinks about how others will respond to what he is thinking of doing, and adjusts what he was going to do to take account of that imagined response. The model here is the chess game, in which I think about moving my pawn here, stop to think what my opponent will do if I do that and, seeing that he has a good response to my planned move, try something else.

This does not assume that the actor imagines the response of others correctly, that the chess player will never be wrong about what the opponent is going to do. In fact, it is most likely that those responses will not be imagined correctly. Instead, there will be inaccuracies which will require the actor to adjust what he is doing to take account of the new information provided by others’ reactions.

But, of course, there are never only two people involved. The actor never thinks only about one person sitting across the chessboard. Instead, the actor takes account of all the people involved in the action undertaken. Even in the chess match, there are onlookers, other players who are potential opponents on other days, officials of chess organizations, family members, etc., etc. In constructing a line of action, the actor more or less simultaeously takes into account the potential responses of all these people.

Note that I spoke of a line of action, rather than a response. Responses are never isolated acts, they are part of developing lines of activity, long arcs of action in which this process of noting things happening in the environment, envisioning responses to them, adapting those responses in the light of anticipated possible responses, and all that is repeated over and over and over.

The picture is, of course, infinitely complicated when we recognize the reality that all these other people who are involved in the development of one person’s line of activity are themselves engaged in the same process of scanning the environment, imagining possible actions and possible responses to it, and developing a line of action. So the image we should have is of a multitude of people all doing this, and out of that coming something that gets done, a collective action.

And a collective action that is, in some sense, effective. This is not to say that the perspective insists that people are always doing "the right thing" and accomplishing their goals. That would be quite unrealistic, because in fact things hardly ever work out as people intend, as Mamet noted. What "effective"means, in this context, is that people more or less arrive at a situation in which they can decide that they are satisfied with what has happened at least some of the time. Or that at least we can understand how and why that does or doesn’t happen. (Blumer used the word "morale" to describe the ability of a group to realize its goals and Tamatsu Shibutani (1978) embodied that idea in a brilliant study of Japanese-Americans in the United States Army in World War II.)

"Interaction," so understood, is not a mystical notion. It’s very straightforward and realistic. Blumer used to give us this exercise: take any ten minutes of your own experience and try to explain it in the language and concepts proposed by his three bêtes noires. This seemed trivial to us until we tried to do it, Then we discovered that we could not find, to take the approach that seemed most reasonable to us at the time, an explanation "in the culture" for the details of what we did and thought in the most ordinary situation: shaving, washing the dishes, crossing the street. "Culture" had no prescriptions for how I put on my pants, other than to get them on somehow so as to be "properly" dressed, and no prescription for how brown my toast should be.

Of course, the other approaches had something to offer. So many smart people could not have been completely wrong. And they weren’t—not completely. The other approaches could be understood usefully once they were located in the framework suggested by an interactionist perspective. Thus:

The kernel of truth in instinct theory is that behavior does have a biological base, which includes needs and basic capabilities for action. The emendation offered by interactionism is that, yes, we all experience hunger for food, the need for sexual activity, but these needs must,  first, be learned as desires that can be satisfied in a certain way, and this is learned in interaction of the kind described with the environment, which includes other people; and that they can be satisfied only by reaching some kind of agreement with others as to how that can be done, an agreement reached through the process of building up lines of conduct I’ve described.

The kernel of truth in S-R theory is that much of human behavior does have a somewhat automatic character, but only under special circumstances. I can behave automatically only if the responses of others can be dependably relied on to be just what they have always been, not to change. There are such areas, in which what others will do is so routinized that my body can take over—the way we cross the street at traffic lights, everyone avoiding bumping into everyone else, is an excellent example.

The kernel of truth in cultural explanations of behavior is that there is indeed such a thing as culture, shared understandings more or less known to all the participants in some collective action, to which they can all refer in anticipating what others are going to do. If we know what those shared understandings are, we can guess, pretty well though not perfectly, what others in a situation may do. What has to be added is that these shared understandings are only the beginning of the negotiation that constitutes interaction, the guidelines to which the participants can refer to as they develop the collective line of action they engage in. They may in fact do things just as they did them the last time, but that has to be recognized as a possibility, not as something guaranteed. William Graham Sumner’s  explanation (in his classic work, Folkways) of the development of culture deals with a process that goes on all the time. The chess analogy would be that the rules of chess provide the framework that makes a playable game possible, but do not dictate the moves anyone makes or how they make them.

There is, of course, stability and regularity in human action. People do not act randomly, so a major question for an interactionist perspective is how this happens, given all the indeterminacy the position insists on. The answer is that such mechanisms as culture and learned responses "work" when the situation allows it. And that happens when a form of collective action arises, through the processes I have described, and creates stable expectations that everyone in the situation can attribute to everyone else, so that what has been "agreed on" as culturally appropriate is what everyone takes for granted and acts on. This means, in turn, that when A guesses at the reaction of B to his tentative act, he will mostly guess right. And if he makes something other than the conventional guess, he will probably guess wrong, his act will be blocked and interrupted, and he will have trouble. If everyone acts conventionally, things will run more smoothly. People learn that culturally suggested responses "work" and so continue to make  them, and that means that these responses work for everyone else as well.  It’s circular. So "inertia" plays an important role in keeping collective action stable. At least until something intervenes: circumstances change, someone is willing to take on the additional trouble entailed by doing things differently, etc.

Though the interactionist perspective is not at all mystical, it does leave many questions unanswered and takes some important things as obvious, even though they aren’t.

The most important topic taken for granted is just how all this works. What, in realistic detail, is the process by which people arrive at a common perspective that allows them to engage in effective collective action? It’s all very well to talk about taking the role of the other, but research that has tried to address that question directly has not produced much of interest.

Conversation analysis—and this may not be what its proponents would want to say about it and its virtues, but is what I find crucial—supplies that lack. It shows how people make tentative gestures that signal their intentions to others, how others come to understand those gestures and then make gestures in return, in a kind of bargaining. Until, finally, they arrive at what to do or, perhaps better said, find themselves doing whatever is going to get done. Deirdre Boden (1990) put it slightly differently, saying that conversation analysis showed how interaction was talk, and that the very processes of talking  that CA studies—turntaking and the like—produces what other sociologists see as social organization, social structure, and institutions:

. . . [S]ocial order and social structure are not external to action but rather produced in and through the local structures of interaction. (p. 250)

Close analysis of everyday conversation reveals just that coordination of action [George Herbert] Mead and [Herbert] Blumer were so sensitive to, and locates it precisely in the orientation of one actor to another in the most pervasive of all social acts [i.e., conversation]. (p. 253)

The several examples given in her article show precisely how, in the course of mundane conversation, concerted collective action comes about. (Pp. 254, 256, 257-8, and pp. 263-4)

Finally, this way of thinking about social life and collective action has consequences for all the outstanding questions about how social science can and should be conducted.  I’ll take just one example: the problem of prediction.

Social scientists have always wanted to make predictions, to use their knowledge of how society works to foretell the future, to say what will happen next, to say how problematic situations will develop and what their results will be. They have lusted after the assurance with which physicists and chemists predict what the results of a chemical reaction will be, what will happen when we drop a weight from a height, all the many topics these specialists in other fields predict with such ease. But we have never succeeded in doing that. Even the most "scientistic" of the social science disciplines, economics, has failed dismally to predict what will happen to national economies, to the stock market, to firms. Sociology has never been able to do what it would like to do: foretell the next outbreak of civil violence somewhere, predict the chances of married couples remaining married or of released prisoners committing new crimes, to mention a few of the things whose future social scientists have tried so hard to guess.

What boosters of social science usually say, confronted with this terrible record of failure, is that we haven’t had enough time to develop such a predictive science but that eventually we will. I don’t believe it. There are two kinds of reasons why we won’t (and, in principle, can’t) predict the course of social life.

First, more important empirically, is the sheer difficulty of taking into systematic account the millions of things that actually are involved in any social situation: the immediate circumstances of the interaction (what might be called the Goffman level of interaction); the organizational constraints and opportunities of a specific social organization (a school, a factory, a neigborhood); the larger regional and national and international realities that people in a situation may not be aware of but which nevertheless constrain what they can do. It is never a question of which of these is "more" or "really" important. They all play a role (in a way that is multiplicative rather than additive) and a change in any of them changes the course of the interaction’s development and its outcome. But we don’t know how to take all that into account at once and it does not seem likely that we will ever be able to.

Let us suppose that, with big enough computers and vast armies of data gatherers, this difficulty is overcome:. There would still remain two difficulties in principle. First, we can’t say how or why people, as they go through the process of weighing alternatives and forging a line of activity, make this or that choice. Second, and more decisive, even if we were able to predict how A would respond to a situation, we have also to deal with B’s response which is part of what A will have to respond to (not to mention, C, D, E . . . N). There may be discoverable laws which let us predict what A will do, but there are no discoverable laws governing when A’s path will cross B’s (and C’s . . . etc.).

Take an example favored by some statistical analysts: automobile accidents. It may be predictable that, if I drink more than a certain amount of alcohol, I will be likely to make bad decisions when I drive and thus will be more likely to hit another car. But it is not predictable that, at the moment when my judgment becomes so clouded, another car will come along in such a relation to mine that my error of judgment will produce an accident. The event we call an "accident" requires the "cooperation" of many other variables of weather, time of day, timing of responses, etc. Each of these follows its own course of development and there is no reason to think that there are laws linking the course of one activity to that of the others. So: no predictions.

This is the bare outline of what an interactionist position means with respect to sociological work. There is much more to say and much more has been said in the many monographs and research reports produced from this point of view.


Blumer, H. (1937). Social Psychology. Man and Society. E. P. Schmidt. New York, Prentice-Hall: 144-198.

Boden, D. (1990). People Are Talking: Conversation Analysis and Symbolic Interaction. Symbolic Interaction and Cultural Studies. Eds., H. S. Becker. and M. McCall. Chicago, University of Chicago Press: 244-274.

Shibutani, Tamotsu. 1978. The Derelicts of Company K: A Sociological Study of Demoralization. Berkeley: University of California Press.