Social interaction is made up largely of the talking that people do when they
get together. Talk is an elusive object of study, in spite of the fact that
a good deal of it exists. It is also a rather sensitive subject. Even a friend
might find it hard to put up with a dissection of the following kind: "I
was just noticing how much you talk. In the last 10 minutes I noticed that
you made a total of 114 remarks, while I made a total of 86. According to
my count you gave about twice as many opinions as facts. Although I agreed
with you 15 times and didn't disagree at all, I noticed that you stammered
once and blushed twice." I first began to develop a systematic procedure
for analyzing social interaction when I became interested in trying to account
for the success of Alcoholics Anonymous in helping apparently hopeless drinkers
to stop drinking. Although I attended meetings and talked with many members,
I did not feel free to ask all the questions I wished. Consequently I fell
back on observation and began to develop crude methods for recording who did
what, who spoke to whom, and how.
Eventually even this quiet occupation began to appear sinister and the effort was abandoned. But by this time my fascination with the process of social interaction had developed to the point of no return. I decided that I must pursue my studies in the more favorable conditions of a laboratory.
------------------------
From Robert F. Bales, "How People Interact in Conferences," Scientific American, 0 1955 by Scientific American, Inc., 192, 31-35, all rights reserved. Reproduced with permission of the author and publisher.
------------------------
A number of laboratories for the study of social interaction within small groups and organizations have been started in the last 10 years-in hospitals, clinics, special research centers and military installations. The studies and experiments I shall describe were conducted in one of the earliest of these laboratories, established in 1947 at Harvard University.
The laboratory consists of a large, well-lighted room for the group under study and an adjoining room for observers, who listen and watch from behind windows with one-way vision. The subjects are told at the beginning that the room has been constructed for the special purpose of studying group discussion, that a complete sound recording will be made and that there are observers be hind the one-way mirrors. The purpose of the separation is not to deceive the subjects but to minimize interaction between them and the observing team.
After much research we developed a standardized task from which significant generalizations could be drawn. A group of persons (ranging from two to seven in number) is asked to discuss a complex human relations problem of the sort typically faced by an administrator. Each member of the group first reads a five-page presentation of facts about the case to be discussed, but each is left un certain as to whether he has been given exactly the same range of facts as the others in the group. The members are not introduced to one another or coached in any way; they must develop their own organization and procedure. They are to consider the facts and report to an administrator, as if they were his staff, their joint conclusions concerning the problem and what should be done about it.
They are allowed 40 minutes for the discussion. The group is observed for four such sessions.
On the other side of the one-way screen the observers systematically record every step of the interaction, not omitting such items as nods and frowns. Each observer has a small machine with a moving paper tape on which he writes in code a description of every act-an act being defined essentially as a single statement, question or gesture. Acts ordinarily occur at the rate of 15 to 20 per minute. The recorded information on each includes identification of the person speaking and the person spoken to and classification of the act according to pre determined categories. There are 12 categories, covering positive and negative reactions, questions and attempts to solve the problem by the offering of information, opinion or suggestions.
------------
TYPES OF ACTS in social interaction may be classed in four main categories: positive reactions, problem-solving attempts, questions, and negative reactions.
The averages for 96 group sessions show that 56 percent of the acts fall into the problem-solving category.
--------------
PATTERN OF ACTION of individuals in a discussion is illustrated statistically.
When a member takes the floor, his first remark (dotted curve) is likely
to be a reaction to the preceding speaker. His next remarks (solid curve)
tend to be problem-solving attempts.
As this table shows, on the average about half (56 percent) of the acts during a group session fall into the categories of problem-solving attempts; the remaining 44 percent are distributed among positive reactions, negative reactions and questions. In other words, the process tends to be two-sided, with the reactions acting as a more or less constant feedback on the acceptability of the problem-solving attempts. The following is a typical example of the pattern of inter change:
Member 1: "I wonder if we have the same facts about the problem?
[Asks for opinion.] Perhaps we should take some time in the beginning to find out." [Gives suggestion.]
Member 2:
"Yes. [Agrees.] We may be able to fill in some gaps in our information.
[Gives opinion.] Let's go around the table and each tell what the report said in his case."
[Gives suggestion.] This example illustrates that a speaker's first remark is likely to be a reaction, and if he continues speaking, the probability is very high that his second act will be a problem-solving attempt. The lower chart on the opposite page sums up this finding statistically: about 50 percent of the time a member's first re mark in a series is a reaction; if he continues, about 80 percent of the succeeding comments are opinions or other offerings classed as attempts to solve the problem.
---------------
1 States primary observation: I observe a particular event, X. 0
2. Makes tentative induction: This particular event. X, may belong to the general class of object , O. e ,......) _
3. Deduces conditional prediction: If this particular event, X, does belong to the general class, 0, then it should be found associated with another particular event, Y.
4. States observation of check fact: I observe the predicted particular event, Y.
5. Identifies object as member of a class: I therefore identify X-Y as an object which is a member of the predicted general class of objects, O.
6. States major premise relating classes of objects: All members of the general class of objects, 0, should be treated by ways of the general class, W.
7. Proposes specific action: This particular object, X-Y, should therefore be treated in a particular
PROCESS IN REACHING A GROUP DECISION is analogous to the operation of a large-scale communication and control system such as the air-defense network. The steps consist of observing an object or event, comparing it with several possible identifications, considering the associated facts and, once its nature is understood, taking the appropriate action.
--------------
When we examine the reactions, we find that positive reactions commonly outnumber negative ones about two to one during a session. It is as if after every negative reaction, the members of the group feel they must make another problem-solving attempt which meets with a positive reaction "just to catch up," and net forward progress is felt to be sufficiently secure only when a repetition of the problem-solving attempt meets unopposed acceptance. It may be that members employ repetition, or near repetition, as an error-checking device to determine whether the others "really agree." Social interaction, in common with many other goal-seeking control mechanisms, seems to depend upon error and correction of error for guidance.
The process of attempting to arrive at a group decision through discussion is in many ways very like the operation of a large-scale communication and control system such as an air-defense network. I recently compared the two processes in collaboration with John Kennedy of the Systems Research Laboratory at the Rand Corporation.
In the military case there are three functions to be performed: surveillance of the air by radar, identification of planes as friendly or unknown and direction of fighters sent out to intercept unknown planes. These are something like the three problems confronting our groups in the standard interaction task: assembling the given information on the case, evaluating it and proceeding toward a solution as the goal. Now the stepwise operations involved in the air-defense sys tem may be tolerably well described as an interlocking series of seven types of in formation-processing operations. Here x stands for the path of a plane tracked by radar, and 0 represents the class of objects unknown. If no known flight plan of a friendly plane coincides with x-a fact represented by the symbol y - then x must belong to the class O. Since there is a general rule, W, that all unknown planes are to be intercepted, the conclusion is that a specific order, tv, should be given to intercept x.
Such a decision, involving many groups and interlocking processes, is obviously a very complicated affair, socially as well as technically. The job of the decision-making organization is essentially to build and maintain through means of communication and evaluation a sufficiently complex and commonly accepted symbolic structure to guide or control the stages of behavior of all the operating units. Effective decision making is basically a continuous process of building and maintaining a structure of cultural objects which in their totality constitute the common culture of the organization affected.
The seven types of acts, or stages, just described are very general: they apply quite as well to the interaction of five experimental subjects in the laboratory group, trying to decide in 40 minutes what the administrator in their case should do about his problem, as to the large-scale operations of an air-defense network. Not all of the elements in the process are primarily logical in character.
They involve elements of perception, memory, association and perhaps inductive insight. All sorts of motivational and evaluative pressures affect the process. The steps make sense not as a formally perfect chain of logic, but rather as a set of symbol transformations which help to guide, although in an imperfect way, a process of decision-making behavior. Error checking is an integral part of this fallible process.
The reason for calling attention to the seven-step structure of the process is that it may help to explain the unequal ratios of suggestions, opinions and information offered in the problem-solving attempts of the groups in our tests. As the first table shows, of every seven problem-solving attempts, on the average four are opinions, two are offers of information and one is a suggestion. It seems significant that in the idealized seven-step outline of the air-defense operation two steps have the interaction form of giving information, four intermediate steps have the interaction form of giving opinion and only one step, the final one, has the form of giving a suggestion.
From the transcription of a group discussion it is often possible to recon struct complete seven-step chains leading to agreement on specific points and the final conclusion. In a general way there is even a tendency for the steps to proceed in a regular order in time. During a session the rates of giving information tend to be highest in the first third of the meeting and to decline in the next two thirds. Rates of giving opinion are usually highest in the middle portion of the meeting. Rates of giving suggestion are generally low in the early period and reach their high point in the last third of the meeting.
-------------
GROUP PROCESS toward a decision is characterized by a change in the frequency of different types of social acts as the meeting wears on. Information giving decreases while suggestions and positive and negative reactions increase.
Rates of both positive and negative reactions tend to rise from the first third of the meeting to the last third. These increases may be connected mainly with social and emotional problems of the group process itself. The ratio of negative to positive reactions tends to be higher in response to suggestions than in response to factual statements. The decision point is a critical bottleneck in the pro cess. Once the decision point has been passed, however, the rates of negative reaction usually fall off and the rates of positive reaction rise sharply. Joking and laughter, indicating solidarity and tension release, become more frequent. With the problems of the task and common values stabilized for the time being by the decision, the interaction process apparently turns to re-stabilizing the emotional states of the individuals and their social relations to one another.
There is a good deal of evidence that the process of social interaction, like other processes involving feedback, tends to fall into oscillation as it "hunts" around a hypothetical steady state. Over a small time span the action tends to al ternate every few acts between the problem-solving attempts of one person and the social-emotional reaction of some other. But this rapid oscillation is not quite rapid enough to keep all elements of the process in perfect balance. There is a drift toward inequality of participation, which in time has cumulative effects on the social relationships of the members. The reason for this drift may be seen fairly easily. When a person has completed one act, the chances are a little better than even that he will continue for another act. After each succeeding act his probability of continuing drops, but never as far as if he simply flipped a coin at each point to determine whether to continue or to yield the floor. In fact, relatively speaking, he exceeds this chance probability by a larger and larger fraction with each succeeding act.
We have already noted that when a person continues several acts in succession the probability is very high that he is giving information, opinion or suggestion-in other words, specializing in problem-solving attempts. We may also infer from the seven-step theory of problem-solving attempts that the tendency to continue for several acts in succession is probably due in part to a felt need on the part of the speaker to provide inferences and check facts which will result in the acceptance of a more advanced step in the series, with an accepted suggestion as the goal.
This tendency toward inequality of participation over the short run has cumulative side effects on the social organization of the group. The man who gets his speech in first begins to build a reputation. Success in obtaining acceptance of problem-solving attempts seems to lead the successful person to do more of the same, with the result that eventually the members come to assume a rank order by task ability. In some groups the members reach a high degree of consensus on their ranking of "who had the best ideas." (The members are interviewed by questionnaire after each meeting.) Usually the persons so ranked also did the most talking and had higher than average rates of giving suggestions and opinion.
While one person becomes a specialist in advancing ideas, another is apt to be developing a specialization on the reactive side. The men most commonly rated "best liked" typically have higher than average rates of showing tension re lease (mainly smiling and laughing) and showing agreement. It is not impossible for the man ranked at the top in ideas also to be best liked, but apparently it is difficult. In one set of experiments the top idea man had about an even chance of also being best liked at the end of the first meeting, but by the end of the fourth meeting his chances were only about one in 10. The best-liked man is usually second or third in the participation hierarchy.
The task specialist seems to "lock onto" the person who is most responsive to what he is saying and address more remarks to him than to the others. In turn, the best-liked man talks more and agrees more with the top-ranking idea specialist than with any other member. The idea specialist and the best-liked man often form a mutually supporting pair. However, the best-liked man may attract the idea specialist even though they are not always in agreement. Indeed, in order for a person to become established in the minds of other members as a social-emotional specialist, it is probably more important that he be representative of their reactions, both positive and negative, than that he should ardently support everything the task specialist says. Apparently reactions that are emotionally gratifying to other members tend to be generalized by them into liking for the person who expresses the reactions.
Giving suggestions, necessary as it may be for accomplishment of the task, is more likely to arouse negative reactions than is giving information or opinions. This tends to put the task specialist in a vulnerable position. The group commonly develops a certain amount of negative feeling toward him. Not only is he likely to lose the status of being best liked, but he may lose his position as task leader unless he is sensitive to the problem and is well supported by other members. Even in a group which ends its first meeting with a high consensus on who has the best ideas, the second meeting is apt to see a challenge to his leader ship, with a rise in rates of disagreement and antagonism and a precipitous drop in his popularity. But then, in a group where the original consensus was high, a peculiar thing seems to happen. Apparently as progress toward accomplishment of the task slows down, some members rally around the leader again and his popularity tends to rise. By the third meeting the rates of disagreement and antagonism go down. The task leader may not retain all the liking that was transferred to him in his time of need, but the net effect of the hunting kind of oscillation that takes place is a tendency to maintain the original rank order of task ability.
In a group that starts with a low degree of consensus on who has the best ideas, the developments usually are more dismal. There tends to be a high turn over in the top ranks throughout the four meetings, with one would-be leader re placing another. In such a group the man ranked as having the best ideas is less apt to be best liked. Furthermore an additional specialist is likely to appear-a man who talks more than anybody else but is neither best liked nor most highly respected for his task ability.
It appears probable that whether the members will agree on who has the best ideas depends to a large degree on how well they agree on basic premises or norms--what we may call the "common culture." If such consensus is not present, at least implicitly, at the beginning, it may take a long time to build. While consensus on major values does not solve all the problems of arriving at a stable social organization, probably no stable organization is possible without this control factor. If it is lacking, the interaction process becomes primarily a means for the expression of individual emotional states.
Our studies have made clear that social stability is an extremely complex achievement: it takes time and patience to arrive at a common culture extensive enough and sensitive enough to regulate strong counter motives, to promote task accomplishment, to harmonize social relationships and to rejuvenate itself when ever the conditions demand. A clear recognition of the complexity of cultural control of behavior should encourage us to believe that interminable series of meetings around the conference table, international and otherwise, are perhaps worth while after all.
Also in Part 5:
- Communication Theory: Social Context
- How People Interact in Conferences
- Group Composition and Structure
- Achieving Change in People: Some Applications of Group Dynamics Theory
- Interpersonal Networks: Communicating Within the Group
- Traditions of Research on the Diffusion of Innovation