Abstract – Cognitive dissonance is considered as emerging between the social identity of persons and that of their acts. An analysis is made of the paradoxical consequences of a double bind: Those who are A are supposed not to do B and are also supposed not to think that those who are A would be allowed to do B. The Cohen-Rosenberg controversy is presented here, revised on this basis, and illustrated by the two authors’ experiments. It is claimed that the psychosocial aspect of social identity is complemented by its socioeconomic aspect. Indeed, the valuation of an identity is always a judgment of the extent to which this model should be reproduced. The more tolerant or the more ruthless manner of imposing value models of social identity is determined by socioeconomic factors On the other hand, the socioeconomic positions may be specified by psychosocial factors. The psychoeconomic connection in social identity is accentuated in post-capitalist societies, turning human faculties and needs into factors to be produced and reproduced by the economic system.
Some social psychologists consider that the question of social identity “is nothing but that of modes of organization for a given individual of his representations of himself and of the group to which he belongs” (Zavalloni, 1973, p. 245). For others (see, for example, Sarbin & Allen, 1969) it is what the individual does from his position in the social structure that defines his identity, rather than what he thinks about it when comparing himself to his group.
These latter could argue that one has a social identity of, for example, a working person when he regularly carries out an activity in working and in claiming the remuneration for it, rather than because of a representation that he has of himself or others have of him. And, similarly, it is not being considered as a hedonist person that identifies someone socially as such, but his acting freely and in eventually assuming the necessary pecuniary sacrifice for it.
But what about the identity of someone who works (for example, whitewashing a fence) and assumes a sacrifice for this activity! Or the identity of that other who acts freely (in playing, for example, football) and claims the remuneration for this very activity?
Although these questions sound absurd, however, we know the story (imaginary, but too real) of Tom Sawyer who led his playmates to pay in order to have the pleasure to whitewash a fence. Now, was the social identity of these children that of a working person when, on that hot Saturday afternoon, bathing in the river would have been a much more attractive activity?
And we know, too, of the famous Hungarian football captain of the team of the “belle epoque” to whom people credit the saying “Good pay, good play, bad pay, bad play”. Does this mean that this sportsman had the social identity of a hedonist player when, at a time of austere amateurism, he claimed a remuneration in proportion to the work carried out?
Looking for indicators of social identity, one may start by preferring acts to representations. But one soon realizes that it is the representation of an act rather than the act itself that is the matter here, since one cannot identify socially a person committing an act without identifying socially the act committed by this person. Is whitewashing a fence necessarily work, and playing football a pleasure? Yet, the act of a representation here may be the act itself in question.
If one plays football and is paid for this activity, the cognitions referring to these two facts will be in dissonance that is considered by cognitive dissonance theory responsible for creating in the individual’s mind a tension that is more or less painful and that can be reduced only by modifying one of the cognitions to the point where it becomes consistent with the other, for example, by modifying the social identity of the activity in order to present it as work. It is the same for the case where one accomplishes a job in whitewashing the fence and lets oneself be led at the same time to pay for doing this activity.
This supposition has been tested repeatedly in laboratory experiments. Deci (1975) gave riddles to students to solve, one group being paid for this activity while another was not. During breaks, those not paid could not resist going on with the puzzle solving, while those paid rested after their work. In another experiment, nursery school children lost their interest in toy A when promised to be “rewarded” for playing with it by permission to play with toy B, and vice versa.
At this point, the question arises concerning the nature of the cognitive field which determines that two cognitions are consistent or dissonant. In this classic form of the cognitive dissonance theory, Festinger (1957) did not raise this question, proposing simply that the dissonance between cognitions A and B emerges if A implies psychologically non-B. Later, he specified the conditions necessary for creating dissonance between two cognitions: “Whenever one has an information or a belief that, taken alone, ought to push one not to commit an act, this information or belief is dissonant with the fact that one has actually committed this act” (Festinger, 1963, p. 18).
But, how can an idea incite one to commit an act? What does “implies psychologically” mean ? To take a classic example, if one thinks that all human beings are mortal and that Socrates is a human being, one finds oneself brought by these two ideas to have yet a third one: Socrates is mortal. If, in spite of this incitement, one thinks that Socrates is immortal, this produces a cognitive dissonance that has the form of a logical error. But he who works and at the same time pays for the pleasure of working commits no logical error, and neither does someone who plays and is paid for playing.
Strictly speaking, in this case of a paid player (as opposed to the person paying/07 the pleasure of working) there should not be any cognitive dissonance, according to the above Festinger formula. If one has the information or the belief of being paid for play, one should not be pushed at all by this to not do the activity. We shall examine this curious matter later on.
To bring us nearer to an answer, Aronson reformulated the theory (Aronson & Mettee, 1968; Nel et al., 1969; Aronson et al., 1975; Aronson, 1976). According to his suggestions, the information or belief which would push me not to commit an act is the cognition of my social identity incompatible with such an act. Aronson takes into consideration more general dimensions of social identity, such as reason and honesty.
If I have the cognition A, “One makes me pay for work done by myself’, and the cognition B, “I bring about this activity”, it is not necessary that A psychologically implies non-B. It is therefore not necessary that a cognitive dissonance emerge between A and B. On the contrary, if I hold the cognition A, “I am a reasonable person”, and the cognition B, “I work and, more, I pay to work”, then the dissonance becomes inevitable, since a person whose identity is described by A cannot commit an act the corresponding identity of which is defined by B.
According to the idea that cognitive dissonance can emerge between the definition of the social identity of the act and that of its author has been revealed as very important in explaining certain apparent irregularities of this phenomenon. In the beginning, one supposed, for example, that to believe X and to say non-X was susceptible in itself to introducing the dissonance. However, to explain this statement sufficiently in everyday life, the reward or punishment dimension has been mentioned: getting the former or avoiding the latter would provide an external justification compensating for the tension of the dissonance.
Lacking such a justification, the tension would tend to be reduced by bringing the afflicted subject to believe what he said. This hypothesis (Festinger & Carlsmith, 1959) has been confirmed by many experiments dealing with forced compliance for a contra-attitudinal advocacy. When the reward or punishment received in these experiments is just enough to force the subject to plead against his attitudes, he is pushed to believe what he said. But when the punishment or reward is larger, the tendency of the subject to believe what he said is weakened. However, there are as many experiments that disprove this hypothesis demonstrating that the liability of the subjects to adjust their beliefs to their words is directly proportional to the importance of the reward or punishment in question.
Now, neither an inverse nor a direct proportionality between the amount of the reward or punishment and the tendency to adapt the thought to the word is given, first, for the simple reason that one may not feel at all the necessity of co-ordinating one’s thought and one’s words. Once again, it is not between a cognition A, “I believe X”, and a cognition B, “I say non-%’, that the cognitive dissonance manifests itself, but between the cognition A, “I am honest”, and the cognition B, “While believing X, I lead others to believe non-X”. It is for this reason, in experiments during which the experimental manipulations prevented the subject from defining his social identity in conformity with A (see, for example, Aronson 8r Mettee, 1968) or that of his act in conformity with B (Nel et al., 1969), that the “normal” display of cognitive dissonance is then perturbed.
Being among the most general dimensions of social identity, honesty and reason are still socially concrete. “To be reasonable” amounts to this: “To choose the most advantageous alternative”. And “to be honest” amounts to “not to prevent others from choosing, in conformity with established rules, their most advantageous alternative”. This means, in the last analysis, that honesty and reason turn out to be characteristics of the middle class in a capitalistic society. (Without examining this statement in more detail let us only consider intuitively the difference between such a “reason” or “honesty,” on the one hand, and that of Brutus or of a Petrograd proletarian in 1917.)
Now, if it is true that the cognitions “I believe X” and “I say, convincingly, non- X” demonstrate a cognitive dissonance only because a cognition defines their relation for the acting person by socially defining this person, it is also true that the dissonance between the cognitions defining the social identity of the act on the one side (“In believing X I lead others to believe non-X”) and that of the acting person on the other side (“I am honest”) exists only by a supplementary cognition defining, so to speak, the social identity of the social identity itself (“Honest people do not lead others into error”).
Thus, the complete formula for cognitive dissonance is as follows:
1. I am A;
2. I do B;
3. A does not do B,
where A is any social category and B is any relevant social act. “Any” means that the formula can convey even contents as concrete as this:
1. I am an authentic Moslem;
2. I drink wine;
3. An authentic Moslem does not drink wine.
For all kinds of concrete incarnations of the above three-piece formula, there exist three types of reducing cognitive dissonance adjusted to each of the above items, respectively, and re-defining social identity.
Type 1.Realize that one is no more (or that one has never been) A. I am no longer an authentic Moslem since I drank wine. I am not honest because I pleaded, to convince others, that the police had their reasons to have penetrated the university campus and to have killed four supposed demonstrators, at the same time being convinced that no reason could exist for such disgrace (Cohen, 1962). The cognitive consistency is recovered, but at the price of losing social identity, a price too high for the counterpart, such that one pays only at exceptional moments of individual and/or social identity crisis.
Type 2. Reinterpret B. This is the sphere par excellence for reducing cognitive dissonance. It wasn’t wine, but vodka that I drank, consequently, I can still consider myself an authentic Moslem. It wasn’t work I did, but an amusement, so I can keep considering myself reasonable when I paid to have the pleasure of whitewashing the fence, or honest in being remunerated for playing football, since it wasn’t for play, but labour. And it is the same for honesty in a situation of arguments contrary to attitudes: if I believe what I say, then I do not mislead others in error by intention, consequently, I can maintain my identity of an honest person.
Actually, relations at this point are more complicated. Besides conditions concerning the form, honesty, and in the same way, reason or any other social quality, also has criteria related to the content. For honesty, formal criteria are given if one does not say what one does not think. The question of content criteria still remains as to whether this very thought is compatible with honesty.
In this context, we have to re-examine the famous controversy between Cohen (1962) and Rosenberg (1965). Cohen invited his subjects to justify the murderous intervention of the police force during a demonstration on the Yale Campus. As far as honesty is implicated, this social identity of a person is lost in any case, since he starts pleading justification of the intervention, either because of a form of bringing other people to believe something important that is not believed by the person himself, or by the content of really holding such a belief.
Thus, for this experience, there is no possibility of reducing a cognitive dissonance referred precisely to this social identity.
On the other hand, the form of arguing against one’s own convictions is incompatible with the social identity of a reasonable person as well, while this time the same content (an advocacy for police intervention) is not particularly inconsistent with that identity. Now, it is exactly for the cognitive dissonance referred to the social identity of a reasonable person that it holds true that the more the reward is guaranteed or the punishment prevented by this very act, the more the pains of a cognitive dissonance are compensated. If one advocates against his own beliefs one runs a risk of losing his identity of a reasonable person, but to do so for an ample reward or for an escape from a painful punishment is just the strategy depicting somebody as really reasonable. Thus, it is by no means surprising that Cohen found an inverse ratio between the size of reward/punishment, on the one hand, and the willingness of someone, driven by a cognitive dissonance, to adjust his beliefs to his words, on the other.
As to Rosenberg’s experiment, the above two factors were related to each other quite differently. This time, subjects had been invited to advocate very unpopular arrangements of the University authorities concerning the University’s football team. As to the honesty matter, this time it has the same form condition: to believe whatever is said. However, as regards the content conditions, nobody is prevented from being an honest person only because he does believe, in conformity with what he has said, that a University’s football team could be restricted by authorities (while in Cohen’s experiment everybody was prevented from it by the content of his belief about the National Guard’s murderous act).
Thus, in this experiment, there does exist the possibility of reducing the dissonance between two cognitions – “I am an honest person” and “1 believe X while having others believe non-X” – by the modification of this latter cognition.
We should remember that the greater the dissonance is, the more powerful is the drive to perform these modifications. That is the point where the reward/punishment matter intervenes. As far as the identity of a reasonable person is concerned (as in Cohen’s experiment) the former serves as a direct index of the latter: the more profitable the freely chosen act turns out to be the more reasonable the person manifests himself by this choice. Now, the opposite is true when the dissonance concerns the identity of an honest person: the more profitable a dishonest act is the more dishonest it is. For this reason, the better paid Rosenberg’s honest subjects were (as opposed to Cohen’s reasonable subjects), the greater was their experienced cognitive dissonance and, for this reason, their willingness to adjust their beliefs to the statements they had previously made.
That was what Rosenberg actually found: he started his experiment in order to falsify cognitive dissonance theory and re-establish the explanation of facts by behaviorism. It is highly symptomatic that the whole cognitive dissonance theory, being interested exclusively in the formal aspect of its phenomena, tried to parry the conclusions of his experiment. If, however, contents of social identity are taken into consideration, Rosenberg’s attempted falsification turns out to be a powerful verification of this theory.
It is the same fixation of this theory (originating from that of Lewin which in turn derives from that of “Gestalt”) on mere form that may be held responsible for the way in which it treats the above three-piece formula in type 3. It is at this point that it would be the most promising to attack, since it is this cognition in the three-piece formula which is undermined the most directly by cognitive dissonance. This is the case because, in spite of what this form pretends, there appears an A (namely me, I who am A) who does do B. Why consider that an orthodox Moslem does not drink wine if there is one (me) who does do it? If it is about the natural identity of objects one has no reticence in proceeding this way:
While having the belief (3) “The glasses of a given set do not break”, the evidence
(1) “This concrete glass belongs to that given set”, and the empirical experience
(2) “This concrete glass is broken”, one can be brought to adjust his belief (3) rather than his evidence (1) to his experience (2).
It is therefore surprising that cognitive dissonance theory does not take into consideration this way of reducing the dissonance. Why not reduce dissonance of, for example, a dishonest act by concluding that “Some honest people do lead others into error”. It is as if the cognitive psychologist said “Those who deliberately deceive others are in fact dishonest people”, or “He who acts against his own interest is really unreasonable”. Actually, it is not said, to the degree that this implication seems evident. Still, the same theory argued since the beginning with empirically observed data of subjects who neglect the most real facts of nature (such as, for example, a connection between lung cancer and the use of tobacco, or a danger of earthquakes in the area where one lives). Would the facts of social identity be more real than those of nature and, at that, of such a life importance?
Far from that, the facts of nature cannot be modified by cognitions: to go back to the preceding example, to class or not class an object among glasses of a set to notice or not notice that it breaks, modifies in no way the fact of belonging or not belonging to the glasses of this set nor that of being or not being fragile. On the contrary, it is true, as formulated by Georg Lukacs (1976), that consciousness has an ontologica1 status in the society, meaning for our present study that cognitions that reflect facts of social identity are also facts of this identity.
Thus, one carries out actions, among them socially relevant ones such as deceiving others or revealing the truth to them, drinking or not drinking wine, etc. At the same time, one may happen to think about what has been done and its social meaning, but those acts of thinking are themselves acts, too, and as such they may, like any other act, be relevant for one’s social identity. Namely, bringing an action against item 3 of cognitive dissonance is an act of thinking that is the most relevant for this matter. Thinking one may commit dishonest acts and still deserve honour is another dishonest act. Can someone who drinks wine consider himself an authentic Moslem? Certainly not, since he does something that is prohibited by Islam. Next, may someone who still considers him as an authentic Moslem be considered as an authentic Moslem. Certainly not, since he thinks something that makes nothing of the sacred interdicts of Islam.
To be fixed, the criterion of belonging to a category of social identity must be set at two levels at the same time: one of socially relevant facts and another meta-level of representations of these facts that are also socially relevant facts.
Let us go back to the above three-piece formula for cognitive dissonance. We have seen that item 2 introduces an ambiguity in identity representation. From item 3 I can conclude that “I am not A since I do B” (being given that A does not do B). At the same time, from item 1, I can conclude that “A can do B since I do B” (being given that I am A). This ambiguity could introduce arbitrariness into the definition of social identity which would be from now on a matter of consideration.
Let us consider, for example the following statement of Tajfel (1981): “We shall adopt a concept of ‘group’ identical to the definition of ‘nation’ proposed by the historian Emerson (1960) when he wrote: ‘The simplest statement that can be made about a nation is that it is a body of people who feel that they are a nation; and it may be that when all the hive-spun analysis is concluded this will be the ultimate statement as well’ (p. 102).” (pp. 229-230).
What is particularly appreciated by Tajfel in this “definition” is that by it, “members of a national group are considered as such when they categorize themselves with a high degree of consensus in the appropriate manner, and are consensually categorized in the same manner by others. His statement is essentially a social psychological one: it is not concerned with the historical, political, social, and economic events which may have led to the social consensus now defining who is ‘in’ and who is ‘out’. But there is no doubt that these events were crucial in the establishment of the nature of this consensus, and equally true that the consensus, once established, represents those social psychological aspects of social reality which interact with the social, political and economic events determining the present and the future fate of the group and of its relations with other groups” (Ibid).
However, it is undecided whether such a type of social, political and economic events incites someone to draw a conclusion from item 3 or, on the contrary, from item 1. Let us suppose that events in a population are marked for a long historical period by cooperation. For this reason will a group be formed (being given the principle according to which those helping each other are at one with each other)? Or, for the same reason will there be formed a large consensus about the mutual dependency producing mutual hate (being given the experience shared by everyone of a frustration by the impossibility of going without others)?
Now, if one would venture to reduce dissonance by type 3, this would make the dissonance reappear at a meta-level:
1. I am A;
4. I think that A can do B;
5. A does not think that A can do B.
The attempts to reduce the meta-level cognitive dissonance (that is superimposed upon the one represented in the formula given earlier by the modification of each of the cognitions would produce a very particular configuration.
For the configuration concerning item 1, we saw above that modification signifies the definition of one’s social identity. The superimposition of this second three-piece formula on the first adds a constraint to that of abandoning one’s identity because of what one does: the constraint to abandon it because of what one thinks. I must recognize that I am no longer an authentic Moslem because I drank wine, but if in spite of it I claim identity of an authentic Moslem it means I consider violable the inviolable principles of Islam that imposes upon me a second constraint to give up my authentic Moslem identity. In the same way, while having committed a dishonest act, one can only claim the identity of an honest person if he is, in accordance with this dishonest thought, dishonest. It is this very double bind (cf. Watzlawick et al., 1967) that brings those who are subjected to it to an identity crisis ending eventually in a modification of the represented identity.
If, furthermore, it was item 4 that one tried to modify, we would regain item 3 and the original dissonance founded on it. Finally, the modification of item 5 would bring us to an infinite regression: to think act B compatible with the social category A, then to think that act of thought compatible with membership in this category, then to think the same thing of the second act of thought, etc.
This double bind is that of an ideology. For as far as it is concerned, the arbitrariness described above cannot exist any more. The induction from a fact can only proceed toward the definition of social identity as if their relationship was also given as a fact. (Let us remember what was said above: “Those who deliberately deceive others are in fact dishonest people”; or “He who acts against his own interest is really unreasonable”.)
True enough, here it is the real social identity that is concerned, in the sense that it is independent of judgments (“true” or “false”) concerning this identity. However, the reality of social identity is different from the facts of natural identity. The way in which nature treats natural identity can be observed by ethological phenomena, such as the proximity or distance keeping behavior of animals (Hall, 1969). The critical distance depends, besides the present activity, on what one could call the natural social identity of fellows. Animals, in the conditions associated with a certain type of activity (feeding, mating, migration, fighting, etc.) let themselves be approached or seek the proximity of a certain category of equals while at the same time keeping a distance from those who do not belong to this category. Supraindividual formations of this nature are organized and made possible by a system of signals produced by individuals.
However, the criterion by which they signify individuals belonging to social categories arises from the genetic program of the species. Thus, once established, categorial limits will be respected unanimously by each individual of the population, independently of each individual’s categorial belonging.
On the contrary, the criteria of the social identity of man are imposed only upon those who set a value on that identity (on the beginning of the definition of social identity, see Kocski & Garai, 1978). Thus, if it seems evident to us that someone who uses illegitimate means to keep others from taking into account their own legitimate interests is dishonest, this is by no means a reflection of natural criteria of belonging to the category of honest people. It is merely the proof of our intention to belong to that category: to be honest one must think in a precise way about what one must do to be honest. On the other hand, if we simply take notice of the criteria of a Moslem identity without finding it evident that a wine drinker cannot have it, it is one proof that we have no intention of identifying ourselves as Moslems.