Classics in the History of Psychology

An internet resource developed by
Christopher D. Green
York University, Toronto, Ontario

(Return to index)


David M. Levy (1941)
New York City

First published in Psychological Review, 48, 356-361.

In Frustration and Aggression [2] the situations in which frustration occurs always call for aggression, in terms of an attack on a frustrating agent. If no frustrating agent is present, then some object must be created for the purpose of relieving aggressive tendencies that arise in the frustrating situations. The need varies according to the tolerance to frustration. If the tolerance is poor, then some aggressor must be fabricated. A thwarted individual would then displace the aggression onto a system or onto any group representing a constellation of ideas that evoked hostility, no matter how mild, in the previous experience of the individual. This is all a familiar type of psychodynamics, namely the release of tensions arising from frustration and the use of an available object for its expression. To say, however, that aggression arises as a result of any frustrating experience is a generalization that requires scrutiny. There are any number of frustrations that do not evoke aggressive response in the sense of discharging hostility against a social object or its surrogates.

There are for example, a number of experiments in which animals are frustrated and in which such aggression does not occur. In my own work on the sucking behavior of dogs, an experiment was made in which two puppies were given adequate milk from the bottle, but fed so quickly that their sucking needs were never satisfied. In contrast to the two 'control' puppies who were given adequate satisfactions both of feeding and of sucking, the experimental animals showed no problem in aggression that could be directly traced to the sucking frustration. The result was rather a type of perverted sucking. They sucked each other, [p. 357] their own paws, objects, and later on, after eating, they licked the plate interminably. In terms of general personality description, one of the experimental puppies could be described as more aggressive than the controls, the other less so. But these differences, it could be shown, were modes of reaction that occurred in the beginning of the experiment and were probably reinforced by the particular difficulty of the situation. I do not wish to elaborate this point, except to say that the puppy who was originally aggressive became more so, in certain respects, after the sucking difficulty was established, whereas the other puppy became more submissive.

In chickens in which pecking frustrations were produced, there was also no evidence of increased aggression. In the experiment 100 chicks were fed from troughs in the usual way, but prevented from pecking off the ground by covering it with a raised wire mesh. The control group of 100 in the adjoining half of the same chicken house were fed from troughs, but not prevented from pecking from the ground. The chickens on the wire pecked each other's feathers but, as the experiment revealed, the pecking was not due to increased aggression but to increased pecking needs. This same type of situation has been shown for other animals and for human infants also. The sucking frustrations in infancy cause finger sucking or sucking of other objects, as in the case of the experimental puppies, rather than increased aggression. There is no proof that the so-called weaning traumas of infancy cause more aggression or even more phantasies of hostility and the like than in other children. The same maybe said of all those frustrations that have to do with bowel and bladder control.

A distinction may be made between the type of aggression described, especially in regard to sucking habits, and the situations of the type described by Dr. Dollard and his colleagues. The former may be called physiologic, the latter social types of frustration. However, Dr. Dollard has included the type of physiologic frustration I have mentioned as typically provocative of aggression. His generalization could be easily amended. It is a question, however, as to whether numerous situations in which the individual does not [p. 358] deal directly with frustration readily translatable into terms of an aggressor, typically stimulate the aggressive rejoinder, for example, frustrations arising out of one's own inability to solve a mathematical problem and the like. The fact that tension may arise in any such instance, and that this tension is released by some motor action, whether tapping a pencil or pacing the floor, does not mean it is an aggressive act in asocial sense. Furthermore, acts that typically call for aggressive behavior in certain individuals may affect others differently. To say that in such cases the aggression should follow but is repressed would require considerable study. Aggression is one of the ways of responding to frustration in a social situation. Presumably it varies in the strength of its impulse and its execution in different individuals. In the phantasies that occur during frustration, or in the choice of behavior to satisfy the particular tensions that arise, various possibilities occur. To state that only one possibility, namely the aggressive act, is the logical response to frustration, all others being forms of extemporizing, needs further proof.

The response of a child to the new baby is a very good example of a frustrating situation commonly evoking aggressive behavior in the form of an assault on the baby or mother, or both. However, though this pattern is seen most frequently, there are instances in which it does not occur. For example, a child may respond to the coming of the baby with, primarily, a desire to possess it, to have it for one's own. This is not an aggressive response in the form of an assault on a social object, yet the reaction may be a very strong one.  The new baby may call forth a very strong maternal protective attitude, especially when the prior child is a girl, say, eight or ten years older than the new-born. In this situation there is a frustrating experience highly reduced as compared with one in which the age difference is only two or four years.  Nevertheless, such situations may reveal frustration and yet show a maternal response as a primary determinant. It maybe argued that the maternal response represents jealousy of the mother and hence a concealed aggression against her. That is to say, one may always argue in favor of the theory of aggression as against any other form of behavior in a [p. 359] similar situation. As yet, are have no definite proof that that is always the case.

In regard to sibling rivalry, however, the aggressive response to the new baby is so typical that it is safe to say it is a common feature of family life. As seen in 'control situations' the aggressive act in its various forms is depicted so clearly that the dynamic process is worth describing. In the control situation dolls are used representing the baby at the mother's breast and an older child who stands for the patient.  If it is a boy the question used to set off the behavior is: "And then the brother sees the new baby at the mother's breast. He never saw him before. What does he do?"

In reviewing the patterns of over too S-R experiments of children ranging from two to thirteen years, it is most useful to conceive of the act as an ongoing social process, a dynamic unit of behavior, with various influences brought to bear upon it in every phase. The completed primitive performance is an act in which the child attacks the baby doll and destroys it by biting it, tearing it with his fingers, or crushing it with his feet. If the experiment is repeated, there is, in most of the instances at ages three, four and five, a fulfillment of hostile activity of this type. In others there are varying approaches to this end-point, easily observed and measured.  In the beginning of the act when, presumably, the impulse to attack is felt, one observes, varying forms of inhibitions to the impulse, so that the act does not come out. These maybe in the form of pauses, of saying, "I don't know what to do," of attempting to change the play into some other form, to play with other objects, or even to get out of the playroom.  Sometimes the first act is to slap the doll standing for the subject. When a three-year-old girl was asked, "Why did you do that!" she said, "Because she was bad. She wanted to hit the baby." This type of response indicates that the hostile impulse had to be dealt with before the act occurred, that the impulse itself was judged to be bad, that the thought of the act had to be dealt with by punishment as though it were already fulfilled. The inhibition of the impulse may take the form of assumed stupidity like, "I don't know what you mean. I don't know what you want." Such inferences are [p. 360] based on the fact that without any explanation, merely by saying "Go ahead" and repeating the experiment, the hostile act in such cases follows.

Without elaborating the meaning of the various kinds of inhibitions that take place before the act overtly occurs, it may be sufficient to say that we are dealing with repressions and, I believe, the equivalent of superego injunctions; that is to say, the child is in its impulse to act identifying itself with the attitude of a prohibiting parent.

Once the act goes into execution we see a number of efforts to deflect its aim so that the object of hostility will not bereached. At this point a common form of inhibition is a blocking of the act, as, for example, a slapping movement made at the doll which is held back. Another common form is displacement, whereby an object close to the doll is hit and the doll itself avoided. This may be seen also in the form of non-specific aggression, in which the child shoots at the ceiling, or at various objects in a direction entirely different from that of the baby doll.

So far we may say of the modifying influences that occur when the impulse to act is felt, the inhibiting influences tend to block the act when they occur at its initial point (the impulse) and, once the act goes into execution, to reduce it to a gesture or shunt it off.

Even when the object is directly hit, modifications of the attack at the target appear. For example, the hit may change into a touch. Instead of being struck at, the baby is just removed, or dropped. At this point, too, the attack may be disguised. The child takes the baby away and says, "It has to go to the hospital for an operation," and the like. As, presumably, the behavior of the child is released, the attack on the baby becomes free of all modifying forms and the object is destroyed in the manner described.

Even when the baby doll is attacked freely and crushed, the act is not necessarily completed. We are aware at this phase of a number of activities that prove its incompleteness. For example, a child after attacking the baby and crushing it, begins to defend itself for its behavior. The child says, "It was a bad baby." Another child says at this point, "We [p. 361] don't need two babies in one house." Commonly at this stage the doll standing for the brother or sister is attacked, usually with the same method used on the baby. Another common pattern is seen in attempts to restore the baby, to make it come to life again, to fix it up, to deny what happened, and say, "The baby fell apart. Now it's all together again."  One child at this stage said it was all a dream, it didn't happen. We see in these various performances that take place in the completion of the hostile attack on the baby, restoring behavior, self-retaliatory behavior, attempts at self-justification, and attempts to wash out the act by attributing to it the aspect of a dream or just play. In several cases children had a kind of war dance after achieving their purpose. In line with other patterns it would be interesting to speculate on this type of war dance as a way of warding off anxiety described in psychoanalytic literature.  Through these control situation experiments we see depicted the type of aggression of which Dr. Dollard and his colleagues have written in clear form, influenced in every stage of their study by the emotional problems of the individual. [3]

[1] Read at the 1940 meeting of the Eastern Psychological Association in the Symposium on Effects of Aggression.

[2] J. Dollard, L. W. Doob, N. E. Miller, O. H. Mowrer & R. R. Sears. Frustration and aggression. New Haven: Yale Univ. Press.

[3] Certain applications of the dynamic pattern presented are being made to a number of problems in psychodynamics. In various anxiety symptoms, for example the phobias, we see inhibitions to the act at the impulse. In the ties of children we see inhibitions to the act while it is taking place, for example, ties of the eyelid representing inhibition of sex curiosity, mouth-opening ties representing an inhibition of biting tendensies, certain choreic-like ties of the hand representing incompleted masturbatory behavior. I think the application of this concept would solve a problem presented in Freud's The Problem of Anxiety. In Chapter IV of that book he deals with his correction of a theory of anxiety, to the effect that anxiety doer, not emanate from repressed libido, but from the attitude of the ego. Anxiety is caused by the ego's awareness of danger. This danger may come from without or from instinctive impulses within the organism, that is to say, from libido. Freud states that anxiety is not due to tranformation of libidinous charges as he thought formerly, but the ego's awareness of the danger of these charges. The repression that follows is due, therefore, to the activity of the ego, not to a transformation of libido into anxiety which is then later repressed. However, Freud found his old theory quite adequate in its application to anxiety arising in sexual behavior, in which the sexual act is not completed in a true physiologic sense. In the latter instance he adhered to the view that the undischarged libido is converted into anxiety, recognizing a contradiction of the two viewpoints. The contradiction is solved if we consider the phase of the act from which the anxiety emanates. In the first set of circumstances, the phobias, it emanates when the impulse to act occurs. In the situation of frustrated sexual behavior the inhibition occurs at the close of the act or, at least, when the act has gone to a near degree of completion. Hence, anxiety may arise during the act Pt any phase, and the rule that it arises from the inhibition of the act at any given point may still be maintained.