Gregory Bateson (1941)
St. John's College, Cambridge University

The Frustration-Aggression Hypothesis and Culture [1]


First published in "Psychological Review", 48 [p. 350-355].

I have been asked to examine the framework of Frustration and Aggression[2] from a specific point of view -- that which comes from the experience of studying contrasting cultures -- and for lack of space I shall confine myself rigidly to this point of view, at the risk of the reader assuming that I regard 'culture' as the answer to all our problems.

The greatest virtue of the book is that it is an attempt to simplify a great gamut of phenomena into a very condensed series of propositions. It attempts a formal picture, shaven as bare as possible with Occam's Razor. This simplification has this important virtue that while it makes the book easy to criticise, it also compels the critic to state his objections in terms of the simple formulations. Every such criticism must of necessity be constructive.

When we approach such a system of propositions, it is of no use to say something like this, "Your picture is in black and white, you make no mention of colour and I, the critic, am only interested in colour."

Rather, I think we should approach the formulations with the question "Can the sort of things which I, a student of culture, want to say, be said in terms of these given abstractions?" Let us, if we possibly can, avoid complicating the formulations, multiplying the entities beyond necessity.

It is true that the cultural matrix is not specifically mentioned in the basic definitions of the elements in the frustration-aggression sequence. The definitions are constructed almost as if the individual existed in vacuo. But, as a matter of fact, there are two places in the system in which culture, though not mentioned explicitly, is at least admitted by implication. Before we say therefore that the hypothesis makes no allowance for culture we must see how much latitude the formulations really allow us.

In the first place, culture is invoked by implication when the whole hypothesis is stated to refer specifically to observed human behaviour. This must necessarily mean 'cultural behaviour' since we know of no human behaviour which is not modified in terms of the social milieu in which the subject lives.

There are a few exceptions to this sweeping statement -- spinal reflexes and intrauterine reactions and perhaps some of the reactions of the newborn, but of these it may be said that we do not know enough about them to say that they are subject to cultural modification -- nor can we at present apply the frustration-aggression hypothesis to these reactions. I think it very important that these reactions should be investigated -- but for the present, at any rate, we do not know whether (for example) the temper tantrum of a new born baby is an instance of 'aggression' as operationally defined -- a series of actions having as their reinforcing goal-reaction 'injury to some other organism or organism surrogate.' So, lacking more knowledge, we must leave aside these exceptions to the sweeping statement that all human behaviour is modified in relation to a cultural or social matrix, and assume that the frustration-aggression hypothesis refers simply to sequences of culturally modified acts.

The second point in the formulations, which implies that culture was in the minds of the authors, is the definition of aggression which I have just quoted -- the reinforcement by injury to some other organism or organism surrogate. And in general it is assumed in later chapters of the book that the aggression will be directed in the first instance against the agent who did the frustrating.

Thus we see that the hypothesis is essentially a statement about series of cultural behaviours in interpersonal contexts, and it is evident that the authors regard it as such though they are willing to regard such behaviour as an animistic assault on a typewriter which will not work, a murderous daydream, or a lonely suicide, as extensions from their central theme.

Let me now try to take this central thesis into two strange cultures and briefly state whether the thesis can be made to fit. If we look at the Iatmul of New Guinea, we find that the thesis fits them perfectly, so that I need not waste much time on the details of their behaviour. The Iatmul when engaged in some series of activities which will bring a future satisfaction, will constantly look forward to that satisfaction as a means of diminishing the pains of present effort. And when they reach the satisfaction, they will heighten its value by looking back to the pains which went into the achievement. If they are interrupted, they will exhibit definitely aggressive behaviour. The thesis fits them, but they have added one wrinkle which is not provided for in the formulation -- they have invested aggression with pleasure. For the Iatmul, aggression must be regarded as a self-rewarding action series, self-reinforcing regardless of whether it ends in injury to some other person.

And they go further than this -- they habitually convert their conative efforts into imaginary aggression. The man who is cutting down a tree will excite himself to greater efforts by seeing himself as engaged in active assault upon the tree -- or the child driving out the mosquitoes will smirch them with violent sexual abuse. On the whole, however, I think it would be fair to say that the thesis fits the Iatmul at least as clearly as it does Europe.

But when we try to apply the same thesis in Bali, we get into difficulties, since it is hard to find a series of acts with a clearly defined reinforcing goal response. It is not that the Balinese behaviour disagrees with the thesis, but rather that the contexts in which we might look for the thesis can hardly be said to occur in Bali. The Iatmul and we ourselves see life as divided into sequences of neutral or unpleasant conative acts ending in satisfactions, but the Balinese do not see life like this. They are a busy, active people -- but they are infinitely willing to suffer interruption. We never at any time saw a Balinese annoyed because he was interrupted in the course of some series of acts. They seem to take a very definite pleasure in mere activity in the present -- in the very instant -- either enjoying their own busyness or else ignoring what they are doing, letting their muscles run on automatically with the activity while their attention is given to some unreal world, singing the songs from the last opera which was performed in the village.

Now it is the assumption of the theory that the typical seriation of acts, punctuated by climaxes of satisfaction is basically human and ought to occurs in all cultures, and therefore we must refer to the Balinese children.

We find that by and large the thesis can be applied to the children although it cannot clearly -- at least not often -- be applied to the adults. The children, however, can be frustrated and have temper tantrums when they are frustrated. The problem remains as to how the children are modified so as to render them unfrustratable in these terms. The problem is something like this:
"How is a certain structuralisation of sequences of acts taught to the child?" "How does it learn to see life as composed of smooth series of enjoyable acts rather than as separate sequences of acts where each sequence leads up to some satisfying climax?"

I can offer a partial answer to this question, but to state this answer I must modify the formulations of the frustration-aggression hypothesis. The hypothesis invokes two types of switching from one sequence to another. The first type is called 'substitution' and here the reinforcing act is comparable to that of the original series. Substitution is defined in terms of partial satisfaction of the original instigators. The second type is the switch to aggression about which it is assumed that the reinforcing goal is fundamentally different from that of the original interrupted sequence.

In order to phrase the Phenomena of Balinese conditioning, that there is only one type of switching, I shall have to assume -- that, in fact, the aggression sequence (the temper tantrum) is fundamentally only another case of substitute response.

So far, I have only simplified the formulations by reducing the number of entities, but I must add one complication. In order to bring the aggression sequence under the heading of 'substitution' I must assume that human acts are primarily and essentially interpersonal acts. I would say that the common element between eating the ice cream cone and hitting the mother is that both are events in a behaviour sequence involving the child and the mother. I would say that the receiving and perhaps the eating of the ice cream is for the child a pleasant small love climax in his relationship with his mother, while the temper tantrum is a hate climax in the same relationship. Either way, he gets his climax, and there is this much to be said for equating the two phenomena that among male primates and men we find a pretty strong tendency to confuse love making with aggression.

Now in the Balinese mother-child relationship, we find that the mother constantly enjoys titillating the child's emotions -- giving it a taste of behaviour sequences which the child might expect to end in climax. The mother enjoys the sequence but the climax does not occur. At the moment when the child either flings his arms around her neck or bursts into tears, the mother's attention seems to have wandered; she is in a brown study or she is speaking to somebody else.

In this way, I believe the Balinese child is driven not to expect or look for climax in his acts, but to take his pleasure where his mother took it -- in preliminary steps with no defined goal -- and to live in the immediate present not in some distant goal.[3]



  1. Read at the 1940 meeting of the Eastern Psychological Association in the Symposium on Effects of Frustration. The fieldwork, on which this paper is based, was done by the author as a Research Fellow of St. John's College, Cambridge, 1936, and William Wyse Student in Social Anthropology, Cambridge, 1937-1939, in collaboration with Dr. Margaret Mead, of the American Museum of Natural History; with grants in aid from the National Committee for Mental Hygiene, through a grant from the Committee for Research in Dementia Praecox, founded by the Thirty-third Degree, Scottish Rite, Northern Masonic Jurisdiction, and the Social Science Research Council. For an account of the scope of this research see Researches in Bali, 1936-1939. Trans. N. Y. Acad. Sci. Series ii, 2, No. 1, 1939.
  2. J. Dollard, L. W. Doob, N. E. Miller, O. H. Mowrer & R. R. Sears. Frustration and aggression. New Haven: Yale Univ. Press, 1939.
  3. After writing this paper, I had a very interesting conversation with Dr. Mittelmann about the rare contexts for rage among adult Balinese. I mentioned the following contexts: (1) At cockfights it is common to see something like rage expressed verbally against any infringement of the rules of procedure or against a man who bar and who confesses (when he has lost his bet) that he had no money. (2) A thief (especially a cattle rustler) is usually mobbed and killed if he is caught red-handed. These contexts are interesting and I think that consideration of them will amplify the thesis of the paper. We may summarise that Balinese show rage whenever there is some sudden overt threat to the stability of that intricate series of conventions within which they live. It is also interesting that cockfighting and gambling are, for the Balinese, in some sense regressive activities, to which men become addicted under stress of misfortune, and that cockfighting with its clearly marked climaxes of excitement between the cocks and among the spectators produces a psychological atmosphere a good deal different from that of daily life. Among the Balinese, the common metaphor which is used to imply sexual attraction between boy and girl is "they stare at each other like two cocks with their feathers up."