A Suicidal Symptom in a Child of Three1

The analysis I am about to describe is of Rachel, aged three, who came for treatment because of an acute inhibition of eating. I am going to try and give an account of what she seemed to be doing in her play and show some of the evidence leading up to my main hypothesis of what she was actually trying to do when refusing to eat. When I was actually working with the child it certainly seemed to me that the hypothesis I am about to put forward explained a large number of the facts of her behaviour; but when I came to consider how to present the material I did not feel certain that I might not have made certain theoretical assumptions that I could not substantiate from the evidence. It is true that the child got better, but I did not feel this was sufficient proof that what I had tried to tell her was necessarily the true explanation of what she was doing. In fact, I began to consider the whole question of the sense of conviction of the truth of one's interpretations which I suppose every analyst has, at least much of the time, when conducting a successful analysis. And I came to the conclusion that in fact I could not prove my hypothesis at all from the material, for though the material in the analysis may seem to provide convincing proof, for the analyst conducting it, of the truth of his theory, for anyone else I think it can only provide illustration of the theory. I think this must be so, since the material presented to anyone else must always be a selection from the great richness of varieties of behaviour (including gestures, manner, tones of voice) and must therefore always be selected on the basis of some theory; thus one can never prove that one's selection of the material is unbiased and that one has not omitted other facts which would prove some different theory. Thus I thought that the only way of evaluating the conclusions, as distinct from illustrating them, would be by trying to show their utility in explaining other facts, beyond the analysis, a