The paper expresses the authors' views on the growing interest in primate social cognition, particularly among descriptive primate ethologists. Its characteristics are the hope to extract cognitive interpretations from field anecdotes, the free use of intentional language, and the untested and so far untestable idea that primate intelligence was selected in social contexts. We believe that 1) To understand how the animal itself represents the structure of its group or its habitat is perhaps the most ethological ethology there is and well worth pursuing. The study of social cognition, in particular, has long been neglected. 2) However, it requires of ethologists that they learn from established cognitive science and integrate its categories with their own. This is an interdisciplinary enterprise. 3) A traditional inductive study begins with anecdotes, which then are translated into hypotheses, which in turn are subjected to empirical tests including experiments. Sociobiology began to publish hypotheses without tests; the social cognition move now goes on to publish anecdotes without hypotheses, with a strong penchant for anthropomorphic interpretations in terms of social manipulation. This is little more than applying human prejudice. Phylogenetic and cognitive insights will come from testing alternative levels of organization in an animal's social knowledge about the same behavioral interaction. The experiment is the largely unavoidable method. Examples are given. 4) The speculation of the social origin of primate intelligence is tentatively interpreted in two possible directions. A version based on ROZIN's (1976) view that generalized mammal intelligence evolved from context-specific "Adaptive Specializations" seems the more accessible to ethological thinking and method.