Mental causation takes explanatory priority over evolutionary biology in most accounts of eating disorders. The evolutionary threat of starvation has produced a brain that assists us in the search for food and mental change emerges as a consequence. The major mental causation hypothesis: anxiety causes eating disorders, has been extensively tested and falsified. The subsidiary hypothesis: anxiety and eating disorders are caused by the same genotype, generates inconsistent results because the phenotypes are not traits, but vary along dimensions. Challenging the mental causation hypothesis in Feighner et al. (1972) noted that anorexic patients are physically hyperactive, hoarding for food, and they are rewarded for maintaining a low body weight. In 1996, Feighner's hypothesis was formalized, relating the patients' behavioral phenotype to the brain mechanisms of reward and attention (Bergh and Södersten, 1996), and in 2002, the hypothesis was clinically verified by training patients how to eat normally, thus improving outcomes (Bergh et al., 2002). Seventeen years later we provide evidence supporting Feighner's hypothesis by demonstrating that in 2012, 20 out of 37 patients who were referred by a psychiatrist, had a psychiatric diagnosis that differed from the diagnosis indicated by the SCID-I. Out of the 174 patients who were admitted in 2012, most through self-referral, there was significant disagreement between the outcomes of the SCID-I interview and the patient's subjective experience of a psychiatric problem in 110 of the cases. In addition, 358 anorexic patients treated to remission scored high on the Comprehensive Psychopathological Rating Scale, but an item response analysis indicated one (unknown) underlying dimension, rather than the three dimensions the scale can dissociate in patients with psychiatric disorders. These results indicate that psychiatric diagnoses, which are reliable and valid in patients with psychiatric disorders, are less well suited for patients with anorexia. The results are in accord with the hypothesis of the present Research Topic, that eating disorders are not always caused by disturbed psychological processes, and support the alternative, clinically relevant hypothesis that the behavioral phenotype of the patients should be addressed directly.