Not trying to discredit any particular hypothesis — if I were forced to bet I'd probably bet on something involving physical harm from adversaries — but the ide

Not trying to discredit any particular hypothesis — if I were forced to bet I'd probably bet on something involving physical harm from adversaries — but the idea of using normal controls to investigate a psychosomatic hypothesis has always seemed a little off. Normal controls are reasonable, but wouldn't you want to compare individuals to a group with psychosomatic diagnoses as well? I would probably expect a group of individuals with psychosomatic problems to look different from normal controls also.