Therapists, [Emmeline] Clein writes, insist that the eating disorder is “a second self … capable of mind control.” [...] Doctors would say, “‘this is the eating disorder talking,’ when you try to describe desires or moods,” one patient recalls. “She was praised when she was able to parse an ‘ED thought’ from a supposedly genuine thought,” Clein writes, “and applauded when she narrated her illness in external terms, casting it as a bad friend or an abusive lover.” Anorexia’s treatment thus reinscribes the disease’s own logic: the anorexic self is separate from the “true” self, who must be recovered; threats to the true self must be identified, quarantined and locked away—exorcised, even—rather than, as Clein puts it, “endured” and integrated. Clein, then, calls for a mode of treatment that teaches the patient to recognize the ways in which her agency has been scaffolded by and complicit in her disease. But she avoids any real confrontation with the problems that beset such a suggestion. The autonomy-deficit model has its attractions. In particular, it seeds an easy justification for why we must sometimes treat anorexics against their will. If anorexics are autonomous, shouldn’t they be left to get on with their diets? [...] I don’t know whether Alyssa should have been prescribed the MAID [Medical Assistance in Dying] drugs. But I am sure that she wasn’t lacking in autonomy. There was no demon lodged in Alyssa, wagging her tongue. Her voice is quite her own. The unhappy truth is that sometimes people with no agential defects will choose things that harm them. We can try to escape the hard choice between complicity on the one hand and paternalistic interference on the other by “finding” agential flaws where there are none. To read the anorexic as autonomous, then, need not mean that we stop trying to get her to eat. The anorexic governs and authors herself more fully than her peers. That doesn’t show that we should admire her discipline. It shows that we should be warier of self-governance as an ideal.