Autism and cognitive impairments are problems with the brain. There was no reason for them to be mentioned (in the context of Szasz's work) because they have nothing to do with "mental illness". As for depression, let's hear what Szasz has to say:
"With the passing of the years, my earlier conviction that conflict and disagreement are not diseases has only grown, and with it my conviction that listening and talking are not instances of medical treatment. This does not mean that I disparage listening and talking; quite the contrary. Attending to what another person tells us and being able to hear what is in his heart as well as on his lips is an exceptional skill, growing rarer by the day, especially among psychiatrists; the same goes for addressing another person and being able to tell him -- frankly, simply, and without humiliating him -- what we think of his predicament and his options to extricate himself from it. It was a tragic folly in the past to try to confine this sort of human helpfulness within the bounds of accredited creeds, and to restrict the right to dispense it to clerics; and it is a tragic folly now to try to confine it within the bounds of accredited cures, and to restrict the right to dispense it to clinicians." -- "The Ethics of Psychoanalysis", 1974.
Finally, we can still ask a number of Szaszian questions in his absence. For example, does a troubled woman suffer from an illness called "anorexia" or does she starve herself as a way to communicate to the world her problem in living? By the same token, does a person who overeats suffer from an "eating disorder" or does the person have a problem in living that has nothing to do with illness or disease? (To be sure, obesity can cause disease, like diabetes, but "obesity" is not a disease.) Finally, does a person who drinks alcohol to excess have a disease that "compels" him to drink or does the person have a problem in living? The subject of drug habits is of special interest to me as it was to Szasz so permit me to go on at length, if I may.
Szasz once said somewhere that if a person is told that he or she is predisposed to being extra sensitive to alcohol in that drinking will make the person tipsy more quickly than other people, then such information means that the person has knowledge that s/he can act on: you can choose to drink responsibly or you can choose to not drink at all. Hold on. This isn't to be confused with the fact that some long-term chronic drinkers can die from sudden withdrawal. But that's not evidence that chronic alcohol drinking is a disease; that's evidence of a poisoning, a crucial distinction. But even for those who have poisoned themselves and need medical assistance to wean themselves off of alcohol (or heroin), chronic drug taking is an activity firmly rooted in the realm of freedom and choice and is not due to some dopamine "directive" in the brain gone awry that "compels" people to drink, smoke, inhale, or inject. Herbert Fingarette, who is apparently familiar with Szasz's work, perfectly captures the maddening contradiction in his book "Heavy Drinking": “Heavy drinkers are told that they are unwilling victims of a disease that destroys their ability to manage their drinking and yet they must strive to exert absolute self-control, that only total abstinence can save them.” That's perfect.
In our mistaken belief that heavy drug taking has more to do with the sensitivity of our bodies and less with the character of our minds, theories of drug users in the grip of a disease become not advancements in medicine but a degradation of our ideas of freedom, and a freedom that includes, by definition, the possibility of not doing the right thing -- which can mean going down a tortured path of drug taking whose end point is self-destruction. The moment we can admit to the reality of mind and choice -- that exogenous substances such as marijuana, cocaine, heroin, nicotine, and alcohol can only enter the body when one chooses to let them in whatever degree or amount -- is the moment when the entire edifice on which the drug-addiction-is-a-disease model rests crashes to the ground.
I guess I'll stop here.