Sometimes, he argues, the term is used to refer to peculiarities of thought or behaviour arising from a disease of the brain (illness). And other times, he argues, the term is used to describe something which is very different than a disease of the brain, something he calls 'problems with living' (mental). But neither of these definitions make much sense when followed through logically, he claims.
In the first instance, if mental illness is at root a disease of the brain, then it seems to only confuse matters to call it 'mental illness'. The problem is nothing to do with anything 'mental' but is a problem of the brain. There are many physical disorders that take this form that we would not make the mistake of classifying as mental. For instance in elderly patients who lose their sight, visual hallucinations are common. Or in the late stages of syphilis, panic and hallucinations are common. But it would be peculiar to consider either 'mental' problems rather than brain/physical problems.
The second definition fairs no better. If mental illness is a problem with living then it seems to not make sense to talk of it as a disease. As Szas says, "Since medical action is designed to correct only medical deviations, it seems logically absurd to expect that it will help solve problems whose very existence had been defined and established on nonmedical grounds."
But does this argument exhaust all the possibilities?
Let's accept the two premises above: 1) the root cause of a mental illness can't be a physical problem and 2) medical action can only be used for mental illness if it's a medical/physical problem. Both conditions can be satisfied if 'mental illness' was such that the root cause was that the patient has a 'problem with living', and this in turn affected the brain in such a way that the patient would benefit from medication.
This third understanding seems to solve Szasz problems but it is also, I believe, what many people have in mind when they speak of mental illness. Consider how mental illnesses are treated: through a combination of therapy and pills aimed at altering brain chemistry. The pills alleviate the symptoms but the patient is recommended that they continue to work out the underlying issues in therapy before they stop taking the pills. (Why would they need therapy if the pills stopped the underlining problem?)
Of course, none of this is to say that there isn't a dark side to how mental illness is thought of. There is still a problem of classification, and the potential abuse of broadening what is mentally ill too far. After all who decides what counts as a mental illness, and what is merely a personality? Indeed, all ideas have an effect on the brain in at least some mild way. But we can begin to see here that a satisfying definition of mental illness could exist, if we can show that the unwanted personal problems in a patient have made the brain ill in a way that is already recognised as an illness in medicine more generally. For example, one is ill if they hallucinate whether the cause be mental illness, syphilis or visual impairment. On the other hand, there is no illness outside mental illness that has, say, excitability as a symptom (or whatever other absurd 'symptoms of mental illness' are currently being employed).
