Well, I highly recommend Szasz's book _Insanity: the Idea and its
Consequences_.  I can't say for sure on this, but apparently there
is an overwhelming tendency among "real" doctors to define disease
in terms of objectively verifiable cell damage (perhaps parallel to
TV-repairmen's definition of "broken" in terms of not turning on?).
I've had the chance to ask one or two doctors about this, and they
said that that is true, for whatever it's worth.  Apparently,
pathologists have no interesting controversies about what is a
"disease,"
and they have clear criterion for establishing them as such.

But perhaps the real problem is that people are taking philosophical
stances under the pretense of science.  For our intuition (mine
at least) is that there are two very different kinds of things:
_inabilities_ and _unwillingnesses_.  A person has a disease if
they are _unable_ to do something they otherwise could; a person
has a pseudo-disease, dubbed a mental illness, if they are _unwilling_
to do something that psychiatrists think they ought to do.  The
philosophical difficulty underlying is probably that most doctors
are determinists who believe that free will is illusory; hence
apparent unwillingness is merely another form of inability.

Of course, the problem is then that _all_ undesirable (undesired
by shrinks?) behavior is a fortiori a disease.  Not a plausible
conclusion, though of course many embrace it.

Now perhaps it will be argued that I am setting up a straw man.  For
could we not take an intermediate position, according to which
most apparently unwillingness is voluntary, but the extreme tails are
actually inabilities?  I don't claim that this view is incoherent.
But here are my reasons for taking the "extreme" view that all
behavior not traceable to brain damage and the like is fully voluntary.

I begin with a wonderful line from the Roman poet Terence:

Homo sum, humani a me nihil alienum puto.
I am a man, and nothing human is alien to me.

What am I getting at here?  Simply, that there is no "insane"
behavior that is so foreign to me that I cannot see it as one
of the countless manifestations of the depths of the human soul.
Adolf Hitler orders the death of millions of Jews.  I cannot
empathize with his decision; but I _understand it_; for I understand
that hatred and cruelty are one facet of the human personality.
As I child, I was a little bundle of passions; and I remember
feeling hatred that could have led me to take life were I not a
powerless child.  Jeffrey Dahmer raped and tortured boys, then 
murdered them and ate them.  Again, we may view his action as
unthinkable, but is it?  Isn't sexual obsession yet another facet
of the human personality?  One needn't feel the same impulses to
understand that a different person may have different impulses,
and that they may choose to act on them.

Think of any great character from literature.  Wouldn't the whole
lot of them be labelled "mentally ill"?  Literature, with its
depictions of the depth of the human personality, with its depictions
of the reaction of ordinary people to extreme circumstances.  But
isn't it the essential characteristic of great literature that it
makes us see the world through the eyes of another person, makes
their behavior and choices seem natural to us, makes us see ourselves
in another?  When we read _the Brothers Karamazov_, for example,
we experience Dmitri's desire to kill his father as natural;
and we experience his choice _as a choice_.  Similarly, 
Smerdyakov's choice to murder his father solely in order to prove
to himself that moral law does not bind him: we experience his choice
as a choice, because we recognize his action as a manifestation of
a desire to rebel that all of us understand.  And finally, when
Smerdyakov hangs himself from guilt -- is that any less understandable
to us?  Is action followed by intense remorse so hard to understand
as part of the human experience?  Surely anyone who has deliberately
hurt one he loves has felt the urge to suicide; or at least _we can
understand_ such an urge.

These examples are merely meant to jolt your intuitions.  Before
us we have two hypotheses: the determinist one that at least
extreme behavior is really a kind of inability, even though it appears
to be merely an unwillingness.  And the hypothesis of radical free
will, which claims that there is no human behavior so bizarre and
wicked that we cannot readily understand it as a free choice once
we try.

But -- what about studies showing that schizophrenics have 50%
more of some chemical in their brains?  When I see claims like these,
I can only intone "reverse causation" again and again.  Perhaps
making unusual choices _causes_ the presence of unusual chemicals,
rather than the other way around.  Certainly if we accept introspection
as a form of observation -- and why shouldn't we? -- we have lots
of support for this idea.  I suppose a reductio ad absurdum of my
view here would be that you could even claim reverse causation for
brain tumors.  True enough; I would distinguish this case in virtue
of the fact that no one has been shown to be able to produce a brain
tumor into existence on demand, whereas surely they could do so
with some brain chemicals.

I have something related to say about the ability of drugs to change
behavior, which is usually supposed to show that somehow the original
condition was a "treatable disease." I just think that this idea
is (if you pardon the expression) crazy.  Suppose we give a happy
person a drug that makes him miserable: does this show that happiness
is a disease?  Similarly, why should a drug that makes a sad person
happy show that sadness is a disease?  Or suppose we give a 
so-called "hyperactive" kid Ritalin, and he calms down.  How does
that show that "hyperactivity" is a disease?  Would calmness be
shown to be a disease if a drug could turn calm kids hyperactive?

--
But what should we make of these claims that "what is a disease" is
culturally determined?  Well, there are two interpretations of this
statement.  The first one is anthropological: people acquire their
views of disease from other people in their culture.  The second is
epistemological: the _justification_ of a statement like "X has a
disease" is cultural.  Now the first view might be coherent; but
the second surely is not.  For how could the mere agreement of
a bunch of people that "X has a disease" make the proposition true?
Only, to be sure, if "diseased" is just a disguised way of saying
"to be pitied" or "lacking valued abilities."  But if this were the
case, why not get rid of of the word and say what we really mean?

But I submit that this option is rather counter-intuitive.  For
suppose I proposed the following bare-bones definition of disease:
a person has a disease if his body contains bacteria or viruses
which will kill him within 24 hours.  Of course this definition does
not exhaust  our common-sense concept; but surely it is not "culturally
determined" whether a person has a disease in this sense.  Now
I suggest further that the above is _sufficient_ for
"having a disease" in the _common-sense_ meaning of the term;
hence there are at least some cases where the presence of disease
is not culturally determined.

Or to reverse the perspective: surely a witch doctor or Soviet
psychiatrist's view that "Bill has a disease" is just plain wrong.
They may have the backing of their culture, but nevertheless they err.
Don't they?
--
You make an interesting use of Rawls' concept of the original position
to justify involuntary commitment, drugging, etc.  I assume you get
this out of section 39, where he tells us that:

     The problem of paternalism deserves some discussion here,
     since it has been mentioned in the argument for equal liberty,
     and concerns a lesser freedom.  In the original position
     the parties assume that in society they are rational and able
     to manage their own affairs...But once the ideal conception
     is chosen, they will want to insure themselves against the
     possibility that their powers are undeveloped and they
     cannot rationally advanced their interests...[Mention of
     children and the mentally disturbed as prime examples,
     but continues as follows] It is also rational for them to
     protect themselves against their own irrational inclinations
     by consenting to a scheme of penalties that may give them
     a sufficient motive to avoid foolish actions and by
     accepting certain impositions designed to undo the unfortunate
     consequences of their imprudent behavior.  (section 39, pp.248-249)

Now the case of the mentally ill should be particularly problematic
for Rawls, because he explicitly states that the choice of principles
of justice must be neutral between different "conceptions of the good."
But why shouldn't the values and choices of the putatively mentally
ill be interpreted as a different conception of the good?  This
is especially clear when the so-called mentally ill are perfectly
lucid but behave "strangely."  For example, supposedly mentally
ill people who are voluntarily homeless; why should their "conception
of the good" be weighted differently than the similar desire of
the Franciscan monk to take a vow of poverty?

You do touch on the interesting issue of childrens' rights.  Suffice
to say that there is a big difference between practicing paternalism
on people whose intelligence and knowledge are drastically limited,
and practicing it on another adult who has normal abilities but
behaves strangely.  I think that hypothetical consent may make sense
as a standard for young children and the retarded, but extending it
further seems laden with grave difficulties.

--
I conclude this rather long comment with my normative views.
To begin with, I believe that medical care should be based
upon the consent of the patient; and moreover, so should pseudo-
medical care.  Just because I don't consider "mental illnesses"
to be real diseases doesn't mean that people shouldn't be able
to buy drugs to try to change their behavior.  But as least as
important: just because someone doesn't like your behavior, and
possesses a drug that can change you, doesn't mean they have a right
to involuntarily drug you.

Now of course "mental illness" does exist in one sense: there is
very strange behavior, some of it harmful to oneself, some harmful
to others.  If harmful to others, then I favor appropriate criminal
penalties, much as I support the rights of devil-worshippers until
they start sacrificing children, at which point I favor standard
punishments.  If harmful to themselves, then of course I favor no
penalty at all.  The sorrow that self-destructive people bring to others
simply has to be accepted and dealt with by voluntary persuasion
and the like.  (As Szasz says: "The truth is, the mentally ill are
not disturbed; they are _disturbing_.  They are not sick, they are
_sickening_")

Probably the two obvious policy implications that drop out here are
(1) I oppose involuntary commitment and (2) I oppose the insanity
defense.

Now of course the difficult case arises with children.  To begin
with, I believe that we should stop trying to make ourselves feel
better by labelling the drugging of problem children as "treatment."
Once we face that harsh fact, we will have to judge whether it
is a good idea to change bad behavior using drugs, taking into
account that the child's lack of recourse raises the burden of
scrutiny which we must impose upon ourselves.
--
To conclude, I believe that one of the greatest errors of modern
thought is the medical metaphor.  The medical metaphor adds 
nothing to our knowledge of human choice and behavior; indeed,
it takes away the deep and penetrating knowledge that we once had
when we looked at every human being as another free-willed being.
Calling wild mood swings "multiple personality disorder" doesn't
teach us anything; but it is not without purpose.  Its purpose
is to de-humanize the person, to strip him of his dignity,
and to justify coercion.
-------------------------
I would think that mathematical _potential_ is determined by
your brain.  There is a little to how good I could get at math,
no matter how hard I tried.  "Being a genius" is rather different;
you have to take advantage of some of your mathematical ability
by applying some effort before you could be considered a "genius."

Maybe I should map out my argumentative flow chart a little better.
1. Mental illness by definition cannot be a condition chosen at the
moment at which it is occurring.  (A physical illness could be
chosen at some time in the past, such as when I inject smallpox into
my veins.  But it isn't choosable at the moment at which I am
suffering from it.  I think imposing the same restriction on our
concept of mental illness makes sense.)
2. Some conditions considered to be mental illness are straightforwardly
brain diseases.  E.g. Alzheimer's disease, and some effects of
syphillis.
I accept these without a problem.
3. So the interesting class are mental illnesses which don't appear
to be brain diseases.
4. Now to begin with, it isn't clear that this idea is even coherent.
Is the idea of free-floating mental damage, independent of brain
damage but not chosen by the mind, a coherent idea?
5.  Let's suppose that it is coherent.  Nevertheless, should we even
consider it?  E.g., the idea that some people appear to have minds
but don't is perfectly coherent; but nevertheless I think it to be
absurd, just based on my own study of my own mind.  Similarly, might
we rule out as absurd the existence of a subset of the populace without
free will?
6. Let's suppose we don't.  Now we are at my final argument, which
I perhaps find most persuasive.  Terence summed it up: "I am a man,
and nothing human is alien to me."  Put a little more formally,
when we observe a person, we can apply one of two hypotheses.  The
first is that they are compelled to act as they do; they don't
experience
their actions as choices; they are mentally ill.  The second is
that they do experience their actions as choices; and however unusual
their choices, we can nevertheless _understand_ their choices.  This
is where I draw upon the power of literature: for it depicts the
extremes of human action is a manner that makes the actors
understandable.
And my argument here is that I have never found a case where I
thought that the second hypothesis wasn't better than the first.

I might add a final point: could many of the things called mental
illnesses be so adjuged without making implicit _value judgments_
about what kind of life you ought to live?  A pathologists can tell
you you have a tumor without judging whether it is good to have a
tumor (if you are Hitler, it is probably good, right?).  But
a psychologist couldn't tell you that you are hyperactive unless
he made a value judgment that you should be less energetic and
flighty.