None of your examples entail building a mind. Also there's a kind of irony in my example, since as a branch of computer science it is independently funded and seemingly unrelated to mental health.
You have a model of the brain. You do not have a model of the mind. You assume that by simulating a brain with significant details, a simulated mind will emerge. I find that's a big pill to swallow.
Nah, I don't assume it. Hence 'may' and 'attempts'. But the explanation is simple enough. If you can discover what it takes to code a mind then you're also going to learn some of the systemic faults that minds, qua minds, can develop. Of course, I could be wrong about this. It could be that mental diseases are all purely down to hardware issues (brain health). But it seems like a bad bet. For example: addiction, or at least specific types of addiction, depend on people's culture and choices.
Sure, it isn't ethical to experiment w/o consent. But in order to program an AGI, you need first to conjecture an explanation of how the mind works. One may be able to deduce from that explanation some of the ways in which minds can go wrong. Also, assuming consent is given, it will likely be easier (and less physically invasive) to observe the internal state of an artificial person.
( ) Hypnotism ( ) Psychoactive drugs ( ) Functional programming ( ) Meditation ( ) Phrenology