If you imagine therapy as a way to condition some sort of gradient descent on emotions and awareness you can already see the problem.

The implication of your statement is you eventually reach some global optima. The reality is you become over-aware of some things and under-aware of others. This is the local optima and you've been caught in a bowl. Therapy "works" when your local optima is "good enough" for your own definition of done. However, it can often take several "bumps" out of those local optima to find it and once again you haven't really "optimized" like you are implying.

I would think the focus on optimization of emotional and awareness skills would simply lead to more, not less, anxiety. It sounds like the same problem people have with always being online and being a good "global citizen". In this example, like your example, when you feed your learning algorithm data about some war in a far off land you necessarily reduce the weight on your immediate surroundings.

Therefore, I believe it's impossible to "optimize" such things without making significant learning losses. Better to succumb to the brownian motion of life - imo.

I think you are overestimating how much an average person can change itself. I’ve seen therapy helping to surmount some hurdles, but completely changing a person, in a positive way, rarely. Again keyword: average.

> It does not lead to obsessive/compulsive behaviour if you simply optimize your emotional and awareness skills first.

I don't think you understand what optimizing entails.

Optimization is inherently antithetical to balance, in principle.

Optimization is looking for local/global maxima. Balance is releasing the need from seeking local/global maxima entirely.

I mean, you could try to seek balance through optimization. Good luck with that.