Teach 'em to read, and see that they enjoy it. Then teach them how to learn on their own. Nothing will ever shut down that mind short of a bullet.
That training must have changed since I went through it. They used to train you to get the mask on quickly, the to get it on in a contaminated environment, then they MADE you take it off and breath that crap for about a minute before they let you out. Everybody ran(ok, stumbled, or blindly blundered) out coughing. Ah, good times,...
Precisely - learn reading, learn basic maths - both will teach you enough about logic and "how to learn" that you're unstoppable from there. Granddad then taught me that anything of importance can be described mathematically (granddad graduated high school, no college, and was the smartest man I'd ever met! Still is, but we lost him a few years ago...) and that was a lesson I took to heart.
Strengths in school for me? Maths, Sciences, and Shop. I took
every damned Shop class that was offered in high school - metal, plastics, wood, engine, auto body, ... Maths all the way out to calculus, three semesters each of Chem and Phyzzies, throw in Astronomy, Aerospace (essentially FAA ground school taught by a CFI,) geology, "Earth Sciences" (mostly intro geology, coupled with geography and cartography - we also learned to survey and draw contour maps, and had to draw a map of that end of the city as well.)
Phyzzies was fun - since it was a general course, we got to play with fluid power, optics, stroboscopy, ... It was a hair short of a general MET course, if anything.
And the common thread through all of those courses was the same thing I'd been thinking for years - "If it can't be described mathematically, it ain't a science." No matter what PoliSci people try to tell you (and Gawd only knows how many times I've gotten into
this argument!) if there is no mathematical model, it's just not a science. Calling a tail a leg doesn't make the name fit.
Flipside: If it can't be described mathematically, it's not very important. Mathematics is the only language with a null semantic content, therefore it is the only truly "universal" language. As soon as semantics get involved, you start to screw things up - that's how you run into "errors in translation."
A mathematical model is predictable. That's why I can envision - at a microscopic level - just what is happening once I put a tool to a piece of metal when there is relative motion involved.
Gas Chamber training? I think I cranked them off when I was the last one out (someone has to be last, it saves trouble if it's me...) and didn't charge blindly out the door. Yeah - I was suffering the effects, but it wouldn't do to let them know how much I was suffering. I mustered up all of the Stoicism I could manage...