Here's a calibration/terminology reference I suggest you read
FWIW I don't like the term, but it's common in ASD material so we're stuck with it:
Theory of mind - Wikipedia
It comes from studying animal abilities and "intelligence".
It would be worth 10 or 15 minutes looking at videos of crows' intelligence to get a feel for it.
Fun though: given that ASD has probably been present "forever", why are we using a term from modern research into animals? If I ever find someone capable of discussing "Critical Theory" I'll throw that at them at some point to see how they react /lol.
BTW: researchers nearly always anthropomorphisize apes so most of that content is rubbish from our perspective. But material on parrots (psittacines) and corvids, even informal fun videos, is usually good.
The simple human-centric example is the way adults control how they say things to young children (think 5 or so, but the same thing is used all the time with everyone).
You need a model of their knowledge and "conceptual substrate" so you can figure out how to "step down" the content and style of what you say.
You may have done this yourself, but not thought about the process, because it's "standard issue" in human wetware (and nearly XY/XX symmetrical which should be interesting in the "War between the Sexes", but isn't.
I used to call Theory of Mind a kind of empathy, because I think it's closely related (maybe uses the same wetware).
But I'm back to "theory of mind", hopefully temporarily, because "empathy" has been subverted and weaponized of late.
To some effect though: these days you can use the
way people confuse empathy and sympathy as a "tell" /lol.
As for mask/non-mask: If you select for people (including NTs, who accept "Aaspie Deadpan" body language) I think you're cutting out a lot of people to no benefit.
On the plus side, IMO it's one of the reasons ASD's get on well with other ASDs. It's easier to communicate because the
amount of counter-factual content, indirection in speech, and non-verbal signaling is so much lower, which cuts way back on the "CPU cycles" you have to dedicate to figure out or filter that stuff.
I use it as a moderately reliable indicator for 'ASD0" people - i.e. people with some of whatever we all have in common, but not necessarily enough to ever get a diagnosis.
It's not just us that does this though - normal precautions should be applied OFC.