Following on from MFS ..
(a) sims are computers and should work well in the validated model regions, especially if the operator spends a bit of time keeping them tweaked well .... but, necessarily, they are a bit of a guess outside.
For most considerations, there is a value to be had from whatever generic training may be available in non-validated regions but that relies on whether, and to what degree, the box presents a quasi-realistic story to the pilot.
We need to be aware that sometimes the sim can present quite false information which may be of counterproductive training value which ought to be avoided.
As a for instance, some years ago I was involved with a 732 sim which had to have a rudder model update per FAA direction as the unit was used by US AOC holder crews.
Due to everyone else's taking bigger steps backward than I was capable of (had nothing to do with the time it was to be programmed, I guess ) ... I ended up playing driver for the before and after testing which the sim OEM folk wanted done ..
Looking at things like crossover speeds, and other low speed bits and pieces such as VMC and general yaw/roll responses, the "before" model was extremely unconvincing and of no training value, the "after" pretty interesting to say the least and, I concluded, of considerable use for extra exposure training for crews ... chalk and cheese.
Gut feel is that stall exposure in the typical sim is of questionable, and perhaps, counterproductive, value ..
(b) ... as opposed to a good dose of UA exposure training which, in my limited view, is real good value for familiarisation with the instrument panel pictures, even when one discounts the body sense limitations. For this reason, I have opted for doing the exercise off motion so that we don't cause the jacks too much angst ...
It really doesn't take long for the typical pilot to get the hang of getting the bird back under control in the UA scenario.
If I recall back to my initial flying training days (Victa 115) ... at the end of the PPL training, I had a bunch of hours to burn off (RAAF ATC flying scholarship for those old enough to remember them .. not too sure whether I should have given those hours back .. aahh ... no way, José) and one exercise we did was circuits and aeros under the hood .. probably 10 hours or so in total.
Confusing for the first bit but, after a couple of hours, limited panel basic aeros were pretty straight forward. (I can still remember, very vividly, the first blind landing .. waiting for the ground to catch up with the altimeter ... good value stuff, I thought).
(c) one needs to keep in mind that the design standards have varied and developed over the years and some aircraft were never certificated to be pulled into the stall regime .. ie one does it in the real world at one's interest.
For instance, the Mojave folk at NTPS tell the tale of a USAF trainee who did this on, as I recall, a B58 ... found himself, shortly thereafter, in an inverted spin .... point of the tale is that, prior to doing anything "unusual" one ought to do some homework in the TCDS and related Standards to get a handle on what might have been applied to, and looked at during, the original certification ....