BB - Recovery from a full stall can be done (and is done during conversion training) on the simulator, and is reasonable representative of the aircraft.
However, what JT says about GIGO is absolutely right. The point I was trying to make is, if you go outside what the simulator is programmed to do, the response of the simulator is likely to be inaccurate and therefore the instructor might well be training the wrong response. One has to be very careful not to indulge in what I used to call 'anti-training', i.e. pilots must not come away from a training session with the idea that the aircraft will behave in a benign way in certain extreme conditions when, in reality, it might exhibit some vicious characteristics.
Yes, training should be done to demonstate the approach to the stall, the recognition of this condition, the flight characteristics at the stall and how to recover with minimum height loss. But pilots should also be trained in the recovery from a full stall, i.e. immediately to reduce AoA and then gently to apply power. They also need to know that applying power too vigorously might cause engine(s) on one side to spool up too quickly causing sideslip and a departure into an 'untested' region of the flight envelope Even more important, they need to know that in many unintentional stalls the situation is likely to be compounded by a severe nose-up trim condition. If they are also flying an aircraft with underslung engines and they apply too much power too quickly the nose-up pitching moment, combined with the out of trim condition, might leave them with too little elevator authority to recover. This latter problem can be demonstrated accurately on the simulator, and it should be done.
But I still caution that all instructors MUST know the limitations of the simulator and not try to draw conclusions from manoeuvres which are outside what has been accurately programmed.