T28 & Cynical,
You can't seriously expect that a new student is able to feel the aircraft drifting on takeoff due to prop slipstream? I certainly don't and I bet if you close your eyes, you won't either (not for one second suggesting that you attempt it).
Look at the attitude (wings level (roll), coaming height ie where the horizon crosses the windscreen divider (pitch), reference point (yaw) if you want to break it down for a beginner) and if it isn't where it's supposed to be, change it! Who gives a crap whether it's a Yak, Cessna, Tiger Moth, Piper or an Antonov. If you are in the mindset of remembering which way it's going to swing to work out which rudder pedal to push, you're one step closer to stuffing it up already. You apply the correct control to maintain the desired attitude.
An attitude trainer (probably a better term for the device) is not the be and end all of learning to fly, as you seem to be perceiving the arguments supporting as being. It is merely a tool that prepares the student for the coming flight. If anything it will actually improve the attitude recognition and maintenance as the student will not have the sensation of movement giving them cues as to deviations from an attitude.
Why do we actually use synthetic trainers for IFR training?
1) They are cheaper than the real thing,
2) Flying under the hood is very easy to "cheat" as opposed to actual IMC (shadows, peeking at the bottom left corner of the windscreen) and a synthetic trainer removes all of those little cues including motion.
As an aside, I believe JAR regs don't allow hoods or foggles. They use angled louvres and another technique I've heard of was using a polariod film on the windows with the student wearing opposed orientation lenses. Perfect IMC!
This encourages the student to (hopefully) scan correctly in order to maintain the desired attitude. Both of these reasons are just as valid for VFR attitude recognition, with the added bonus (as opposed to IFR training under a hood) that it is easier to sit beside the student and see where they are looking (outside vs. inside).
I guarantee if I was given twins (two identical people, that is) to train and I flew one of them purely in the aircraft and one with at least every second flight practised beforehand in an attitude trainer, the second would be at their GFPT with less hours in the air. (This of course, may not equate to cheaper, as the attitude trainer still costs money, and the flight cost savings probably wouldn't offset the reduced flight hours, then there's the fact that a commercial pilot still must have 150/200 hours anyway. Different story with the motion simulators as hours can be credited eg. up to half of the NVFR can be done in some types.)