If there's a one-off feedback elimination pass the digital signal processor in the aid probably has what it takes to do adaptive feedback elimination.

I'm curious about what algorithms are being used. There's a lot of hands-on experience in music and audio engineering of eliminating noise and increasing perceived volume. I have a depressing feeling that audiology is off in a separate world, and the two disciplines aren't talking to each other as much as maybe they could be.

That's a good point about headphones, because the folds around the ears physically steer sound and add directional information that won't be picked up with a point microphone or reproduced with a pair of point headphone speakers.

You can fake the effect using either a plastic head with fake ears (it looks odd, but it works surprisingly well) or mathematically using something called a Head Related Transfer Function. (Great name...)

I suppose potentially you could improve noise filtering by increasing the effect of the HRTF. You'd get more directional hearing - which wouldn't always be useful, but could be an improvement in some situations.

(Just speculating...)