I don’t think the open paradigm is about training your brain to need less SNR. It’s just about providing SNR in a different way. Noise reduction is still applied to speech in the OPN for better SNR, it’s just done in a different way, dynamically as opposed to statically. The dynamic noise reduction employed by the OPN is applied momentarily to speech when it’s present, but when speech is not present, it lets the other “noise” come in. The static noise reduction is the traditional way of many hearing aids to block out the “noise” all the times, even when there’s no presence of speech.
I’m guessing that the Oticon More expands on this principle, but with DNN, it is not limited to being able to apply dynamic noise reduction on speech in the front only, but now it can apply to speech in any direction because it can now dynamically come up with a dynamic noise model instead of relying on a location-based (sides and back) noise model like it does on the OPN.
If it can determine through DNN more accurately which sound is which like it claims, it can probably discreetly decide which sound is more desirable to bring forth and which sound to fade out in a sound scene, therefore achieving a better balance. But the sound scene is evaluated dynamically, so the adjustment to the sound scene is constantly instantaneous, giving you the impression consistent of the “open” paradigm.
All this is just a guess from me based on what Oticon is saying. But the bottom line is that I really doubt that it is trying to train the brain to reduce reliance on SNR. It’s probably just applying SNR a lot more smartly and dynamically so that it happens so quick you don’t even notice it.