The claim isn't EM in general, it's specifically microwaves or sonic waves.
Sonic waves would have to be of relatively low frequency in order to not be reflected or otherwise attenuated, so you would need low frequency waves, which, no matter what kind of interference you try to use, will never have effects moreso localized than the wavelength itself. If you were trying to get resonance with the skull at n<8, you'd be well within the audible range, too.
As for microwaves, the same issue is still there, I can't think of any interference pattern for waves of lambda>1mm that would somehow have effects that would have no detectable thermal characteristics yet be extremely localized to only a few specific parts of the brain.
Same for resonance, you'd need a wavelength comparable to the dimensions of the skull, and for that you couldn't have effects as localized as damaging only the optical nerve.
It's not out of the question that very high-powered microwaves could have some kind of electrical impact on the brain, but those impacts would not be localized to only the optical nerve and instead hit the entirety of the brain, and if you were able to get a machine that could create strong electrical currents throughout the brain the idea that this could happen enough to degrade a nerve without being somehow noticeable, for days to weeks is extremely implausible. Plus, a thin line would have not to be crossed where a lot of waste heat could be dumped in the room.
In all experiments done, effects were either wildly different or extremely rapid, and are thought to come from a completely different mechanism of rapid variations in temperatures causing physical vibrations. But again, this would need energies in the order of multiple watts applied to the human brain for quite a while, and would definitely be noticeable.
Not only that, but this would be incredibly easy to detect, you'd only need an antenna. The energies necessary for any of these effects are exceedingly high.
It just doesn't make sense, in many many ways.