Representational drift, if validated, tells us that a memory is a set of relationships, not the specific neurons that embody those relationships. This sentence might be rendered in electrons or ink, but it has the same meaning.
Reading an article about this reminded me about an old concern with drugs that aim to treat Alzheimer's by reducing amyloid accumulation in neurons. Drugs like the recently approved (and seemingly minimally effective) monoclonal medication aducanumab. The root problem is that we don't know why neurons accumulate amyloid. There's been a growing suspicion over the past few years that amyloidization might in some way be helpful.
I wrote about one way this might play out in a twitter post which I've revised here:
Representational drift reminds me of a theoretical problem with aducanumab and amyloid therapy for Alzheimer’s dementia. It begins with recognizing that we don’t know why neurons accumulate amyloid.
Many suspect amyloid has a physiological reason to appear in neurons. Suppose, for example, amyloid is the way old crappy neurons are "retired" from forming memory relationships. Amyloidization would then be the brain equivalent of marking a SSD region as unusable.
A system like this would have 2 kinds of bugs. It might be too aggressive or not aggressive enough.
If the retirement mechanism is too aggressive then neurons will be amyloidized prematurely. They could have still formed useful memories, but now they're dead. The brain can only produce so many neurons so it runs out prematurely. Early dementia develops. In this case a drug that cleared amyloid could help -- as long as it wasn't too aggressive. The balance may be fine and hard to get right.
If the retirement mechanism is too permissive then a lot of flaky neurons accumulate without much amyloid. Dementia follows from this too -- but it might look clinically quite different. In this case a drug that cleared amyloid would make the dementia worse! Even more flaky neurons would accumulate.
Even if the balance is just write we do run out of viable neurons. Even a very healthy centenarian has only a fraction of the cognition they once had. Again, in this case, an amyloid clearing drug would make the brain worse.
If this was the way the brain worked then an amyloid reduction drug would make some dementia worse and some better. The net effect would be quite small -- even if the medication worked perfectly and was dosed correctly.
All speculative. Come back in 5 years and see how it turned out.