> why would we expect a specific sensation to always activate the same set of neurons over time?

For at least two reasons. First, the perception of smell is extremely stable in humans. You can immediately recall a smell you smelled like a child and where you were and what you were doing.

And related to that, smell is a very strong association factor for memories. Meaning you much more easily remember something if it smells like before, rather than if it just sounds or looks like before.

If the neurons of this association drift all the time, question is how come those associations stick through the decades. As a programmer, you know that if you start moving associated data in your database around, it's very easy to break your data structure (dangling foreign keys etc.). A brain isn't digital but connected things must stay connected somehow.

I like how to a layman, every finding is so obvious, but to a scientist it's not obvious because it doesn't match the other things they know about the brain.

I mean

if you don't remember a smell, how would you notice? It's not like your brain will throw up an error message. The memory just won't come to mind.

I think the point is that there are smells that we do remember for a lifetime, which must be explained. Either these stable memories aren't mediated by the cells in piriform that Schoonover and Fink recorded, or there is a mechanism that supports drifting yet stable distributed representations.

Or sometimes you get drift, and then you don't remember a smell, and this happens all the time - which is why if we re-smell something, we don't necessarily hit the same neurons, but we also remember old smells, because those didn't drift. Or is that excluded by the experiment?

My understanding was that these experiments didn't completely rule out the possibility of a stable sub-population. Also, the olfactory bulb projects to other areas besides piriform, and those might support long-term stability. My extremely speculative unfounded conjecture is that piriform is actually a novelty encoder, and plays a role in encoding new memories in a changing environment. But, I think the data don't quite support this at this time.

Stable storage mediums also work by refreshing data over time, eg DRAM. I can't think of one where data must be refreshed by moving it around physically, but surely there is one.

Delay line memory.