Same here. But I wouldn't use the word 'holographic', since it's now too associated with junk science and New Age crap. Even if it does happen to nicely capture

Same here. But I wouldn't use the word 'holographic', since it's now too associated with junk science and New Age crap. Even if it does happen to nicely capture the meaning. Put simply: if you have two related concepts, say "car" and "truck", then it makes a lot more sense to me that they'd be represented by similar weights among a collection of neurons, than if each concept got its own single neuron (or handful of distinct neurons scattered throughout various functional parts of the brain.) If you did that, you'd need to explicitly encode the various similarities and differences. It would take time for priming to travel from one thing to all the related things. There would be no or limited redundancy. Forking off a concept (eg with language, learning a synonym) would have to be some explicit process. With an aggregate representation, you get all of that automatically. When you're thinking "car", you're 70% also thinking "truck". You can evolve your understanding freely without breaking anything, you can split and merge representations, etc. It kind of seems obvious to me, which is no proof that it's correct. I'm reading Kahneman's "Thinking, Fast and Slow" right now, and it's a brilliant book yet I feel like this representation would convert many of his puzzled observations into unavoidable consequences. Separately, I think there's less of a distinction between memory, recall, and actual experience than this paper makes out. They feel the same because in the brain, they are (mostly) the same. And there's more of a distinction between memories of different things. I expect to find very different brain regions involved in memories of pain vs movement vs vision vs procedural knowledge etc., because much of the memory will be in the neurons directly linked to the relevant sensors or actuators. (As above, the memory is the experience, or at least it overlaps substantially.) Sure, there will be overlap across all those, but that's not because it's the One True Memory Region, it's just that there's a generic component to any memory (or rather, anything that we would refer to as "a memory".)