I recall reading in On Intelligence that brain activity when someone is imagining something is identical to actually perceiving/performing it. For example, imagining oneself playing piano causes the motor cortex to light up as though it was actually playing piano, and imagining a scene causes V1 to light up.
However, I haven’t seen any mention of a mechanism for this in discussion around HTM. It seems as though you’d need proximal feedback from higher level regions for this to work. Digging through some neuroscience texts, here’s what I’ve found so far:
Layers 5/6 project to Layers 1/6 in lower-level regions
There seems to be evidence of neurons in L6 that project to L2/3, but no mention of whether this is to proximal or distal dendrites.
It also seems necessary for some kind of recurrent proximal feedback somewhere. After all, if that wasn’t the case, shutting off sensory input would mean sensory regions would shut off, meaning no input to higher level regions, which would shut off too, eventually leading to the entire brain shutting off. Sensory deprivation chambers evidently aren’t a death sentence and seem as though they might even increase activity in some sense.
There’s also the possibility that this problem is solved by thalamic feedback too. I know Numenta is looking into the thalamus now, so any ideas from that research is welcome here too.
You may read about “mirror neurons” and Jeff’s words on the connections between the sense and somatosensory cortex.
I suggest that it may be useful to think of this in a more holistic way based on the measured distribution of function in language. Semantic representation seem to be distributed over wide tracts of the brain. Please read this paper and note the wide distribution of representations over many areas of the brain.
Directing your attention to figure 1, language seems to be grounded in the parts of the body that is most likely to be associated with it in your personal experience. Since most of us encounter the world with our body these concepts are rooted in that part of the body. Once the concept is formed recall will reactivate those neurons when that concept is manipulated in your mental space; the recall involves these neurons again.
Neurons have spontaneous firing, which means firing without the experimenters providing stimuli intentionally. I don’t know how much of it is automatic, but at least some of it during anesthesia is probably from oscillations. Oscillations aren’t so slow or organized during normal functioning, though.
Distal inputs could influence spontaneous firing.
It’s hard to tell what a projection does, so feedback to proximal dendrites might not send the right information for imagination. My guess is that there are a bunch of different feedback signals from different layers to distal dendrites involved in imagination, and maybe also some proximal ones.
It would be great if there’s an article about layer-specific activity during imagination. There is a lot of research on similar things like anticipation, behavior, and attention.