They talk a lot about emotions, but the actual paper title is more informative: it’s the decoupling of dendritic learning from somatic spiking during REM sleep.
Only read the summary as the paper is closed access, interesting. Is this a case of dendrodendritic synapse learning… ?
I think it’s synaptic learning by feedback of dendritic spikes, even when the soma doesn’t spike. There was a recent paper on that, don’t have it off the top of my mind. I am guessing the soma doesn’t spike because it needs cholinergic input, and ACh is blocked during REM? Haven’t read the paper either.
Unsure about the ACh, I’m presuming this paragraph also points out one of the huge number of gaps in my knowledge…
“This decoupling reflects a shift of inhibitory balance between parvalbumin neuron–mediated somatic inhibition and vasoactive intestinal peptide–mediated dendritic disinhibition, mostly driven by neurons from the central medial thalamus.”
More reading and research ahead…
I dunno if this is what it’s talking about (also keep in mind there are probably multiple kinds of these cells).
VIP (vasoactive etc.) cells inhibit martinotti cells. Martinotti cells inhibit the apical dendrites of pyramidal cells, and receive input from pyramidal cells. Martinotti cells receive facilitating synaptic input, possibly selecting for bursts (or something else, like persistent activity or a firing rate code, I dunno). So overall, VIP cells might tune-down competition for burst-mode firing in pyramidal cells. They can be activated by motor cortex, or by subcortex during running, and might do spotlight attention*.
*(https://www.sciencedirect.com/science/article/pii/S2211124721000875, https://www.sciencedirect.com/science/article/pii/S0092867414001445, and https://www.jneurosci.org/content/jneuro/36/12/3471.full.pdf. As part of a deeper dive also see “Nonlinear dendritic integration of sensory and motor input during an active sensing task,” there’s a free pdf on google scholar, and L5tt cells are attentional (paper summaries)).