3 ms·
Does excitatory/inhibitory map cleanly on to the higher level abstractions? I'm no neuroscientist, so maybe it does, but that seems like a bad assumption to me
by schmidtleonard 2y ago
Does excitatory/inhibitory map cleanly on to the higher level abstractions? I'm no neuroscientist, so maybe it does, but that seems like a bad assumption to me because in digital logic "active when low" is extremely common. A spurious-suppression system that caused inhibition when low would be perfectly reasonable and compatible with the observation that excitation caused an increase in spurious behavior.
- neom 2y agoThey don't really map well. The brain obviously operates on more continuous principles rather than binary states. The relationship between excitation/inhibition and emergent effects may be more direct and graded in biological neural networks vs to the "active when low". Buuuutt stiilll, 5-HT2A receptor and its downstream effects on glutamate transmission play a central role in mediating the subjective effects of psychedelics. Very very many studies have consistently linked 5-HT2A activation to the perceptual, cognitive, and emotional characteristic of the psychedelic state. Disruption of the default mode network and other changes are somewhat important maybe, the 5-HT2A receptor appears to be the focus of action for producing these effects. The point being, neurochemical interactions at the molecular level are the most important aspects of the LSD interaction, so.., now sure how that works in the context of synthetic AI, I don't know much about ML/NN.
- schmidtleonard 2y agoI still don't see the reason why glutamate or 5-HT2A receptor excitation (in the chemical sense) couldn't be involved with inhibition reduction (in the psychological sense).