[HS]

We can build a neural net (mechanical or simulated), thus replicating electrical impulses. Is it necessary to replicate chemical processes in the brain? For AI, ASI, maybe not. For AGI that seems necessary. Anyone know how?

[Korrelan]

Regarding the simulation of ‘chemicals’ within a neural model…

I have personally so far, not found a need for actual chemical reactions within a neural model but I do simulate the presence of chemicals/ neurotransmitters (NT) throughout the model.

My 3D vector connectome models are defined within a volumetric voxel space.  Levels of NT’s within a voxel are simply defined by variables.  I use a similar voxel model to an octree, so imagine a 3D vector model within this type of volume.

https://en.wikipedia.org/wiki/Octree

The levels of NT surrounding a cluster of neurons within a voxel space effects the firing rates of the neurons/ electrochemical synapse.  A percentage of synapses don’t terminate on dendrites, they eject their NT’s into the surrounding fluid/ space.  It acts like a localised global variable for that specific cluster of neurons, affecting their overall properties, firing rates, PSD, etc. 

NT’s have a myriad of uses,  for example… during sleep the levels of certain NT’s define the areas of high activity during the last awake/ training cycle, etc allowing for greater levels of neurogenesis to occur specifically where required, these NT’s are then cleared ready for the next cycle.  The voxel space can also be used to simulate localised EMF fields, blood flow, etc.

I have experimented with using NT’s to effect moods, emotions, etc.  I can understand how these mechanisms work but… it makes my models extremely unstable, I plan to go back to this at a later date.