Kanter et al claim that the old Hebb that learning takes place in synapses, is mistaken. Instead, the learning would take place in dendrites and much nearer to the neuron and only few parameters would determine the outcome unlikes in Hebbian approach in which thousands of parameters - synaptic strengths determine the outcome. Furthermore, weak synaptic connections - most of synaptic connections are weak - would be more significant as believed.
What the new view about learning could mean from the viewpoint of quantum brain paradigm according to TGD? In this vision magnetic tube pairs having define connections of a dynamical network having neurons at nodes. The connectivity/topology of this network is changing all the time. At deeper level supra currents and dark photons would be responsible for signalling and the function of nerve pulses would not be communication but to change the topology of the network via the activation of synaptic contacts. Neurotransmitters would be like relays in old fashioned telephone network.
If Kanter et al is right, dendrites would learn instead of synapses. Should one talk about dendritic strengths instead of synaptic strengths? Also weak synapses - most synapses are weak - would be important. What happens to "neurons that fire together wire together" paradigm?
Consider first as background TGD vision about neuroscience. The following article summarize the recent developments: DMT, pineal gland, and the new view about sensory perception, Is it possible to reverse Alzheimer's disease? , Emotions as sensory perceptions about the state of magnetic body?, Artificial Intelligence, Natural Intelligence, and TGD.
- In TGD picture axons and dendrites would be accompanied by pairs of flux tubes carrying opposite magnetic fluxes. This is required by their super-conductivity based on spin zero Cooper pairs - this is quite general model of high Tc superconductivity in which the flux tube pairs are made possible by anti-ferromagnetism.
- Reconnection of flux tubes is the basic topological mechanism changing the topology of the network. It corresponds in string theory the basic vertex for closed strings.
- One can represent axon and dendrite by two parallel lines with opposite directions representing flux tubes with opposite fluxes.
- Consider first axon and dendrite (or axons and axon, or dendrite and dendrite, etc...). What synaptic connection could mean in this picture? I wis I could draw. One has a pair of lines A+A-. One has B+B- has U-shape. B+ simply turns back as B-.
Then reconnection takes place. Nothing happens for A+. A- splits to two piece A-(1) A-(2) and the end cap of B+B- U-shape is cut off.
B- reconnects with A-(2) and B+ reconnects with A-(1). One obtains V shaped structure with edges of V represented by pairs of lines with opposite directions: nowhere opposite arrows meeting each other. Synaptic strength tells the probablity for the formation of this structure, which represent change in the topology of the network.
The reconnection for flux tube pairs makes the earlier topological picture more complex. The communication channels defined by flux tube pairs can branch or fuse so that the network structure is much richer. Supra-currents or dark photon signals from two sources can superpose. Also more complex entanglement patterns become possible.
- What about the new notion of dendritic strength? It should tell the probability that there indeed exists a flux tube pair connection between neuron and the rest of the network. This connection can be however split by reconnection. Parallel lines with opposite fluxes pinch together and transform to two U-shaped structures: two U:s face-to-face.
Dendrite strengths tells how stable the parallel flux tube pair is against this reconnection. In TGD model of superconductivity it tells how stable supracurrent "wire" is and transition from small scale super-conductivity to genuine super-conductivity occurs when long flux tube pairs become stable.
- The claimed findings would say that the dendritic connections are most important for learning and certainly they are so: without dendritic connection at flux tube level, no signals enters neuron. Neuron becomes a hermit isolated from the rest of the brain.
But also synaptic strengths are important although not important from the point of view of single neuron but from the point of view of the topology of the entire network: the qualitative features of this topology distinguish between spatial thinking involving 2- or even 3-D networks and verbal cognition involving linear networks: this explains why right brain signs and left brain talks. Dendritic strength as a measure for the stability of the connection of neuron to the network and synaptic strength for the ability to change topology of the network temporarily.
- Hebb's statement could be rephrased as follows. Distribution of synaptic strengths would determine which neurons can wire together and dendritic strength would determine the probability with which neuron can fire together with others.
For a summary of earlier postings see Latest progress in TGD.