If I understand things correctly a cell can be brought to a polarized state which in indicates a predicted state. So let’s say my cell takes in 0.3 and becomes polarized but not firing because fire threshold is 0.5.
Does the cell in biology stay polarized for a longer time period or does it depolarize rather quickly? Have anyone seen information about this pol/Depol process? If I remember correctly received spikes along the dendrite must be close to another in distance or rapid in time for the cell to fire, if they do not cause a fire event do they just “vanish” then?
It depends on things like the place on the cell and the type of response. Generally, I think the depolarization just ends pretty quickly, but some responses last a while. These responses have their own thresholds or requirements, but I don’t know of any common ones which produce responses for more than one or two hundred milliseconds without the cell ever firing.
There are also some rarer types of connections between cells, which have metabotropic receptors. Metabotropic receptors can produce signals lasting seconds. I’m not sure exactly what they do, though.
I have had long walks talking to myself about these metabotropic critters. One amusing line on these receptors is as follows: They smear the current activation pattern out in time. If this function was happening to you right now you would be “enhancing” the current perceptions into a sort of “here and now.” You know, what you are feeling at this moment. Well - every moment.
Functional description - project current activation into the future. If similar fields are firing at the same time they would interact.The past sensation could be made to interact with the present sensation. Allowing this to drive learning rates for a given pattern would enhance episodial memory.
Two patterns that were activated in turn could interact - I could see a memory of the change in patterns could be a form of logic and perform all the rules of internal logic by extracting and learning that from the sensed patterns.
Then again on these nights, the fireflies were out and it was magical. So even if not a bit of this pans out the long night walks were not a complete waste.
As far as my understanding goes, the duration is only as long as it takes for the next pattern to come in and change the activation states of the neurons. After that, according to HTM, there is no functional use of the depolarisation. This happens in some msecs. So the duration in an implementation can be decided to be the time it takes for the next pattern to get computed. But this should stay consistent and the system performance should be extremely reliable and again, consistent.
Regarding the spikes that don’t cause an electrical activity, I think they just dissipate in the dendrite. Functionally they are as good as non existent, I think.
Also, please elaborate on the 0.3 value. I think a cell will get depolarised only if it’s active distal synapses have a permanence value exceeding the threshold. So assuming the firing threshold you mentioned is this threshold in question, the synapses that cause depolarisation(predictive state) should be above this threshold.