NeuroUpdate

We like few-celled organisms, not bugs!
Post Reply
DaCellMan
Posts: 7
Joined: Sun Jan 07, 2018 3:47 pm

NeuroUpdate

Post by DaCellMan » Sat Jun 22, 2019 5:05 pm

One-way Signals
A setting to make cells that are only able to pass signal to a certain cell/s. This will allow for programming of the organism to be easier.
Mementocytes
Mementocyte will serve as a memory cell and would give organisms the ability to learn from experiences they've had. The Mementocyte detects all the individual neuro cells and their signal and compares the current signal to the memory. Using substance signal dependance, they could trigger the storage of the memory (amount of signal/s) and the mementocyte would release signal based on the similarities between the stored memory. The signal from the mementocyte would be used to avoid that situation or seek to recreate it. Many things would be much easier for us to create such as: Behaviour change, Behaviour learning, "Personalities" etc.
For example if a predator notices that hunting together ,by chance, gives them more mass then the Mementocyte will release signal and will affect the behaviour by trying to get as close to the situation as possible.
User avatar
countingtls
Posts: 15
Joined: Mon Jun 10, 2019 10:36 am

Re: NeuroUpdate

Post by countingtls » Sat Jun 22, 2019 6:40 pm

DaCellMan wrote:
Sat Jun 22, 2019 5:05 pm
One-way Signals
A setting to make cells that are only able to pass signal to a certain cell/s. This will allow for programming of the organism to be easier.
Mementocytes
Mementocyte will serve as a memory cell and would give organisms the ability to learn from experiences they've had. The Mementocyte detects all the individual neuro cells and their signal and compares the current signal to the memory. Using substance signal dependance, they could trigger the storage of the memory (amount of signal/s) and the mementocyte would release signal based on the similarities between the stored memory. The signal from the mementocyte would be used to avoid that situation or seek to recreate it. Many things would be much easier for us to create such as: Behaviour change, Behaviour learning, "Personalities" etc.
For example if a predator notices that hunting together ,by chance, gives them more mass then the Mementocyte will release signal and will affect the behaviour by trying to get as close to the situation as possible.
I like the idea of memory neuron cell it surely can make creatures much interesting. The formula of (output = a x signal + b) is fundamentally similar to those used in Artificial Neural Network (Output = Weight x Input + Bias), just that normally the learning algorithm can "learn" these weights/bias. Right now by using one-point mutation evolution to tune these parameters is essentially just like real world machine learning process through genetic algorithm for optimization. It would be fun if we can design a creature that learn in its own lifetime. The memory is essentially stored in a and b values of a cell (if we treated those cells like ANN neurons). We just need mechanism to update them, and even pass their current "learned" parameter values to other neuron cell of the same mode.

I think the issue of selective transmit signals to particular cell, has more to do with "not" passing signals to certain cell. It is like we need some kind of "insulating" cell equivalent that will "block/absorb" signals, like a keratinocyte equivalent in neuro-transmission. It will help a lot for allowing more compact nerve system without interfering with each other, and they can use the same signal without the need to introduce more signal types. Real nerve system do have many of those to insulate, recycle, even help signal to propagate, like glial cells Astrocyte, Oligodendrocyte, etc.
User avatar
RedShift
Posts: 26
Joined: Mon Dec 17, 2018 10:04 am

Re: NeuroUpdate

Post by RedShift » Sun Jun 23, 2019 2:01 pm

countingtls wrote:
Sat Jun 22, 2019 6:40 pm
DaCellMan wrote:
Sat Jun 22, 2019 5:05 pm
One-way Signals
A setting to make cells that are only able to pass signal to a certain cell/s. This will allow for programming of the organism to be easier.
Mementocytes
Mementocyte will serve as a memory cell and would give organisms the ability to learn from experiences they've had. The Mementocyte detects all the individual neuro cells and their signal and compares the current signal to the memory. Using substance signal dependance, they could trigger the storage of the memory (amount of signal/s) and the mementocyte would release signal based on the similarities between the stored memory. The signal from the mementocyte would be used to avoid that situation or seek to recreate it. Many things would be much easier for us to create such as: Behaviour change, Behaviour learning, "Personalities" etc.
For example if a predator notices that hunting together ,by chance, gives them more mass then the Mementocyte will release signal and will affect the behaviour by trying to get as close to the situation as possible.
I like the idea of memory neuron cell it surely can make creatures much interesting. The formula of (output = a x signal + b) is fundamentally similar to those used in Artificial Neural Network (Output = Weight x Input + Bias), just that normally the learning algorithm can "learn" these weights/bias. Right now by using one-point mutation evolution to tune these parameters is essentially just like real world machine learning process through genetic algorithm for optimization. It would be fun if we can design a creature that learn in its own lifetime. The memory is essentially stored in a and b values of a cell (if we treated those cells like ANN neurons). We just need mechanism to update them, and even pass their current "learned" parameter values to other neuron cell of the same mode.

I think the issue of selective transmit signals to particular cell, has more to do with "not" passing signals to certain cell. It is like we need some kind of "insulating" cell equivalent that will "block/absorb" signals, like a keratinocyte equivalent in neuro-transmission. It will help a lot for allowing more compact nerve system without interfering with each other, and they can use the same signal without the need to introduce more signal types. Real nerve system do have many of those to insulate, recycle, even help signal to propagate, like glial cells Astrocyte, Oligodendrocyte, etc.
For insulating a signal substance, a lipocyte can do that, it blocks every type of signal substances.
Randomly Mutated (o)
User avatar
countingtls
Posts: 15
Joined: Mon Jun 10, 2019 10:36 am

Re: NeuroUpdate

Post by countingtls » Sun Jun 23, 2019 4:56 pm

RedShift wrote:
Sun Jun 23, 2019 2:01 pm
For insulating a signal substance, a lipocyte can do that, it blocks every type of signal substances.
Yes, but not selectively. The issue is selectively transmit signals to targeted cells or not transmit signals to certain ones, as well as inherent/memorize updated/learned parameters in a neurocyte or to other neurons, which none of the current cells can perform these, and in nature we do see similar functions in nerve systems.

Right now, we technically can design a kind of neural network with lipocyte as insulation, but they will take up so much space it's impractical. And without selective insulation, we are forced to only have at maximum 4 neurons in one hidden layer and one output with very limited functionality.

Another problem is cross wire problem, where two signal passageways can never cross each other's path, and since right now it is in 2D space, this is another serious limitation for getting more integrated sensors to gather signals together in one place and process them in another place.
Post Reply