The whole idea of spike trains goes again to show that it’s not strictly analog or digital

I begin to see what you mean

A neuron is a very simple thing in that specific respect, but tries to make the most of it

Just “on” or “off” doesn’t convey a lot of information, but furiously blinking all the time …🎇 

Yes, like you said, that is like a telegraph, or more akin to code

Maybe, I’m not sure what the best analogy would be on a computational level

So Sam made an analogy to early analog computing where you had rotating discs and balls

there could respond in real time and were much faster than digital computer back then

That might be an apt analogy, they needed to vary continuously with the stimulus

“spiking” more frequently with the intensity or orientation etc.         

OK, so reasonable analogy then?

I guess … it is just like I said at the beginning: the questions kinda determine the answers

So I’m abstracting too much? 😅

By imposing the framework of “either analog or digital”, yes  😅       

I’ll try to convey some more nuance in the article then …

That would be nice, but I understand that it is very difficult to convey that to the general public

Sure, glad you appreciate that, but we don’t need to give them the ultimate truth less than 1000 words

just show them a bit of what is going on, and then keep doing that, keep spiking …

Ha! ha! 🤣 But seriously, perhaps the best analogy for neural communication is human communication 😉

I might use that analogy if you don’t mind! 🤔

Always happy to help!  😊                                                                        

And I’ll be back for more, don’t worry! 😄

Yes, it makes little sense to just zoom in on one single neuron         

Information and information processing is distributed over the whole brain?

Well, not all of it all the time. Specific pathways and modules exist for specific types of information

Such as the “visual pathway” and stuff like that?

That’s a good example. The retina already detects and filters some signals right at the beginning

and then the associative cortex deals with the end product: there’s serial and parallel aspects at all levels.

Ok, so “it’s complicated”, again 😟

I’m afraid so! If I am not mistaken Sam once told me about his artificial neural networks and I was quite shocked at their simplicity

Simplicity? 🤔

Well, it is not just a simplistic model of how a neuron works, but also the networks were organized in a totally biologically implausible way

So they forgot essential features?

More like essential constraints: not every neuron is connected to every other neuron! But that is what a lot of artificial networks do

Oh, right, that is a major difference of course!

Now I think he did tell me that there were new developments, there always are, which organized things in a more reasonable way

I think Sam is working with “spiking neural networks” nowadays?

That sounds promising. People tend to think neurons only fire now and then, but in fact they fire all the time

What do you mean?🤔

It’s not like they send on one impulse and then just sit there, as long as you are hearing the sound, seeing the color, etc. they fire!

Oh, ok, so there’s lots of spikes, not just one big yes or no signal?

Indeed, we normally speak of “spike trains”, more like a telegraphed message with lots of yesses and noes 😊

OK, this is very helpful 😉

So an electrochemical signal travels along the axon, that is, along the membrane of the cell itself …

Right, and that then triggers chemical messengers that convey information to other cells.                    

And this can happen in various ways too? 😊

Yes, that is another form of categorization: what the target is.         

Ok, so depending whom the cell wants to speak to, it releases different chemical compounds?

Something like that. Keep in mind that all cells communicate, not just nerve cells

Please elaborate!

So a cell can target itself, can target its close neighbors, or can target other cells far away.

we call that “autocrine”, “paracrine”, or “endocrine” signalling.            

OK, but that’s all done with “neurotransmitters”?

Not really, no. Endocrine signalling is done by hormones                  

Cells release a compound into the bloodstream which then affects cells far away, in other parts of the body.

Oh, wow,😳 that definitely something that isn’t normally included in “neuromorphic computing”

Nope, I guess not, but it definitely is “information processing” between cells.

We might need to do an article on that sometime … but let’s focus on neurons for now.

OK, indeed the standard case in that context is go/no go based on action potentials

Yes! And for the purposes of computation, we can safely regard those as 1s and 0s? 😊

Up to a point, because cells can excite or inhibit other cells, making it more or less likely they will fire

And it more often than not is many to many communication           

What do you mean?🤔

One cell releases neurotransmitters to multiple other cells while at the same time multiple cells affect one cell

OK, so we definitely need to stress the holistic and network aspects😉

The answer varies depending on the question? In what sense?

Well, you were asking about “neuromorphic computing”, so you’re interested in the computational aspects                          

Yes, that’s right. 😊

But Sandra, as a doctor, might be interested in completely different aspects of a neuron

Like: what is wrong with it, why isn’t it firing, is there damage to the cell membrane, etc. 

Oh, ok, sorry I thought you meant something completely different!

So from a computational point of view, you’d just be looking for information                                 

And since we’re making assumptions we’re missing things

Exactly. If you look for 1s and 0s, you’ll look at the bits that could be 1s or 0s and forget about the rest            

Ha, bits, quite literally 😅 … And that’s why we look at the electrical part?

Yes, a stimulus, a current, something that transmits information    

And what do we miss then?

Hard to know where to begin really … communication between cells isn’t just firing off electrical signals 😊                  

OK, so you mean the chemical aspects of communication between nerve cells?

Yes, how signals spread along the axon and then how that triggers the release of neurotransmitters across the synapse (Neurotransmitter)

I’ll need to include a short dictionary with the article I’m afraid … 😅

That would be seriously useful I think, I just can’t avoid some of the jargon 😅

Hello Cho, I wonder if I could bother you with some questions? 😊

Hi John, sure, what do you need? 😉                                                      

I’m working on an article on neuromorphic computing

Well, I’m good for the “neuro” part, but computing …not so much😅

No problem, I already talked to Sam, and the “neuro” part is precisely why I come to you now 😊

Ah, OK then I do think I can be of assistance!                                       

Excellent! Sam really confused me with his talk of analog and digital or continuous and discrete 😣

Yes, that is something that comes up a lot when talking about neurons             

So can you help me with that? What kind of information or signals do neurons exchange?

Let’s start with “it’s complicated”                                                             

I was afraid you’d say something like that …😅

First of all, there are many different kinds of neurons                        

Oh! Yes, I recall we talked about that! “Little big brain“

Very good! 😊

Secondly, people tend to focus on the electrical aspect, not on the chemical one

What do you mean?

When a neuron fires, it isn’t exactly the same as flipping a switch

So it isn’t digital or discrete?

Not straightforwardly no, but it also depends on how you look at it?

Is this going to be about quantum and observers and stuff like that?

Ha! No, not at all. 😅 

I just meant that the answer might vary depending on the questions you ask