Can we use individual spikes as a reliable carrier of information?

[Pythagorean]

I'm currently reading Spiking Neuron Models by Gerstner and Kistler:

Since all spikes of a given neuron look alike, the form of the action potential does not carry any information

But I've also come across this in a review of Spikes:

I'd have liked more at this point about how the neural code gets used in network computations, or, conversely, how to use the tricks we've learned for breaking the code to get a handle on what the networks are computing and how, but this the authors largely reserve for future research (they have great hopes for the hippocampus and the motor cortex), along with a better understanding of the statistics of natural stimuli. They close with a "homage to the single spike," as a trustworthy and reliable carrier of a substantial amount of information, sometimes even responsible by itself for conscious sensations. "The individual spike, so often averaged in with its neighbors, deserves more respect."

Question(s)

Is this really true? Are all action potentials of a given neuron the same?

Is that justification for action potential shape not "carrying any information"?

If not, where is the assumption valid and where is it not?

 

[bobze]

I'm currently reading Spiking Neuron Models by Gerstner and Kistler:But I've also come across this in a review of Spikes: Question(s)

Is this really true? Are all action potentials of a given neuron the same?

Yes and no--That's the short answer. It depends on the type of action potential (AP) were talking about and the nerve its happening in. Most long unmyelinated neurons use active conduction. What does this mean? To be honest its a whole semester's worth of reading , but I'll try and sum it up.

A resting membrane potential (RMP) sits around some negative electrical potential for a membrane (E_M)-- This is normally around -65 to -90 mV in most human cells. When a signal input raises that E_M to the threshold potential (TP), often around -60 mV, then an AP fires.

It fires because on the membrane are voltage-gated sodium ion channels (VGSC henceforth). The change in voltage (ie; ΔE_M) opens the channels allowing sodium to rush into the cell. This depolarization reverses the cell's membrane potential (making it positive) and drives the membrane potential, E_M, toward the sodium potential E_Na.

To repolarize the cell you have delayed voltage-gated potassium channels (VGPC), which open over a certain positive E_M. This let's potassium flow out of the cell driving the membrane potential, E_M, back toward the potassium potential, E_K

This wouldn't be enough in itself to repolarize the membrane. VGSC, spontaneously close after 1-2 milliseconds, going from an open to inactive state. Once in an inactive state they are unavailable to be reopened at all (stopping the potential for "backwards" transmission, making AP unidirectional). The inactivation of VGSC (stopping sodium influx), coupled with the delayed opening of VGPC (instigating potassium efflux), coupled with the resting potassium channels (always open, allowing potassium efflux) are what reset the membrane to it's RMP.

The open to inactivation state of VGSCs is a fast and spontaneous (in most channels) process, while the inactive to closed state (yes, three states they exist in) is slower. Those VGSCs are only available to open, from the closed state.

I'm getting there I promise :rofl:

The point is, because of the concentration differences maintained by cell's sodium-potassium ATPase pumps (actively pump 3 sodiums out and 2 potassiums in) chemical work is being done. At equilibrium, the concentration gradient for any given ion is balanced by its electrical gradient (electrical work is done), which is only minutely changed during an action potential. For example the intracellular concentration of potassium might go from 140 mEq to 139.999...9999, etc.

What does this all ensure? That when an action potential fires "down" a membrane, the amplitude of the potential is essentially the same every time. Something engineers might give some kind of nifty name too like, a loss-less signal conduction.Of course I said Yes and no, nature is never that straightforward with us

There are other types of signal conduction, like passive ones. Which like the name suggests, function passively across a cell membrane. These don't use voltage gated channels and the APs spread across a membrane depends on its space constant, λ. Which is something measure and is defined as the distance when signal's amplitude has fallen to 37% of its initial amplitude. Right away that should clue you into something, that in the case of passive conduction, the amplitude of the AP doesn't stay the same.

You can think about it like a long wire used to communicate with that isn't sheathed and protected. You'd loose signal over the distance. Much like the problem the first trans-Atlantic telegraph cables had (Solved by a witty grad student, better known as Lord Kelvin )

Passive conduction like this is seen primarily in two places; short neurons and in the cell bodies of longer nerves. That is to say, at a post synaptic junction (the spot where the transmitting neuron interacts with the receiving neuron) the transmitting neuron induces a passive conduction of active potential across the body of a receiving neuron, which shrinks as it cross the cell body (a place mostly devoid of those voltage-gated channels we talked about). So long as the amplitude of the passive signal doesn't fall off too much (in other words λ is large) then the electrical potential that reaches the axon Hillock, will be enough to generate an active potential ("all the same") which will travel down the axon to the next synaptic junction.

Edit: Disclaimer, it gets even more complicated than that when you throw myelin into the mix. That was a general overview of action potentials. It differs for different cell types (say a motor neuron, Purkinje cell and ventricular myocyte). The channels, types of ions and electrical potentials all change as well. When I said things like "voltage-gated sodium channel", understand that is "generalized" description of a channel.

Is that justification for action potential shape not "carrying any information"?

If not, where is the assumption valid and where is it not?

Action potentials themselves don't carry information, they're simply the change in membrane potential of a cell. To add information into the mix you'd need more than one nerve and differential firing (the nerves don't all fire at the same time). Then a device devoted to the collection and "decoding" of these differential firings, maybe you'd even give it a fancy name like "brain" or "central nervous system" :tongue:

Hope that helped

 

[Pythagorean]

thank you for a lengthy reply! I've taken the graduate neurobiology course at my university so I know about the basic, classical picture of action potential mechanics. I'm currently doing research with the Morris Lecar model (which assumes Ca recovery to be instantaneous) but I'm currently studying it more from a mathematics/physics perspective than a biologist's perspective.

Passive signals weren't covered; in fact, nothing besides the action potential along the axon were mentioned.

Action potentials themselves don't carry information, they're simply the change in membrane potential of a cell. To add information into the mix you'd need more than one nerve and differential firing (the nerves don't all fire at the same time). Then a device devoted to the collection and "decoding" of these differential firings, maybe you'd even give it a fancy name like "brain" or "central nervous system"

Well, my first reservation is with the word "carries information". I mean, that's meaningful to a homunculus like you and me, but do our neurons really care?

What I'm really asking is whether the shape of the action potential physically effects how the soma/dendrite processes the incoming signal. I'm assuming the action potential pertains only to the axon so that passing through the synapse, the shape of the action potential is lost (but even still, I have some sticky feeling that if you passed a wider envelope action potential through an axon, so that it spent more time at a high potential, it would affect all the chemically stuff happening in the synaptic junction).

But what about, for for instance, gap junctions, who couple diffusely rather than synaptically? A diffusive coupling between neurons would have to pay respect to the shape of the action potential, wouldn't it?

 

[Kglocc]

Passive signals weren't covered; in fact, nothing besides the action potential along the axon were mentioned.

Well, my first reservation is with the word "carries information". I mean, that's meaningful to a homunculus like you and me, but do our neurons really care?

What I'm really asking is whether the shape of the action potential physically effects how the soma/dendrite processes the incoming signal. I'm assuming the action potential pertains only to the axon so that passing through the synapse, the shape of the action potential is lost (but even still, I have some sticky feeling that if you passed a wider envelope action potential through an axon, so that it spent more time at a high potential, it would affect all the chemically stuff happening in the synaptic junction).

But what about, for for instance, gap junctions, who couple diffusely rather than synaptically? A diffusive coupling between neurons would have to pay respect to the shape of the action potential, wouldn't it?

Do our neurons care? I'd like to hear that discussion in full, but I'll just say yes, they do care and God created apoptosis for the ones who dont...

Just to warn you, I have limited knowledge of the subject, (not familiar with Morris Lecar Model)
but from how I understand it you are asking if a wider/longer action potential would affect the electrical synapses (i.e. passive diffusion of electrical current), by passing more/less current through the gap junctions. I do not think that a longer action potential would have any effect on the chemical synapses, (i.e. unloading of neurotransmitters from synaptic vesicle to post-synaptic membrane) as they would already be an all or none event.

For my own clarification, gap junctions are either pre- or post-synaptic, the inbetween part being where the axon and action potential is.

To quote bobze, The action potential "...differs for different cell types (say a motor neuron, Purkinje cell and ventricular myocyte). The channels, types of ions and electrical potentials all change as well".

The most vague yet concise answer that I can come up with is "yes".
"...differences in action potential duration have important implications for the gradient of repolarisation..."
http://cardiovascres.oxfordjournals.org.oca.ucsc.edu/content/27/10/1795.abstract

The waveform of the action potential has probably become the most efficient that it can possibly be for the neural circuit it operates in. So changing the length will most likely have overall negative effects due to repolarization and "whatnot"... Don't take my word for it.

 

[gerben]

Are all action potentials of a given neuron the same?
Is that justification for action potential shape not "carrying any information"?

The action potential (spike) is the change in the electric potential difference between inside and outside of the cell membrane. This change, we know, is caused by the flow of ions into and out of the cell. Thus, different shapes of individual spikes reflect an underlying difference in the compound inflow and outflow of ions. However the difference in total ion flow between two slightly differently shaped spikes pales in comparison to the difference between ion flow caused by one spike and that caused by two spikes. This is why neuroscientists are generally interested in spike rate and not in the exact shape of the individual spikes (which are obviously never exactly the same even though very similar).

No mattter whether it's chemical synapses or gap junctions the rate of incoming spikes has a much larger effect than the differences in individual spike shape.