Hello everyone.

Finally our rebuttal comment has been accepted for publication in Physical Review X, almost 2 years after our first submission. We could not reproduce the observation of penetration of action potentials by the Heimburg group. Instead we consistently observed annihilation of colliding action potentials.

Further we show that their measurements of action potential velocity was flawed and indicate that their measurement electrodes were so close to the stimulation electrodes, that 1) they could not contain two action potentials 2) passive propagation may explain part of their observations. 

Also, they did not demonstrate the All-or-none property of action potentials, which again suggest that they observe either activation of multiple nerves, or they generate passive electric signals. Attached is preprints and supplementary information.

Cheers!

Rune

U can find details abt this in earlier editions of Guyton 

A quantitative description of membrane current and its application to conduction and excitation in nerve  (A. L. Hodgkin and A. F. Huxley) 

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1392413/

Dear Rune:

This is a very interesting and tough question. Years back, we did recordings in cat brain stem using collision test, but the work was never published. Check papers by Allen, Tsukahara, Udo and others from 1960-70s; you may find some relevant information there. I remember there was a paper in Exp. Brain. Research in anesthetized cat preparation, which dealt with this topic in detail.  The test has been used a lot for dissecting out the supra-spinal pathways and projections before better tracing methods have been developed. My guess is you will not find a clear cut answer as these studies measured conduction waveforms from large set of axons, that leaves a lot of room for speculations. From my own experience, close to complete annihilation of  the population spike waveform can be achieved if stimulation strength and positioning of electrodes are optimized properly. Hope this is helpful.

Regards,

Saak

Pierrot-Desseiligny developed a method for assessing recurrent inhibition, which was based on the collision technique. It have always looked a bit suspicious to me. May be you can find some references to the older work in their paper

PIERROT-DESEILLIGNY, E., BuSSEL, B., HELD, J. P. & KATZ, R. (1976). Excitability

of human motoneurones after discharge in a conditioning reflex. Electroenceph.

dlin. Neurophypiol. 40, 279-287.

Regards,

Maria

I don't know about invertebrates but, with respect to mammalian species, I don't know how one could account for the results of the many collision experiments in antidromic stimulation studies if annihilation does not occur. 

When the experiment is properly set up, it's simply an observable fact that the stimulation evoked spike doesn't show up in the recording electrode if a naturally-occurring spike occurs at the right time. It DOES show up if there was no naturally-occurring spike. What could account for that, other than annihilation? 

Thank you guys. Yes, Mark I agree with you. I am well aware of the collision test which have been applied in tens to hundreds of previously published studies. And I agree that the most parsimonious interpretation is that action potentials annihilate. I cannot come up with any other interpretation of those results. 

The collision experiment is quite easy to do in earth worms, stimulating in both end and recording in 2/3 way the body, which I have done my self to verify the findings of the publication in PRX. I performed the experiment in 5 earth worms and they all consistently showed annihilation. I have attached recordings from a sample earth worm. The red trace is what we should see if penetration happened. The top two traces are the AP generated either from left or right (ortho- or antidromic). The traces below are the increasing stim strength of both, making sure the most distant site has lowest threshold. We never observed both AP in same trace. 

Nevertheless, this experiment is difficult to perform in most other animals, because you have to be able to exclude the activation of other axons, and most nerves consist of more than one axon. 

This is why I am looking for the first publications which directly addressed the collision of action potentials, because they must have been conducted in the early days. 

cheers!

Rune - 

Antidromic collision should produce annihilation.  As Saak mentioned, the collision test is commonly used (I use it myself) for determination of supraspinal projections.  Of course, exclusion of activation of other axons is not possible, but the typical heuristic for determining if the neuron/axon you want to be testing was annihilated is based on timing.  Stimulation (not collision) should elicit an action potential with a consistent time delay.  Using collision testing with a lag below or above the neuron/axon's time delay should make the action potential disappear or appear, respectively.  Inconsistent timing of action potentials either during stimulation or the collision test is indicative of activation of other axons with collateral connections to the one you're recording from.

A couple of things about that paper seem odd - the conduction velocities look very low for myelinated axons.  I'd also expect to see (but don't) multiple collision traces overlaid to demonstrate that the antidromic latency is stable across multiple stimulations. If it isn't, the activation is likely due to collateral stimulation.

Check out Bishop et al. (1962) - I believe that article goes into more details, and might help you track down the original article.

Erik

Hello Rune,

even if I don't have so much experience about this topic, once I tried the collision experiment using the leech Hirudo medicinalis. This animal allows you to make such experiment in a very simple way, you just need the appropriate equipments and you have one axon in one nerve for a given cell.

Anyway, what I can tell you from my experience (as you said is difficult to have other interpretation) is that the two AP simply annihilate eachother. Moreover using this collision test you can even study the duration of the refractory period of the AP (I am sure you know that) so, to observe the anti-dromic AP in the soma the delay should be at least longer than the absolute refractory period.

I hope it can be usefull.

Francesco

Thanks Francesco-  Do you know a paper where such collision test is described in leech?    

The two colliding AP's will annihilate each other. This has been verified many times. This also means that an AP that reaches a nerve terminal should not reflect back because if it does then it would block incoming AP's. The nerve terminal would then not respond to a high frequency train of incoming AP's.

Rune,

I agree with Mark and Henrique about annihilation. However, something to keep in mind is that changes in geometry along the conduction path (e.g. varicosities, branch points, axon-soma junction) can delay conduction. So if delays are on a similar time scale as de-inactivation, then in theory, annihilation may not occur. In fact, there is theoretical evidence of action potential reflection. In any case, you may want to see John Eccles’ work from the early 50’s.

Thank you guys. Yes I also agree with you that annihilation occurs (like my own experiments show) - and thank you for all the input.

What I am really asking for is specific references to previous work that has addressed this. Henrique says annihilation has been verified many times, but what are the references? If you know the best most clear articles addressing this, please share. I am putting together a rebuttal to the physical review X paper where the specific references are necessary.

best, Rune

This is an excellent review for your rebuttal:

Lipski (1981) Antidromic activation of neurones as an analytic tool in the study of the central nervous system. J Neurosci Methods 4(1):1-32

Thank you Taylor, This is very helpful!

I generally agree with the other correspondents and would certainly recognise that Tasaki's beautiful single fibre study clearly revealed collision at work. Kuffler used it in an experiment to allow him to study the effects of the small motor fibre system on muscle spindle afferent discharges -don't have the reference to hand and Murdoch Ritchie( 50's or 60's Journal of Physiology) and another collided naturally evoked  action potentials with the compound action potential evoked by electrical stimulation to reveal which group of fibres in the latter carried the information in the former- references may follow

I'm going to jump in here with a quick comment:

One of the authors (Heimburg) has been pushing a crank theory that the HH model is wrong and that action potentials are actually 'solitions' (sound waves). I have collected objections here: http://www.quora.com/Are-nerve-impulses-solitons

Collision annihilation is, of course, a routinely observed phenomenon. There are few papers after 1949 simply because there is no question to be investigated.

I just came across this discussion now. I absolutely agree with William.There are just no questions to be investigated, but collision is routinely monitored in many labs.

The same group has now another paper that is on a preprint server, in which they claim that action potential can cross without annihilation.

http://arxiv.org/abs/1502.07166

When you look closely you see that the collision experiments are severely flawed. They are stimulating different axon in their collision experiment (Fig. 4, left - look at the amplitude difference between ortho- and antidromic spikes [check scale]). Naturally the APs don't collide since they are traveling in different axons.

Their interpretation of the magnetic force microscopy results is also fishy..

Here are two papers that show that annihilation occurs in identified axons:

Daur, N., Nadim, F. & Stein, W.* (2009) Regulation of motor patterns by the central spike-initiation zone of a sensory neuron. Europ. J. Neurosci.,30(5): 808-822., Fig. 4B (*)

Follmann, R., Rosa, E. & Stein, W. (2015). Dynamics of signal propagation and collision in axons. Phys. Rev. E 92, 032707. http://dx.doi.org/10.1103/PhysRevE.92.032707

The last one we did in response to the PRX paper that claimed that no annihilation occurs. They could not be more mistaken.

Hi,

I'm sad to see this group is still pushing their theory. 

Solitons are very cool... they are non-linear waves that lose very little energy as they propagate... I can see how people get fascinated by them and wonder if the AP can be like that waveform, but the non-linear action potential wave is not a soliton because the underlying differential equations are just very different. 

The classic and clear description of the collision of antidromic and orthodromic action potentials are given by my past supervisor Prof Janusz Lipski, who currently works (retiring) at University of Auckland in New Zealand. His classic paper has i think has over 500 citations is called:

Lipski (1981) Antidromic activation of neurones as an analytic tool in the study of the central nervous system. J Neurosci Methods 4(1):1-32.

We together used this method alot in our in vivo intracellular work in adult rats. For example see

Article Morphological study of long axonal projections of ventral me...

Article A comparative study of pre-sympathetic and B??tzinger neuron...

Thank you Refik for your references. Yes I know Lipski's 1981 paper very well. We've also cited it in our manuscript comment, which is still in review now more than a year later. Thanks for the other papers too.

Dear Rune,

i wish Janusz was using RG and he would probably give you a much better answer than mine. He is pretty much retired now. Therefore I thought I better answer your question, he was my first electrophysiology teacher and he taught me the collision test, when we were doing in vivo experiments from 7am till midnight.

best wishes, Refik

Some people already hinted at it, but for HH model to be valid one must do  a single-neuron measurements. Something the paper obviously did NOT! Must go over in more details, but it could explain findings !?

I do not know the experimental result, but from a pure mathematical perspective, particularly for the FitzHugh-Nagumo model (a simplification of the HH model), whether the colliding waves annihilate or cross each other is not definitive but depends on the parameters. See http://citeseerx.ist.psu.edu/viewdoc/download;jsessionid=40E5010121E21A3BE7FF7CEC244CB650?doi=10.1.1.216.3108&rep=rep1&type=pdf.

Hello Everyone

Here is a little update on the status. Our comment manuscript have finally been accepted for publication in Physical review X. We are now waiting for the authors of the original paper to comment on our publication. So stay put, it should come out soon. I'll write an update once that happens.

Hello everyone.

Finally our rebuttal comment has been accepted for publication in Physical Review X, almost 2 years after our first submission. We could not reproduce the observation of penetration of action potentials by the Heimburg group. Instead we consistently observed annihilation of colliding action potentials.

Further we show that their measurements of action potential velocity was flawed and indicate that their measurement electrodes were so close to the stimulation electrodes, that 1) they could not contain two action potentials 2) passive propagation may explain part of their observations. 

Also, they did not demonstrate the All-or-none property of action potentials, which again suggest that they observe either activation of multiple nerves, or they generate passive electric signals. Attached is preprints and supplementary information.

Cheers!

Rune

Well, I decided to remove the preprints, since the editors of PRX have requested not to distribute prior to publication. The date for release is April 25. If you want preprints send me an email, and I'll give it to you. 

Dear All

Our comment is now out. You can find it here:

https://www.researchgate.net/publication/315475289_Comment_on_%27Penetration_of_Action_Potentials_During_Collision_in_the_Median_and_Lateral_Giant_Axons_of_Invertebrates%27

or on the PRX web site: 

https://journals.aps.org/prx/abstract/10.1103/PhysRevX.7.028001

Note there is also a reply to our comment. The link should be next to our publication.

Article Comment on 'Penetration of Action Potentials During Collisio...

Really nice, I would love to have the final manuscript. For someone just getting into the physiological studies business there is a lot to learn. Maybe it is worth commenting what the moral of the story is. Physiologic studies are hard ... both theoretical and practical matter .. !?

Interesting quesion Petar. I think there is not that much to learn that we didnt already know. Maybe this:

* The scientific process is not perfect. Scientists and editors are people who make mistakes like everyone else.

* It pays to read the literature. Most likely there are someone else who have thought of your idea already-  If we dont read the literature we will get into trouble eventually.

* Last, science today is so much about productivity and attention. Solid and careful science does not get the reward it should. I think we should try to implement a reward structure for rewarding the thorough work rather than the fast and easy work.

Does this make sense?

Rune

Thanks Roman! I've already seen your paper. Good work!