Script and References for This Neuron Shouldn’t Exist

The following is the script, credits, and references, for the video This Neuron Shouldn’t Exist, over here on YouTube: https://youtu.be/RQ4AMIFWnN4

Long ago, on a random schoolday, I was standing in the lunch line when a friend informed me that there existed, on my hands, particular fingers that when extended by themselves produced a rude gesture that would make our teacher angry.

I said something to the effect of, but probably less articulate than, “why? what could possibly be offensive about a finger?” And to demonstrate my point, I fell right into my friend’s trap. Extending both of my middle fingers at the precise moment that our teacher walked by. My friend was right, the gesture did make our teacher angry. I was confused, and so after I was done getting chewed out, I asked why I wasn’t allowed to hold up my middle fingers. To which the teacher responded, “that’s the rule.” Naturally, [with LAUGHTER] I followed that up with asking, once again, “why?” To that, they only offered “just… follow the rules, okay? Stop making trouble.”

It’s been twenty-something years since then, and in that time, my relationship with rules has changed somewhat, because these days, I’m a scientist.

And one of the foremost rules of science is that scientists love rules. Rules help people to understand the world via simplification.

It was a lot simpler for my teacher to tell me to follow the rules than it would have been for her to explain that the middle finger is considered a rude gesture due to centuries of tradition, whether extending it upwards or perhaps “horizontally… just as one might when checking an ovipositor of a recumbent bird” (Nelson, 2017, p. 70).

Unfortunately for science, nature doesn’t care if we are able to understand her. Where there are rules, there are often exceptions.

Recently, an anonymous 45-year-old woman broke a fundamental rule of science with her brain (Blakely & Januszewski, 2021; Shapson-Coe et al., 2024b, p. 1).

Neuroscientists like me are still grappling with whether our precious rule has rare exceptions, or if her brain is just making trouble.

Allow me to explain!

“MY BRAIN HURTS” INTRO MUSIC

Ha. Ha ha ha ha.

CAUTION: This video contains CONNECTOMICS, the inherently confusing science of brain cells exchanging information across physical space. Symptoms of confusion may include:

-laughing alone

-up past bedtime

-pants on head

-mispronouncing discretion

-and/or wiggle arms

Viewer disc-reet-shun is advised

We don’t have a ton of details about this woman, so the following list will necessarily be a bit sparse. The Who is an anonymous 45-year-old woman, the Where is likely Massachusetts General Hospital (Blakely & Januszewski, 2021; Shapson-Coe et al., 2021, p. 3)

the When is sometime before 2021,

and the What is Partial Brain Removal. Don’t worry, she consented to the surgery.

Two big names, Harvard and Google, took a piece of her brain, from the left (Shapson-Coe et al., 2024b, p. 1) anterior middle temporal gyrus, just over a cubic millimeter in volume (Shapson-Coe et al., 2024a, p. 1).

The research team then spent the next 326 days (Shapson-Coe et al., 2024b, p. 3) slicing and scanning that tissue to create an insanely high-resolution set of two dimensional images at nanometer scale, which you are looking at right now. Unfortunately, the brain is very complex, so these pictures are quite confusing, even if you know what you’re looking for. Hey… what are we looking for?

This is Schwanzie, a neuron plushie that I use in my introductory psychology course. On one end he receives information with his dendrites, and on the other end he sends information… out through his axon. What we are looking for in that woman’s brain are dendrites, and axons. Isn’t that right, Schwanzie?

“Hello!”

Shhhhhhhh…

Don’t listen to him. You see, Schwanzie… is a liar. Neurons don’t look like this. He is a vastly oversimplified model. For example, his haircut is made up of just a few dendritic branches, when real multipolar neurons look more like… this.

This is a photo of a 2D slice of a neuron, a subtype called a Purkinje cell (Martone et al., 2002). A Biblically accurate Schwanzie.

But Schwanzie does have at least one thing right. This part, the long neck covered in bumps, is the axon. And Schwanzie… has one axon. You can also see one axon on the Purkinje cell, over here.

That singular axon is a fundamental rule of neurons. It’s so fundamental that you will find it written or at least implied in basically every textbook that involves the brain. Here’s the textbook that I used to teach Behavioral Neuroscience. On page 4, it distinguishes neurons as a cell with “the presence of a single axon.” (Augustine et al., 2024, p. 4)

Sure, they get cute with it sometimes, like on page 10 where they claim that some neurons “have two axons” (Augustine et al., 2024, p. 10). However, they are actually using terminology incorrectly, as I point out in this scribble in the margin from years ago.

See, while neurons are only allowed to have one axon, that axon is allowed to branch as many times as it likes. Here is how it is stated within another textbook: “While there can be thousands of dendritic branches coming from one neuron, there is only one axon” AND “axon: a part of the neuron that sends signals away from the cell body. After leaving the neuron at one point, they can branch several times” (Shapiro, 2024, p. 48).

In sum, when we say that neurons have one axon, we actually allow for that axon to branch using what are called axon collaterals. But, neurons have one point where an axon is attached to the cell body. That point is sometimes called the axon hillock.

///side note: it is possible to have more than one axon, as in the amacrine cells of the mammalian retina such as nNOS-1 AC, but those emerge from dendrites and do not emerge from axon hillocks from the cell body (Greschner et al., 2014, p. 3597; Lin & Masland, 2006, p. 800; Musgrove, 2026, p. 7)///

///another side note: there is another term, axon initial segment, that is also thrown around out there, but it isn’t as much fun and means something slightly different and less measurable via Neuroglancer than axon hillock///

Neuroscientists generally think it’s important that neurons form one such axon hillock. For example, here’s a paper that claims that “The formation of a single axon (SA) and multiple dendrites in neurons is called neuronal polarization, and it is fundamental for proper neuronal network functions,” (Rastogi et al., 2024, p. 2513).

That polarization term is used in reference to “the law of dynamic polarization,” which comes from around the year 1891. In a nutshell, the “law” says that “nerve cells have a single axon” (Guillery, 2005, p. 1283).

///Side note: It has some additional parts, which also have exceptions, but let’s focus on the bit about singular axons.///

///Another side note: Rapport reports in his 2005 book “nerve endings” that during his Nobel Prize acceptance speech, Golgi “attempted to overturn the law of dynamic polarization” (Rapport, 2005, p. 164). Mostly the bit about conducting in one direction from dendrite to synapse. It did not go well for Golgi.///

Alright, well, if having more than one axon hillock per neuron is illegal then somebody better arrest that lady I mentioned because her brain is breaking that law.

When the research team took the two-dimensional slices of her brain and stitched them all together into 3D with assistance from machine learning, they saw that her brain had this. Right here are two axon hillocks, on the same neuron (Google Research & Lichtman Lab, 2021a, 2021b).

That is so illegal, I’m shaking. I’m shaking! Somebody DO somethi—!

If that’s not enough to blow your mind yet, both axons form synapses with the dendrites of other neurons, as you can see in Figure S3 from their publication’s…sss supplemental materials (Shapson-Coe et al., 2024b, p. 37).

Hmm, that figure isn’t very good, there’s just a floating chunk of dendrite there… Don’t worry, I’ll make a better one before the end of this video!

I’m looking at my script and realizing that I should probably explain what a synapse is.

Hello!

Shhhhhhhhh…

By the way, these plushies are from NerdBugs on Etsy, who did not sponsor this content.

A synapse is a small gap, most classically between the axon of one neuron and the dendrite of another neuron. In this case, Schwanzie and, uhh, I guess this one needs a name.

Fortunately, I have options left over from when I asked my students to suggest names for Schwanzie. The runner-up name was “Somadopa.”

The streamlined idea is that the axon on Schwanzie sends an electrochemical signal across a tiny gap, called a synapse, to the dendrite on Somadopa. Why do we care about this? Because all human experiences depend upon synapses between neurons.

Looking at her brain tissue, we are seeing the physical substrate that enabled pieces of her psychological experiences: her thoughts, feelings, & behaviors. In all likelihood, the neuron with two axon hillocks contained pieces of some of her life’s memories.

That’s because this neuron was located in her association cortex from near the hippocampus, where memories get stored in loops between neurons, making use of axons, dendrites, and synapses. That memory of mine regarding learning about the middle finger? Somewhere in my brain, that memory is made up of synapsing neurons much like these.

The particular layout of such neurons is not only what makes us human, but it is also what makes us into individual humans.

Changing the layout of neurons, such as via a brain surgery or brain injury, can change memories, perception, personality, beliefs, and, well, anything and everything about what it is like to be you. I wrote a whole book about this with over 150 examples of people whose psychological lives changed when their physical brains changed, if that’s your cup of tea (Toftness, 2022). But for now, let’s get back to breaking the law.

You can find variations of the law of dynamic polarization pretty much everywhere… which begs the question… is it okay to break the law?

—Palay et al., 1968, “Because each neuron has only one axon…” (Palay et al., 1968, p. 198)

—section 12.2 from OpenStax Anatomy and Physiology 2e “Neurons are usually described as having one, and only one, axon” (Betts et al., 2022, Section 12.2),

—Doral Health & Wellness “Every neuron in the nervous system contains a single axon” (Doral Health & Wellness, n.d., para. 2),

Well, first we need to ask:

Why is “one axon hillock per neuron” a rule of science in the first place? I dug into it, and it’s kind of just one of those things that we’ve been assuming to be a rule. The “law of dynamic polarization” is just a description of what seemed to be true, not really an explanation. More recent evidence suggests that neurons would quite like to grow additional axons, thank you very much.

In fact, growing axons is so appealing that there are cellular mechanisms that specifically suppress the growth of supernumerary axons (Wang et al., 2018, p. 3).

If you turn the suppressing mechanism off, dendrites can turn into axons (Jiang et al., 2005, p. 123).

It’s a classic Gremlins (Alvin, 1984; Dante, 1984), but in reverse. You have to feed dendrites a kinase called GSK-3β every night after midnight or else they turn into Gremlins (Alvin, 1985), I mean, axons.

This gremlin-esque ability can be useful. Labs have successfully converted apical dendrites into axons after, uh, cutting off the existing axon, allowing the cell to create a new one (Hayashi et al., 2002, p. 14).

Wait, what? Apical? The New York Times Spelling Bee insisted to me that that wasn’t a word! Unbelievable (The New York Times, 2026).

Anyway, it’s also true that when we knock out certain proteins in genetically-engineered mice, we can get neurons that form multiple axon hillocks (Wang et al., 2018, p. 2).

And the evidence suggests that the hillocks each are capable of functioning, that is, they are electrically active (Guo et al., 2017, p. 10; Jiang et al., 2005, p. 126).

However, there isn’t a ton of research yet on what is different about the functionality of these multiple-hillock neurons. One study suggests that they go from not firing to firing a bit faster, but not in a way that affects the threshold of firing. (Guo et al., 2017, p. 1). So, either having multiple axons might not affect much, or we just haven’t figured out what’s different yet.

///Side note: Rastogi et al. (2024) claims that multi-axon neurons are associated with down syndrome, but I did not feel convinced at all by their work///

I feel like I’m forgetting something. [snaps fingers and points] Right!

The “Why!” As in, why were they cutting out part of this lady’s brain? She was undergoing psychosurgery as a bold last-ditch effort to control her epilepsy.

In other words, she was having seizures, and medications were not helping enough, so, they resorted to brain surgery as a treatment (Shapson-Coe et al., 2021, p. 21). As I discuss in my chapter on Seizures in Incredible Consequences of Brain Injury: The Ways Your Brain Can Break, psychosurgery can honestly be an effective treatment for epilepsy, it just… usually isn’t the first thing you try (Toftness, 2022).

[[Quote shown on screen]]

“…some types of seizures don’t… respond to medications… Another option is the surgical removal of the seizure focus… The literal removal of part of the brain may sound extreme, and it does have many risks, but it is often a highly effective treatment.” (Toftness, 2022, p. 252)

So, she consented to have pieces of her brain cut out in order to remove what is known as an epileptic focus, which is kind of like the spot in her brain where the seizures are being generated. If the surgery goes well, it’s like removing a battery from a smoke alarm.

Wait, does that mean that the neuron with two axons came from some unusual epileptic focus in the brain of someone with a disease that had been treated with brain-altering drugs, and so it’s some kind of unusual brain composition?

Acknowledging that, the authors of the 2024 article write: “we cannot exclude the possibility that long-term epilepsy, or its pharmacological treatment, had subtle effects on the nanometer-scale structure of the sample” (Shapson-Coe et al., 2024a, p. 1).

That’s what I was referring to earlier when I suggested that maybe her brain was just “making trouble.” If her brain is abnormal in some way, then maybe the law of dynamic polarization is not really being broken?

But that doesn’t seem to be the case. The cubic millimeter or so that Harvard imaged and Google turned into a database is from the area that covered the focus, called the anterior middle temporal gyrus, which needed to be “removed to gain access to [the] epileptic focus in the underlying hippocampus” (Shapson-Coe et al., 2024a, p. 1). As far as the researchers could tell, the brain images are of typical brain tissue such that “the sample was deemed normal” (Shapson-Coe et al., 2021, p. 21, 2024a, p. 1).

Okay then, well, maybe their machine-learning algorithm made a classification mistake, and those are not actually both axons? To gather more information, I asked Harvard if I could have edit permissions for the connectomics dataset—I told them my confidence was like a 4 out of 5—and they said “yeah sure.”

Which led me to this footage of some random evening around 10 p.m. exploring the neuropil while wearing my connectomics-editing hat. Allow me to orient you real quick.

This is a 3D mesh of an axon built from these three 2D planes in the X, Y and Z-axes. You can see the planes if I toggle them on. You can scroll through the 2D layers, rotate the axes, zoom and pan. It’s like flying a tiny spaceship through a brain! You can also toggle specific 3D meshes on and off. If I want to know what this thing is, I can double-click on it to take a look, and find out that it is a dendrite. Just try not to get jumpscared by an astrocyte.

Importantly, turning the 2D scans into 3D meshes relies on machine learning, which can and did make mistakes. That’s why I’m looking closer at the neuron with two axons, to see if anything was overlooked. And I found something.

It quickly became apparent that the neuron did indeed have two axons that do indeed form synapses with dendrites.

But, remember that supplemental figure? This weird floating dendrite chunk doesn’t appear to be attached to anything. I set out on a mission to find the cell body that the dendrite was attached to.

I’ll show you how that went right after I say that this episode was brought to you in part by my patrons over on Patreon, where you can help make videos like this possible and pick up some nice perks along the way. And while you’re at it, if you’re watching this on YouTube, please make sure that you’re subscribed to the channel and to like the video. I hate asking for engagement, but it makes a huge difference to the algorithm, so thank you for your help. Alright, back to the neuron that shouldn’t exist.

“Now where… where is your axon, sir?”

Confidence drops to 3

“I mean, I guess… I mean I guess? Couldn’t tell you which is which off the top of my head.

Confidence drops to 2

What is this thing? Ohhhhhkaaaayyy… What? Do you two have any… synapses? So, that…it—why is it?

Confidence drops to 1

Is that the axon? Is that the axon? Why is it… why are you like this? Are you a chandelier cell?

Confidence increases to 2

What are you? But why? For what reason? I think that’s a different guy! I’m pretty sure it is!

Confidence increases to 3

They are meeeeerging!

Confidence increases to 4

Is that your axon? So this thing should continue, right? I think sooooooo…”

Confidence increases to 5

The public dataset and the raw dataset have different ID numbers for everything, so first I had to hunt down the specific neuron and the related axons and dendrites.

I started out by accidentally clicking on an astrocyte, which took up so much rendering power that I had to reload my browser in order to see anything else.

“Is it you, buddy? Is that you? You got that weird… yeah, that’s you, right? That’s my boi?”

“Okay, so you’ve got, you’ve got two axons. One, two. Two axons.”

“The left one… has an interesting synapse. That I want to try and find. Which is going to take a second because it’s going to have a different ID number. Oh man. This is like my final boss. Oh, good lord, okay.”

“We need to turn left, and then right. Left… then right… Okay, and then we make it to another… very strange-looking axon [collateral] right here.”

“Yes, okay, fantastic, and I need to follow… surprisingly, I need to follow this branch right here, which then curls up and it has a weird foot… right here. This foot right here.”

“Okay, there is the foot, and I need to find your synapse partner. The resolution here is beautiful. That looks promising. Hello… are you what I’m looking for? Yes. Fantastic, okay.”

“Right here. It’s weird that it just ends. So there’s more to it. And if you look at their official PDF, their supplementary materials, the image that they have of this two-axon neuron has this incomplete dendrite in it. So we need to… fix that. Need is a strong word. I’m going to fix that. Because I know there’s more here.”

I follow the dendrite and rotate through the two dimensional slices to find plausible pieces of dendrite that should be connected.

“I don’t think it’s this, is it?”

Eventually, I found a continuation; this big blue piece.

“A continuation of the dendrite, right?”

Y-yeah, that’s right.

“Yeah I think so, and especially with that added, like, yellow part. Like, where else would he be going?”

Then I click on another astrocyte.

“Good lord, no!”

I then merge the big blue pieces with the little orange piece, and get a lavender piece. That’s just math.

Because the dendrite goes off in two directions, I had a 50/50 shot of picking the correct direction to follow the dendrite back to its cell body. Either I would find the cell body or I would find a dead end.

“We got blue this time. What in the world?”

Naturally, I found the dead end first. Unfortunately, there was no way to confirm it was a dead end, but eventually I turned around and tried the other way.

“My God, you have so many synapses dude!”

“Are you… my dendrite? That’s not a continuation. It might just stop over here. That is legal. Thing is, is that dendrites can kind of just do whatever, you know?”

“Whelp [sighs] begs the question though: what’s on this end? ‘Cuz you gotta attach to something.”

“I hate astrocytes.”

“Got like a light blue this time.”

“There should just be a spot where you continue.”

“I mean, that looks promising… Oh good lord.”

I had found a potential candidate for a cell body that the dendrite might attach to. All I needed to do was find evidence that the two dendrites were actually the same continuous dendrite.

“So, there’s a synapse between this big purple foot and this green…”

I should also probably note that I am up past my bedtime in this footage, so I am getting increasingly sleepy, and loopy.

There’s not between my blue guy and the green axon. That’s a synapse. You can almost see the vesicles boiling in there. But now I guess the question is: Is that, right there, I think it is actually, hang on, look at that, I think that’s a bridge between my…”

“That. It doesn’t get much clearer than that, right? Right there. Right? Like that’s a continuous… that’s a continuous little thingie. So my argument is that the purple and the blue are actually the same object?”

Yeah.

“If that’s true… right there! Right there! Right? I think so.”

“Heh. Heh heh heh heh. I think so.”

“Yeah, no, that’s… that’s correct. There he is.”

“[snaps fingers and gives thumbs up] [sighs] Got it! And we’re back to green. Actually, that’s a happy ending.”

“There we go. Well, that’s half the battle. Because we still have axon number two.”

“I think it’s the thicker of the two, yeah?”

“I mean it’s… in a really wacky part. If it is this little footsie tootsie. Well, first of all, is it this little footsie tootsie? Yyyyea—maybe. This face up here, right? Is it this? Looks promising. Yeah, that looks correct. Yeah, that’s him.”

“This one in the middle has active synapses with this one, this blue one, and this sort of lime green one down here.

And at that moment, I became the first person in human history, to be able to actually see a multipolar neuron with two axon hillocks synapsing with two other complete neurons.

“That’s nuts.”

And these are just three neurons of over 16,000 in the sample (Shapson-Coe et al., 2024b, p. 52). What else is left to discover in just this small piece of brain?

“There’s an axon emerging right there; synapses. There’s the other axon which does a crazy thing over there. Uhh, which synapses.”

If you’re interested, the uneditable neuroglancer tool is available for you to try for free. And if you want to go even deeper, at the time of this writing they are still accepting applications for editing the dataset, too. “…a dataset ~1.4 petabytes in size” (Shapson-Coe et al., 2024a, p. 1).

That’s a really big dataset for only a tiny piece of brain.

And according to Google Research, it would take hundreds of exabytes for nanoscale-accurate scans of a complete human brain (Google Research, n.d.).

Should we be worried that Google is building connectomics databases with real human brain data? What are they planning on doing with that, anyway?

“Does it really take two axons to accomplish whatever you’re trying to accomplish?”

“Just floating in the night, doing whatever, except not really, they’re jam-packed together with all those other axons and dendrites. I know firsthand, sorting through these.”

In the end, I guess Schwanzie is even less accurate than I thought. Maybe I can fix him?

“Oh noooooooooo!”

“Ohhh!”

“Waaaah!”

“Mmmm!”

“Nomp!”

Hahahaha hahahaha hahahaha hahaha That’ll be $200,000.

“Helloooooo!” x2

Shhhhhhhh…

Thank you for watching this episode of ARTexplains… leave a comment about what you thought of this episode, and I’ll see you for the next one. Don’t forget to subscribe and like the video and all the other engagement nonsense. Put it on a playlist, tell a friend about it, uhh, keep being yourself, and have a good one. Bye bye.

Audio Credits:

Intro Music is an excerpt from a song titled “My Brain Hurts” from the eponymous album “Fine Art” (1978) — Used with permission of musician Colin Mansfield: https://colinmansfield.bandcamp.com/

“Airport Lounge” Kevin MacLeod (incompetech.com)