Michael Levin has a crazy idea. Bioelectric signals among cells (all cells, not just neurons) are responsible for controlling large-scale outcomes in organisms. Things like left/right symmetry (and asymmetry), body form, cell differentiation, and regeneration. You can think of it roughly like the “software level” in organisms, an abstract representation which guides the “hardware level” (DNA).
Most non-biologists are surprised that all cells communicate electrically. Non-neuronal cells communicate more slowly (on the order of minutes instead of milliseconds), but they do it through a similar mechanism: gap junctions. In fact, Levin thinks that neurons evolved from these more primitive mechanisms.
Bioelectricity seems responsible for many astounding things: if you change the voltage in a single cell in a frog embryo, an entire eye will grow on its leg. If you change a group of cells in the adult frog's gut, leg, or tail to that same voltage, an entire functional eye will form. This is all without any direct modification to DNA. And you can get even odder things still: two-headed worms, two tailed worms, growing legs on non-regenerative animals.
But as of yet, we don’t know how bioelectricity orchestrates all of this macro-scale complexity. We can copy and paste the voltage pattern of “head-shape” onto a worm's amputated tail, and lo and behold, we get a head where a tail should be. But what about that electric signal, exactly, caused a head there?
Unfortunately the answer is that we still don’t really know. This paper addresses some possibilities, though.
The first is that the code might be a direct representation of body form. For instance, the bioelectric pattern in a tadpoles face more or less one-to-one corresponds with its adult form (it’s not exactly one-to-one because it grows in size, but you can see a frogs face in the late-stage embryo when you image its voltage).
The second possibility they cover is that bioelectricity codes for specific anatomic structures. This comes from the study mentioned above where a specific voltage in a single cell in an embryo caused an entire eye to grow on a leg. Maybe the code is composed of a few of these “buttons” – a single voltage which executes a complex program, e.g. “make an eye.”
More evidence for this possibility is that placing a tadpole with an amputated tail in an electric bath caused it to regrow (tadpoles are not naturally regenerative at this stage). They applied this same technique (same voltage) to an amputated frog leg and got similar results.
Of course, these might not be mutually exclusive, or the only options.
Some other interesting tidbits from this paper:
Bioelectricity is abstract. It doesn’t matter if you get the voltage from potassium, sodium, from creating more gap-junctions, whatever – the result is the same.
There are electrical gradients at all levels in organisms (within cells, across cells, in tissues, in entire bodies) and these can differ tremendously from level to level. A single cell can represent quite a bit of electrical information (as in the image below):
My thoughts/takes:
The “button” possibility seems on the whole pretty strange to me. It seems like bad design to have the possibility of entire organs growing anywhere on your body if the voltage in one of your cells gets messed up. And yes, evolution doesn’t generally use engineering principles in design, but still – this seems like a big error that you could easily stumble into and hence want to avoid. Maybe it’s for some reason quite hard for that cell to have that voltage in embryo?
More than that, though, it seems like this can’t be the whole story. Deer antlers regenerate new structures depending on how they were injured in seasons past. There is some active computation going on, not just a static button that reliably creates the same structure. The button story is also non-linear, which again seems like a bad design principle for large-scale outcomes and is also argued against by Levin and colleagues in a later paper.
The one-to-one story seems more compelling to me (although of course it’s not exactly one-to-one because the organism grows). This mapping is much more easily modifiable, e.g., in the antler case you could straightforwardly add the tine to your representation.
However, I’m confused about how this relates to “Picasso tadpoles,” ones where their face is completely scrambled, but then reconfigures into an appropriate adult form. This is true of almost all embryos, and it’s true of regenerative species (like this Hydra, a small multicellular creature, which can reassemble its entire form after being blended up).
Is the bioelectric code for face also scrambled in the Picasso frogs? How does it come back together? This is something I wish I had a better handle on. One would think that electrical gradients are being generated by the cells, and yet you want to explain the embryo rejiggering with reference to some abstract, global representation which guides them back into shape. But where is that global representation coming from? How are cells still maintaining that after they’ve been scrambled? Levin sometimes talks about basins of attraction for electrical gradients, maybe it’s something like that?
The one-to-one story also isn’t as obvious in other species. For instance, the bioelectric pattern for head shape in planaria isn’t apparent to the naked eye (as it is for tadpoles). Still, you’d expect some conservation of this code across species.
There’s still a lot of work to be done here, in charting all of this out. There are several puzzles like the ones above that constrain the problem. What’s a representation that’s easily modifiable on the fly, that can rejigger itself once scrambled, that can trigger entire organ formation, etc.? I suspect that some time reflecting on these various bits of information at a computational level might yield interesting insights.
I also expect that directly probing it via experimentation will be fruitful – something I’m hoping to make happen.