Evolution doesn’t give a damn what you think a brain region is called
A cortex by any other name
We do love naming things. Nouns are the first words we learn: mummy; daddy. Or if you’re my kids: “digger” and “Bob” (the cat) — because who needs parents when there’s a big ginger cat to play with? That love for tagging an object with a label never goes away.
Neuroscientists are no different. In large brains, like ours or rats or ferrets, we give hundreds of names to their constituent parts. Some name vast swathes of brain tissue: cortex, brainstem, midbrain. Some name tiny subdivisions of those swathes, like the almond shaped subthalamic nucleus (literally, the cluster of neurons underneath the thalamus). In tiny brains, like maggots or roundworms, even individual neurons have names. (Boring ones like “P4”; sadly, there is no neuron in C Elegans called Boris).
By naming these bits of brain, we can communicate effectively, we can talk about brains and know we are talking about the same bit. We can navigate where we are in the brain by knowing the names of the brain regions. We can even identify ourselves by the bit of brain we study: “I work on cortex”; “I work on hippocampus”; “I couldn’t decide what I wanted to do when I left school; now I work on the nucleus ambiguus”. Naming bits of brain greases the wheels of scientific communication.
But it is also a dangerous game. If we confuse the convenience of naming with the reality of the brain, we end up in deep trouble. Because evolution does not give a crap what we call a brain region.
The danger is that naming a bit of brain makes us think it is a discrete thing, a bit that just lifts right out, which we can study and marvel at in isolation. But how evolution (and development) conspire to disperse and wire together neurons pays no attention to these names.
Consider primary motor and somatosensory cortex. They sit next to each other. But they appear in different chapters in textbooks. Entirely separate fields of research have grown up around them. Work on primary somatosensory cortex looks at how the activity of its neurons represent touch. Work on motor cortex looks how the activity of its neurons represent movement. These are starkly different: work on somatosensory cortex focuses on its inputs; work on motor cortex focuses on its outputs. They are treated so separately that papers merely showing the flow of activity from one to the other end up in high-profile journals.
But these bits of cortex are next to each other. Either we believe that there are border guards who turn away the motor cortex neuron axons at the crossing with somatosensory cortex, and the same from the other direction. Or we have to assume that these two names loosely delineate a continuous network of neurons by the fact that a small set of neurons in somatosensory cortex get direct sensory input, and small set of neurons in motor cortex connect to the spinal cord. Most of the neurons in these bits of cortex neither get sensory input from the thalamus nor project to the spinal cord. They are wired to other neurons all over cortex, and very much to each other.
I mean, look at this nutter:
Evolution has deemed it necessary for this neuron in motor cortex to send its output wires, its axons, all over the shop. To the somatosensory cortex. To prefrontal cortex, so that the apparent seat of “executive function” — cogitating, planning, turning things over in your mind — apparently needs to know what your foot will be doing shortly. To the ectorhinal cortex (no, me neither). All over the striatum. And to the other side of the brain — yes, the brain has two sides. We call it a motor cortex neuron; it is studied for how it represents simple movements; but it is not a neuron that sits neatly tucked up in one neatly labelled part of the brain — it is part of the brain’s sprawling, tangled network.
Cortex is divided up into many different areas. Old schemes for dividing up the areas just used numbering. These numbers have been handed down to us from epic studies by unfathomably dedicated neuroanatomists in the early twentieth century. They finely sliced up brains, stained the slices to pick out the neurons, and then minutely examined the changes in the density, size, and shape of neurons across the brains. Each place where there was a change in one of those things got a number.
The temptation is to then think that these subtle differences in neurons correspond to different functions. We followed that temptation by turning many of those numbers into names: primary motor cortex; secondary motor cortex; primary somatosensory cortex; visual cortex. But in the human cortex Brodmann found 52 areas. von Economo and Koskinas found 107. Hmmm. One may reasonably ask: if just dividing up the cortex by changes in the shape, size, or density of neurons actually tells us something about what the different areas do, how come the numbers of areas were so different?
A recent massive fMRI study used a bunch of ways to isolate different parts of cortex, and ended up with 180 different regions. Far more then the slice-and-stain guys. Which is right? None of them. Those lovely differences in cells are immaterial to actual function. Any given function we care to name — seeing, jumping, talking, putting our DVDs in thematic order because we’re not culturally illiterate barbarians — engages many areas of cortex, and many areas of the brain, at the same time. The names and numbers are misleading, are irrelevant to how evolution has driven the wiring of the brain to perform a function.
Cortex has names for its layers too. Names straight out of a Dan Brown novel: 1, 2, 3, 4, 5 and — wait for it — 6. In rodents, some areas have six layers, and some have five because they are missing layer 4. And as the textbooks will tell you, it stands to reason that motor cortex and somatosensory cortex have different functions because motor cortex doesn’t have a layer 4.
This is an object lesson in the problems of naming: we could not name a clear cell “layer 4” in motor cortex, because we couldn’t see one. But there is one, and it’s been there all along. What would you bet me that it will turn out the rat prefrontal cortex does have a “layer 4” after all — a set of neurons that are specifically targeted by input from thalamus and then connect to other neurons in the cortex without projecting outwards — just not in a distinct layer we can see by colouring in neurons? Evolution doesn’t care that we can name five layers in one region, or six in another. In only cares what they do to keep the animal alive.
To evolution, the brain is just a gigantic bag of cells, wired together. The purpose of that gigantic bag of cells is to contribute to the survival of the organism in which it resides, to surviving long enough to reproduce. Those that reproduce, win; those that don’t, don’t.
If a random mutation causes or changes the wiring of some neurons to another group of neurons, and that mutation improves the chances of having offspring, it will likely spread through the population. If that random mutation adds connections from prefrontal cortex to visual cortex; from cortical interneurons to a structure outside cortex; from motor cortex to midbrain dopamine neurons, then it will happen. Evolution will not feel sorry that it’s just ruined another set of textbooks.
And it is just a giant bag of cells wired together. Our best evidence that it is not — that we can cling to our names of all the bits — come from studies where we cut a bit out or turn a bit off. When we cut out area X and we see a “deficit” in behaviour Y of an animal (like tying its shoelaces), then we think “aha! Area X is for tying shoelaces”. No. For starters, we never see a complete and permanent end to behaviour Y. We normally see that the animal is simply worse at doing or learning Y — not that it cannot do Y at all. The brain can carry on doing Y just fine, thanks, just not as well — there is massive redundancy in the brain. Like what you’d find if it was a giant bag of cells, wired together.
Moreover, seeing that behaviour Y gets worse logically tells us little about what area X is actually doing. It just tells us that damaging area X causes problems. Which on reflection isn’t surprising as you just ripped a chunk out of the brain. The logical fallacy is simple to demonstrate. I am right now going to make a new startling new prediction of how the mouse brain works: the ventrolateral medulla is necessary for mice to learn to associate pictures of Benedict Cumberbatch with food. Cut out the ventrolateral medulla, and a mouse will not learn to associate pictures of Benedict Cumberbatch with food.
Because it will be dead. The ventrolateral medulla contains the neurons which control the rhythm of breathing. Cut it out: no breathing. Ergo, no learning. Is the ventrolateral medulla a crucial brain area for learning? No. But by damaging it, we damage something vital to the process of learning. Thus, cutting a bit that we’ve given a name can have an effect on that named thing, and we learn nothing at all. Except that we have damaged a big bag of cells, wired together.
What’s more remarkable is when cutting bits out has no effect. If we cut some bits of brain out before learning we see an effect — learning is made slower or worse or both; but when we cut them out after learning, it has no apparent effect whatsoever. These bits of brain have become completely redundant. Again: giant bag of cells, wired together — there are many ways within that bag of cells to solve the problem at hand, enough for the brain to just stop using a bit of itself altogether.
And people get hyper-excited about finding odd signals where they were utterly unexpected. Like finding that reward changes activity in primary visual cortex. Or finding that sound is encoded in the hippocampus. These are great, interesting studies. But to be surprised by them is to fall into the naming fallacy. Evolution does not know nor care that we called this chunk of neurons the “hippocampus”. It is just another bag of neurons, connected to many other bags of neurons.
Names. Names are vital. Names mean we can all be sure we’re talking about the same thing. They identify whopping great phalanxes of neural tissue; they identify individual neurons; they let us compare brains of different creatures. But we have to handle them with care, less we confuse the label with the contents. Evolution only cares about the contents.
And development doesn’t give a damn what we call a brain region either. But that’s a tale for another time.
Want more? Follow us at The Spike