National Geographic

The brain is full of Manhattan-like grids

London’s streets are a mess. Roads bend sharply, end abruptly, and meet each other at unlikely angles. Intuitively, you might think that the cells of our brain are arranged in a similarly haphazard pattern, forming connections in random places and angles. But a new study suggests that our mental circuitry is more like Manhattan’s organised grid than London’s chaotic tangle. It consists of sheets of fibres that intersect at right angles, with no diagonals anywhere to be seen.

Van Wedeen from Massachusetts General Hospital, who led the study, says that his results came as a complete shock. “I was expecting it to be a pure mess,” he says. Instead, he found a regular criss-cross pattern like the interlocking fibres of a piece of cloth.

For years, scientists have been able to trace the outlines of individual neurons by injecting them with telltale chemicals that migrate along their lengths. But this technique can only be used in dead brains, and it’s small in scale. To get the big picture, Wedeen turned to diffusion magnetic resonance imaging (MRI), a technique that uses magnetic fields to detect the water flowing along our neurons. By tracking these streams, Wedeen mapped the brain’s white matter fibres – the tracts that carry signals from one area to another. They are the original information superhighways, and Wedeen could see huge groups of them at once.

He studied the human brain, as well as those of four primates – the rhesus macaque, owl monkey, marmoset, and galago (or bushbaby).  He started with a single large white matter tract – the equivalent of a motorway. Here’s one in the macaque’s brain, coloured in blue.

Then, Wedeen looked for all the fibres coming off this main tract. Here they are in red, orange and yellow. You can see them branching off perpendicularly in a single curving sheet.

Wedeen did this over and over again, creating stunning images like these – a riot of right angles, arranged in sheet after colourful sheet. “I was astonished,” he says. “[The pattern] was present in every part of every brain of the different species. It was always there.” Wedeen has even found the same patterns in the brains of cats, rats, possums and other animals, although those data have not yet been published.

Crossed wires

Opinion is divided on the new study. “It’s really ingenious what they’ve done,” says Tim Behrens from the University of Oxford, who is particularly impressed with the idea that the white matter forms interwoven sheets. “It’s really quite convincing,” he says. “There’s no way that the sheets are there by chance.”

David van Essen from Washington University in St Louis agrees, but he and Behrens both say that Wedeen’s technique is more sensitive at measuring right angles than other angles. They feel that the right-angled connections of the white matter remain to be proven.

Partha Mitra from Cold Spring Harbour Laboratory is more critical, describing the paper as “lots of pretty pictures, but not something that will fundamentally alter our views of how the brain is wired”. He notes that the structures of the grids don’t actually say much about how parts of the brain are connected to one another, since the origins of the fibres are still a mystery. He is also says that Wedeen didn’t “ground-truth” his maps against ones obtained from tracer studies in other animals.

It is certainly true that the maps, while beautiful, have to be interpreted carefully. For a start, diffusion MRI is an indirect technique, so the lines it produces are not really depicting specific neurons. “They’re mathematical constructs showing the most probable trajectory of the cells,” says Wedeen.

And while diffusion MRI gives a much broader view than tracer chemicals, it lacks the same resolution. It reveals the general structure of the brain’s road network, without showing where individual alleys and streets are. So each colourful line represents thousands of cells. Wedeen also cautions that, no matter what the diagrams might suggest, the white matter isn’t arranged in bundles or “discrete noodles”. The brain is full of cells running in parallel, and the lines just represent areas where they are most heavily concentrated.

Origin of the grids

Wedeen’s maps may not reveal all the details about the brain’s network, but it does show how that network is structured. “If you look at brain connections in an adult human, it’s really a massive puzzle how something so complex can emerge,” says Behrens. “If we can establish any sort of organisation, we get a clue about how these things grow. If it obeys some rules, you could start to work out how it follows those rules. You have something to hang onto.”

Wedeen thinks that the origins of the orderly sheets arise during our early embryonic days. As we grow from a featureless ball of cells, some molecules become concentrated at specific ends. These gradients set up three invisible axes that determine left from right, top from down, and front from back. And these axes, all perpendicular to one another, guide the growth of our first neurons to create neat grids.

Later, things get more complicated. Some fibres execute 90 degree turns, and some entire grids will curve and warp. But the same underlying pattern holds. This simple system can still produce a brain of staggering complexity, but it makes it easier for neurons to find one another.

Wedeen also suspects that right angles have been an important feature of in brain evolution. The simple nervous system of a flatworm, for example, looks like a ladder, with two main fibres running down its body and rungs connecting them. In fact, Wedeen speculates that such angles may have been necessary in early brains.

Our neurons are surrounded by insulation sheaths made of a substance called myelin, which shields the electric currents that run down their length. But these sheaths are a recent evolutionary innovation. “Electrical engineers have suggested to me that in some earlier brains, which don’t have myelin, neurons would have electrical problems if they crossed at anything other than a right angle,” says Wedeen. “They’d be more likely interfere with each other.”

The human brain is both larger and messier, but Wedeen is hopeful that his study will help us in our quest to understand its connections. He and van Essen are both involved in the ambitious Human Connectome Project, which is trying to “map the wiring diagram of the entire, living human brain”. “The field is just beginning,” says Michael Huerta from the National Library of Medicine. “As more data are collected… I would be astonished if additional general principles didn’t emerge regarding the way the brain is organized, functions, develops and evolves.”

Van Essen says that the grid idea “is likely to be an incomplete and imperfect representation of the fabulously complex wiring of the human brain.” But Wedeen isn’t claiming otherwise. He hopes that it will provide a valuable starting point. “Everyone knows that imaging isn’t equivalent to mapping every cell. There’s an endless expanse of potential detail that we still wish to know. But this [pattern] is a new piece on the chessboard.” He pauses. “And it really is a chessboard.”

Another pause. “And the piece is a rook.”

Reference: Wedeen, Rosene, Wang, Dai, Mortazavi, Hagmann, Kaas & Tseng. 2012. The Geometric Structure of the Brain Fiber Pathways. Science http://dx.doi.org/10.1126/science.1215280

All images courtesy of Wedeen and Science.

There are 10 Comments. Add Yours.

  1. Noah Gray
    March 29, 2012

    I’ll be brief — What Partha said.

  2. Hank
    March 29, 2012

    Induction at right angles.

    Do they follow the right-hand rule?

    This may be for electromagnetic benefit, no?

  3. Georg
    March 30, 2012

    The German
    technical term for that is “Kreuzschienenverteiler”,
    a basic pattern of telefone exchange systems.
    I don’t know the English word for that.
    Georg

  4. Mary Beth Davison
    March 30, 2012

    May I use this article and your photos as a teaching tool? We’re performing a sheep brain dissection this week in lab (API) and your article and photos are so right on as a review exercise to confirm comprehension of structures and their position in the CNS. Thank you for your consideration. Great photos, not just pretty pictures. Demonstrates connection pathways….

  5. Luke Parrish
    March 31, 2012

    Evidence that the brain is more comprehensible supports the notion that in the near future we can repair brains from some limited amount of damage. This in turn supports the notion of cryonics, supercooling the brain with as little damage as possible to halt deterioration for future repairs.

    A lot of people find this a radical idea (you’d in a sense be bringing the dead back to life), but this is something which apparently turns out to be morally imperative if it has a good chance of working, at least from the perspective that we need to always care for the sick rather than let them die. A brain is a person, and the fact that the heart has failed, or parts of the brain are starting to fail, does not imply that it should not be cared for if a means to do so is available.

    The only reason we aren’t routinely cryopreserving brains at the moment of clinical death is because we aren’t sure we can bring them back. This article is a useful data point in making that determination. As long as the brain has discernable structure, we can probably do most of the relevant repair work to that structure — whether physically with nanotech, or digitally in a simulated version of the brain.

    Some of the person’s individual memories might be lost (depending on various factors such as how redundantly they are stored), but as long as the bulk of them can be conserved and recovered in the repair process we should think of the person as having survived just as any other survivor from a traumatic brain injury.

    Regardless of the implications for cryonics, this is good news for research into neurological problems of all kinds. What we can simulate and understand on an abstract level, we can more likely find a cure for.

  6. Mia
    April 1, 2012

    This makes me wonder if the reason we think grids are “logical” and set up our cities and charts that way has to do with our inner setup… I’m curious to see how spider neurons are mapped out.

  7. Paul M. Sheldon
    April 4, 2012

    Years ago, I tried to understand engineering image processing noise as part of an evidence machine. I read Bela Julesz try to sleuth out the visual system in the brain and motivated myself for understanding what had been a mess by spending long periods of pain looking at anatomy slices and reading Latin words for things in them.

    These new techniques of seeing where the water goes and fMRI try to look at details how things connect and communicate taking time, dynamical activity, more toward understanding.

    I hope this stimulates more sleuthsike Julesz to write stimulating stories to help the community grasp the details of understanding brains.

  8. James T. Dwyer
    April 5, 2012

    Excellent article – the most complete report I’ve found of this study. If I understand (I may well not) what we are seeing in these images is not neural passageways or axons themselves, but their inference indicated by their pathways illuminated by water in a magnetic field.

    In that context, I suggest that the apparently interconnected fibers, seemingly intersecting at right angles, may be completely unconnected: the longer passages carrying signals longitudinally between different regions of the brain, while the apparently branching fibers are the passages laterally carrying signals between the right and left halves of the brain.

    Perhaps I’m wrong, but I doubt that nerve fibers carrying ‘long distance’ signals would branch off, carrying those same signals laterally. If its suggested that there is some switching of signals occurring along a crossbar type of network, I think there would have to be some addressing scheme implemented within an encoded signal – I find that difficult to imagine…

    These images appear to me to represent more the pipes carrying network wiring between buildings on a campus and between floors within a building. They may represent some general distribution structure but do not identify the information pathways much less any network mappings of distributed brain function.

    I too agree with Partha Mitra’s sentiments.

  9. Pablo Soriano
    April 8, 2012

    Unless I got this terribly wrong, this “novel” study makes no difference to the regular work and techniques used on a daily basis in neuroimgaing labs, at least the one I worked in. There’s commertial (and opensource) software for DTI statistical fiber tracking, and fiber maps are regularly built for some studies.

    Lots of fancy images, and nothing new on connectivity, though the pics will admittedly catch the eye of those who did not know about the technique

  10. Giuseppe Palombo
    April 21, 2012

    I thank very profoundly each and every one participating in any way to the furtherance of the knowledge of the brain. My hope is that soon some “brain mechanic” can fix some “faulty brain”.

    SOATTOE will second all of you.

    And thank you, Science and one and all.

Add Your Comments

All fields required.

Related Posts