What a beautiful mind you have.
March 29, 2012 9:27 PM   Subscribe

 
The brain appears to be wired more like the checkerboard streets of New York City than the curvy lanes of Columbia, Md., suggests a new brain imaging study.

Hey, I'm from Columbia, have to wonder how many people understand this reference. In high school I made an entire street map of Columbia, by hand, and at one time knew almost every street by name like a London taxi, due to the map, and work in the pizza business. Anyway, Columbia does sort of look like the twisted folds of a brain with fractal properties. It also is where a lot of people who work at the NSA and NASA and other brainy places call home.
posted by stbalbach at 9:47 PM on March 29, 2012 [5 favorites]


This grid structure appears to guide connectivity like lane markers on a highway,

Anyway, Columbia does sort of look like the twisted folds of a brain with fractal properties

Yeah, yeah. All very noble. I'm guessin' that had they peered into my skull with their machine thingy, they'd now be describing the brain's structure as being more like sink-holes in a swamp.
posted by PareidoliaticBoy at 9:57 PM on March 29, 2012


I heard this on NPR today and while this is exciting it, I suspect, as usual, it loses a large bit of its nuance when it gets to the popular press.

We have already been mapping the brain for some time and we have pretty clear images that the individual neurons are shaped more like trees and this is the first I have heard about perpendicular 2-d connections and the finding seems to be quite a simplification.

Here are some more relevant links.

Mapping the Brain, MIT press release

The Connectome

BrainMaps.org is an interactive multiresolution next-generation brain atlas that is based on over 20 million megapixels of sub-micron resolution, annotated, scanned images of serial sections of both primate and non-primate brains and that is integrated with a high-speed database for querying and retrieving data about brain structure and function over the internet.

Eyewire, crowd sourced neuron mapping, You can help!

And last but not least The associated Ted talk by Sebastian Seung
posted by psycho-alchemy at 10:08 PM on March 29, 2012 [8 favorites]


The technology used in the current study was able to see only about 25 percent of the grid structure in human brain. It was only apparent in large central circuitry, not in outlying areas where the folding obscures it. But lessons learned were incorporated into the design of the newly installed Connectom scanner, which can see 75 percent of it, according to Wedeen.

So this brain map is only showing 25% of what's there? Even knowing that there's a lot more than they show, the brain's wiring is way, way less dense and complicated than I would have guessed.
posted by jason_steakums at 10:09 PM on March 29, 2012


The brain appears to be wired more like the checkerboard streets of New York City

Other researchers describe the brain topography as more like the "Iggy-Pop-beef-jerky-sinews" of the streets of the West Village.

A neanderthal interviewed was like, "well, duh."
posted by noaccident at 10:14 PM on March 29, 2012 [1 favorite]


Boy, they sure hit the mark when they said "grid like New York"! I don't know many places where parallel streets cross each other, but surely in a city where 4th and 10th streets cross, that'll work.
posted by Goofyy at 10:19 PM on March 29, 2012


"Of course. A child could do it."
posted by Strange Interlude at 10:19 PM on March 29, 2012 [1 favorite]


The brain appears to be wired more like the checkerboard streets of New York City

Well, that's another strong argument AGAINST Intelligent Design.
posted by oneswellfoop at 10:21 PM on March 29, 2012


Interesting. Thanks for posting this.
posted by Tell Me No Lies at 10:27 PM on March 29, 2012


Ahhh-Ha I sees ya problem

POINK

SLAM

All fixed, that's be thirty bucks, come right back if ya don't think it's fixed right.
posted by mattoxic at 10:39 PM on March 29, 2012


This is very cool. Diffusion Tensor Imaging (DTI), which is the precursor to this, is pretty standard in MRI labs and happens to be what I work on at the moment. With a few click of my mouse I can see the white matter tracts that connect the different parts of the brain to each other and to the spinal cord. It kind of blows my mind.

DTI is cool but it's resolution limited. An MRI scan will have resolution on the order of 1mm x 1mm x 1mm (a box called a "voxel") and if you have tracts crossing inside a voxel you'll just get noise. Since brain circuitry is obviously below the 1mm^3 scale, DTI tractography lets us see the freeways in the brain, but not the side streets.

I'm new to the field and haven't read this paper yet, but it looks like DSI (Diffusion Spectrum Imaging) has somehow managed to break through that resolution barrier and show these really cool pictures. I'm impressed.

A bit about DTI, since I assume DSI uses a similar principle:

To put it very simply, the magnetic field in an MRI lines up the spins of the protons in your brain. A radio frequency field is applied which flips the spins to the opposite direction. When the RF stops the protons will switch back to being inline with the magnetic field and each proton releases energy as it does this. The released energy is what is seen by the MRI machine and reconstructed into a picture of your brain.

But that's just a simple scan that will map out brain anatomy and such. We can actually be a bit more clever. When we're lining up the spins of the protons in the magnetic field, we can create a magnetic field gradient. This means that the magnetic field is not the same at every point in space. The direction of the spin of each proton in the brain will therefore depend on its position in the detector -- we've location tagged every proton. But we then apply the reverse of the gradient to line everything back up in the same direction.

Now, for a static brain this gradient/anti-gradient wouldn't matter. Everything would be back and lined up how it started. But. If there is diffusion in your sample (ie water molecules randomly moving with respect to each other) you'll get protons which were aligned in one direction mixed up protons which were aligned in another direction. When the anti-gradient is applied, if a proton isn't in the same position it was when the gradient was applied it will be out of phase with nearby protons. If two nearby protons are out of phase with each other they will destructively interfere when they release their energy, leading to signal loss in areas with lots of diffusion. HUGE swaths of physics have been glossed over. Apologies to any medical physicists in the house.

Now. Nerve fibers are long and thin. This mean that water in and around the nerve fiber is going to tend to diffuse longitudinally as opposed to transversely. Say you set your gradient in the x direction, and you observe signal loss at point A. You just discovered that at point A water tends to diffuse along the x direction. If you then set the gradient along the y direction and don't observe the same signal loss at point A, you discovered that at that point water prefers to diffuse along the x direction. If you see a series of points where water prefers to diffuse in the same direction, you just discovered a tract -- a bundle of neurons connecting one part of the brain to another. You need to take gradients in at least 6 different directions to get the full 3D picture of the direction and strength of diffusion in each voxel.

"Tractography" is what these guy are doing (and doing very well) -- mapping out the tracts in the brain. But we can use this information for various interesting things. One of the things I look for is the degree of anisotropy in different parts of the brain. I don't care about where the tracts are going, just the degree to which one direction is preferred over the other directions. In a healthy brain you get high anisotropy -- lots of long fibers mean lots of strongly preferred directions. But if you've had a brain injury or potentially a condition like MS or ALS, you might lose some of those tracts and we see that by observing a lower anisotropy than in a healthy brain. So this stuff isn't just for pretty pictures -- it's very helpful in diagnostics as well.

I hope someone can chime in about how this new technique works in more detail.
posted by no regrets, coyote at 11:22 PM on March 29, 2012 [32 favorites]


Amazing. It was clear that there were going to be some discoveries in this field, but I think a regular grid is about the last thing anyone could have predicted.
posted by Segundus at 1:04 AM on March 30, 2012


This explains how my brain is a sieve, and the uniformly large sizes required to be enmeshed in there.
posted by StickyCarpet at 1:14 AM on March 30, 2012 [1 favorite]


Yah, this is pretty damn cool. I'm not in imaging myself, so I can't really evaluate how much of this is flash vs substance, but this strikes me as a pretty big result.

Just to put this in context (although I am not neuroanatomist!), neuroscientists generally break the brain down into grey and white matter, where gray matter is primarily cell bodies (think processors) and white matter is primarily axons (think connectors/communicators). Now if we just focus on the cortex (the part that we generally associate with intelligence as opposed to low level motor control and hormones, etc) Gray matter actually has a surprisingly simple structure. The processing part of your brain exists essentially as a sheet covering your brain, six neuron layers think everywhere*. Yes, it's basically a simple 2d structure.

Now when non adjacent parts of your brain want to get talking, they throw out these long axons which go through the brain to connection to just about anywhere. Since the shortest path between two points is a straight line, we might expect that the axons would all just b line it to whatever region, which would result in the white matter of your brain being an incomprehensible tangle.

But it turns out it's not. The nerves (bundles of axons) which extend between different parts of your brain seem to always cross orthogonally - at right angles. This is a very interesting finding. At the end of the Science paper they speculate about reasons - simplifying axon path finding, helping to regularize timing issues - but this is an imaging paper, not a theoretical one. They've found a structure which is really cool, and now it's open to speculate about why it's so. It's what science is all about really.

* And of course actually connections are really just thrown out to anywhere. A lot of neuroscientists are using graph theory to find various relays and hubs in the structure of the brain. I think these days if you study neuroscience you'll still tend to get a big mess of biology thrown out you, but at the moment big strides are being made in breaking down the functional architecture of the brain. In twenty years introduction to neuroscience may involve as much graph theory as biology, and that's so awesome. But that's also me with my A.I. hat on, so don't place any bets on that claim.
posted by Alex404 at 2:55 AM on March 30, 2012 [3 favorites]


I used to be a proponent of DTI. But now, when people post to Ask MetaFilter asking for help with their brain structure relationship issues, I think I'm just going to respond: DSIA.
posted by knile at 2:58 AM on March 30, 2012


Umm, should have checked that over better:

- Six layers think = six layers thick.
- Asterisk should have been at the beginning of the next paragraph.
- Asterisk point at the end should be, connections *aren't* really thrown out anywhere.
posted by Alex404 at 2:59 AM on March 30, 2012


Neural pathways that get used a lot tend to get built up more.

Neural pathways that provide access to many other neural pathways have a better chance of getting used more.

Neural pathways that cross many other neural pathways are better able to provide access thereof.

It makes intuitive sense, really. If nobody guessed it, it's because biology is generally counterintuitive.
posted by LogicalDash at 3:39 AM on March 30, 2012


And this is your brain on 'shrooms. Any questions?
posted by sfts2 at 4:27 AM on March 30, 2012


I'm just surprised this hasn't been found out before. Is there a reason this kind of structure can't be seen if you examine a brain under a regular microscope?
posted by ymgve at 4:27 AM on March 30, 2012


I'm just surprised this hasn't been found out before.

i) There are references in the article to people suspecting that this may be the case, but this is the first result which demonstrates it on this scale.

ii) The nerves can stretch from end to end in the brain. They're fine, but long.

iii) Invasive techniques (e.g. Microscopy based) won't get you this big picture, because either you're looking through a hole in the head into a very local region, or your looking at some brain slice. And if you think about it, the chances of a 2d brain slice capturing a 2d grid structure are very small. Especially since these grids aren't necessarily flat.

iv) This is an extremely fine structure. Traditional MRI techniques have a resolution of 1mm cubed, which isn't sufficient to see anything but the largest pathways. You certainly couldn't resolve these orthogonal sheets.

Therefore, this observation had to wait for high resolution non invasive techniques. Which is what they used.
posted by Alex404 at 5:08 AM on March 30, 2012 [1 favorite]


Thanks, no regrets, coyote, for the primer on DTI. If I understand correctly, the trick with DSI is that diffusion tensors are related via a Fourier transform to fiber orientation. There is apparently an additional shortcut to take with so-called q-ball imaging.

I'd also like to point out that Wedeen is a leading researcher in diffusion imaging techniques, but he is only one of several. The area is quite young, and many questions in tractography (which is what's shown here) are very contentious. Objects like the fiber-crossings he discusses are difficult to image and data from them is highly ambiguous. Although the patterns Wedeen has found are an interesting finding, they are still unconfirmed. Although the graphic you're seeing in the linked article looks very concrete and easy to grasp, it started with very abstract data that are difficult to visualize.
posted by Nomyte at 6:16 AM on March 30, 2012


What the hell is up with the comments on that site? Like, for example, this one:
"The grid structure encodes dualities to model paradoxes, which are central to a task-driven system. These dualities are the atoms of memes.

Employing triangular numbers highlights how individual paradox units cooperate in a harmonic way, as birds form a functional flock gestalt."


I'd suspect some astroturfing bot but there's no spammy link so what's the point of throwing in a totally nonsensical statement?
posted by caution live frogs at 7:59 AM on March 30, 2012 [2 favorites]


I remember something Jeff Hawkins said in his TED talk years ago, paraphrasing, that people always say the brain is incredibly complex, more complex than the universe, blah blah, but that's just nonsense. We have a lot of data on the brain but we don't have any theories and we don't understand the data, so it seems more complex than it is

Now it looks like we're finally focusing our observations enough to get clear, simple detail. This seems like an incredibly important result for AI research and trying to understand consciousness. Basically a lot of the apparent complexity of the brain is just a simple structure being squished around in your skull. The structure of connections between cells are actually very simple
posted by crayz at 8:15 AM on March 30, 2012


The brain is like a telephone switchboard...
posted by Crabby Appleton at 9:43 AM on March 30, 2012


I don't know many places where parallel streets cross each other, but surely in a city where 4th and 10th streets cross, that'll work.

In San Francisco, 16th St. crosses 7th St.; and 3rd St. runs parallel to 4th, but crosses 20th-26th Sts. It's still a grid.
posted by mrgrimm at 9:53 AM on March 30, 2012


In any city where the grid is deformed by trying to make it follow a natural feature like the shore or bank of a body of water, you'll get anomalies. You don't even want to know about New Orleans.
posted by localroger at 10:58 AM on March 30, 2012


This reminds me of the simultaneous elation and deflation that accompanied the sequencing of the human genome: it too was vastly smaller and less complex than thought, which was interesting, but it meant that people had to stop thinking of themselves as their genes. So, too, there are all kinds of hints that people have been identifying far too much with their brains.

Related.
posted by stonepharisee at 12:35 PM on March 30, 2012


« Older something about bells, balls and bulls   |   Manuscript Found On The Coast Of Yucatan Newer »


This thread has been archived and is closed to new comments