Jeff Lichtman: Connectomics: Mapping the Brain | Harvard Department of Physics

I uh I'm Melissa Franklin I teach physics here and um uh we have a new lecture series uh once a month Wednesdays uh public lectures from the Harvard faculty Harvard including um science faculty faculty from the medical school that's it uh and anyhow I'm really really glad you came it's a a terrible day it's windy and rainy and horrible and it's a wonderful turnout for that kind of day um so I just want you to know that the reason we had you sign in was just so we could figure out who you are and maybe

you want to come again to different lectures in this series which hopefully will go for the next 30 or 40 years um until we're all no longer able to come anyhow I want to introduce uh Professor Jeff lickman uh he's gonna give a fantastic talk I've seen it I've seen it anyway he uh I just wanted to tell you a little bit about him um he was younger and then he got older when he got older he moved he went to Bowden College uh that's in in a in a state north of here um and

then after he graduated from there he went to Washington University which is in uh St Louis where he did an MD PhD um and the PHD was in something like neuros science then he liked it so much in St Louis uh that he stayed there for 30 years um then when his hair turned white he came to Harvard actually I don't know that uh that's it's true uh where he where he's uh started up this great um uh um Neuroscience initiative uh and he's going to talk about I think he's interested in how information gets

stored in the brain and a lot of people are interested in that and and then what happens to it after it gets stored and then what happens to your brain after the information is there uh the thing I find incredibly fascinating about this field is that uh nobody knows which approach to trying to figure it out is going to work so it's really really exciting um and so Jeff is going to talk about his particular approach and here he is so thank you very much for braving the elements tonight uh I consider myself a very

dedicated scientist but I wouldn't have come so I am impressed and grateful uh that there's more my wife is here but I'm grateful that everyone else is here as well in fact I'm not even sure she feels was appropriate to come given the weather um I am a neuros scien scientist and uh I've been one for most of my adult life uh and because of that the language I speak uh has words in it that uh may be unfamiliar to you and and I I strongly suggest that if I say something and you you just

don't understand what I'm saying just shout out a a question uh interrupt me uh so that I make sure I'm not saying something that you can't can't follow uh neuroscientists of course study the nervous system um and I think you're all aware just that the if you read the newspapers or listen to the State of the Union Address or know what's going on in Europe uh there's a huge amount of interest uh in the brain right now and uh it's a fair question why uh the nervous system gets so much attention you know there's like

uh three or four departments in every Medical School associated with the brain psychiatry neurology there's usually a neuroscience Department uh for the medical students there's just a really remarkable amount of interest in it but there's not like a kidney department or a lung department or a liver department and uh it's worth considering why why this might be um a special organ system and and I want to begin by saying that uh all organs all parts of the body have the same general uh relationship between a physical organ a structure and what the organ does its

function and and this is a theme uh that biologists are very interested in the the relationship between the physical structure of something and its function uh in fact the state-ofthe-art uh in biology right now is is a field called structural biology which sounds like a very broad field but it's actually quite a narrow field it's a field that focuses in on the atomic structure where the atoms are in single molecules uh to explain how that molecule does its function so molecules like enzymes that tear up other molecules or molecules that bind to DNA uh are

molecules where their physical structure uh is analyzed at the level of single atoms to understand how that structure gives rise to function and this is where uh a lot of state-of-the-art biology is taking place of course there was a time when uh humans were interested in the relation between structure and function but it was long before we knew anything about atoms uh I think the first time people began thinking about this is probably when they opened up the carcasses of animals they were going to eat or uh looked inside the carcasses of human beings and

notice that there are all these aggregations of tissue that had uh kind of a familiar look because the same aggregation could be found in individual after individual even in different species uh all vertebrates have a liver all vertebrates have lungs uh all vertebrates on the land at least have lungs and um there was a time when uh you could see these organs but what their structure was was not so well understood and certainly their function was not understood either there was at at the Early Times probably the uh first organ that kind of got a

structure function explanation that is the structure explained the function was the heart where without anything but a naked eye looking at a frog heart beating or perhaps the heart of another animal beating near at at the time of death last few beats you could tell that this was a pumping organ that was moving blood around the body but but the liver or the kidneys it wasn't so easy to to relate the structure of those organs to their function uh and the big breakthrough uh in in understanding the relation between structure and function was the invention

of microscopes because with microscopes one could see for the first time and this is hundreds of years ago now that each of the organs had its own cellular Motif that is the cells were put together in a particular way that was so standard that once you saw that cellular Motif and understood it you could find it again and again in any animal so if I know what the liver of a human being looks like I also know what the liver of a of a zebra looks like uh or even a frog they have a similar

relationship of nerve cell of cells not nerve cells and in uh each organ there was a special cellular Motif in the kidneys which filter U the blood there is this complicated tubul system and once you understand that tubul system it's kind of easy to uh infer what is going on in kidneys uh as a filtering system in the lungs there's this uh s uh vessels that carry air that get very close to vessels that carry blood and that allows gas exchange between the outside atmosphere and the lungs uh so for each of these organ systems

we've made tremendous progress in understanding the cellular motifs that underly function and with that came a huge bonus which is that for most of these organs uh all or of these organs ex except for one which I'll talk about in a moment not only did you understand the normal structure function relationship but you discovered that diseases of those organs almost always had an abnormality in the cellular Motif there was something wrong with the cellular Motif so when people's livers aren't working well if you look at a little biopsy of liver you will find it's either

a cerotic liver or another kind of disease of the liver uh in other disease like hepatitis inflammation of the liver in kidney there are many diseases of kidney each of them have their own particular appearance of a cellular Motif and so we've not only learned the normal structure function but for diseases of these organs we have an underpinning of what's wrong and that leads to ways of thinking about therapy now let's think about the brain where all of what I've just told you is not so true the brain has lots of diseases and these diseases

believe it or not many of them are incurable not simply because they're hard to cure but because we have no idea what's wrong we don't know what's wrong because the underpinning of the cellular structure of the brain is not well understood at all this is a tremendous challenge uh for human beings and it means that the nervous system is for some reason much harder to understand than any of the other organs and I'd like to start uh this conversation with you by going through some of the reasons why the nervous system compared uh to all

the other organ systems in the body why the relationship between the structure and the function of the nervous system is much more complicated I'm doing this because I want to justify what will seem to you a kind of crazy goal which is to actually get the structure of the brain at a level that we could understand things that seem like they're worth understanding but once you start seeing what we have to do to get that structure you're going to say this is crazy crazy and so I'm I have to justify it I think by giving

you a sense of why the nervous system is such a special case the first thing about the nervous system that is different from other organ systems is that the cellular motifs are insanely complicated in part because the number of different kinds of cells that make up the brain far exceed the total number of different kinds of cells in the rest of the Body for example there is a state-of-the-art question right now about how many kinds of cells there are in the retina not how many cells total but how many different kinds of cells make up

the retina which is part of the nervous system and as nervous system goes it's perhaps the most organized part of the nervous system everything is laid out in nice layers everything is distributed in a nice tiled way and yet it is a a research question of many of my colleagues to count how many different kinds of cells there are and this uh huge structural diversity I'm going to give you a sense of of just how many types of cells there are in the retina uh in the next slide that generates a not only a large

number of cells that you have to sort of figure out how they're connected to each other but because of that what the nervous system can do is much greater than what other organs can do you can imagine you could write a chapter you couldn't but most some physiologists could probably write a chapter and tell you everything that a kidney does it may be a 100 Pages or so it wouldn't tell you every disease of the kidney but the normal function of a kidney could probably be described in in 100 maybe 150 Pages how big a

book would you have to write if you wanted to say everything a human brain could do could you write a book like that you couldn't it in part because every every week uh we're doing things with our brains that we're never done before my children are a good example of this they're using their brains to do things with tools that I can't use the you know the the brains are constantly kind of evolving especially human brains to do new tasks and this is partly related to the fact that the structure of the brain is much

more complicated so so let's start by looking at the kinds of cells in the retina so this is a a managerie if you will of about 50 different kinds of cells in the retina that everyone accepts is there and and they fall into uh photo receptors there are some that respond best to green light some best to red light some best to Blue Light so there are three kinds of photo receptors then there's a kind of cell called horizontal cells and the structure of these cells show you that there are different kinds some that are

sort of symmetrical and some that are very asymmetrical the cell bodies of these cells are these little swellings here then there's another kind of cell called bipolar cells and you can see by where the cell body is relative to these branches there's many different varieties of bipolar cells it's not a Continuum each of these is a class and this is the number of classes that people have seen then there's a kind of cell called Amrine and look at those they're just a wide variety this is all in one retina and then gangan cells which is

the cells of the retina that go into the brain and send visual information into the thalmus there's a wide range of those as well so here's 50 cell types and No One Believes that's the total number everyone believes there will be more cell types that some of these can be broken into two 3 four maybe 10 categories someday we will know how many cells there are in the retina but that's not going to help us one bit with the cerebral cortex where it has its own complete managerie of cells or the cerebellum or the brain

stem or the amygdala or the spinal cord so this is a big problem about getting the cellular motifs and getting the structure that underlies function so that's problem number one problem number two is that the brain is very different from other organs in another very interesting way some of you probably know people who have donated a kidney to let other people live why does a person donate a kidney because they can without killing themselves because if you've got one kidney you've got kidney enough in fact you don't even need one kidney you can do with

even less than a whole kidney you can do with one lung you can lose a lung you can lose part of your liver doesn't matter you can survive that fine but if I take half of your brain any half no matter how you want to cut it you're going to notice and that is because what is the difference between the brain and these other organs that if you look at a kidney you find it's very boring once you see that cellular Motif it's iterated hundreds of thousands of times over and over and over again one

part of the kidney does the same thing as another part the lung is made up of these alvioli that come up right close to blood vessels one part of the lung does the same thing as another part you lose part of your lung you still can breathe but the brain is different because it's organized at many different size scales there's importantly different information at every single part of the brain and I want to just go through that with you by sort of zooming in from the most macroscopic picture of the brain down to the most

nanoscopic or microscopic picture so we'll start with a brain uh this is is a human brain in cross-section this is the back of the brain and this is the front of the brain and this is about 10 cm maybe 20 in 25 in 2T uh about 10 cm wide here uh and even at this level we know that the brain has many many different parts for example if you're trying to plan uh for what you're going to do to get back home after this lecture and you're thinking about that rather than about what I'm saying

your frontal cortex is being involved uh as you're looking at what I'm showing you you here we're activating the occipital lobe where vision is when you get up so you don't fall over your cerebellum is working on balance each of these parts has specific roles to play in your function you lose the front part of the brain uh you may see fine but you can't plan for the future you lose your cerebellum as an adult you can't balance and the brain stem uh where breathing takes place is also essential obviously the spinal cord where motor

activities are so at this gross level already the brain is divided into very different functions and this is of course a minuscule portion of the thousands and thousands of things brains do if we go down an order of magnitude from 10 cm to a little piece of brain now where we're looking maybe at about a centimeter of brain you find that the cerebral cortex itself which is just shown here is divided into an outer part where the nerve cells are and an inner part where the cables that connect nerve cells from one part of the

brain to another part are the in inner part is called the white matter made pink here just to confuse you and the grayish white outer part is the gray matter and so even in in this little area there's nothing uniform about it the outer part of the brain has a different function than the inner part if we zoom in on the outer part the cerebral CeX as it's called and go now from a centimeter uh down to a millimeter so this is like say a millimeter by a millimeter and then label a single nerve cell

and there are many many there's hundreds of billions of nerve cells in your brain but if we look at one nerve cell in a region like that this is what it looks like these cells are crazy they're just way way more complicated uh than the cells anywhere else in the body the cell has a cell body like cells elsewhere but these cells have huge numbers of what you might think of as antennas these are their dendrites that collect information from other cells and the fact that this cell is distributed over more than a millimeter of

length that's just for its reception of information from other cells but this cell also has a process sticking out of it called an axon which then sends this cells information to other cells and that axine can spread its branches all the way to the other Hemisphere or down the spinal cord or to the thalmus it can go centimeters so this one cell covers huge areas of the brain and every brain cell has a similar kind of distributed Network and this allows cells to receive and send information to a vast number of other cells if we

zoom up from a millimeter to a tenth of a millimeter which would be 100 microns and just look at the dendrites of these cells we find that the dendrites themselves are divided into two categories and that is if you zoom in on one dendrite you find that the dendrite has these little spines sticking out of it and then there are regions in between where there are no spines and it turns out that the spines are the place where cells receive information via synapses from neurons that are trying to excite this cell and make the cell

talk to other cells and when there are other kinds of inputs in the brain that try to shut this cell off called inhibitory inputs and those are mostly connections made on the shafts the non spiny part of dendrites so there's this other div division of labor that's very important that you see when you start looking at a tenth of a millimeter now if we want to zoom in on one dendritic spine and go down to 10 microns we see that again there is a remarkable amount of interesting structure so let's look at that region there

and to see that we have to now go to an electron microscope we're now looking at things where the resolution limits of light microscopes get in the way the defraction limit of light makes it difficult to resolve all the information so this region here I'll just color it in for you this is one of these dendritic spines there's a dendrite and it has a little process coming out and a little swollen end and it has this little thing in here you probably wouldn't have noticed if I didn't point it out called the spine apparatus which

is in there and it's an important part of the spine so that's the spine that's receiving information from an axon from another cell and the axon is shown here in cross-section this is an electron micrograph so this is just cut sliced through a thin piece of brain and here this thing is an axon and it's filled with these little circles and those circles are synaptic vesicles and each of those vesicles is filled with neurotransmitter and when the neuron that this axon came from is excited it dumps the chemical neurotransmitter in these vesicles right against the

membrane of the dendrite right here at the synaptic connection and so this is a synapse in the brain this is where all the communication between one cell and another takes place there are about a trillion or more synapses in the brain so there's many of them and you can see that they're not bidirectional they're unidirectional most of them the axon sends information to the dendrite so here is one little place where one neuron is making a connection with another neuron to communicate some information to it and the brain is just filled with that and this

is one of the challenges of of Neuroscience is to understand what all these synapses are doing now I want to give you one sense of why this is such an extraordinary problem if we look at this kind of resolution up at the brain level there are tools like functional magnetic resonance imaging some of you may have heard of this it's a technique that measures local blood flow in the brain to see what parts of the brain are working when you do particular tasks fmri is done with a resolution of about a cubic millimeter that is

the smallest spot you can resolve with fmri is about one cubic millimeter a millimeter on each side a little box that's a cubic millimeter on a side and we have learned very important information about the brain using this technique because as I said the brain is distributed so functions are distributed to different places this is a very powerful approach and a lot of people when they see these images say oh that explains how the brain works but let's think about the resolution of an electron micrograph it is its voxal size is 100 cubic nanometers nanometers

are a thousandth of a micron a single nanometer and this box the voxels the resolution of this image is a trillion fold smaller than the resolution of this image and if you want to understand the brain you've got to understand details at the level of the electron microscope at the level of the light microscope all the way up to the level of the fmri that's a trillion fold range and no matter how you slice it that is a big number and that's just to get the information about how the brain is organized you have to

deal with a trillion fold range in resolutions so those are the easy problems s yeah there's a harder problem and and this is the one that I'm most interested in and it is related to something that I'm sure most of you know uh which is that all structure that we think about when we think about bodies and organs are are built based on a genetic blueprint your genes build the structure and the structure generates the function that's how kidneys livers lungs all the organs work with one notable exception again and it's the brain and and

let me give you an example of why this is not entirely this whole story and why it's so hard to understand how brains work and for this I need to know is there anyone here who learned to ride a bicycle or tried to learn to ride a bicycle as an adult who never had had experience with bicycle riding as a child anyone here who who has tried to ride a bicycle as an adult for the first time I suspect there is someone but they don't want to admit it because it's not a pretty story they

would be telling I have a neighbor uh who grew up in another country and she moved to Cambridge number of years ago started a family has two daughters uh one a couple summers ago one was like nine and the other 11 and they were beginning to ride their bicycles around our street which is pretty quiet and uh the mother did not ride a bicycle but I think either she was jealous because they seem to be having so much fun or it's just amazing how quickly a kid can get out of the view of a parent

on a bicycle uh for one reason or another she decided she was going to learn to ride a bicycle so a couple of summers ago uh she decided to ride a bicycle as an adult and I watched this I didn't watch all the time had a job but when I was home I paid attention and and watched and it was very interesting as far as I can tell uh this woman is neurologically normal in all respects except when she gets on a bicycle where she looks like she doesn't have a cerebellum she just cannot balance

the thing you know her arms are going backwards and forwards uh and her kids are you know literally writing circles around her and making fun of her it's very embarrassing I'm sure and and for reasons um I think related to the incentive of wanting to be with her kids doing this she she worked on it all summer but I noticed as the summer went on she was doing it less and less and by the end of the summer I didn't see her on a bicycle and I haven't seen her on a bicycle since so that

says her kids learned how to ride a bicycle and she didn't but let let me tell you another story which is about me and I like most of you learned to ride a bicycle as a child and then we moved uh to St Louis where I just didn't have opportunity to use a bicycle and so I I walked I I rode that car a lot and nothing about St Louis there's lots of people ride bicycles there but I didn't ride a bicycle and then in 2004 uh we moved back uh East uh to Cambridge and

in 2005 my wife uh told me I was growing I was very excited but she say not this way you're growing this way and maybe uh you know it's time for you I think she said we we should we should exercise she meant me but and uh we decided bicycles or she decided bicycles would be a good idea so we went to a bicycle store I hadn't been on a bicycle in a very long time and uh you know how it is that bicycle they give you a bicycle to try out where you go outside

and you take this road trip uh only a block long and they watch you very carefully and so you know if you're a nervous person it's a little harrowing anyway to have somebody watch you ride a bicycle for the first time and I hadn't ridden a bicycle in a very long time and I got on the bicycle and and sure enough I was having trouble with balance but it was only for about 8 seconds and then suddenly it all came back and I was riding perfectly fine now what is the difference between me and the

woman uh who cannot ride a bicycle as an adult it's not that I'd been practicing bicycle riding for uh years I I there were many years a decade or so or more when I wasn't riding a bicycle at all and yet somehow it was still there my bicycle riding and and so the question I would ask you is what is the difference between the brain of a person who can write about bicycle and the brain of a person who can't what do I have that she doesn't have or is she got something that I don't

have that by getting rid of it I can now ride a bicycle nobody really knows this is a fundamental question but to bring it back to this point I don't have a gene for bicycle riding I didn't get bicycle riding from genes where did I get the bicycle riding from I got the bicycle riding from practice from the use of my nervous system rehearsal and practice I've changed some the structure of the brain and that structure then has given me a new function this is a loop biologists hate these things we like these linear things

because you knock out a gene it changes the structure changes the function but in a situation like this cause and effect are very confusing and we don't like this but this is true this is H happening all of you if you can button your shirt if you can speak a language if you can ride a bicycle your brain has stably changed itself based on experience and what is the form that experience has taken in your brain is a deep question but because it can last for such long periods of time even without rehearsal one argument

might be it's related somehow uh to the structure of the brain which is the interconnections of nerve cells so if you want to understand something like that you're going to have to map out the connections at some deep level I don't think like a brain of somebody who rides their a bicycle it has a big pathway connecting two parts of the brain that someone who doesn't ride a bicycle has it's probably very subtle and to map out these uh subtle differences one would need to map all the connections humans have a penchant for mapping things

and the big map we've made uh for humans is the Human Genome Project where with genomics we've mapped out our DNA yes I I would say it this way that it's not that an adult can't learn new things uh at least I hope that's not the case although my children would argue with me about this they seem to think I don't learn as much I'm not as adaptable as them it's certainly harder for adults to learn than children and if you want to learn a new langage language as an adult without an accent it's almost

impossible where for a child they don't even have to work at it at all it's natural so that our ability to learn difficult tasks becomes more and more limited as adults I know you don't want to hear this people my age hate hearing this we like to think we're plastic our brains are constantly changing but what we're learning as adults is a minuscule fraction of what we learned as children and we it's much harder for us to learn new things you know my kids don't ask me too many questions anymore in part because they say

there's no point because they know exactly what my answers will be and and that's a very depressing thing to hear but I remember thinking the same about my parents that there was a point when I it was just no point in telling them or maybe telling them but asking them they would almost always react the same way and so I I wouldn't want to make it this black and white that adults can't learn anything obviously I read the newspaper and I feel like I know something today that I didn't know yesterday yesterday but I think

uh I filter everything I learned through what I already expect the world to be like it's very hard to make an adult Republican into a Democrat you would agree with that or vice versa you agree with it no okay so um if we want to map out uh these connections we would need an omix like genomics but it wouldn't be genomics it would be conomics this is new term probably for most of you I brought the dictionary definition and this word here is just how you would make a function into a structure you'd make a

physical engram of it so I made up this word intizam your structure your functions into physical structure this is a deep mystery what does bicycle riding look like what does your grandmother your memory of your grandmother look like in your brain nobody knows but I think if one wants to know that one's going to have to map the connections at the level of individual connections and that would be connectomics so here is the dictionary definition of connectomics or connectomics it's a plural noun but singular in construction like economics is something this is from Mar Webster's

underbridge Dictionary 2019 it's a branch of biotechnology when I first made this it was 200 14 and I've keep having to push this number up it's kind depressing a branch of biotechnology important to see that this is biology but it's technology you can't just say oh I'll map that on you know I'll come in and map it you have to do this in an industrial way so it's really a technology just like the human genome was done by a bunch of machines concerned with applying the techniques of computer assisted image acquisition computers are essential for

this process uh and uh analysis of the structural mapping of sets of neural circuits or to the complete uh either to map some neural circuits or perhaps even map the complete nervous system of an organism using high-speed methods you've got to do this faster you'll never finish as you'll see and organizing the results in databases and by databases I don't mean you know Excel spreadsheets this is uh really a different kind of data this is you know very large data big data Maybe and with applications of the data as in neurology and Psychiatry so I

I began by saying there are all these diseases of the nervous system that don't not only don't we have a cure we don't even know what's wrong and that's because they're diseases where we don't have a physical underpinnings of many psychiatric diseases in adults um and some neurological diseases where we M mainly see abnormal behavior or a patient complains of like migraine headaches but there's no blood tests there's nothing you can do to confirm that other than looking at the patient there's no physical instantiation not that there isn't it's just we can't find it until

we have I think a deeper understanding of the brain so you might say there are pathologies of the wiring diagram connect opathy uh that one would like to get at and and that's going to require uh connectomics if one wants to get the proximate cause of those diseases and fundamental Neuroscience questions such as what does bicycle riding look like uh and and it maybe also the word connectone would be like a genome the full wiring diagram in the brain so that is the uh goal and I'm going to tell you just two approaches we've taken

the first uh was to try to do the nervous system wiring diagram with a light microscope and this is uh to take advantage of the idea that if every nerve cell were a different color and the whole brain were lit up maybe you could get the wiring diagram just by looking at the color of cells connected to each other this may seem fanciful but uh some of you may know that thanks to uh a jellyfish that Biol lumines uh a green color because of a protein in it that's fluorescent that is you shine blue light

on it and it fluoresces green this green fluorescent protein gave rise to a revolution in the use of fluorescent proteins to understand things and and one of the things that came out of the discovery of the green fluorescent protein which got a Nobel Prize in chemistry a few years back uh was that then it was clear that there were certain corals that were red fluorescent proteins were in them and other carals that had blue fluorescent proteins and you know when people put black lights on their aquarium and get these beautiful carals to to Glow this

is because of the fluorescent proteins in these animals so there are red fluorescent proteins green fluorescent proteins and blue fluorescent proteins and you may know that if there's a protein that's fluorescent there's a gene that makes that protein and once these fluorescent proteins were their structure was understood it was not long before clones of the genes for these proteins were found so we now have genes for red fluorescent proteins green fluorescent proteins and blue fluorescent proteins so we have a gene for a green a red and a blue fluorescent protein and that is really all

you need to get all the colors a human being can see because we only have three kind of photo receptors in our eye one for red one for green one for blue so if you could make every cell have a randomly different amount of red green and blue relative to every other cell each cell would have its own Hue you know this is the way color television works there are only three colors in this projector uh but all the colors you see are mixtures of the amount of red green and blue that are coming out

here and we filter it all through our own eyes so RGB is all you really need so I won't tell you the molecular trick we did but we built these animals we call them rainbow uh animals that have lots of colors in them I'll just give you a sense of what these what parts of the brain look like these are nerve fibers cut in cross-section these are ones running along the plane of focus these are the ones in cross-section you can see there are lots of colors this is in the auditory pathway and here's a

zoom up of a another region of that you very pretty uh they almost look like paintings of some sort uh if we zoom up on that little region you can see the individual fibers these are axons that are very big that are taking auditory information from your ear and sending it into your brain stem and because they're big the information travels very quickly and that's an essential part of hearing and then if you focus up and down on a little data set like this you just focus in and out you see this kind of remarkable

fact that if you look at any one of these objects you can kind of follow it uh as you're as you're going up and down these things are not really moving in the sense that you're just looking at one slice after another with a laser scanning con focal microscope a particular kind of microscope but this allows you to trace out branches over a long distances some of them are running into the plane and out and others are running across the plane so this is in the brain stem where the axons as these little branches are

all axons are very big uh in other parts of the brain like the cerebellum a very special kind of cell called the Peri cell a very large cell again they all come in different colors and if we view this uh by focusing up and down from this direction turning this on its side and focusing up and down you see these kind of remarkable cellular details of how individual cells dendrites these each color is a different cell are running parallel to each other but they interfere with each other slightly and I don't this looks to me

like Fabric or something I don't really know exactly what it is that it looks like but it it's pretty uh to look at and here one more example is this may look like a butterfly but this is actually the spinal cord and these are the neurons in the spinal cord that send uh information to your muscle fibers to cause them to contract they're called motor neurons spinal motor neurons and this particular line of brainbow mice these are genetically these animals inherit these colors so that each animal will have the same kind of color uh distribution

from one generation to the next these cells are then connected to muscle fibers so if you look at the nerves coming out of the spinal cord going to a muscle uh you find that the axons coming out again are all different colors and this uh was very gratifying to see because it meant we could follow them long distances and and but it's important to realize we did not invent this idea this was invented uh by people trying to figure out how computers work they they put colored wires inside computer hookups so you could trace a

wire from one place to another and that's basically the same thing uh that was done here here's just one more picture of these wires so over very long instances you can follow each individual wire going to the periphery if you look at these axons where they get to muscle fibers you can see the synapses and this again is a three-dimensional data set but now rather than focusing through I'm just spinning it for you and what you are looking at here take a moment to get used to seeing this is that these axons end in these

pretzel shaped objects that appear to be clasping invisible cylinders the invisible cylinders would be running up and down those are the muscle fibers which are not labeled in these brain bow mice but these are the neuromuscular Junctions where nerve cells connect uh to muscle fibers and the color tells you the origin which particular nerve cell gave rise to each of them there's no significance to the color except that when the color is the same those two neuromuscular Junctions came from the same neuron in the spinal cord which could be a centimeter away but I don't

have to trace it all the way back because the colors of the same and I know equally certainly that this is a different neuron and that's a different neuron and that's a different neuron because the color is different in each case using techniques like this we can get the whole connectome of a little piece of an animal like the muscle we can get every wire connected to a muscle so this is a very small part of a muscle but this is a a slightly larger muscle it's the muscle that Wiggles the ear of a mouse

it's called the intercalaris muscle it has only 15 different axons that come into the muscle and what we just looked at in the previous picture was an area about this big so these are the individual neuromuscular Junctions the muscle fibers are running up and down like this and this is one animal where we've analyzed the branching of every single axon in this muscle that's just a joke up there but this is the wiring diagram of that muscle and the color is each representing a different nerve cell there are about 15 axons in here and this

immediately allowed us to do something something that we could never do before which was ask is this wiring diagram stereotyped animals use the muscle the same way with the left ear and the right ear there's no earedness in in in mice which is what animal we're using here the left ear does the same thing as the right ear the nerve comes in in a mirror symmetric way into both muscles the muscles are bilaterally symmetric they're pulling the ear back towards the midline so the question is is this wiring diagram the same in the left side

and the right side and if it's different is the left side of one animal look like the left side wiring diagram of another animal is there any stereotypy so I ask you are we going to find a stereotyped wiring diagram here anyone want to guess same obviously same genes on both sides and I should tell you if you did this kind of experiment in an insect you would almost certainly find a highly stereotype branching pattern so how about anamal are wiring diagrams going to be stereotyped no anybody say yes you say yes yeah well you're

wrong every instantiation was unique every wiring diagram we looked at the every left was different from every left every left was different from every right everyone was different despite the fact that genes are the same and the left and right side somehow the wiring here has been unfettered from the tyranny of the genes and and I like to put it in that positive way that somehow genes tyrannize nervous systems they make nervous systems do the same thing from animal to animal except for things like mammals where a lot of the wiring diagram is kind of

figured out on the Fly it's figured out as the animal is developing and therefore there's no genetic regulation of exactly where individual branches go some evidence to support this is that this wiring diagram in muscle is extraordinarily inefficient icient it has lots of suboptimality in it LO useless Loops premature Branch I'll give you a quick example of this this is the a wiring diagram julu uh was a graduate student when he did this and and he's labeled one of the axons here a different color so you can see it clearly I'm just going to focus

in on this box right here this is the nerve coming into the muscle and this blue are all the axons running this way there are two axons however that take this left branch one of them goes both ways this blue one here is a branch of an axon that goes both ways but this red axon in this particular muscle this this particular animal's muscle decides only to go the left branch and you know I I am I'm not judgmental who am I to judge anyway this seems like okay if it wants to do that it

can but look what it does it goes all the way up here and then it makes a hairpin turn now it's going backwards in the wrong direction if it goes any further it's it's going to leave the muscle then it makes a hair pin turn and goes out again and especially notice this Branch here which generates this synapse here there was an easy way to get there just there but it's not there so and here's another example this neur muscular Junction here is due to this Branch Point here so it seems like it should Branch

here and just go that way but it doesn't it branches here and then it co- faules with itself all the way up here then it peels off crosses over itself to inate that Junction and this is one example of things we saw in every animal everyone was different everyone had its own particular peculiarities but they all had peculiarities like this and this suggests there's something unfinished about our nervous system relative to the nervous system of animals that have been around much longer than mammals uh where things are maybe more perfect uh less there's less suboptimality

but of course the nervous system always does the same thing they're not going to learn new things and so an interesting question is why is this variability exist and the answer and I'm I'm I'm not going to spend any time on this at all just a quick moment is to tell you it's the main research of my lab but I I don't have time to talk about it tonight is that this wiring diagram one sees in the adult is the product of a developmental period at the time animals are learning to walk and humans are

learning to ride bicycles where the right wiring diagram in muscles and probably elsewhere in the nervous system is far different than the adult wiring diagram and different in the weirdest way possible it's not that there's less connectivity it's that in these young wiring diagrams everything is connected to everything every nerve cell has made branches almost everywhere so here is a sort of cartoon of this that the neurons in the spinal cord which end up each inating a separate neuromuscular Junction with a single branch in babies every single neuromuscular Junction has got lots of ination from

every neuron and I I'm only showing four neurons here but there could be 15 or even more neurons and and hundreds of muscle fibers and you have this kind of all toall connectivity everybody is talking to everybody and then as development proceeds and in a mouse this is only a couple of weeks the vast majority of these branches permanently disappear and you end up with this simple-minded nervous system so you could think of it this way that when a child is Young anything is possible and then through practice television parents teachers experience with gravity the

vast majority of what the person could become is totally eliminated and you're left with a kind of simple system where each nerve cell has a particular task to do but how it turns out is different in every animal and every person so this is a very destructive view of Education it says education is there to kind of eliminate alternative ways of thinking and that the more we know the more narrow-minded we are and you know most children would tell you adults seem to be much more like this and they are much more open and so

maybe this is a good metaphor for what's going on elsewhere but most people are not interested in muscle uh but you're interested in brains and memory and including me and so could we use the same kind of techniques uh to study this in the cerebral cortex uh so here is a brain bow of the cerebral cortex the nerve cells are different colors but what matters is the connections between them which is all this feltwork here if we zoom up uh to look at that felt workor at a little higher resolution you see it's really hopeless

there's just way surprise there's way too many wires to trace the the problem in a more specific way is that the kind of Imaging we're doing here the thinnest focal section we can look at this which is called an optical section the thinnest Optical section is so thick that many wires are running on top of each other in our finest Focus so we can't actually follow every wire in most parts of the brain I showed you those giant axons you could follow but where there are dendrites and fine axons it becomes much too difficult now

we can make brain bow animals where only a subset of brain cells are labeled this is for example a technique where only the inhibitory neurons in the brain are labeled now the density is much less these big black blobs are the paramal neurons that are receiving inhibitory synapsis so you you can see you could trace out most of the inhibition in an animal like this but that's not what we want what if we want to see all the connections what do we have to do so what we realize is we have to kind of give

up for the time being on this brainbow approach and go to a higher resolution technique which is electron microscopy and cut brains thinner than the optical section thickness and I want to show you our slicer for the brain it's a weird Contraption it looks like a movie projector and you'll see it has some analogies to a movie projector this is a film tape and this is a piece of brain here and and what we're doing is collecting the brain section by section I'll just show you what's actually happening here piece of brain is put into

a block of plastic it's hard and then the brain is sliding up and down like this on this big Chuck against a Diamond Knife and the Diamond Knife slices off a section that's about 30 nanm thick that's about a thousandth the thickness of a hair so these are just about invisible and they float on water and then a conveyor belt is picking them up one by one so you take a three-dimensional block of brain and you turn it into a linear array of single sections then you take the array and of those tape and you

break cut the tape into pieces you paste the pieces onto a flat silicon wafer and you keep doing that until you have the whole brain piece uh in a library of Wafers and so here's what a wafer Library looks like of 10,300 sections of the thalmus this data set is about 100 terabytes so a terabyte is a th000 gigabytes some of you may already be using terabytes so this is a pretty big data set and and I'll show you what these Wafers look like uh Bobby Castor uh is develop a way of Imaging these using

an electron microscope and I asked him to show uh one of these uh Wafers to you by holding his hand really still so bobb's holding his hand still here and we're going to zoom in on one of these Wafers and each of these is a section of brain as we zoom in uh we're going to get to a piece of cerebral cortex he has to step into the electron microscope at about this moment these are blood vessels these big white things these little white surf are the nerve cells of the brain and these white streaks

you're seeing are AB dendrites the black enclosed objects are melinated axons and as we zoom up further and further and further finally we get to the point where we have a synaptic terminal and axon making a synapse onto a dendritic spine with a spine apparatus and so that's one section but uh this is thousands of sections and if I just give you a sense of this displays come on yeah so here is uh one section after another of several thousand uh nothing seems to be happening in this movie even though each is a picture of

an xsection and that's because these are 30 nanometers thick so only gradually are you going to see that blood vessel slowly disappear it takes about a thousand sections to get through a single nerve cell each of these white things is a single nerve cell at this level you can't really see the wiring diagram but you can appreciate that the same data is appearing in SE after section we have to zoom up higher and this is what it looks like if you zoom up higher you can see these melinated axons moving again they're not really moving

this is a nerve cell here and there's another couple of nerve cells that will appear such as this one right in the middle these things moving around are the wires that are running through the volume and because they're moving at different angles through the volume they appear to be moving in directions this is a big dendrite moving off of that cell there these little gray objects inside these cells are that are sort of oval in shape are called mitochondria at this level you see all these wires moving you also see that this looks like it

was made in 1903 and and that's because the tape is not very good yet you know it's not this is a blood vessel by the way you know we we have a little further to go to make the quality better but you really have to zoom up higher because at that level although you get a box of brain uh you can't really see all the synapses these are nerve cells here we have to go higher resolution and if we look at that data at higher resolution and that's what's shown here you start to see that

in between the large objects moving are lots of little objects moving those are axons and little dendritic spines and there are synapses all over the place those little things with little gray circles in them are synapses and they're everywhere in this data set the kind of data we're taking now is approximately a terabyte a day our images are about 100,000 K 100,000 by 100,000 each image and we take about 100 images a day so it we're moving but it's still a slow process but what do you do with this data you what you can notice

is that with your eye you can easily follow an object from one section to the other in fact you could probably get a 5-year-old with a good set of crayons and a very large coloring book of these to color the same object in the same color from section section and if they did that they would be segmenting this out as a wiring diagram and here's basically what you'd like to do and and Daniel berer uh who has been working on this project has developed a kind of digital coloring book which is basically you go in

and you say oh that's an interesting object I'll color it in red on each section I keep doing that and if you keep doing that you can then generate that object in three dimensions as you'll see so there is a d with dendritic spines sticking out of it and here's an axon that happens to inate two of those spines running across so that is a hand segmentation that is a human being is doing the coloring but they're coloring it in with a computer and and now you have you know that that's the axon that innervates

that dendritic spine but you don't just want to do two things you know you basically want to do the entire data set and and so let me just show you that here so here is a entire data set colored in segmented by this program of hand segmentation and and so this is ultimately what would give us wiring diagram information once you have this uh data you can then render it in three dimensions this is exactly the same data but now just rendered in three dimensions those are the dendrites with all their spines and then we're

going to fill in all the axons there you have it totally useless I mean it's useless for many reasons I'm sad to say one is that the material is orphaned in all directions you don't know where anything came from the second is more depressing and that is this piece that took a long time to reconstruct is there in this section and that section sits there in this image and this image sits there in that image and that little Green Dot I don't know if you can see it it's a smallest dot I can make is

bigger than what was reconstructed now you say well what about a cubic millimeter which would be a voxal of an fmri image how big is that it's that big so there's a problem and and when we started we were taking these images at the pace of about a half a million pixels per second and at that rate to just to take the pictures of the entire cubic millimeter is 2.24 centuries and um 2,000 terabytes I could find no graduate students interested in this project although interestingly a lot of postdocs were interested and I I think

this just tells you what the job market is like this would be stable employment so that is a a problem so you know we had to go faster and and uh now we're not going 0.5 million pixels per second but we're going 20 million pixels per second second terab a day we can do a cubic millimeter in a PhD thesis time a one PhD thesis unit uh and soon we're going to have we already have the machine but it's not quite automated yet uh we'll be able to go 40 million pixels per second with a

single beam scanning electron microscope that uh goes very fast and then we could do a cubic millimeter in 2.8 years that's still pretty slow for one one voxel of an fmri image we want to go fast still and in a few years probably uh about a year from this spring we will have a new machine in the lab that goes over a billion pixels per second allowing us to do a cubic millimeter of a of a mamalian brain in 3 weeks this machine is worth looking at um it it's it's like an electron microscope with

a single scanning beam except it has 61 beams in it uh and Zeiss is making this for us uh it's as if you have one machine that is like 61 microscopes it goes 61 times faster and should allow us to do billion pixels per second or faster this looks uh impressive it's more impressive when you see it next to a normal-sized human being it really is a humongous machine it's gigantic not only in size but in price and uh this is what it looks like without its clothes on it's a really impressive piece of technology

so we are going to be able to get the data pretty quickly but I'm sure some of you are wondering what about the coloring in part the segmentation problem how you going to get a lot of children to color this in and and this was our problem that we could color it in by hand but it's slow I talked to some Engineers here Hans Peter Fister is a en engineer in the School of Engineering and he said that's just engineering you know really uh we C we can make make it work and after five years

they developed an algorithm that was quite good you know this is colored in entirely Now by machine and it's colored in in uh the right things and then if you look at this movie it's done here about 8,000 of what a single segmentor could do uh 8,000 times faster so this took uh about a day to do it would take 8,000 people to do the same data so I just want to end by giving you a sense of what one uh is what we're trying to do with this uh by just showing you a few

little movies so this is a column of nerve cells in the cortex and this is still not all the cells there but we decided to get every single element in one part of one cell and this may look like a minuscule small area but it's the largest area this about 700 cubic microns where we have now identified every synapse every axon every synaptic vesicle uh every dendrite in this little region around the dendritic spines of one uh dendrite we wanted to see how many other things are in the vicinity of one dendrite and uh this

will show you what's in that little cylinder and it's uh quite depressing because there's so much there there's 700 axons there's about 55 different dendrites that interact in that little region and there's gal cells and every one of these things extend out in all directions this extraordinarily large amount of stuff uh that we're itemizing um and in there are really nuggets of interesting data you know if you just look at the main dendrite you find some of the axons make multiple synapses on multiple spines that's what these blue arrows show uh for one for for

this one dendrite and then because um I am uh surrounded by students undergrads and and high school students who are interested in in getting involved in nurs science and are you know eager for letters of recommendation they're willing to do things that I'm not willing to do and one of those things is to actually look at the synaptic vesicles in every one of these synapses to see is there something common about the synaptic vesicles in one axon versus another and so uh this is a huge amount of work uh and I'll show you this is

not an artist rendering each of these little yellow dots as a synaptic vesicle uh in a different synapse and you know this is a lot of work this is letter of recommendation letter of recommendation letter of recommendation so but it is impressive uh you see a lot I just show you this one final view of this same data set you know someday my view is that um this is the level one is going to have to probably look at the nervous system to see how information is instantiated and stored physically I hope someday uh diseases

like schizophrenia will look like like something at this level of resolution now of course you know when you look at this you can only feel a certain amount of uh humility that uh this is a lot more complicated and certainly a lot more beautiful uh than most of the thoughts that come out of our brains uh so it's like the machine is a lot more impressive than what we use it for and I I don't know what the lesson is but I think there's some truth to that I just want to end by saying that

although it s I may have made this sound like I did this work uh this work is industrial and it requires a lot of people uh working not only a lot of students but a lot of collaborators from other labs other universities and a lot of support from from many different sources so thank you very much for your [Applause] attention last slide you were showing uh vesicles on the D right vesicles on axons that are inating that dendrite that are synapsing on the dend so does it does the geometry of the vesicles matter in terms

of function I mean what exactly is the I mean why exactly should we be interested in in in studying the vesicles I think the thought nobody knows but but the thought is that some synapses have lots of vesicles other synapses have rather few and it's Poss possible that this is a measure of how potent those synapses are okay that's one thing we're hoping to see thank you so um given the projection of the speed increases that are going forward and the endless supply of cheap labor when do you see sort of the intersection of all

this um uh getting to the whole brain of a mamalian subject so I think a a a mouse brain or even a small mammals brain like a European Tre shrew uh is something that we're going to be thinking about in the next four or five years um a human brain is a thousand times bigger than a mouse brain and that uh these techniques can't do a whole human brain at that resolution yet of course it could it would just require a lot of money but you need to genetically modify the human to do that right

you'd have to kill the human right so you'd need a volunteer and not only do you have to kill them you'd have to kill them and start infusing them with gutter alahh and from alahh the moment of death you know so you wouldn't want to kill somebody you what you would want is is a a human who decided to dedicate their body to science and are willing at the moment a doctor declares them dead um to infuse them with something that would certainly kill them if they were still alive and there's a lot of problems

with doing that in a hospital for example if I was in the bed next door there's always European research so about this say what there's always European research right so but but uh you know there is a there are brain Banks of diseased human brains that are fixed in from aldhy they usually there's several hours between the moment of death and the time the brain was taken out we're going to start looking at those pretty soon and then uh I have colleagues who are working in neurosurgery who are neuropathologists who often get pieces of brain

that they can dunk not whole brains but little pieces of human brain that can be dunked into formaly very quickly so we'll get a sense of what human brains look like long before we try to do something as amazingly difficult to imagine now you know it would be about uh two million pedabytes of data a human brain and that that's much larger I think than the digital content of the world right now so we're not ready at least in my lab for anything that big thanks yeah I mean I think when you get your full

wiring diagram if you ever do it's going to be like bz's uh Universal Library where all the facts in the world are represented but you can't find anything of of Interest or meaning because there's so much of it I mean as a neurophysiologist I am always a little back when people talk about the brain and don't even mention Spike trains or information in the spike trains there's an informational order that's uh somewhat not not completely but somewhat independent of the structure uh and if you look in the early auditory system and in the mntb where

where you showed a slide it the action and all the coding of the information is in the spike trains and the structure of the spike trains and we know you know we know the the wiring diagram to a first approximation so the question is and and the thing I'm always frustrated with is you know if we could read the anatomy in a way that could tell us how the system works what what could we learn so what what how are we going to figure out how the system works from a wiring diagram right so this

is of course a very obvious and important question I don't want to make make light of it or say this is Trivial I didn't talk about the neuromuscular Junction enough to show you an example where the wiring diagram in development which is very different from the adult actually is highly ordered and it's ordered in a linear order connectional Matrix that is the physical instantiation of the size principle that is the firing order turns out to be played out in the wiring connectivity once we made that decoding any muscle I go to in principle I can

understand the firing pattern just by looking at the wiring diagram now now you say well that's a special case and I say no well no all the things you learn must be put in to a form that you can read it out again based on connections there has to be a connectional underpinning so that's what I'm searching for which are these sort of motifs where the subtle differences from one to another are not as important as the principle of the way information is organized if you could decode like that you would be able to read

out information now the challenge is of course information coming into the auditory system is coming in from the outside it's not in the nervous system but if you have a memory of a voice or a memory of a Melody that is in the nervous system and it has to be in the connectivity I it's not in the firing trains if you haven't been thinking of a song for a year and then you play and then you sing the song it's not because you've been playing that firing train over that time it's that somehow the connectivity

between the cells allows that firing train to well I actually teach psychology of music and the neuros pychology of music and and the there are Notions of temporal memory traces which are also mediated by structure but that could be a way that you could dredge up a Melody after several decades the thing is is that we have to think about both the the the informational order and the spike trains and the structure of the thing in a complimentary complimentary ways that mutually affect each other so the loop I mentioned is what I literally mean that

that I'm not saying structure is more important than function function generates structure and structure generates function it's a loop you can't ignore structure and you can't ignore function one has to be able to decode one relative to the other but the spike trains are it's like it's like what if we were going to study the genetics molecular genetics and we didn't know about the genetic code and we were going to go in and look at the molecular structure of chromosomes and then further and further in without knowing what the nature of the code was there's

an informational order there that's complimentary to the structure it function is another aspect of that but but we have to always be thinking about about the nature of the informational order Vis A the structure so I I completely agree with that yeah I agree thank you yeah yes H I have a more general question but it was inspired by your very nice uh introduction to this subject I mean how does the brain evolve how does it evolve yeah since I have the impression that genes can create everything since the beginning so and and of course

what you learn in your life it's not it cannot be given to to your sons so uh I I had the impression that apart from increasing the size of the of the brain there is no other way to to evolve well I think humans show that there's a very weird way to evolve which is to jettison most of the fundamental information that other animals are imbued with at Birth and require that to come in through experience humans are interesting in that we come into the world knowing less about the world than other animals and we

take far longer to reach mature State what you know what is it it's a year till a baby walks 15 more years to the driver's license three more years before they leave the nest you know they're 18 when they leave the nest what other animals leaves the nest at the age of 18 so so so the the magic of being a human being is not that we've evolved all this fancy stuff it's that we come in with this huge brain and we can take advantage of experience to mold the circuitry such that we have skills

that are related to the world we find ourselves in rather than to our genetic Heritage whereas most other animals are kind of stuck we we don't transmit this to to other Generations so I think of darwinian EV Evolution so well of course I mean you your kids speak langar my life I don't I don't give to my to my son so you don't what your sons I don't give to my son so it's you don't give to your sons what I learned what kind of parent are you of course you give to your sons so

I don't make the genes of my of my my sons better this is the magic of the human nervous system you give to your sons experience that gives them a structural brain that evolves much faster faster than genetic Evolution that's why humans today are doing different things than your parents were doing or your grandparents were doing there's no other animal like us it's because we are so slow in growing up you we have all this information that makes our structure as opposed to our genes it's that Loop between structure and function it's this and gramz

in my view that makes us so special even if the genetic box you make great should we think