October 24, 2018 • 37 mins
Alex Vaughan is a scientist and entrepreneur, founder of MapNeuro, Inc. He’s been working at Zador Lab on groundbreaking new ways of tracking neurons and mapping them. In this episode, Alex talks about the molecules in your brain and how they control you. he describes how viruses are used as a delivery mechanism for brain mapping barcodes. You’ll also find out how your brain is like a subway.
It’s not easy to map the brain, but we’re starting to understand how, thanks to many fields of science contributing to our current technological toolset. Listen in to figure out how all fields of science are connected, and how business and science can be used together to better understand the brain and intelligence.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Matt (00:07):
Hello and welcome to the Numenta on Intelligence podcast.
I am Matt Taylor and today I'mgoing to be bringing back the
Interview with a Neuroscientist,a series that I started on
YouTube.
If you haven't seen that, I haveseveral different interviews
with neuroscientists who havecome by the Numenta offices for
one reason or another, includingDavid Eagleman, Jon Michaels,
(00:28):
David Schneider, Carmen Varela,and Eric Jonas.
Today I'm going to continue thatseries and we are going to talk
to Alex Vaughan of the companyMapNeuro.
He has been working at Zador Laband has recently been recognized
for his work, creating tools tohelp map the brain.
So really interesting stuffgoing on there.
So hopefully you enjoy thisepisode of the Numenta on
(00:52):
Intelligence podcast.
Today I have with me AlexVaughan from Cold Spring Harbor
Labs and a company calledMapNeuro, which we're going to
talk about both of those thingsat some point.
Would you like to give yourselfa bit of an introduction with
what brought you into the fieldof neuroscience?
(01:12):
What are you excited about?
Alex (01:13):
Sure.
Uh, so my entry to neurosciencewas actually a little atypical.
I came in from kind of standardanimal biology behavior, not
anything having to do withmedicine.
I actually, the first animal Iever studied was a spider way
back at Cornell and Ron W., andwe were studying spider behavior
and cricket behavior andbasically trying to understand
(01:34):
how these weird little animalstick and, and kind of what makes
them perform all of their crazybehaviors.
And so I moved on from that andI ended up doing a PhD at
Stanford where I, uh, where Idid something kind of much more
directed in the sense that Istarted working on fruit flies
and fruit flies are this amazingmodel system in which we have
(01:57):
over the course of a century'sworth of work, gotten ourselves
essentially perfect geneticcontrol over a single organism.
Matt (02:05):
Right.
Alex (02:06):
So we can create mutations however we want.
And by the time I came into thefield, we could also use this,
use these genetic tools to lookat neuronal activity and to
manipulate neuronal activity inthe fly.
And fly brains are very special.
They have relatively few neuronscompared to humans and every
neuron does something unique andspecific you can find the same
(02:29):
neuron in one fly versus anotherfly.
You can't do this in a person.
So a lot of what I did was wewere studying courtship behavior
and flies and I would kind of gothrough and systematically turn
on, some neurons, turn off someneurons, look to see how it
manipulated courtship behavior,uh, and that.
And after that I started workingin a, in a more of the field of
(02:51):
decision making in mammals andstudied kind of how animals
think, how they make decisionsand moved from there to working
on making maps of the brain andthat's a lot of what we do now.
And what, uh, what I can talkabout, what we're doing at
MapNeuro is basically buildingtechnologies and tools for maps
in the brain.
And in my current work, I'mtechnically still a postdoc and
(03:13):
Tony Zador's lab at Cold SpringHarbor.
And Tony has developed a bunchof technology over the years for
what he calls molecularconnectomics, which is basically
a bunch of tools for tracingneurons throughout the brain
that work really well, likemuch, much better than, than
previous tools.
So that's what we're building.
Matt (03:30):
And all the way back to fruit flies.
I mean, you were working earlyin your career with genetic
tools, applied geneticsessentially to do neuroscience.
I find that really interesting.
I'd like to dive into that withyou as one of our subjects we
talked about today.
What's more interesting from abroad standpoint are these,
these molecules in your brainand this idea of molecular
(03:51):
structure.
Alex (03:52):
So I would say the analogy is this, that in flies, because
we have genetic tools, we canmanipulate every neuron in the
brain, uh, with arbitraryspecificity and more or less do
whatever we want to them.
And this, this lets usunderstand the brain and if the
fly were a patient sufferingfrom some sort of neurological
disease, this would let us fixthings relatively easily.
(04:15):
In humans we have nothing likethat.
Nothing like that, control andin fact the only things we have
in humans that let us sayprovide therapies are basically
neuromodulatory drugs and theseare drugs you've heard up.
So they're things like serotoninand dopamine that people
associate with reward andpleasure and addiction.
(04:37):
And there's about half a dozenof these and the interesting
thing about them and what makesthem actually a little bit like
the case in flies, is thatthey're in special circuits,
special circuits there is thatthere's kind of one or two areas
in the brain that expressserotonin and they distribute
the serotonin from a smallpocket of neurons all the way
throughout the brain.
And everything that Serotonindoes, it does through these
(04:59):
neurons.
Matt (04:59):
So these molecules are flowing through the brain.
They're moving.
Alex (05:04):
Yeah.
So there's a bunch of neurons inone area, but most notably the
dorsal raphae that expressserotonin.
And then they have these longrange projections that reach all
the way throughout the brain.
They can reach inches andinches, which is huge for a
cell, all the way throughout thebrain.
Matt (05:22):
Wait, within a cell or through cells?
Alex (05:22):
Uh, so they, the molecule, the neurotransmitter of
Serotonin is expressed in thecell and within this cell it's
usually contained in what'scalled a vesicle.
Uh, so this is basically alittle, a little transport
module, a little pod, more orless than what they do is they,
they'll synthesize it usually inthe cell body, which is in one
location.
They'll package it in one ofthese little pods and then it'll
(05:45):
be transported all the way downdown the axon to somewhere very
far away, for instance, to yourprefrontal cortex, which is the
other part of the brain thatdoesn't make its own Serotonin,
but in which Serotonin iscritical for thinking and
emotion and, and that sort ofthing.
And then at the right time, thisneuron will release the
(06:05):
Serotonin and the Serotonin thenacts as a neurotransmitter.
And what that means is basicallyit's passed from one neuron to
another to change the activityof that second neuron.
So there's a bunch of these thatdo really canonical things-
serotonin and dopamine andnorepinephrine, acetylcholine
and these ones are particularlyimportant because for the most
(06:25):
part, every drug we have in thebrain that helps people with
depression or helps peopleimprove their memory if they
have Alzheimer's, every one ofthese targets, either the
synthesis, the distribution, orthe degradation of these
neuromodulatory chemicals.
Matt (06:41):
Really important molecules.
And are these important becausethey just have certain molecular
characteristics that allow themto carry information or do you
think there's a lot more that wedon't know about?
Alex (06:54):
Yeah, so it seems like there's a couple things going
on.
One of the reasons they'reimportant is that they seem to
play a specialized role incomputation.
Uh, there are, there are kind ofbrain-wide changes that you want
to make in order to alterbehavior.
And one of the ways that-
Matt (07:13):
Like global changes.
Alex (07:14):
Global changes, and norepinephrine is a great
version of this.
So norepinephrine you can findthroughout your body, if you are
nervous, say you're giving apodcast or something and you're
a little a little excited, youwould, you would have a little
more norepinephrine than usualthat's involved in the flight or
flight fight or flight response.
Matt (07:31):
A stimulant, right?
Alex (07:32):
Yeah, so in the brain, we think that what it does is it
controls attention, and we don'tknow a ton about this, but this
is, this is the best theoryright now.
And so essentially when theseneurons and the locus coeruleus
release norepinephrine into thebrain, they'll kind of wake up
brain areas and say, payattention to this.
Pay attention to the sensoryinput that's coming in now and
act on it as quickly aspossible.
(07:54):
And what it looks like, thistiny little set of cells here
that, that releasenorepinephrine, what they do is
they signal global changes inthe brain and they so they, it's
a very precise timing.
So now is the time when you gotto wake up and pay attention and
do something, but it's more orless global.
So, and the idea is basicallythat you need to be able to
(08:17):
distribute this kind of signalto the entire brain at once.
And these, these brain regionstend to specialize on that.
Matt (08:22):
And you take advantage of these molecular circuits in your
company MapNeuro in reallyinteresting way through viruses.
And I'd love for you to talkabout how do you use viruses and
these molecules and how they,those circuits go through the
brain to help map the brain.
Alex (08:39):
Yeah.
So one of the things we'rereally interested in doing, uh,
in, in our work is actuallyhelping the field of
neuroscience as a whole movebeyond these four or five
chemicals.
Matt (08:50):
Great.
Alex (08:51):
Uh, and a lot of the reason that we want to move
beyond them is there's a 99point nine percent of the brain
or something like that doesn'texpress these chemicals, is
doing different things and hasconnectivity maps that are hard
to untangle.
And what I mean hard tountangle, I mean, if you really
want to know where a single oneof the 100 million neurons in a
(09:14):
mouse brain goes, like it startshere and it has this long
production to a bunch of otherareas,
Matt (09:19):
Axon and all of its dendrites.
Alex (09:21):
If you really want to know that, the way that you do that
is you, you label one neuron inone animal and then you
sacrifice the animal and youvisualize, once you've kind of
put the brain under amicroscope, you visualize the
anatomy of that neuron and sothat's, that's the way that you
can kind of get resolution overwhat single neurons are doing.
(09:43):
Uh, and people who've done thisfor many years in particular for
the neuromodulatory circuitsthat we were talking about and
they've also tried to do it formany of the cortical circuits
that are involved in learningand memory and motor control and
sensory processing andeverything else.
But the problem is basically thetools that we have are one
neuron at a time or maybe a fewthousand neurons at a time,
(10:05):
they're nowhere near efficientenough to create reliable maps
of the brain that address the100 million neurons you find in
a mouse brain.
Matt (10:14):
You have to understand what the populations are doing.
Alex (10:18):
Yeah, and you have to understand the, you have to
understand the map.
So one, one analogy that I liketo give is in neuroscience,
people have been studying many,many different regions of the
brain to see what they do,right?
Uh, and so they can do this,either you have lesion
experiments where you disruptactivity in one part of the
brain or you can simply recordactivity in the same part of the
(10:39):
brand and see what it does andbehavior.
And so using this approach, wehave things like, we have motor
areas, we have sensory areas, wehave areas involved in decision
making and we know where theyall are.
And you can think of these likesubway stops in a city, right?
There are all these differentplaces and if you want to go to
the spot where you're gonna,where you're gonna really, uh,
make decisions, you go to theprefrontal Cortex, that's where
(11:01):
you go.
But it's like a subway map wherewe know where all of the stops
are and we have no idea how toget between them.
That's the big problem thatwe're, that we face.
And there's been an enormouseffort in the last 10 years in
particular to try to create someof these maps.
Uh, and so one actually oneoutstanding project has been the
Allen brain connectivity atlast.
(11:21):
So the Allen Brain is uh, uh,the Allen Institute up in
Seattle, founded by Paul Allenand filled with a tremendous
collection of neuroscientists.
They've done a project wherethey label neurons in one region
and then they look at theprojections of all of those
neurons to a bunch of otherregions.
So if you label, say theprefrontal cortex, you might see
(11:42):
projections back down to sensorycortex or to motor regions that
are helping kind of output adecision.
Um, and that might be a dozendifferent different areas.
Matt (11:52):
Yeah, because you're doing a lot in the prefrontal cortex.
Alex (11:54):
Yeah, you're doing a lot in the prefrontal cortex.
Uh, and so what they did is itthrough a really heroic project,
they did 500 of these mapstracing different areas of the
cortex as they projected kind ofacross each other and in this
dense network of connectivity.
And they can make this beautifulmap that shows you where all of
the connections are.
Matt (12:14):
I think I've seen the visuals of this.
This was really impressive whenit came out a few years ago.
I'll probably try and link to itin the show notes.
Alex (12:22):
That sounds great.
They have an outstandingvisualization people, I have to
say, we've learned a lot fromthem.
Uh, so, uh, there's one, so itwas an enormous amount of work
and it's been really useful.
So now if you're looking in theprefrontal cortex and you want
to know where these neurons go,you want to know who they're
sending information to, whichregions they're distributing
information to, you can justlook and figure it out.
(12:43):
And there's one drawback of thisapproach, which is that in doing
that you no longer haveresolution over single cells.
So you're labeling thousands ofneurons simultaneously.
And when you look at theprojections, you see them all
simultaneously and they're allvery difficult to differentiate.
Matt (13:02):
Yeah, you can't tease them apart.
Alex (13:03):
Yeah.
So going back to the subwaymetaphor, it's a little bit like
saying, you know, I live in NewYork.
It's like saying all of thepeople at Wall Street who were
getting on the subway aresimultaneously going to central
park and the empire statebuilding in Brooklyn.
All right, that's just not true.
Individual people go todifferent places and your
individual neurons go todifferent-
Matt (13:21):
You don't have the resolution to understand the
true map there.
Alex (13:24):
Yep.
Yep.
Matt (13:25):
Um, so, but you're working on that.
Alex (13:26):
Yeah.
So we're working on that.
So one of the things that, uh,that Tony Zador at Cold Spring
Harbor has been working on formany years, are ways to label
neurons individually, socially,you can still use them for
tracing.
And one simple idea is you mightlabel some neurons red and some
neurons green and then you cansee where the red neurons go and
(13:47):
the green neurons go.
And what we did is kind ofabstract that concept out, such
that we label neurons with a DNAbarcode.
And so what this means is, iswe, we make a virus similar to
the virus that the Allen BrainInstitute used where they
express a green fluorescentprotein to label neurons.
What we express instead is ashort sequence of, uh, of DNA
(14:11):
such that every neuron we infectis uniquely labeled.
Matt (14:14):
That's a huge difference.
Before you would have had todetangle all of these different
reds and greens and this is red,but which neuron did it really
come from?
Alex (14:23):
Yeah and you know that that actually has been a really
effective approach for, for manyyears, Josh Blackman, among
other people, a pioneer to atechnology called brainbow where
they have lots of colors.
They're beautiful
Matt (14:36):
They are.
Alex (14:36):
They're astonishingly pretty pictures.
Uh, and using that approach youcan trace maybe a dozen neurons
at a time, maybe 20.
It was actually, it was reallyuseful for dissecting the fly
brand.
It's proved less useful in themouse brain just because the
mouse brain has so many more.
So instead of using four colorsor six colors or eight colors,
(14:57):
what we do is we abstract theidea of colors into DNA.
So every base of DNA has fourpossible letters that can be
associated with a g, t, a or c,and you can think of those as
four colors.
And because we use a sequence ofDNA that's 30 bases long, we
kind of have four colors andthen the next base we have four
(15:19):
more and four more, four moreand four more.
Matt (15:21):
So you can have a unique identifier for pretty much any
neuron.
Alex (15:23):
Yeah, so we have, so we use a, a barcode as we call it,
of length of 30.
And I get those four to thepower of 30 possible color
combinations, which turns out tobe enough to label something
like every sand grain on earth,uh, which, uh, isn't that much
more than the number of neuronsin your brain, but it's probably
enough.
Matt (15:43):
But you could theoretically map an entire
human brain.
Alex (15:46):
You could, yeah.
With, uh, with this approach,uh, on an enormous amount of
work.
And some luck, you couldtheoretically map the human
brain.
Matt (15:53):
Tell us about the work involved here because it's not a
trivial task even doing it thismethod, right?
Alex (15:59):
Yeah.
So it's actually, it's a, it'svery different than a
traditional tracing experiment.
In a traditional tracingexperiment, you would find some
way to label some neurons, youwould say, in infect them with a
virus that expressed a greenfluorescent protein and then
they would all be green and thenyou would take that brain,
sacrifice the animal, put itunder a microscope and you would
(16:20):
visualize it.
And this is amazing, right?
You have perfect spatialresolution, essentially as fine
details as you want to see.
If you want to see the finedetails of a synapse between two
neurons, you can just kind oflook and, and you can see the
what are called presynapticboutons that are, that are at
the neuronal outputs and postsynaptic dendrites that are kind
of the inputs to the that nextone, you have amazing spatial
(16:43):
resolution.
And what we do is we throw allof that out as quickly as
possible.
So if we're, for instance,looking at the projections from
prefrontal cortex to otherareas, we, uh, we grind up the
area where we made ourinjections.
Matt (16:57):
Oh, you don't take these slices and have a beautiful
rainbow like picture.
Alex (17:01):
We've, for the most part, we were getting back into that a
little bit, but, but in thesimplest part of the experiment,
and I should say that this, thiswhole approach was pioneered by,
by Tony Zador and a phenomenalGrad student in his lab named
Eustis.
Uh, and in particular, whatEustis's insight was, was you
don't have to keep spatialinformation.
(17:22):
You can infect a bunch ofneurons to label them uniquely.
And then at the injection site,if you grind it up, you'll see
all of these barcodes and maybeyou infected a virus that had a
million barcodes and you labeleda thousand neurons.
So there were a thousandbarcodes that you end up finding
in that sample.
Once you go through thesequencing process.
(17:43):
The way we do this experiment iswe take a virus that allows us
to uniquely label neurons and wedo an injection into, say, the
prefrontal cortex to label theseneurons.
And every neuron is uniquelylabeled with a barcode.
The barcode, while the animal isstill alive, the barcode is
amplified by the cell and thenit gets distributed along the
(18:05):
neurons.
So we have a protein come andgrab onto this, this DNA
barcode, it's an RNA form atthis point and drag it along the
process all the way out to itsmost kind of distal extent.
It's the synapsis that it'smaking on other neurons and
other brands.
Matt (18:19):
Is this a part of the molecular circuits you're taking
advantage of is dragging these-
Alex (18:24):
Yeah, so one of the things we've tried to do is basically
actually kind of precisely thetransport machinery that I
mentioned earlier that movesvesicles from place to place.
We can piggyback on, on thatkind of transport machinery to
move our barcodes from place toplace.
So that's actually a reallyimportant thing that ended up
working surprisingly well.
Matt (18:44):
It's fascinating.
I think inside the cell, all thethings that are going on and us
being able to take advantage ofthat to increase things like
this.
Alex (18:53):
A lot of that comes from really, really hardworking
scientists who figure it out.
Fundamental cellular processes,uh, that seemed maybe esoteric
at the time, turned out to bothhave medical relevance and to
lead to new technology.
Matt (19:09):
And we're talking about virology and genetics and very
specific areas, a niche areaseven within virology that could
come to fruition andneuroscience and potentially no
one ever thought that would beapplicable here.
Alex (19:21):
Yeah, yeah, exactly.
Matt (19:23):
Very interesting.
Alex (19:23):
So yeah, so to, to, to finish the experiment that we
ended up doing, we, we sequenceat the site where we did our
injection and that, that tellsus which neurons interacted with
which barcodes and when we wantto know where each of those
neurons projects, we sequencethe barcodes that made it to
other areas.
So if there are 10 barcodes inthe injection site and in one of
(19:45):
the distal targets, we findthree of those barcodes, then we
know that there were threeneurons that from the injection,
so that projected all the wayover there to that other area.
And if we see one of thosebarcodes in another area, then
we know that that single neuronsplit and sent to axons to two
different areas.
And that's really where we get akind of a real uptick in the
(20:10):
amount of information.
So instead of the traditionalapproach where you have bulk
projections and again, everyperson is simultaneously going
to Central Park or the EmpireState Building, we're or to
Brooklyn, we now know whereevery neuron goes on an
individual level.
Uh, and this has led to some,it's obviously we focus a lot on
the technology, but it's led tosome biological insights.
(20:33):
So actually going back to what Iwas saying earlier about alarm
signals in the brain.
The first experiment that Eustisdid was to trace the projections
of cells that expressnorepinephrine in the locus
coeruleus and to trace them on asingle neuron basis.
So where does every single oneof those neurons go?
And the previous hypothesis wasbasically that they were all
(20:55):
widely distributed through thebrain, sending a global kind of
alarm or attention signal thatwould wake up the whole brain
simultaneously.
And that's not what Eustis saw,kind of, much to our surprise.
He saw that this individualareas where a uniquely
innervated by one neuron.
So there would be an area, andor a better way of saying it is
(21:17):
a single neuron can control onearea and doesn't project to
other areas.
So there is some sort ofanatomical specificity.
There might be processes incircuits that allow you to kind
of wake up the earth, theauditory part of your brain,
wake up the visual part of thebrain and modulate the brain in
much more detail than we'vedone.
Matt (21:35):
Specifically in response to specific stimulus potentially
as a part of a learned behavior.
So yeah, that's veryinteresting.
So you already mentioned thatyou worked at Tony Zador's lab
at Cold Spring Harbor, butthere's also this MapNeuro part
of it.
And I wanted to talk sinceyou're, you're sort of on both
sides of the fence here.
(21:55):
You're, you're, you're a postdocat Zador lab and you're also
kicking off this business butyou just got this award and some
grant money for it withMapNeuro, which is exciting.
Um, so what is this, what's thedifference between, as a
scientist a scientist and abusiness person?
You know, what's the differencebetween the lab environment and
where your priorities are thereversus the business environment
(22:15):
and how do you play thoseagainst each other?
Alex (22:17):
Yeah.
Well, the first thing I shouldsay is that, uh, on the business
side, I'm still learning as Igo.
I'm very new to this.
I'm trained as a scientist.
I like to joke that anyone whowants to work with me, he gets
to make the mistake of letting abusiness be run by a scientist.
Um, but, uh, but it's beenreally fun so far and, and kind
of really, really interesting.
The main idea that we're, thatwe're going for with this
(22:39):
business is that we think thatthe technology that's been built
in this Zador Lab over the lastseveral years has legs.
And it has legs in, in twosenses.
One is that we find that thereare lots of other scientists who
want this information, thisinformation about the single
neuron projections.
And we-
Matt (22:59):
That's one of the reasons you're here, we're certainly
interested in that.
Alex (23:02):
Yeah, exactly.
And I spent a lot of my timedoing, having this kind of
conversation and trying tofigure out what we can do now
that we couldn't do before.
And that's actually one of thegoals of MapNeuro is simply to
help scientists do this kind ofexperiment.
And the reason for that is justthat while we think we've done a
great job making this experimentstraightforward and routine in
(23:23):
our hands, it turns out thatpeople have different skillsets
and some people are, forinstance, very good at analyzing
things but not so good atmolecular biology.
Some people are good at workingwith animals but not so good at
molecular biology.
Uh, and we are a lot of the goalof MapNeuro is to fill those
gaps and to help people do thiskind of experiment without
(23:44):
having to-and is the worst casescenario-having to have a single
person working for a year toreplicate something that's
already going on in someoneelse's lab.
So it's essentially ameta-collaboration.
It's, it's an effort to, uh, toprovide, uh, uh, as much
collaboration with the field aspossible to help people do a
(24:04):
kind of a new scale ofconnectomics.
Matt (24:06):
So as you're looking as, as MapNeuro grows in structure,
you're looking for other labsthat want to collaborate on this
type of work, it seems likepretty closely with them because
you want to understand whattheir priorities are and exactly
how you can help develop ways ofmethods of doing what they need
to do.
Alex (24:23):
Yeah.
And getting back to kind of whatI've learned over starting this
business is you really have tolisten to what people need and
it turns out that you can't justmake up what you think people
should need or what it would beconvenient for them to need and
then offer that.
And the example is basicallyjust that we have the first
thing we set up was, was afacility to do the core map seek
(24:46):
experiments to do kind of thecore molecular biology.
Uh, and then we offered it tothe world and it turns out
basically that that is notenough, not enough to really
help people.
Uh, and there's a bunch ofreasons for that, but the
biggest reason is that it'sreally hard to adopt new
technologies without fullyunderstanding all of the caveats
(25:07):
and all of the difficult partsof the experiment and all of the
ways it's going to fail that youmight not even find out about
until the experiment has done.
And so people are both reallyexcited about new technologies
but also reticent to put anenormous amount of their own
effort into them.
Matt (25:24):
So you're also, you not only need to offer them the
technical services for themapping, but also your expertise
and how are they, how could theytackle this.
Alex (25:31):
Yeah, precisely.
A lot of the conversations thatwe have that are super
interesting for me are learningmore about the kind of
scientific questions they wishthey could address and then
telling them which of those wecan actually, uh, we can
actually get at with our, ourtechnology.
Um, but you know, it's been,it's been really rewarding
(25:52):
because we see fields open upthat haven't had access to many
of the genetic tools, forinstance, that I talked about in
flies or even in traditionalnice.
And one of the things we'redoing now is we're doing
experiments in, um, in animalsthat are usually not studied
very much because our approachcan be used in mice.
(26:12):
It can be used in rats, it couldconceivably be used in birds, it
can be used in other, otherspecies of animals.
Uh, and so the hope is basicallythat by taking away this one
pain point of the hard parts ofthe molecular biology and
figuring out how to do theexperiment as a whole, we can
kind of make it widelyaccessible for everyone.
And, but there was a lot oflearning.
(26:33):
There are a lot of figuring outwhat people wanted and figuring
out what the pain points werefor them.
Matt (26:39):
Well, it's as, as new labs sort of adopt this and create
own experiments, of coursethey're going to get data that
lots of other people are goingto find really interesting.
Alex (26:48):
We hope so.
Alex has been here for the pastday and I saw you give a
presentation yesterday.
They're just looking at thatpresentation, the ability for,
for us for Numenta to understandnot just the matter of the
neuron, but what layers it goesto is, is really important.
Especially we're talking aboutpopulations, you know, because
all of our theories about thecortical column and what layers
(27:08):
are doing, what, so gettingbetter resolution about this
population of cells in thislayer of projects to this other
layer and this layer, that'sreally important to us because
it helps validate what we'rethinking and raise new questions
that we haven't thought aboutbefore.
Like why is this happening?
And that sparks innovation whenyou, when you run into that
thing, even if something saysyou were wrong, at least you
(27:30):
know, you can go, you can golook for other doors.
It's always interesting when weshow that kind of data, it
always make someone a littleupset.
Matt (27:40):
Yeah.
Alex (27:40):
And I think what's going on is that, uh, the data that we
tend to show is pretty highresolution.
There are lots of neurons, weknow their projections, we know
in the case you're talking abouttheir exact spacial location and
in the field of neurosciencethere's been a lot of implicit
knowledge about where neuronsare and where they go and it's
built up over a century ofreally careful work, but it's
(28:04):
also, it's combined a lot ofdifferent information in ways
that don't, that don't alwayspan out.
Matt (28:10):
It's a bit tribal.
Alex (28:12):
Yeah, and people have longstanding beliefs that it
must be this way and it turnsout that when you can actually
test it, uh, you will validate90 percent of those beliefs and
then in a very irksome way,validate about about 10 percent
of them.
So it's always a good sign whensomeone gets a little frustrated
and then wants to do anotherexperiment.
Matt (28:31):
Right, absolutely.
Alex (28:33):
That's the goal, right?
Matt (28:33):
I mean we're searching for the truth, you know, but you got
to find out how things reallywork.
I mean, that's, that's our wholemission is figuring out how it
works so we can try to build itin software.
So, uh, about MapNeuro, you'vetold us a lot about sort of how
it works and the idea of youwant a map or not high
resolution maps of where neuronsare projecting to.
(28:56):
What are the questions that thistechnology might help us answer
what, what things is this goingto help with in the future?
Alex (29:02):
Yeah.
So we are hoping, I would say toaddress three things and one is
simply, there is a fine detailof neuronal circuits in the
brain that if we can understandthose will, I think give rise to
not just an understanding of thebrain but new technologies.
(29:22):
And actually this is kind of theguiding principle of one of the
main funding agencies that, thatfacilitated a lot of this work,
which was the IARPA micronsproject and IARPA as you know,
is affiliated with the UnitedStates intelligence community
and what they wanted.
Oh, and we were part of this andas well as many other people
(29:43):
using different technologies,uh, they wanted mechanisms to
trace neurons in the brain tomake machine learning better.
The idea is that for things likedeep learning and neural
networks, they're hard.
They're hard.
It's hard to make them learn,it's hard to get them kind of
off the ground, but if we cantransport information over from
the human brain or the mousebrain onto those artificial
(30:04):
neural networks, then maybewe'll get a lot better at it.
Matt (30:06):
We could create our networks to more closely match
what we're seeing in brainsexperimentally.
Alex (30:11):
That's one hope.
The second is that we, we reallywant to just move neuroscience
forward as a field, and I thinkone of the big technological
stumbling blocks of neurosciencehas been that there are 100
billion neurons and we don'tknow where they go.
When you say it like that, it'sjust insane.
Like what's the most importantthing about a neuron?
(30:32):
It's where it projects and whoit talks to.
Do you know that for any of theneurons you work on?
No.
The, you know, the brain issuper hard in the sense and so
we really hope to make, makeprogress.
Matt (30:43):
It seems like you're in the right space right now
because people want thisinformation and there's a, uh,
there's been a lot more interestin just in neuroscience in
general over the past 5-10 yearsfor it seems from what I've,
what I've seen and especiallywith the new hubbub about grid
cells and stuff and how that allworks with everything else that
(31:05):
we're trying to incorporatefrom, from the past knowledge of
neuroscience.
So you're in a great place with,with MapNeuro and I'm really
looking forward to seeing whatyou guys come up with.
Alex (31:14):
Thanks very much.
Matt (31:15):
You mentioned to me a paper that you're currently
working on.
I think it's on bioRxiv rightnow.
It was about, it was aboutbehavior and sparse coding in
the frontal lobe.
And uh, I mean we can't go overthe whole paper, but what
interested me about it is thatit's all about how sort of
populations of neurons thatyou've shown observing them
(31:36):
through this, this technologyyou're talking about can show
aspects of the behavior that'shappening right now and you can
decode based upon what neuronsare firing, what they mean, what
they're meaning.
Can you explain that?
Alex (31:50):
Yeah, I'd love to.
So this is actually a previouswork that I did when I was in
the lab of Adam Kepecs who'salso at Cold Spring Harbor, uh,
and working with an outstanding,uh, experimentalist Junya
Hirokawa, he recorded, uh, theactivity of neurons in a
particular part of the ratbrain, the orbital frontal
cortex in the rat brain whilethe rat was doing a complicated
(32:12):
behavioral task.
So the rat had to, uh, had tosmell an odor and, and basically
figure out did it smell morelike strawberries or did it,
smell more like watermelon.
And then we would vary thedifficulty of that task by
giving the rat mixtures.
And to abstract this out, wewere basically putting the rat
in a situation where it had tomake decisions that integrate a
(32:33):
bunch of information.
So integrating sensoryinformation about kind of what,
what the, what decision youshould make, but also
information about whether heshould expect a big reward, if
it's correct or a little rewardand a and other kinds of
information like that about thetask.
And one of the things we werelooking for was basically we, at
(32:54):
the highest level, we weretrying to ask the question of
whether neuronal activity in thebrain makes sense.
That's the simplest thing andthe baseline hypothesis that
people have had in particularfor this part of the brain, kind
of frontal, the frontal lobe andthe decision making parts of the
brain.
(33:15):
The idea was was thatessentially all of the available
information that an animal hasis utterly and catastrophicly
mixed in a way that if you lookat the activity of any
individual neuron, it doesn'trepresent anything in
particular.
You know, it might, this neuronmight be active when it smells
watermelon, but it might also beactive when the sky is blue and
(33:37):
it might also be at be activewhen the animal expects a big
reward and all sorts of kind ofuseless combinations.
And the theory in some sense hasbeen that these useless
combinations are great becausethey allow you to make arbitrary
associations, right?
Like maybe want to make anassociation between a really
good reward that you expect andthe sky being blue.
And that, that information hasto overlap somewhere.
Matt (33:59):
You can do it.
Alex (34:00):
Yeah, you can do it.
So what we showed instead isthat at least in this part of
the brain, in animals that arevery well trained on a task and
kind of very, very used to onetask-
Matt (34:13):
Consistently-
Alex (34:13):
So we found neurons that represent psychological concepts
in a really consistent andcoherent and understandable way.
Matt (34:24):
Meaning abstract concepts.
Things that are not associatedwith physical sensations even.
Alex (34:28):
Yeah.
So they're, they're, yeah, theyprecisely right.
They're not linked precisely tosensations, although there are
neurons that represent thosesensations, sensation of the
watermelon versus the strawberryor whatever.
But more psychological.
So the, an example of this, youfind neurons that, uh, that
encode not just what decisionthe animal made, but essentially
(34:49):
the probability that thedecision was correct.
Matt (34:52):
Oh yeah.
Alex (34:52):
So if the animal goes right and it's high probability
this neuron fires, a lot of itgoes right.
If the end, what makes it makeschoice A and the animal has had
a lot of evidence in that, inthat favor, so the animals
almost certainly right, thisneuron fires a lot because the
animal makes choice b.
and it's also a good decision.
The animal had a lot ofevidence.
This neuron also fires, but ifthe animal makes either choice
(35:14):
without very much evidence thanthe, the animal is uncertain or
should be uncertain and thisneuron reflects it by firing
very little.
So that classically is called aconfidence neuron.
And this is one of the thingsthat Adam, Adam Kepecs has
worked on over many years, isidentifying representations of,
of not just the decision you'remaking, but a kind of
meta-cognitive confidence inthat decision.
(35:37):
And we all know that when wemake decisions we have
confidence or not confidence inthese decisions, but the idea
that representations of, of thatkind of an abstract concept and
other abstract concepts that arerelated to decision making would
be identifiable and kind ofcoherent and individual neurons,
is the real, the realbreakthrough, I think, of that
paper.
Matt (35:56):
I think it's a really cool finding here because, uh, you
know, we, we, I mean we alwaystalk about how things in the
world are represented sparselyby neurons in our brain, whether
it's and especially in thesensory cortex areas.
Alex (36:12):
Yep.
Matt (36:13):
But we've always said, you know, as part of our theory that
conceptually the same thing ishappening, you're, you're, um,
you're representing concepts andbehaviors and that sort of thing
and the same methods as yourepresent sensations.
And so, um, it's, it's nice tohear that because it makes sense
that you would build in yourfrontal lobe on top of all of
(36:33):
those sensations coming in andyou're going to build your
representations in behaviorsusing those and with the whole
behavior motor loop involved aswell.
Um, so that's really excitingstuff.
Well, Alex Vaughan, thank youfor coming in and being on the
Numenta On Intelligence podcast.
It's been a pleasure having youhere.
Alex (36:51):
Thanks very much for having me.
It's been great.
Matt (36:56):
Stay tuned for the next show where we'll continue the
series of interviews withneuroscientists and I will be
talking to Blake Richards, theassociate fellow of the Canadian
Institute for Advanced Research.
That's the very next episode ofNumenta On Intelligence.
Thanks for listening.
Hello and welcome to the Numenta on Intelligence podcast.
I am Matt Taylor and today I'mgoing to be bringing back the
Interview with a Neuroscientist,a series that I started on
YouTube.
If you haven't seen that, I haveseveral different interviews
with neuroscientists who havecome by the Numenta offices for
one reason or another, includingDavid Eagleman, Jon Michaels,
(00:28):
David Schneider, Carmen Varela,and Eric Jonas.
Today I'm going to continue thatseries and we are going to talk
to Alex Vaughan of the companyMapNeuro.
He has been working at Zador Laband has recently been recognized
for his work, creating tools tohelp map the brain.
So really interesting stuffgoing on there.
So hopefully you enjoy thisepisode of the Numenta on
(00:52):
Intelligence podcast.
Today I have with me AlexVaughan from Cold Spring Harbor
Labs and a company calledMapNeuro, which we're going to
talk about both of those thingsat some point.
Would you like to give yourselfa bit of an introduction with
what brought you into the fieldof neuroscience?
(01:12):
What are you excited about?
Alex (01:13):
Sure.
Uh, so my entry to neurosciencewas actually a little atypical.
I came in from kind of standardanimal biology behavior, not
anything having to do withmedicine.
I actually, the first animal Iever studied was a spider way
back at Cornell and Ron W., andwe were studying spider behavior
and cricket behavior andbasically trying to understand
(01:34):
how these weird little animalstick and, and kind of what makes
them perform all of their crazybehaviors.
And so I moved on from that andI ended up doing a PhD at
Stanford where I, uh, where Idid something kind of much more
directed in the sense that Istarted working on fruit flies
and fruit flies are this amazingmodel system in which we have
(01:57):
over the course of a century'sworth of work, gotten ourselves
essentially perfect geneticcontrol over a single organism.
Matt (02:05):
Right.
Alex (02:06):
So we can create mutations however we want.
And by the time I came into thefield, we could also use this,
use these genetic tools to lookat neuronal activity and to
manipulate neuronal activity inthe fly.
And fly brains are very special.
They have relatively few neuronscompared to humans and every
neuron does something unique andspecific you can find the same
(02:29):
neuron in one fly versus anotherfly.
You can't do this in a person.
So a lot of what I did was wewere studying courtship behavior
and flies and I would kind of gothrough and systematically turn
on, some neurons, turn off someneurons, look to see how it
manipulated courtship behavior,uh, and that.
And after that I started workingin a, in a more of the field of
(02:51):
decision making in mammals andstudied kind of how animals
think, how they make decisionsand moved from there to working
on making maps of the brain andthat's a lot of what we do now.
And what, uh, what I can talkabout, what we're doing at
MapNeuro is basically buildingtechnologies and tools for maps
in the brain.
And in my current work, I'mtechnically still a postdoc and
(03:13):
Tony Zador's lab at Cold SpringHarbor.
And Tony has developed a bunchof technology over the years for
what he calls molecularconnectomics, which is basically
a bunch of tools for tracingneurons throughout the brain
that work really well, likemuch, much better than, than
previous tools.
So that's what we're building.
Matt (03:30):
And all the way back to fruit flies.
I mean, you were working earlyin your career with genetic
tools, applied geneticsessentially to do neuroscience.
I find that really interesting.
I'd like to dive into that withyou as one of our subjects we
talked about today.
What's more interesting from abroad standpoint are these,
these molecules in your brainand this idea of molecular
(03:51):
structure.
Alex (03:52):
So I would say the analogy is this, that in flies, because
we have genetic tools, we canmanipulate every neuron in the
brain, uh, with arbitraryspecificity and more or less do
whatever we want to them.
And this, this lets usunderstand the brain and if the
fly were a patient sufferingfrom some sort of neurological
disease, this would let us fixthings relatively easily.
(04:15):
In humans we have nothing likethat.
Nothing like that, control andin fact the only things we have
in humans that let us sayprovide therapies are basically
neuromodulatory drugs and theseare drugs you've heard up.
So they're things like serotoninand dopamine that people
associate with reward andpleasure and addiction.
(04:37):
And there's about half a dozenof these and the interesting
thing about them and what makesthem actually a little bit like
the case in flies, is thatthey're in special circuits,
special circuits there is thatthere's kind of one or two areas
in the brain that expressserotonin and they distribute
the serotonin from a smallpocket of neurons all the way
throughout the brain.
And everything that Serotonindoes, it does through these
(04:59):
neurons.
Matt (04:59):
So these molecules are flowing through the brain.
They're moving.
Alex (05:04):
Yeah.
So there's a bunch of neurons inone area, but most notably the
dorsal raphae that expressserotonin.
And then they have these longrange projections that reach all
the way throughout the brain.
They can reach inches andinches, which is huge for a
cell, all the way throughout thebrain.
Matt (05:22):
Wait, within a cell or through cells?
Alex (05:22):
Uh, so they, the molecule, the neurotransmitter of
Serotonin is expressed in thecell and within this cell it's
usually contained in what'scalled a vesicle.
Uh, so this is basically alittle, a little transport
module, a little pod, more orless than what they do is they,
they'll synthesize it usually inthe cell body, which is in one
location.
They'll package it in one ofthese little pods and then it'll
(05:45):
be transported all the way downdown the axon to somewhere very
far away, for instance, to yourprefrontal cortex, which is the
other part of the brain thatdoesn't make its own Serotonin,
but in which Serotonin iscritical for thinking and
emotion and, and that sort ofthing.
And then at the right time, thisneuron will release the
(06:05):
Serotonin and the Serotonin thenacts as a neurotransmitter.
And what that means is basicallyit's passed from one neuron to
another to change the activityof that second neuron.
So there's a bunch of these thatdo really canonical things-
serotonin and dopamine andnorepinephrine, acetylcholine
and these ones are particularlyimportant because for the most
(06:25):
part, every drug we have in thebrain that helps people with
depression or helps peopleimprove their memory if they
have Alzheimer's, every one ofthese targets, either the
synthesis, the distribution, orthe degradation of these
neuromodulatory chemicals.
Matt (06:41):
Really important molecules.
And are these important becausethey just have certain molecular
characteristics that allow themto carry information or do you
think there's a lot more that wedon't know about?
Alex (06:54):
Yeah, so it seems like there's a couple things going
on.
One of the reasons they'reimportant is that they seem to
play a specialized role incomputation.
Uh, there are, there are kind ofbrain-wide changes that you want
to make in order to alterbehavior.
And one of the ways that-
Matt (07:13):
Like global changes.
Alex (07:14):
Global changes, and norepinephrine is a great
version of this.
So norepinephrine you can findthroughout your body, if you are
nervous, say you're giving apodcast or something and you're
a little a little excited, youwould, you would have a little
more norepinephrine than usualthat's involved in the flight or
flight fight or flight response.
Matt (07:31):
A stimulant, right?
Alex (07:32):
Yeah, so in the brain, we think that what it does is it
controls attention, and we don'tknow a ton about this, but this
is, this is the best theoryright now.
And so essentially when theseneurons and the locus coeruleus
release norepinephrine into thebrain, they'll kind of wake up
brain areas and say, payattention to this.
Pay attention to the sensoryinput that's coming in now and
act on it as quickly aspossible.
(07:54):
And what it looks like, thistiny little set of cells here
that, that releasenorepinephrine, what they do is
they signal global changes inthe brain and they so they, it's
a very precise timing.
So now is the time when you gotto wake up and pay attention and
do something, but it's more orless global.
So, and the idea is basicallythat you need to be able to
(08:17):
distribute this kind of signalto the entire brain at once.
And these, these brain regionstend to specialize on that.
Matt (08:22):
And you take advantage of these molecular circuits in your
company MapNeuro in reallyinteresting way through viruses.
And I'd love for you to talkabout how do you use viruses and
these molecules and how they,those circuits go through the
brain to help map the brain.
Alex (08:39):
Yeah.
So one of the things we'rereally interested in doing, uh,
in, in our work is actuallyhelping the field of
neuroscience as a whole movebeyond these four or five
chemicals.
Matt (08:50):
Great.
Alex (08:51):
Uh, and a lot of the reason that we want to move
beyond them is there's a 99point nine percent of the brain
or something like that doesn'texpress these chemicals, is
doing different things and hasconnectivity maps that are hard
to untangle.
And what I mean hard tountangle, I mean, if you really
want to know where a single oneof the 100 million neurons in a
(09:14):
mouse brain goes, like it startshere and it has this long
production to a bunch of otherareas,
Matt (09:19):
Axon and all of its dendrites.
Alex (09:21):
If you really want to know that, the way that you do that
is you, you label one neuron inone animal and then you
sacrifice the animal and youvisualize, once you've kind of
put the brain under amicroscope, you visualize the
anatomy of that neuron and sothat's, that's the way that you
can kind of get resolution overwhat single neurons are doing.
(09:43):
Uh, and people who've done thisfor many years in particular for
the neuromodulatory circuitsthat we were talking about and
they've also tried to do it formany of the cortical circuits
that are involved in learningand memory and motor control and
sensory processing andeverything else.
But the problem is basically thetools that we have are one
neuron at a time or maybe a fewthousand neurons at a time,
(10:05):
they're nowhere near efficientenough to create reliable maps
of the brain that address the100 million neurons you find in
a mouse brain.
Matt (10:14):
You have to understand what the populations are doing.
Alex (10:18):
Yeah, and you have to understand the, you have to
understand the map.
So one, one analogy that I liketo give is in neuroscience,
people have been studying many,many different regions of the
brain to see what they do,right?
Uh, and so they can do this,either you have lesion
experiments where you disruptactivity in one part of the
brain or you can simply recordactivity in the same part of the
(10:39):
brand and see what it does andbehavior.
And so using this approach, wehave things like, we have motor
areas, we have sensory areas, wehave areas involved in decision
making and we know where theyall are.
And you can think of these likesubway stops in a city, right?
There are all these differentplaces and if you want to go to
the spot where you're gonna,where you're gonna really, uh,
make decisions, you go to theprefrontal Cortex, that's where
(11:01):
you go.
But it's like a subway map wherewe know where all of the stops
are and we have no idea how toget between them.
That's the big problem thatwe're, that we face.
And there's been an enormouseffort in the last 10 years in
particular to try to create someof these maps.
Uh, and so one actually oneoutstanding project has been the
Allen brain connectivity atlast.
(11:21):
So the Allen Brain is uh, uh,the Allen Institute up in
Seattle, founded by Paul Allenand filled with a tremendous
collection of neuroscientists.
They've done a project wherethey label neurons in one region
and then they look at theprojections of all of those
neurons to a bunch of otherregions.
So if you label, say theprefrontal cortex, you might see
(11:42):
projections back down to sensorycortex or to motor regions that
are helping kind of output adecision.
Um, and that might be a dozendifferent different areas.
Matt (11:52):
Yeah, because you're doing a lot in the prefrontal cortex.
Alex (11:54):
Yeah, you're doing a lot in the prefrontal cortex.
Uh, and so what they did is itthrough a really heroic project,
they did 500 of these mapstracing different areas of the
cortex as they projected kind ofacross each other and in this
dense network of connectivity.
And they can make this beautifulmap that shows you where all of
the connections are.
Matt (12:14):
I think I've seen the visuals of this.
This was really impressive whenit came out a few years ago.
I'll probably try and link to itin the show notes.
Alex (12:22):
That sounds great.
They have an outstandingvisualization people, I have to
say, we've learned a lot fromthem.
Uh, so, uh, there's one, so itwas an enormous amount of work
and it's been really useful.
So now if you're looking in theprefrontal cortex and you want
to know where these neurons go,you want to know who they're
sending information to, whichregions they're distributing
information to, you can justlook and figure it out.
(12:43):
And there's one drawback of thisapproach, which is that in doing
that you no longer haveresolution over single cells.
So you're labeling thousands ofneurons simultaneously.
And when you look at theprojections, you see them all
simultaneously and they're allvery difficult to differentiate.
Matt (13:02):
Yeah, you can't tease them apart.
Alex (13:03):
Yeah.
So going back to the subwaymetaphor, it's a little bit like
saying, you know, I live in NewYork.
It's like saying all of thepeople at Wall Street who were
getting on the subway aresimultaneously going to central
park and the empire statebuilding in Brooklyn.
All right, that's just not true.
Individual people go todifferent places and your
individual neurons go todifferent-
Matt (13:21):
You don't have the resolution to understand the
true map there.
Alex (13:24):
Yep.
Yep.
Matt (13:25):
Um, so, but you're working on that.
Alex (13:26):
Yeah.
So we're working on that.
So one of the things that, uh,that Tony Zador at Cold Spring
Harbor has been working on formany years, are ways to label
neurons individually, socially,you can still use them for
tracing.
And one simple idea is you mightlabel some neurons red and some
neurons green and then you cansee where the red neurons go and
(13:47):
the green neurons go.
And what we did is kind ofabstract that concept out, such
that we label neurons with a DNAbarcode.
And so what this means is, iswe, we make a virus similar to
the virus that the Allen BrainInstitute used where they
express a green fluorescentprotein to label neurons.
What we express instead is ashort sequence of, uh, of DNA
(14:11):
such that every neuron we infectis uniquely labeled.
Matt (14:14):
That's a huge difference.
Before you would have had todetangle all of these different
reds and greens and this is red,but which neuron did it really
come from?
Alex (14:23):
Yeah and you know that that actually has been a really
effective approach for, for manyyears, Josh Blackman, among
other people, a pioneer to atechnology called brainbow where
they have lots of colors.
They're beautiful
Matt (14:36):
They are.
Alex (14:36):
They're astonishingly pretty pictures.
Uh, and using that approach youcan trace maybe a dozen neurons
at a time, maybe 20.
It was actually, it was reallyuseful for dissecting the fly
brand.
It's proved less useful in themouse brain just because the
mouse brain has so many more.
So instead of using four colorsor six colors or eight colors,
(14:57):
what we do is we abstract theidea of colors into DNA.
So every base of DNA has fourpossible letters that can be
associated with a g, t, a or c,and you can think of those as
four colors.
And because we use a sequence ofDNA that's 30 bases long, we
kind of have four colors andthen the next base we have four
(15:19):
more and four more, four moreand four more.
Matt (15:21):
So you can have a unique identifier for pretty much any
neuron.
Alex (15:23):
Yeah, so we have, so we use a, a barcode as we call it,
of length of 30.
And I get those four to thepower of 30 possible color
combinations, which turns out tobe enough to label something
like every sand grain on earth,uh, which, uh, isn't that much
more than the number of neuronsin your brain, but it's probably
enough.
Matt (15:43):
But you could theoretically map an entire
human brain.
Alex (15:46):
You could, yeah.
With, uh, with this approach,uh, on an enormous amount of
work.
And some luck, you couldtheoretically map the human
brain.
Matt (15:53):
Tell us about the work involved here because it's not a
trivial task even doing it thismethod, right?
Alex (15:59):
Yeah.
So it's actually, it's a, it'svery different than a
traditional tracing experiment.
In a traditional tracingexperiment, you would find some
way to label some neurons, youwould say, in infect them with a
virus that expressed a greenfluorescent protein and then
they would all be green and thenyou would take that brain,
sacrifice the animal, put itunder a microscope and you would
(16:20):
visualize it.
And this is amazing, right?
You have perfect spatialresolution, essentially as fine
details as you want to see.
If you want to see the finedetails of a synapse between two
neurons, you can just kind oflook and, and you can see the
what are called presynapticboutons that are, that are at
the neuronal outputs and postsynaptic dendrites that are kind
of the inputs to the that nextone, you have amazing spatial
(16:43):
resolution.
And what we do is we throw allof that out as quickly as
possible.
So if we're, for instance,looking at the projections from
prefrontal cortex to otherareas, we, uh, we grind up the
area where we made ourinjections.
Matt (16:57):
Oh, you don't take these slices and have a beautiful
rainbow like picture.
Alex (17:01):
We've, for the most part, we were getting back into that a
little bit, but, but in thesimplest part of the experiment,
and I should say that this, thiswhole approach was pioneered by,
by Tony Zador and a phenomenalGrad student in his lab named
Eustis.
Uh, and in particular, whatEustis's insight was, was you
don't have to keep spatialinformation.
(17:22):
You can infect a bunch ofneurons to label them uniquely.
And then at the injection site,if you grind it up, you'll see
all of these barcodes and maybeyou infected a virus that had a
million barcodes and you labeleda thousand neurons.
So there were a thousandbarcodes that you end up finding
in that sample.
Once you go through thesequencing process.
(17:43):
The way we do this experiment iswe take a virus that allows us
to uniquely label neurons and wedo an injection into, say, the
prefrontal cortex to label theseneurons.
And every neuron is uniquelylabeled with a barcode.
The barcode, while the animal isstill alive, the barcode is
amplified by the cell and thenit gets distributed along the
(18:05):
neurons.
So we have a protein come andgrab onto this, this DNA
barcode, it's an RNA form atthis point and drag it along the
process all the way out to itsmost kind of distal extent.
It's the synapsis that it'smaking on other neurons and
other brands.
Matt (18:19):
Is this a part of the molecular circuits you're taking
advantage of is dragging these-
Alex (18:24):
Yeah, so one of the things we've tried to do is basically
actually kind of precisely thetransport machinery that I
mentioned earlier that movesvesicles from place to place.
We can piggyback on, on thatkind of transport machinery to
move our barcodes from place toplace.
So that's actually a reallyimportant thing that ended up
working surprisingly well.
Matt (18:44):
It's fascinating.
I think inside the cell, all thethings that are going on and us
being able to take advantage ofthat to increase things like
this.
Alex (18:53):
A lot of that comes from really, really hardworking
scientists who figure it out.
Fundamental cellular processes,uh, that seemed maybe esoteric
at the time, turned out to bothhave medical relevance and to
lead to new technology.
Matt (19:09):
And we're talking about virology and genetics and very
specific areas, a niche areaseven within virology that could
come to fruition andneuroscience and potentially no
one ever thought that would beapplicable here.
Alex (19:21):
Yeah, yeah, exactly.
Matt (19:23):
Very interesting.
Alex (19:23):
So yeah, so to, to, to finish the experiment that we
ended up doing, we, we sequenceat the site where we did our
injection and that, that tellsus which neurons interacted with
which barcodes and when we wantto know where each of those
neurons projects, we sequencethe barcodes that made it to
other areas.
So if there are 10 barcodes inthe injection site and in one of
(19:45):
the distal targets, we findthree of those barcodes, then we
know that there were threeneurons that from the injection,
so that projected all the wayover there to that other area.
And if we see one of thosebarcodes in another area, then
we know that that single neuronsplit and sent to axons to two
different areas.
And that's really where we get akind of a real uptick in the
(20:10):
amount of information.
So instead of the traditionalapproach where you have bulk
projections and again, everyperson is simultaneously going
to Central Park or the EmpireState Building, we're or to
Brooklyn, we now know whereevery neuron goes on an
individual level.
Uh, and this has led to some,it's obviously we focus a lot on
the technology, but it's led tosome biological insights.
(20:33):
So actually going back to what Iwas saying earlier about alarm
signals in the brain.
The first experiment that Eustisdid was to trace the projections
of cells that expressnorepinephrine in the locus
coeruleus and to trace them on asingle neuron basis.
So where does every single oneof those neurons go?
And the previous hypothesis wasbasically that they were all
(20:55):
widely distributed through thebrain, sending a global kind of
alarm or attention signal thatwould wake up the whole brain
simultaneously.
And that's not what Eustis saw,kind of, much to our surprise.
He saw that this individualareas where a uniquely
innervated by one neuron.
So there would be an area, andor a better way of saying it is
(21:17):
a single neuron can control onearea and doesn't project to
other areas.
So there is some sort ofanatomical specificity.
There might be processes incircuits that allow you to kind
of wake up the earth, theauditory part of your brain,
wake up the visual part of thebrain and modulate the brain in
much more detail than we'vedone.
Matt (21:35):
Specifically in response to specific stimulus potentially
as a part of a learned behavior.
So yeah, that's veryinteresting.
So you already mentioned thatyou worked at Tony Zador's lab
at Cold Spring Harbor, butthere's also this MapNeuro part
of it.
And I wanted to talk sinceyou're, you're sort of on both
sides of the fence here.
(21:55):
You're, you're, you're a postdocat Zador lab and you're also
kicking off this business butyou just got this award and some
grant money for it withMapNeuro, which is exciting.
Um, so what is this, what's thedifference between, as a
scientist a scientist and abusiness person?
You know, what's the differencebetween the lab environment and
where your priorities are thereversus the business environment
(22:15):
and how do you play thoseagainst each other?
Alex (22:17):
Yeah.
Well, the first thing I shouldsay is that, uh, on the business
side, I'm still learning as Igo.
I'm very new to this.
I'm trained as a scientist.
I like to joke that anyone whowants to work with me, he gets
to make the mistake of letting abusiness be run by a scientist.
Um, but, uh, but it's beenreally fun so far and, and kind
of really, really interesting.
The main idea that we're, thatwe're going for with this
(22:39):
business is that we think thatthe technology that's been built
in this Zador Lab over the lastseveral years has legs.
And it has legs in, in twosenses.
One is that we find that thereare lots of other scientists who
want this information, thisinformation about the single
neuron projections.
And we-
Matt (22:59):
That's one of the reasons you're here, we're certainly
interested in that.
Alex (23:02):
Yeah, exactly.
And I spent a lot of my timedoing, having this kind of
conversation and trying tofigure out what we can do now
that we couldn't do before.
And that's actually one of thegoals of MapNeuro is simply to
help scientists do this kind ofexperiment.
And the reason for that is justthat while we think we've done a
great job making this experimentstraightforward and routine in
(23:23):
our hands, it turns out thatpeople have different skillsets
and some people are, forinstance, very good at analyzing
things but not so good atmolecular biology.
Some people are good at workingwith animals but not so good at
molecular biology.
Uh, and we are a lot of the goalof MapNeuro is to fill those
gaps and to help people do thiskind of experiment without
(23:44):
having to-and is the worst casescenario-having to have a single
person working for a year toreplicate something that's
already going on in someoneelse's lab.
So it's essentially ameta-collaboration.
It's, it's an effort to, uh, toprovide, uh, uh, as much
collaboration with the field aspossible to help people do a
(24:04):
kind of a new scale ofconnectomics.
Matt (24:06):
So as you're looking as, as MapNeuro grows in structure,
you're looking for other labsthat want to collaborate on this
type of work, it seems likepretty closely with them because
you want to understand whattheir priorities are and exactly
how you can help develop ways ofmethods of doing what they need
to do.
Alex (24:23):
Yeah.
And getting back to kind of whatI've learned over starting this
business is you really have tolisten to what people need and
it turns out that you can't justmake up what you think people
should need or what it would beconvenient for them to need and
then offer that.
And the example is basicallyjust that we have the first
thing we set up was, was afacility to do the core map seek
(24:46):
experiments to do kind of thecore molecular biology.
Uh, and then we offered it tothe world and it turns out
basically that that is notenough, not enough to really
help people.
Uh, and there's a bunch ofreasons for that, but the
biggest reason is that it'sreally hard to adopt new
technologies without fullyunderstanding all of the caveats
(25:07):
and all of the difficult partsof the experiment and all of the
ways it's going to fail that youmight not even find out about
until the experiment has done.
And so people are both reallyexcited about new technologies
but also reticent to put anenormous amount of their own
effort into them.
Matt (25:24):
So you're also, you not only need to offer them the
technical services for themapping, but also your expertise
and how are they, how could theytackle this.
Alex (25:31):
Yeah, precisely.
A lot of the conversations thatwe have that are super
interesting for me are learningmore about the kind of
scientific questions they wishthey could address and then
telling them which of those wecan actually, uh, we can
actually get at with our, ourtechnology.
Um, but you know, it's been,it's been really rewarding
(25:52):
because we see fields open upthat haven't had access to many
of the genetic tools, forinstance, that I talked about in
flies or even in traditionalnice.
And one of the things we'redoing now is we're doing
experiments in, um, in animalsthat are usually not studied
very much because our approachcan be used in mice.
(26:12):
It can be used in rats, it couldconceivably be used in birds, it
can be used in other, otherspecies of animals.
Uh, and so the hope is basicallythat by taking away this one
pain point of the hard parts ofthe molecular biology and
figuring out how to do theexperiment as a whole, we can
kind of make it widelyaccessible for everyone.
And, but there was a lot oflearning.
(26:33):
There are a lot of figuring outwhat people wanted and figuring
out what the pain points werefor them.
Matt (26:39):
Well, it's as, as new labs sort of adopt this and create
own experiments, of coursethey're going to get data that
lots of other people are goingto find really interesting.
Alex (26:48):
We hope so.
Alex has been here for the pastday and I saw you give a
presentation yesterday.
They're just looking at thatpresentation, the ability for,
for us for Numenta to understandnot just the matter of the
neuron, but what layers it goesto is, is really important.
Especially we're talking aboutpopulations, you know, because
all of our theories about thecortical column and what layers
(27:08):
are doing, what, so gettingbetter resolution about this
population of cells in thislayer of projects to this other
layer and this layer, that'sreally important to us because
it helps validate what we'rethinking and raise new questions
that we haven't thought aboutbefore.
Like why is this happening?
And that sparks innovation whenyou, when you run into that
thing, even if something saysyou were wrong, at least you
(27:30):
know, you can go, you can golook for other doors.
It's always interesting when weshow that kind of data, it
always make someone a littleupset.
Matt (27:40):
Yeah.
Alex (27:40):
And I think what's going on is that, uh, the data that we
tend to show is pretty highresolution.
There are lots of neurons, weknow their projections, we know
in the case you're talking abouttheir exact spacial location and
in the field of neurosciencethere's been a lot of implicit
knowledge about where neuronsare and where they go and it's
built up over a century ofreally careful work, but it's
(28:04):
also, it's combined a lot ofdifferent information in ways
that don't, that don't alwayspan out.
Matt (28:10):
It's a bit tribal.
Alex (28:12):
Yeah, and people have longstanding beliefs that it
must be this way and it turnsout that when you can actually
test it, uh, you will validate90 percent of those beliefs and
then in a very irksome way,validate about about 10 percent
of them.
So it's always a good sign whensomeone gets a little frustrated
and then wants to do anotherexperiment.
Matt (28:31):
Right, absolutely.
Alex (28:33):
That's the goal, right?
Matt (28:33):
I mean we're searching for the truth, you know, but you got
to find out how things reallywork.
I mean, that's, that's our wholemission is figuring out how it
works so we can try to build itin software.
So, uh, about MapNeuro, you'vetold us a lot about sort of how
it works and the idea of youwant a map or not high
resolution maps of where neuronsare projecting to.
(28:56):
What are the questions that thistechnology might help us answer
what, what things is this goingto help with in the future?
Alex (29:02):
Yeah.
So we are hoping, I would say toaddress three things and one is
simply, there is a fine detailof neuronal circuits in the
brain that if we can understandthose will, I think give rise to
not just an understanding of thebrain but new technologies.
(29:22):
And actually this is kind of theguiding principle of one of the
main funding agencies that, thatfacilitated a lot of this work,
which was the IARPA micronsproject and IARPA as you know,
is affiliated with the UnitedStates intelligence community
and what they wanted.
Oh, and we were part of this andas well as many other people
(29:43):
using different technologies,uh, they wanted mechanisms to
trace neurons in the brain tomake machine learning better.
The idea is that for things likedeep learning and neural
networks, they're hard.
They're hard.
It's hard to make them learn,it's hard to get them kind of
off the ground, but if we cantransport information over from
the human brain or the mousebrain onto those artificial
(30:04):
neural networks, then maybewe'll get a lot better at it.
Matt (30:06):
We could create our networks to more closely match
what we're seeing in brainsexperimentally.
Alex (30:11):
That's one hope.
The second is that we, we reallywant to just move neuroscience
forward as a field, and I thinkone of the big technological
stumbling blocks of neurosciencehas been that there are 100
billion neurons and we don'tknow where they go.
When you say it like that, it'sjust insane.
Like what's the most importantthing about a neuron?
(30:32):
It's where it projects and whoit talks to.
Do you know that for any of theneurons you work on?
No.
The, you know, the brain issuper hard in the sense and so
we really hope to make, makeprogress.
Matt (30:43):
It seems like you're in the right space right now
because people want thisinformation and there's a, uh,
there's been a lot more interestin just in neuroscience in
general over the past 5-10 yearsfor it seems from what I've,
what I've seen and especiallywith the new hubbub about grid
cells and stuff and how that allworks with everything else that
(31:05):
we're trying to incorporatefrom, from the past knowledge of
neuroscience.
So you're in a great place with,with MapNeuro and I'm really
looking forward to seeing whatyou guys come up with.
Alex (31:14):
Thanks very much.
Matt (31:15):
You mentioned to me a paper that you're currently
working on.
I think it's on bioRxiv rightnow.
It was about, it was aboutbehavior and sparse coding in
the frontal lobe.
And uh, I mean we can't go overthe whole paper, but what
interested me about it is thatit's all about how sort of
populations of neurons thatyou've shown observing them
(31:36):
through this, this technologyyou're talking about can show
aspects of the behavior that'shappening right now and you can
decode based upon what neuronsare firing, what they mean, what
they're meaning.
Can you explain that?
Alex (31:50):
Yeah, I'd love to.
So this is actually a previouswork that I did when I was in
the lab of Adam Kepecs who'salso at Cold Spring Harbor, uh,
and working with an outstanding,uh, experimentalist Junya
Hirokawa, he recorded, uh, theactivity of neurons in a
particular part of the ratbrain, the orbital frontal
cortex in the rat brain whilethe rat was doing a complicated
(32:12):
behavioral task.
So the rat had to, uh, had tosmell an odor and, and basically
figure out did it smell morelike strawberries or did it,
smell more like watermelon.
And then we would vary thedifficulty of that task by
giving the rat mixtures.
And to abstract this out, wewere basically putting the rat
in a situation where it had tomake decisions that integrate a
(32:33):
bunch of information.
So integrating sensoryinformation about kind of what,
what the, what decision youshould make, but also
information about whether heshould expect a big reward, if
it's correct or a little rewardand a and other kinds of
information like that about thetask.
And one of the things we werelooking for was basically we, at
(32:54):
the highest level, we weretrying to ask the question of
whether neuronal activity in thebrain makes sense.
That's the simplest thing andthe baseline hypothesis that
people have had in particularfor this part of the brain, kind
of frontal, the frontal lobe andthe decision making parts of the
brain.
(33:15):
The idea was was thatessentially all of the available
information that an animal hasis utterly and catastrophicly
mixed in a way that if you lookat the activity of any
individual neuron, it doesn'trepresent anything in
particular.
You know, it might, this neuronmight be active when it smells
watermelon, but it might also beactive when the sky is blue and
(33:37):
it might also be at be activewhen the animal expects a big
reward and all sorts of kind ofuseless combinations.
And the theory in some sense hasbeen that these useless
combinations are great becausethey allow you to make arbitrary
associations, right?
Like maybe want to make anassociation between a really
good reward that you expect andthe sky being blue.
And that, that information hasto overlap somewhere.
Matt (33:59):
You can do it.
Alex (34:00):
Yeah, you can do it.
So what we showed instead isthat at least in this part of
the brain, in animals that arevery well trained on a task and
kind of very, very used to onetask-
Matt (34:13):
Consistently-
Alex (34:13):
So we found neurons that represent psychological concepts
in a really consistent andcoherent and understandable way.
Matt (34:24):
Meaning abstract concepts.
Things that are not associatedwith physical sensations even.
Alex (34:28):
Yeah.
So they're, they're, yeah, theyprecisely right.
They're not linked precisely tosensations, although there are
neurons that represent thosesensations, sensation of the
watermelon versus the strawberryor whatever.
But more psychological.
So the, an example of this, youfind neurons that, uh, that
encode not just what decisionthe animal made, but essentially
(34:49):
the probability that thedecision was correct.
Matt (34:52):
Oh yeah.
Alex (34:52):
So if the animal goes right and it's high probability
this neuron fires, a lot of itgoes right.
If the end, what makes it makeschoice A and the animal has had
a lot of evidence in that, inthat favor, so the animals
almost certainly right, thisneuron fires a lot because the
animal makes choice b.
and it's also a good decision.
The animal had a lot ofevidence.
This neuron also fires, but ifthe animal makes either choice
(35:14):
without very much evidence thanthe, the animal is uncertain or
should be uncertain and thisneuron reflects it by firing
very little.
So that classically is called aconfidence neuron.
And this is one of the thingsthat Adam, Adam Kepecs has
worked on over many years, isidentifying representations of,
of not just the decision you'remaking, but a kind of
meta-cognitive confidence inthat decision.
(35:37):
And we all know that when wemake decisions we have
confidence or not confidence inthese decisions, but the idea
that representations of, of thatkind of an abstract concept and
other abstract concepts that arerelated to decision making would
be identifiable and kind ofcoherent and individual neurons,
is the real, the realbreakthrough, I think, of that
paper.
Matt (35:56):
I think it's a really cool finding here because, uh, you
know, we, we, I mean we alwaystalk about how things in the
world are represented sparselyby neurons in our brain, whether
it's and especially in thesensory cortex areas.
Alex (36:12):
Yep.
Matt (36:13):
But we've always said, you know, as part of our theory that
conceptually the same thing ishappening, you're, you're, um,
you're representing concepts andbehaviors and that sort of thing
and the same methods as yourepresent sensations.
And so, um, it's, it's nice tohear that because it makes sense
that you would build in yourfrontal lobe on top of all of
(36:33):
those sensations coming in andyou're going to build your
representations in behaviorsusing those and with the whole
behavior motor loop involved aswell.
Um, so that's really excitingstuff.
Well, Alex Vaughan, thank youfor coming in and being on the
Numenta On Intelligence podcast.
It's been a pleasure having youhere.
Alex (36:51):
Thanks very much for having me.
It's been great.
Matt (36:56):
Stay tuned for the next show where we'll continue the
series of interviews withneuroscientists and I will be
talking to Blake Richards, theassociate fellow of the Canadian
Institute for Advanced Research.
That's the very next episode ofNumenta On Intelligence.
Thanks for listening.
Popular Podcasts
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Intentionally Disturbing
Join me on this podcast as I navigate the murky waters of human behavior, current events, and personal anecdotes through in-depth interviews with incredible people—all served with a generous helping of sarcasm and satire. After years as a forensic and clinical psychologist, I offer a unique interview style and a low tolerance for bullshit, quickly steering conversations toward depth and darkness. I honor the seriousness while also appreciating wit. I’m your guide through the twisted labyrinth of the human psyche, armed with dark humor and biting wit.