[MUSIC PLAYING] Stanford University. Something incredibly cool I
just found out about yesterday, which, oh, I wish I had
known about two days ago, but still, it's useful. You remember those Siberian
foxes and that amazing thing? You breed them for tameness,
purely on a behavioral trait, and come back 30
generations later, and they look like puppies. They have the short muzzles,
and the big roundy eyes, and the cute ears, and
they wag their tails and all of that, punchline
there being, number one, evolution can move really fast. Number two in some
interesting, mysterious way, if you are selecting for
certain behavioral traits-- in this case, one where you
like being around humans and are all cuddly with
them-- what you're also going to select for are a
whole bunch of traits that are associated with
baby wolves in terms of physical appearance. Turns out, there's a flip side
of this going on right now with what are known as metro
dogs, which are dogs that live in the subway system in Moscow. And I don't know, but
they live in there, and a whole deal was made out
of the number of them which get on subways and know
which stops they're going to. And there's an entire
website devoted to metro dogs, which
are in Russian, but it has lots
of nice pictures. But what they're seeing
there is that these are dogs who, essentially,
have been feral and in feral packs for
decades and decades. And what have they been
being selected for? Being able to scrounge,
being, in fact, quite scared of humans, a lot of the
time staying far from them whenever possible because
they are raiding garbage cans, who knows what. And what they're seeing
in the metro dogs is about 30 generations
into it now, their tails are looking
more like the tails you see on wolves. Their muzzles are
getting longer. Their coat patterns
are beginning to get less fancy and
distinctive from all the different strains of dogs. These guys are being selected
for the exact same traits that went into being wolves. It's a really interesting
demonstration. Again, these guys are not
surviving because they've got a different-shaped tail. It's just part of the package. If you were going to have
this canine-like thing and be selected for
being scared of humans, functioning in packs, having a
fairly aggressive temperament, you're going to wind up looking
like a wolf after a while. Same deal, again, very rapidly. So unexpectedly,
Russia turns out to be the motherland of all
sorts of interesting things with puppies and puppies in the
past and puppies in the future and all of that, so
that's wonderful. OK. However, I don't have
any pictures of them that I've brought today,
but go to www.metrodog.edu and see what they have to say. OK, so picking up
on Monday's lecture, I recognize Monday's lecture
was probably the first one officially to be
able to blow half of you guys out of
the water because it was really tough material. Not only was it, as promised,
yet another disciplinary jump, but it was the first
wave at a lot of material that went by fairly fast,
and a lot of it quite subtle. What are the things
to focus on in there? What did we cover
Monday that's critical? Behavior genetics as
an overall approach is not doing the constructing
an evolutionary story. Who's got the best story? We win. It's not molecular
genetics looking at evolutionary change
driven by actual mutations down at the molecular level. Instead, it's looking
for patterns of behavior that go along with relatedness. What we saw was-- the most
completely useless version was, well, here's a trait
that runs in a family, and it runs in the family
more the more related the individuals are--
completely useless, bringing in the first of the
problems that we're going to have nonstop
in the subject, ruling out different
environments. We saw a number of
the classic approaches they have, which is
studying monozygotic versus dizygotic twins,
studying adopted individuals, or studying the gold standard
of identical twins separated at birth, adopted, and then
brought back together again. And in each one of
those cases, that's explaining some degree of the
confusion, eliminating some of the environment, but over,
and over, and over again, this problem of environments
are sneaking in all sorts of interesting ways. Adoption is nonrandom in terms
of where placement occurs. All sorts of stuff like that--
making it harder to tell. What we then transitioned
to was dealing with the most interesting
realm of environment, having influences, which
is this whole world of prenatal environment and
environmental effects there, one version of it being
this whole new field, fetal origins of adult disease,
the fact that early events-- fetal environmental events--
can have lifelong consequences, even multigenerational,
as we saw. And people are understanding
the nuts and bolts of how that might work,
that epigenetic stuff. So we saw that as one--
a very subtle realm of environment slipping in. And we also saw examples
where "environment" could be occurring within
an hour of birth, and where environment
is occurring despite seemingly like similar
things, like the number of math classes you take-- environment
being very critical. What we also saw, just to
make life a whole lot messier and more complicated,
was violating the rule we have all
known since infancy, which is you get equal genetic
input from each parent. The business about mitochondria
coming from your mother-- the business about, as we
saw, more subtle version-- the same equal input
of genes, perhaps, but the regulatory
control is much more coming from the mother by way
of the transcription factors, all of that and the
cytosol of the egg, sperm having nothing interesting
going on from a cell biology level, the egg
having all of this. And where you can even
get something as nutty as inheritance of an acquired
trait, Lamarckian evolution. OK, so that was plowing
through all of that, and I'm fairly certain
that reviewing the last two hours in these
last three minutes has not put anybody
back in the water who was blown out of the water. So go to sections-- this
is difficult material, and it's going to
get worse today. OK, this is hard material. Make sure you read the
extended notes because there's a fighting chance that will be
more coherent than the lecture. OK, so we got to
the point last time around of seeing that,
OK, prenatal effects-- the last thing we focused on
was the actual nuts and bolts of it, epigenetics, all of that. Now we transition to what became
the much more modern, exciting version of behavior
genetics, unless you happen to have yourself
hundreds of pairs of identical twins
separated at birth, which is to go and actually find
the gene, finally beginning to bring together-- to
marry the classical behavior genetics approaches
with molecular biology. And it was starting
around the '80s or so that people began
to be able to work in actual molecular techniques
into this whole field, and it started off in
a very primitive way and has gotten much
better since then. The version that you
can start off with is you know something
about the trait, or as we're really
saying here, you know something about that
there are differences in the trait among
different individuals, and now you go
looking for the gene. How do you do that? The first classical
way of doing that is that you actually have
an observable difference in two individuals,
in two populations, in two branches of
the same family, a difference in an external
manifestation of what's going on with
genes, a difference in phenotype-- appearance,
behavior, some such thing. And what you then do is
you use that approach to try to then find
genetic differences. You get a whole
bunch of examples of folks with one version,
a whole bunch of folks with another version,
and you start looking for where the differences are. Slightly more
constrained version, you don't actually
have the phenotype. You are not looking
at eye color. You are not looking at how
wrinkly somebody's face is. You are not looking
at behavioral traits. You're not looking
at anything external. The next more
focused step is now, you're looking at proteins,
what some proteins do. Here is some enzyme that
turns this into that, and there's two different
ways the enzyme can do it or two different variants on it. And now, we've got
this bunch of people where we take their
blood and look, and they've got this variant. These guys have this version
of how the protein goes about its business. Can we find the
genetic difference? Next step down, next
step more frustrating-- when you can't see
external differences that you're tracking. There is no
phenotypic difference that you've got to
work with, where you don't know what the protein
does, what its function is. Next, lower rent version is to
look at the same protein that comes in slightly
different sizes because that's usually
a hint that there are different functions. The way the protein
works is going to differ, and somewhere along
the way there, you came up with one
version by having some insertion, mutation, or
deletion, or some such thing. You don't know what
the protein does, and you certainly
don't know what that does in terms of the
individual's appearance, behavior phenotype, but at
least you've got something to work with there. So what it is you are using as
your starting point-- easiest external signs. Next is what would be called
intermediate physiological endpoints. Intermediate-- it's
something functional, but it's not functional like
how the person functions. It's functional like how an
individual protein functions. And then, if you're up the
creek with that, at least hopefully going for differences
in the size of the protein, the electrical charge
of the protein. For non-chemistry types, how
this version of the protein interacts with a
water environment versus how that
version-- you don't know what the protein
does, but at least you've got something
to work with. You got a difference. In all of these
cases, the strategy is to then see what are
the gene differences that go along with it? Primitive initial version
would be as follows. You've got, for
example-- in this case, you were trying to find
the gene for some disease. And this is where
the whole field started because it is a lot
easier to go after a disease. Either you have it
or you don't if you pick the right kind
of awful disease, rather than going on a
phenotype like television taste. That one's going to be a lot
harder to find the gene for. The whole field started instead
looking at extremes of really major league,
you-got-it-or-you-don't-have-it diseases. So you've got this family
that has this disease running in the family-- PKU,
phenylketonuria, for example, or Huntington's disease,
or cystic fibrosis. These were some of
the first ones to go. And what you now do is
you've got the big family, and what you want ideally is
a big huge family where about half the people have the
disease and half don't. And the initial way the
whole approach would be done was you get blood samples
from everybody there, and you start sifting
through everybody's DNA, and you're looking
for a stretch of DNA that everybody with a
disease has in common and that none of the relatives
without the disease have. And what you've
got at that point is what would be called a
genetic marker because doing it this way, you're just
sifting through a haystack. You're just trying to pick
up big, crude patterns. The difference that you wind
up seeing between these two populations of relatives
almost certainly-- in fact, as far as I know, never
once with this approach had you just found
the individual gene. You found a stretch
of DNA that's got 10, 20, 30 genes in there. But somewhere in that stretch
is the interesting one. So at least you have
a genetic marker here. You know the
genetic neighborhood where the difference occurs. So this was at one
point cutting edge, a technique called RFLP,
restriction fragment link polymorphisms. Do not write that down. And this technique
was incredibly slick at some point early
on, and it is not easily done because you are dumping
20,000 puzzle pieces over here and 20,000 there,
and you're looking for the one difference
between the two that's otherwise identical. It's that sort of task. Nonetheless, using
that approach, people began to
get disease markers where they got the stretch
of DNA that contained a whole bunch of genes
and who knows what else, but at least the
candidate gene, the gene that was floating
around that was responsible for this
disease, was somewhere in the stretch of DNA. At that point, what would be
done is a number of things. First is people would
obsessively, obsessively check their statistics to see just
how certain they are that this is the genetic marker
for this disease because you were soon
going to be advising people whether to terminate
pregnancies based on genetic tests like those. This is an area where more
stringent rules were imposed before you announce you think
you know how the universe works with respect to this disease. Really, really major
room for things going wrong in terms of it
being then applied clinically, so first big demand
would be you really have to be much more certain--
a much more demanding field than lots of others in biology. The next issue would be
this bioethics problem, which is you've got
somebody from one of these families who may
or may not have the disease. This is a disease that doesn't
get you until age whatever, and you're younger than
that, and this immediately exploded this whole
issue for the first time. Should people get
tested if they have the genetic marker
for that disease, if they are a candidate for it? And this opened up a
whole world of confusion about this, where a key
feature of these debates was the fact that you
haven't gotten the gene. You've just gotten the
candidate stretch of DNA, and there remains a very
small statistical chance that you've got the wrong one. For a bunch of reasons,
there is that possibility. So do you let
somebody take a test? Can people tell the difference
between a yes and no and a statistical probability? What do you do if it's a disease
where the person has just spent their last 10 years
watching their parents slowly die from it, and
there's no cure for it, and it's absolutely horrible? Do you give them a test that
is-- if the bad outcome is going to throw them off
the top of a building with a high likelihood? Enormously complex
bioethics issue. The third challenge, though,
was to step up the science from just having
the stretch of DNA with a bunch of genes in there
to actually find the gene. And in the years since, the
molecular biology has gotten much, much better at moving
past finding a genetic marker, we are 99.9% certain that
the gene that's relevant is somewhere in this stretch,
to finding the actual gene and then being
able to see what's the difference in it
between them and them. What's the difference,
down to base pairs? Where is the mutation? Where's the difference
coming from? That's the world of
sequencing genomes now. That's where the field is at. So that has been enormously
powerful in terms of potentially finding
one single gene. But everything that came through
in the molecular lectures last week should be
pointing out there's not a whole lot
going on in there that revolves around one
single gene at a time. So hold on to really
horrible diseases. Yeah, one gene, single
mutation, but most of what goes wrong, if you want
to start understanding the genetics of psychiatric
disorders, of diabetes, of early-- of late-onset
dementias, things of that sort, it's not going to
be just one gene. So the next huge
advance in the field was being able to do searches
like these for multiple genes. One thing that has been
used is a technique called microarrays, gene microarrays. And they were basically
invented here on campus by a guy over in the med school named
Pat Brown, who is already very famous for
having done this and I suspect has all sorts of
interesting awards coming for him down the line. This spectacular technique-- you
take what is now called a gene chip, and through techniques
that are absolutely boggling and make me feel queasy,
what you somehow manage to do is you get all the
copies of RNA that could be made in a cell
from this individual. If you are completely
new to this and you've managed
to get in this far without having to think
about what RNA is, ignore it. This is just sort of a
subtle detail for people who are interested in this. What you do is you get
a copy of each type of RNA made in a cell. You force the cell basically
to transcribe all of the genes it has. And then you get
this little chip, and you basically glue the
end of each one of these RNAs to it, and you will
have a chip with 20,000 of these little things. And then you use
probes which will allow you to see which versions
of 20,000 of those genes are different in this
individual from that individual. Total boggling, has
forced an entire new world of trying to analyze data,
this new field of informatics because you're not looking
at one gene at a time. You're looking at patterns,
and that becomes a lot harder. That becomes
incredibly important because, again, it's all sorts
of the interesting stuff where it's not going to be one gene. It's a dozen. It's 100. So that has been one
challenging approach. Another is a technique called
QTLs, quantitative trait lowside. What that basically
is built around-- that if you have subtle enough
of tests, you could figure out this is a complex trait that
is modulated by two genes, or by three, or by 17. You can see all of this
going in the direction from techniques that
would let you know you're kind of in the
neighborhood of a gene, to actually identifying a gene
that differs between the two populations, to the much
more useful realistic world of whole bunches of genes,
networks going on at a time. So using these
approaches, people have made lots of progress. The most recent
version of that-- I mentioned the other
day there, one type of mutation, a macro
mutation, is this business about variants of major
stretches of DNA copy length variance, where a
whole stretch of DNA is duplicated twice or
eleventy different times. Can you identify traits
that are occurring because of differences
in the number of copies? Yet another version
of dealing with this whole hierarchical,
multi-level business about gene regulation. OK, take home message
from this is 20, 30 years ago when
all of this started, you can start with a trait, and
through increasingly fancier techniques, you could find
the neighborhood of where the gene is that the
different versions of the gene correspond to the different
versions of the trait-- moving from patterns of that to the
neighborhood of the genes to the actual gene to
whole bunches of genes to macro differences
there, that's one approach. The other approach
of the field has been starting the other way. You've got a gene. You've got a gene, and
thanks to researching it in flies, and worms, and
rats, and all of that, you kind of have a sense
of what the gene does, and you have a reasonable
guess of what sort of traits might be relevant in
a human, and now you run it the other way. You have two different
versions of the gene, and you look for
differences in behavior, or differences in
protein function, or differences in the
size of the protein. You start with the
genetic knowledge. And for our purposes, where
people are mostly looking is at behavioral end points. So what do you see with that? Lots of findings in recent
years, and my feeling is this is one of the more
interesting ways of doing this behavior
genetics-- starting with basic research
with animals. You've got some sense
of what the gene does. How about it in people? Because there is variability. First example, and this has to
do with a really interesting hormone that we will hear
plenty about down the line, a hormone called vasopressin. And vasopressin in rodents
I think I already discussed. Vasopressin in rodents has
a receptor, the vasopressin receptor. And what we talked about
the other week is depending on which version
of this you have, if you happen to
be a vole male, it will determine
whether or not you were monogamous or polygamous. That's interesting. What was also interesting
was that difference-- the two variants were not in the gene
itself but in the promoter, and that winds up meaning that
the promoter caused in one strain the gene to be expressed
in different parts of the brain than in the other version of it. OK, so what we saw there was all
sorts of hints from the animal world that vasopressin has stuff
to do with social affiliation, things of that sort. So what do you now do
if you want to apply this knowledge to humans? The first thing you
need to ask is, well, do we know where the
gene is in humans? And with modern sequencing
in the human genome, you've got it, so you know where
the gene is for the vasopressin receptor. And the next thing
you have to ask is, does it come in
two different flavors? Does it come in two
different flavors, and do they differ
in the same way as the ones people have
studied endlessly in rodents? Yay. Yes, indeed, the receptor
gene is exactly the same in everyone, but there's
two different versions of the promoter--
the same exact story. So that's great. If you've gotten
that far, now you start looking for what
the differences are between people who have one
variant versus the other. And you've got to
do your homework. You got to match them for
age, and gender, and number of cavities, and all
sorts of stuff like that. And when you've got
that under control, you go looking for behaviors. What's the limit of this? By definition, you're
only looking where the light is shining already. You're only looking
where you already have a sense that maybe
you should be looking, unless you're totally
crazy-obsessive and want to try to measure
everything possible in people and look for a difference. In most of these studies, you
kind of know where to look. That immediately means
there's all sorts of areas you're not looking. Nonetheless, with
this approach, people have gone looking in humans
at these different variants on the vasopressin
receptor gene, and you find some really
interesting differences. One was a paper published
a couple years ago in a very prestigious
journal, which got picked up in the papers
all over the universe, showing if you have the
version of the promoter that you find in monogamous
male voles, you as a human male are likely to have more stable
relationships than if you have the other version of it. Whoa. That's pretty bizarre. That's pretty cool. Next thing, a study
that just came out in that same prestigious
journal a couple of weeks ago, and this is one
we're showing, which, depending on which version
you have, how good are you reading facial expressions
in other people? Subtle differences in
facial expressions-- then, there's a whole
other world of approaching, evidence emerging that in
families where autism runs majorly through the families,
people are seeing mutations in genes related to vasopressin
and the vasopressin receptor. So this is one
really cool example of starting with that
approach, and this is a gene whose
variants have something to do with some pretty
meaningful realms of human social connectiveness,
variability stuff. Totally interesting. Another example-- this is
a gene coding for a protein called BDNF, brain-derived
neurotrophic factor. Doesn't really matter. The main point of BDNF
is it prompts neurons into growing new processes. And it turns out, BDNF does
something that ultimately is kind of a drag for you. In one part of the
brain, the amygdala, which-- I think I've already
mentioned the amygdala has a lot to do with fear and
anxiety and that sort of thing, and we will come back to that. Don't panic. But the main thing is BDNF plays
a role in causing individuals-- individual rats-- when they
are highly stressed to make the neurons in the amygdala grow
new connections, new processes. What are you doing? You are training that
rat to be anxious. You are training that rat
to be hyper-responsive to scary things. BDNF appears to play this role. So now, the next step in
this sequence-- and you'll begin to see it's the same
pattern over and over-- people then found, ah, there's a couple
of different flavors of BDNF. There are a couple of
different genetic variants. There's a couple
of different ways in which BDNF pops up in
rats or mice or whatever, differing by a base pair
here, an amino acid there, where you then do your
scutwork and show, OK, this BDNF works faster than that
version of it or it works less but it lasts longer,
or who knows what. You find a functional
difference. There's a genetic difference. There's two different
versions of it, and it actually
makes a difference for how the protein functions. And this has all been
studied in animals, and now you understand what it's
doing in the amygdala of a rat, depending on which
version you have. And then the exact
same strategy-- go look at the BDNF gene
in humans and ask, well, do you see the same difference? Are the same two
flavors in humans popping up as in rodents? Yes, indeed. And then, the exact
same approach again, looking for differences, and
you know where to start looking. What has that
literature been showing? The type of BDNF
variant that you have in the exact same way
that it maps in rodents maps onto your likelihood
of an anxiety disorder, the levels of stress
hormones in your bloodstream, the levels of metabolic
activity in your amygdala, the same exact sort of finding. Next, a whole other
world, which we will learn much more about down the line. This is obviously a quick
survey of a whole bunch of different areas. All of these we will
come back to big time later on in the course. Next area, this
neurotransmitter, dopamine-- you will learn
amazing amounts about dopamine. Dopamine, which
works through a-- you guessed it-- dopamine
receptor-- and in fact, there is a gazillion different
types of dopamine receptors. And dopamine
receptor number four, D4, the dopamine
receptor four gene has something to do with a whole
bunch of interesting traits in rodents. And it comes with variability
in the rodent versions, different flavors, and
humans have variability along similar lines. And dopamine-- what is
dopamine involved in? Pleasure, and anticipation,
and reward-seeking behavior. What two different versions
of the dopamine type four, the D4 receptor gene,
begin to map onto in humans? Levels of risk-taking behavior,
levels of sensation seeking, levels of novelty craving. That's interesting. Another example, a gene
called NPY-- a protein, a neurotransmitter called
neuropeptide Y. Do not get overwhelmed by all of these. It's just the same
pattern over and over. NPY, same deal again-- and
two different human variants have something to do with
differences in levels, again, of anxiety, of metabolic
rates in the amygdala, all of that. And in this case,
the NPY difference is quite interesting
because it's one of those where
the difference is not in the gene structure sequence. It's once again a
difference in the promoter. OK, so what have we
have in all of these? This is a really
powerful approach. The first version
that we heard, you've got a behavioral difference. You've got a
phenotypic difference. You've got a difference in
how two proteins function, or their size. Go looking for what's the
genetic difference between those who have this trait
and those who don't. This version, reverse
engineering it, you start off with a gene that
comes in different flavors. You got a sense of what it
does from animal studies. Do you have the same
variation in humans, and does it map onto behavior? So all of these are amazing. All of these, it
has to be emphasized in a way that will come
back to haunt us also big time-- all of these
explain tiny percentages of the variability in the data. There is nothing
remotely resembling a world in which
if you have flavor vanilla of the D4 receptor
versus if you have flavor chocolate-- if you
have flavor vanilla, you are going to be hand gliding
when you are in preschool. You're going to be
totally sensation seeking. And if you have
the other version, you're going to spend your
entire life collecting stamps or who knows what. It's not that deterministic. No surprise with
this gene stuff. No single gene is going to
be particularly deterministic once you get outside the
world of phenylketonuria and cystic fibrosis and such. They contribute some
explanatory power. So on one hand, amazingly cool,
seeing these genetic links to aspects of human behavior. On the other hand, they're
not very big effects, and this is something
we will come back to endlessly down the line. OK, so this is how the field has
gone about more recently trying to find links between variation
in genes or their promoters and variations in aspect of
behavior, running it both ways. And what all of
this is looking at, what is intrinsic
in everything we've been talking about with
this behavior genetics, is oh, my God, the
cliche finally emerges. Nature and nurture, and
gene and environment, and gene/environment
interactions-- what is becoming increasingly
clear in the field is there's a third
leg to all of this. There's a third leg that
could be extremely powerful and is going to be the basis
for the chaos and complexity lectures down the
line, which is chance, the role of random
events, the role of chance in being a form that
transcends what we would call environment and is
certainly not genetics, a whole different world
where that contributes. OK, where will we see chance
playing out an important role in all of this? Brownian motion-- for those who
don't have a techie background, I will describe it in the
way that I understand it, which is flailing desperately. Brownian motion has to do
with the fact that molecules oscillate and use an intrinsic
movement, oscillation, of molecules. You take a whatever
full of water filled with all sorts of
things-- molecules, whatever-- and they're all
going to be vibrating to a certain extent because
of this Brownian motion. I suspect that was the most
amazingly incorrect description of what Brownian motion
is, but that's the point. That's the main point, simply
that there is this oscillation, which is completely random. It is just an intrinsic
feature of how the world of physics and Albert
Einstein's mustache works. And so you've got this
Brownian motion stuff going on. So where might that play
out in terms of chance? So we got here a cell with our
four powerhouses of the cell here. And on the right-- let's assume
this is a genetically identical cell, and I just got tired
of drawing it completely-- but the same four mitochondria. And what we see here
first off is genetically, these mitochondria differ. One of them has a blue spot. One of them has a red sweater. Wait a second. Every single cell in
your body was built out of the same genome when
you started off life. Ah-ha. Back to that thing
the other day. Mitochondria have their own DNA. Mitochondria are going
about their own business of dividing in the
cell, of replicating their-- replicating themselves
now and then with a mutation. Within any given cell, the
mitochondrial sequence of DNA is going to differ from
one mitochondria to another because they've been
separate organisms that are inside a cell going
about their own evolution, their own mutations,
their own whatever. What you then
have-- what would be called a mosaic of
different genetic profiles of the mitochondria
within a cell. So that in and of itself
is pretty interesting because this is a
whole other world. This is not the DNA
back in the nucleus, and this is not
variability introduced by jumping genes or
accessibility of transcription factors to genes, all that. This is-- meanwhile,
outside the nucleus, there's this whole weirdo world
of these mitochondria that are functioning
all on their own, so the different
mitochondria in the same cell will have different
genetic makeups. So great. So these two cells are
absolutely identical. Each of them have four
mitochondria, and each of them have two red kinds,
one blue kind, and one that gets nothing. So now that cell is going to
go about splitting in two. And what happens is it splits. And in the process of splitting,
what the fluid, the cytosol, the stuff inside where all
the molecules-- including the mitochondria in the
other organelles-- are found. And it splits,
and there winds up being a very simple question. Which mitochondria wind
up in which daughter cell? And what we've got here is in
this case, a split like that. In this case, a split like that. Oh my god, why did that happen? Is that because there is
a gene in this one that says when you form a new pair
of cells, have blue and blank in one and two
reds in the other? And this one is a completely
different instruction manual for how you split cells. No, simple randomness of
where the mitochondria were in the cell when they split. A totally random event. So what you see here
is you start off with two cells that are
genetically identical. And within one round
of cell splitting, these genetically
identical individuals are already
genetically different. How come? Because of instructions
or a blueprint? No. Simply because
molecules, organelles, things like mitochondria
just oscillating around. And along comes a split, and
who is stuck on which side of the divide. So you've got
variability that way. Intrinsic in that is another
type of variability, which the second I
mention it, you will be able to run
with effortlessly, which is floating
around in here also will be transcription
factors, and splicing enzymes, and other enzymes. And along comes the split. And there is not
the remotest chance that half of every single
type of transcription factor is going to go each way, and
half of every type of splicing enzyme, et cetera. It's going to be who's
oscillating where and where the cells splits. So you wind up not only
with different outcomes in terms of mitochondria
and their genetic makeup, you're going to have
different distributions of different types of
transcription factors. All due to this random
oscillatory stuff. So that's one big component
where chance plays a role. And what the chaos
complexity lectures are going to be
very heavily about is just how important
stuff like this is, how important it is in
explaining what goes on, how important it is in
explaining why, no matter how much you know about what
every single molecule is having for lunch next
Tuesday, there's still going to be a huge degree
of unpredictability. All of that is to come. This is just beginning to
establish the idea yes, nature, nurture, nature, nurture. Chance also is the third leg
in there playing a large role. Whether chance is
what's going on when you get a transposable event--
you remember at the point that neural stem cells
are beginning to divide into neurons. That's when they juggle around
their transposable genes most dramatically. Whether that is a completely
random event as to where the copies land when
they come back to Earth, I'm not sure if people know. But I suspect there's an
element of chance in there also. So this is introducing
that notion in there. So great. So now you've got
all these caviats, which is environment
is subtle and coming all over the place,
and early experience, and epigenetic changes, and all
the different ways you could find genes linked to behavior. And don't forget,
variability also has something to do with chance. At the end of the
day though, what all of the behavior
genetic approach is going to come down to at
the end is a single number. We have studied this trait
with adoption techniques, with looking for
genetic markers, with looking for
copy length variance. We have studied this trait
and the variability of it. And thanks to us studying
for a gazillion years, we can now conclude that 53.5%
percent of the variability in this trait is heritable. You come up with this
number at the end. This trait has a
53% heritability. That's the number that
winds up coming out. Scientists report that
this or that trait has variability as
80% heritability. This is the number that
always comes out at the end. What's the heritability? Ranging somewhere
from 0% up to 100%. This is heritability. And this is absolutely critical. Because this is a truly
influential number. The one that people
come away with when they are being taught,
Oh, this trait is genetic. Those who want to
be quantitative will have been told, Whoa this
trait has 80% heritability. It has 90% heritability,
all of that. And it's the wrong
interpretation. And what we're going to
spend a bunch of time on now is looking at what
heritability actually means. The heritability
number-- because it is completely different
from the everyday intuition about what it means. And what it's
usually telling you is how unimportant genes are
in the deterministic way rather than the other way around. OK. So what does everyone assume? What is heritability
telling you? It is telling you
how much do genes determine the average
level of this trait. You have one version of a gene,
and there's the other flavor. And they produce different
average levels of some behavior or something at the end. What heritability
is teaching you, if you've got this
completely wrong notion, it's teaching you
how much do genes have to do with the average
level of that behavior, that trait, that whatever. That's not what
heritability actually means. What heritability as
a number is telling you is not what genes have
to do with the average level of the trait, it's
what genes have to do with the
degree of variability around that average. This is a point where
things immediately seem panicky and start getting
very, very complicated, all of that. We will see it's not that bad. And it's not that bad
in a very critical way. OK, simple first sense
of getting at it. Here's two populations where
you measure something or other. And these three individuals,
they come in at nine, 10, 11. And these come in at one,
10, and 19, or whatever. What's going on here? You average them. They have the
exact same average. What's the difference? There's a lot more
variability around the average here than in this case. What's the wrong
idea that people have that heritability means? It's telling you
how much do genes determine the average here. That's not what it's doing. It's how much do genes determine
how much variability there is. Initially, this seems
like a very subtle point. And what's the big deal? Because at the end
of the day, it's still talking about how
important genes are. What we'll see now
is it's actually a way of seeing how less
deterministic genes are in lots of cases. So this heritability stuff,
talking about variants. Here we have an example. We've got some plant. Some plant with
some gene that comes in three different flavors. And you measure something
or other about the plant. This is how much
water it retains, or of the plant's IQ,
or something like that. You're measuring some
trait, and you're asking, Does it differ as a function
of which version of this gene you have? So you do your study,
and this is what you see. And you say, Whoa, OK. I went and looked at this
plant in a rain forest, and we identified the
genetic versions of it. And look-- very different. The gene that you have
there, knowing that, gives you a lot of
predictive power over what level of
whatever it is that you're measuring this plant is. That's great. So you're going to get
your doctorate out of that, and you get some publications. And it's great, and you
finally stop being a student. And it's terrific. And meanwhile,
some individual who shares like none of
your genes in common on the other side of
the planet, meanwhile there are some S-O-B who's
studying the exact same plant in the middle of
the Gobi Desert. You're sitting here in
the Amazon studying this, and here's this individual
doing the exact same kind of experiment,
saying, Oh, here we have this unlikely plant
that grows both in the Amazon and the Gobi Desert. But here we have this, and
we're studying this here. And what am I studying? I'm studying those
three variants. Those three variants
of this gene. And I'm asking do they
influence plant IQ? The same thing that you
were asking over here. And that individual
was doing their study, and they see that yes,
indeed, the gene influences what version that you have. And here you had A,
B, and C. And here's what they see when
they measure plant IQ. And it looks like that. And what do they conclude? Whoa, look at that. Knowing what version
of the gene you have gives me enormous
predictive power in predicting what the
plant IQ is going to be. And then catastrophically,
tragically, the two of you meet each other and
discover that you're both researching the same plant. And you look at your
numbers, and your heart breaks at that point. Because you look at these,
and what's going on here? What's going on? Let's translate this
notion of heritability, and variance, and all of that. Let's translate it into
a very simple question you can ask, translating
all of this in English. You're interested in what this
gene has to do with plant IQ. And you're allowed to find
out one piece of information. You could know whether the
plant has version A, B, or C, or you could know
whether the plant is growing in the Amazon
or the Gobi Desert. Which piece of information
is going to give you more predictive power? And what you wind up seeing
here is if you can either know this or the
environment, you want to know what
the environment is. The variability-- plant IQ could
come in at 98, 100, 102, eight, 10, and 12. Far more of the variability
in those six numbers is explained by what
environment it's going on in rather than
the genetic difference. That's what heritability
is telling you. And in a study like
this, it would tell you that the heritability number
is actually quite low. Because the amount of variation
due to this is far less explained by the gene
type than whether it's the Amazon or the Gobi Desert. So that's a first pass at it. Why is this really important? For the following reason. You're a scientist,
and you are trying to understand how A, B,
and C influence plant IQ. And you come up with
something nutty and stupid like saying, Well,
I would like to be able to do field seasons in both
the Amazon and the Gobi Desert. And your advisor will
say, or all of your elders will say something like,
No, you can't do that. Because that's just-- you're
not controlling for environment. You're not studying it in only
one controlled environment. You can't go to-- they're
such different ecosystems, all of that. Pick one and go study it there. And what you do is
you go study it there. And as a result of
just studying it here, you come away thinking
that virtually all of the variability
is explained by genes. What have you just done? You explicitly have
designed your experiment so that you can't detect
the environmental role in determining that trait. What counts as setting up the
experiment as the right way for people who think
about this sort of thing, standard sort of
approaches to experiment, what counts as doing a good
job by definition, what you're doing is biasing towards
thinking the genetic input is more important
than it actually is. Because scientists don't
decide to study this trait in two different places. You don't decide
to do a study where you're looking at some trait
in both rats and ocelots, or something. Look, pick one species. Maybe even pick just one gender. Pick one environment. Don't have your rats in one room
where the air conditioning is going nonstop and the other
room you're barbecuing stuff. Control for environment. And by definition,
if you've done the nice, careful,
responsible thing that a scientist's supposed to
do doing this sort of thing, you control for environment. What have you just done? You have just
removed your ability to see the role of environment. And you have just artificially
inflated how important you think the genes are. OK, let's take a
five minute break. We are going to look at
this in a lot more detail. Do not freak out if this is
not immediately intuitively obvious. Questions that just came up. The first one is
pointing out one of those where I probably
should have actually taught something clearer. What are you talking about with
proteins being different sizes? How do you figure out
proteins are different sizes? And the answer is
you make them stand against the wall in their
bedroom with a ruler, and you mark it there. And then you see if
they differ in size. Or if that doesn't
work, what you do is you use various
biochemical techniques. You basically make
something akin to a thick, homogeneous
soup of something or other. And you put the two different
versions of the protein in there, and essentially
you sit and wait and see how far they sink down. And the one that's heavier is
going to sink down further. Anyone who knowns about gels
and the electrophoresis, that's insanely simplified. But that's basically
the notion of it. Put them where they
move through something as a function of
their weight, gravity, and you could then pick
up size differences. The other good
question that was asked is, What did you just say
in the last 10 minutes? This is very difficult. This
is an extremely subtle point. And I'm going to hammer
it in over, and over, again here, in
sections, all of that. Why is it so important? Because number one, this
is a number, heritability, that the general lay
public comes away with interpreting
incompletely wrongly. And number two, the vast
majority of scientists, when they're working in this
field of behavior genetics, design their experiments
so that by definition, they are eliminating
all sorts of realms of environmental influences. So going back to this here. So here we see once again
this simple question. Use this one over and over
translating all this theory in variation and stuff. Always translates into
the same question. I can find out what type,
what version of this gene, or I can find out which
environment this is happening. Which one do I want to know if I
want to be in a better position to guess what's going on? And when it looks like this, you
don't want to know the genes. You want to know the environment
because it's far more powerful. And if you only studied
this in this setting, you would come away saying,
Oh, this variability is entirely explained by
which version of the gene. Oh, this trait, plant IQ,
has 100% heritability. And what you see there is
when you combine it together with numbers like these,
it's 15% heritability. So that's totally
critical to hammer in. So what are some of the
responses at that point by people who will
say, That's ridiculous. If you're saying what
heritability mostly should be teaching us is
how unpowerful genes are, what would be one of
the initial responses? Great. How many plants out there are
growing in both the Amazon and the Gobi Desert. And it's so hard to
study IQ in plants. This is a totally artificial
dichotomy between the extremes of environment. You're like
inflating things now. You're cheating in
the other direction to get the most dramatic
artificial circumstances to inflate your sense of how
important environment is. This is totally
artificially dichotomized. So think about humans, and think
about one single fact, which is we inhabit more
different environments than any other species on Earth. We live in the Amazon, and
we live in the Gobi Desert. And we live in Peoria, and
we live in all these things. And we have more exposure
to different sorts of environments. So immediately, that
argument goes down the drain. OK. So now instead, somebody
argues something different, saying OK, sometimes plants
have IQ in the Amazon and in the Gobi Desert. And I get your point,
your stupid point here that ooh, environment
can make-- No one's going to argue that the difference
between the Amazon and the Gobi Desert isn't important. Oh, yeah, OK. Well, humans, they live
in both, all of that. It's not an
artificial difference. But you notice something? Isn't it interesting
that in these two different environments, C plants
always have the highest IQ, and A plants always
have the lowest. That's telling us
something about that gene. That's telling us, and you're
then saying well, yes, yes, it differs by environment. But we've just learned something
very important about these gene versions. Which is, in totally
different environments, version C gives you
a higher plant IQ. So that's important. Yeah, yeah, environment. But we've just seen how
powerful this gene is. But then you run into the
person at the conference who is studying it in
the Gobi desert, and they put their
data up and it's even worse than in
the last version because it looks like this. And what have you just learned? That you can't say a
thing about this gene. You have just learned the
translation of this sentence. The first critical
sentence we've had over and over is,
if you can only know one factoid,
one about which gene version or what environment,
choose the environment, we have now just learned
a second sentence. A second question to ask. Which is, what does having A, B,
or C have to do with plant IQ? And if the answer
is "it depends," you've just seen this subtlety. If the answer is, "it
depends on which environment you're looking at." if
you're looking in the Amazon, C gets you the best plant IQ. if you're looking
in the Gobi desert, C gets you the worst plant IQ. What have you just shown? What is technically the
definition of a gene environment interaction? And we've just seen going
from, well yeah, yeah, they're very different,
but C is always the best. Isn't that interesting? To a completely
different profile. What does A, B, or C
have to do with plant IQ? It depends. It depends on the environment. That's how you've just
defined, that's your diagnosis, for gene environment
interaction. And what ultimately
one has to argue is that it is impossible to
ever say what a gene does. You can only say what a gene
does in the environments which to date it has been studied in. OK, let's see that
expanding even more. Because you've got this--
OK, let's jump ahead. OK, so this is showing you
now just how totally nutty and counter-intuitive
heritability terms actually are. You ask a question. What's the heritability of
number of fingers on your hand? You know, genes have
to do with the fact that we've got five fingers
instead of flippers, or some such thing. Genes have huge
amounts to do with it. You're not asking about the
average number of fingers, you're asking about
the variability. Remember that again. So what are the
circumstances out there which will give somebody
six fingers instead of five? That's incredibly rare. What about four fingers
instead of five? Oh, industrial accidents. Three fingers instead of five? Lots of industrial accidents. Two? Change jobs or whatever. What are we seeing
here in terms of, how much do genes have
to do with having fewer than five fingers? It's all industrial accidents. Genes have nothing
to do with it. There's no doubt some
weirdo disease out there. But for our purposes,
this is how it works. What have we just discovered? Number of fingers, that
trait has a 0% heritability. That's totally bizarre. That's completely
counter-intuitive. Genes have everything to do
with why the average human has five fingers, but
they have nothing to do with the variability. In that case, it's
entirely environmental. The number of fingers you
have has 0% heritability. Now let's look at
another example. It's 1950 in Eisenhower America. Actually, it wasn't until 1952. But it's a very
different world than now. And one of the things that you
would never ever, ever, never, never, ever see in
the United States would be some guy
wearing an earring. Unless in a very cloistered
part of the country he was a sensitive guy. But for most of America,
this is not what men do. They do not wear earrings. And likewise, in
most of America, if you were a good
red-blooded American woman, you would not go outside
without your earrings on. So now you've got
to say OK, well what causes variability in
earring wearing behavior? And it's entirely explained by
whether you are female or male, which is a genetic trait. What we've just seen
is, whether or not you wear earrings in 1952
has 100% heritability. Totally counter-intuitive. Think through this again
and again and again, because this makes sense. When heritability is a number
about this rather than this, you get a world
where 0% heritability for your number of
fingers and 100% heritability for whether you are
wearing earrings at that time. Because once again,
asking OK, I've got a choice in the matter. I can either know the entire
genome of this individual, or I can know whether
they are in a frat where they close their eyes and
work with a wood saw every now and then. Which fact do I want to know? That's the one that will tell
me what about the environment going on with them. Or now you have a choice. I can either know the entire
genome of this person, or I can know that what
sort of environment they're living in 1952. What do you want to know
there is, male or female. If I know that, I can
completely predict this behavior of earring wearing. So this totally
counter-intuitive thing here, where heritability is telling
just the opposite of what people intuitively think. And as soon as
you deal with that and recognize that and
recognize the way scientists do experiments, is to
try to do things as cleanly as possible--
study it only in one place, in only one setting, only one
circumstance-- you have just artificially
guaranteed that you're going to come away more
impressed with the genes than they actually
deserve to be. So how would this look? Beginning now in more detail. So what we've got here are
a number of different ways in which you can see when
are genes important, when are environment important,
that sort of thing. OK, so we have here two
different traits-- no, we have one trait, two
different flavors of a gene. Flavors A and B. And three
different environments. So here you have the
Amazon, the Gobi Desert, and a roller coaster. And you're measuring plant IQ. And there's two different
versions of the gene there. And you're asking
well, what does gene, what to environment
have to do with it? Your data look like this. What does it tell you? Environment makes no
difference at all. It doesn't matter if you are
in the desert, the rainforest, whatever. Which version of the
gene you have entirely explains variability. So this is what a heritability
of like 100% would look like. Now you do the study,
and instead you get data like these. You'll note just one of
the most important things about being a card
carrying scientist, which is data are plural. Which, talk about
counter-intuitive, earrings are genetic
and data are plural. So now you get these data
and you see as follows. In this environment,
no difference depending on what type of gene you have. In this environment, no
difference, no difference. In each environment,
very different averages. Ah, this is what it would look
like with 0% heritability. There is no difference at all
of the variation explained by gene variation. It's all environmental. So now we have a
version that forces us to put in the same phrase
we heard about before. This is what your
data look like now. And you now ask
the question, well, what does being in environment--
what does your environment have to do with your plant IQ? And the answer is, it depends
on which version of the gene you have. And now you ask,
what does having a certain version of the gene
have to do with your plant IQ? And the answer is, it depends
on which environment you're in. This is, in a sense, the verbal
definition, again, of a gene environment interaction. What do genes have to
do with this trait? Depends on the environment. What does environment
have to do with a trait? It depends on the type of genes. This is what the
data would look like. And here you have
an interaction. And in some fudging
of the numbers, this is what 50%
heritability would look like. What we just saw here before,
our hypothetical example of the messiest
it could get, this is the one where the people come
back and say, yes, yes, yes, environment matters. But you notice
version C is always associated with the highest IQ. And here we have
the version where depending on the setting, if
you have this type of gene, it gets better. If you have this type
of gene, it gets worse. This is one where there's
a dramatic interaction between what the genes
are doing and what the environment is doing. This is one that's going to
have an even lower heritability because this is one where
you are saying big time, it depends. It depends on what
the gene type is. It depends. It depends on what
the environment is. This is, in a sense,
as dramatically as it could be that case. So you constantly have
stuff like this going on. Let's look at an example
where, in fact, you wind up seeing something like this. Back to our iconic mutation
disease from the other day there, PKU. Phenylketonuria. Just to remind you, do not
worry about the details. It's this disease where normally
there's this thing in your body which would be
toxic to your brain unless it's turned
into something else. And this is the
enzyme-- the protein, and thus coded for
by a gene-- this is the enzyme that
turns the scary version into the safe version. And in PKU, you have a
mutation in this enzyme, it's not doing its job. The scary version builds up and
wipes out your nervous system. That looks bad. And what we've got now
is a trait where you're going to say, well, let's see. I am very interested
in knowing, being able to predict whether or not
this person has a fairly normal looking brain or whether
it looks like Swiss cheese. Do I want to know whether
or not they're growing up in Idaho versus Montana? Or do I want to know
whether or not they've got a mutation in this enzyme? Obviously, you want
to know if there's a mutation in the enzyme. Seemingly, heritability is 100%. But now you do
something which people have been doing for people with
PKU for a couple of decades now. Which is, you put little labels
on bottles of food and stuff saying, if this food contains
the scary bad news thing. And if you have
PKU, if you simply don't eat food that contains
that-- if you have what's called a fennel
alanine free diet-- you don't get your brain
looking like Swiss cheese. What have you got there? You've just put in an
environmental influence that has reduced
heritability down to zero. This is a case where we've got
something like this looking. And here we're measuring is,
like, the Swiss cheese index, or whatever of the brain. And what we've got here is, this
is an environment where people know how much of the stuff there
is in different types of food and don't eat it. A simple environmental--
not so simple-- but an environmental
intervention. A behavioral change
in the environment. And you've just reduced
heritability from 100% to 0%. So that is a dramatic version. Now, let's look at a version
of a real life example of this. This would be with-- I think
I've already put this one up. OK, does this work
familiar by now? No. OK, that was at Thanksgiving. OK. What we've got here is a
gene that's got something to do with depression. It's a gene that has something
to do with neurotransmitter serotonin. Details don't matter,
they will come later. But the gene comes in
two different flavors. And one version of it,
based on everything that's known from
laboratory animals, one version should be
the kind that predisposes you more towards depression. So, in this incredibly
important study that was done a
number of years ago, a guy named Caspi, Duke
University, and colleagues, they had been studying 17,000
people basically from birth up to age 25 or
so in New Zealand. And who knows why
in New Zealand, but studying them there,
and following and asking. They took blood so they know
the whole genetic profiles. And looking at, among
other things, by age 25, does the person have
major depression or not. Have they had an
episode of depression. And now let's go look at which
version of this gene they have. And what would everything
predict from all the animal studies, which is you have
the bad version of the gene and you're, as a human,
going to have a higher chance of having a depression. That's not what the
data looked like. Instead, it was one
of those "it depends." It depends on the environment. What you saw was,
it all depended on how many major childhood
stressors there were. Loss of a parent, divorce,
abuse, some such thing. And the absence of
any of those, this was the rate of depression
in the people having the good version. And throw in one childhood
trauma, two of them, three of them, and the
incidence is going up slightly. Now, you've got the bad
version of the gene. And it puts you
more at risk only in certain types
of environments. So here we have a great gene
environment interaction. That's something
straight out of this one. You're saying well,
what does this gene have to do with the
risk of depression? And the answer is, it depends. It depends on what your
childhood environment was like. But now, one additional detail
showing that this is actually a case of this. One that just gets
ignored whenever people talk about the
study because it's cool, but it's actually not
all that important. OK, so here's the people with
the good version of the gene, but it makes this point though. You've got the good version,
and they're down here. And you know what's
going to happen, the bad version is going
to be exactly the same. But you look at the data
closely, and in the absence, the bad version is actually
a little bit protective from depression versus
the quote "good version." What have you just
discovered there? It's one of these. This is a radical gene
environment interaction. Not only is it the case that,
does this put you more at risk? Well, that depends
on the environment. Depending on your environment,
it could put you less at risk. This one is not a big
deal because there's a tiny difference. There's no reason
to pay attention to it, other than just sort of
pointing it out pedagogically. But this is an amazing
example where this can go in the opposite direction. This is a hugely important
gene environment interaction. So what all of this begins
to set you up for is, you've forgot this
heritability number. And people come away
constantly thinking about, it's telling you about how much
the average level of a trait is determined by genes. It's instead telling you
how much the variability. What that means, thus, by
definition, that there's all sorts of nutty
counter-intuitive things going on, where earrings are totally
heritable and number of fingers are not. What that tells you is, by
definition, the way experiments are set up to make things nice
and clean and interpretable, scientists typically remove
environmental variability. Scientists typically
artificially have boosted up the seeming
heritability of a trait. What does that tell you? The more different environments
you study a trait in, the lower the heritability
is going to be. Because you are
going to be getting more and more opportunities
for things to be different. More opportunities to be
able to say, it depends. It depends on which environment. We've just studied it in
99 different environments, and it always looks like this. No difference by
environment, and genes always make a huge difference. And then you go out and,
before dying of fatigue, you study it in the
hundredth environment, and it now looks like
this sort of thing. What you've just learned is,
this is a "it depends" example. And the more environments
you study something in, the lower heritability
is going to be. Translating that into
English, the more environments you study a genetically
influenced trait, the less interesting
and important the genes are going to be. The less interesting
and important they're going to
be in answering, what does this gene have
to do with whatever? And what we're seeing
over and over again is, the only way to answer
it over and over again is, it's going to
be an "it depends." It depends on the environment. Ultimately-- as I said a
little while ago-- ultimately, it winds up being meaningless
to ask what a gene does. Ultimately, the only
really truly scientific way you can answer a
question like that is, what does this gene do in
this particular environment? People make this big deal out
of oh, genes do something, environments do something. And every now and
then, hooray, they interact in some exciting way
in teaching the gene environment interaction cliche,
that's what's happening every single time. And that winds up
being the basis of that quote from Paul Ehrlich
in this department that I put on the
handout, which is great, summarizes this entire
point, asking whether genes or environment have more
to do with some trait is akin to asking whether
height or length have more to do with the volume
of a rectangle. They're inseparable. There is no such thing
as a gene influence outside the context of an
environmental interaction. OK. Let's look more at
what this looks like. Here, this being
one great example. And this being this iconic one. And this is a finding that
is destined to be, like, the most important
in the last quarter century in biological
psychiatry. It is the most
powerful demonstration in the realm of abnormal human
behavior of what your genotype has to do with your behavior. The most amazing, logical,
intuitively reasonable demonstration of "it depends." It depends on how stressful
the childhood environment was. Really great. Bringing things
back to last weeks sort of molecular
lectures, people already know who study this. The two different versions
of the gene, certain classes of stress hormones interact with
the two versions differently. When stress hormone levels
aren't particularly raised, that difference isn't
being manifested. The more of a history
of stress-- and it's obviously more mechanical,
more complicated than just how much you're
exposed to-- but the promoters, the variation here
is in the promoters. The promoters
interact differently with those glucocorticoid
stress hormones. In the absence of the
glucocorticoid stress hormones, because in the absence
of something exciting, your genetic difference
makes no difference. So we've just translated
an epidemiological answer-- what does this gene have
to do with depression? It depends. It depends on your
childhood stress history. We've just translated that into
last week's molecular biology, what does this genetic variance
have to do with whether or not you have depression? It depends. It depends on how much childhood
exposure to glucocorticoids you had. So we've just leaped
from analyzing this in the context of this
field, to translating that into last week. And translating that
into the endocrine lectures that are going to come
next week, beginning to show, OK, you're an epidemiologist. The answer is, what
sort of environment? You're an endocrinologist. The answer is, well, the
hormonal environment. Whether you define it,
it's the same punch line. It depends. It depends on the environment. So this is the best,
most amazing example of this in all of psychiatry. Except, these guys--
Caspian colleagues-- measured something
else and they got just as bestest of an
answer for that. Which is, they looked at
another gene that has variation. One we will hear
about down the line. A gene called MAO,
monoamine oxidase. All we need to know right now
is, a gazillion animal studies suggest that one
version of the gene predisposes more towards
aggression than the other. Whoa. So they've got
17,000 people who've just had 25 years
worth of opportunity to be aggressive or not,
and they got the genes. And now they ask the question,
rather than at age 25, has this person ever had a
major clinical depression? Now they ask, at age
25, has this person ever been in trouble with
the law for some sort of antisocial behavior? Antisocial behavior
is the new jargon for what used to be called
sociopathic behavior. Has this person been
in trouble with the law for sociopathic
behavior, anti-social. And does it vary depending on
which version of the MAO gene you have? And what you get is,
not only is the answer "it depends," the
graph is basically identical to this one. It's superimposable, the
magnitude of the effect. What was this in that case? Whether or not you had, and how
severe your child history was, of abuse. The more childhood
abuse you had, the more having the wrong
version of the gene increased your odds of
having anti-social aggressive behavior. Once again, a major
dramatic "it depends." And it's stunning. You look at these
two different papers, both were published in
"Science" a couple years apart. Landmark studies, all of that. And the two graphs really
are virtually superimposable. The same magnitude of effects. The same answer. It depends on your childhood. So that is a very
strong example. Next one, another one, now you
go back to this gene difference with the serotonin,
the depression world. And now you study it in monkeys
and the variation there, there is a variation in
brain chemistry related to depression, all
of that depending on how stressful the
monkey's childhood was. Same exact sort of finding. What else? Now when you begin to
look at-- going back to that world of dopamine
receptors and thrill seeking and all of that--
and what you look at there is variability
in a certain type of different dopamine
receptor gene. And what you find
there is, if you have a certain
version of that gene, you have less social attachment. There's a whole psychology
of how you measure that. And God knows why. It's associated with
less social attachment if, and only if,
you got brought up by a mother who was
cold and withdrawn. There is once again
a "it depends." My looking at that, it's
nowhere near as clean as this. But it's the same
sort of theme again. Another example of it. There's another
gene called FAD2. And don't even ask me
what it stands for. But which version
of the gene you have has some predictability
over your IQ. That's kind of interesting. It's not a big effect, but
nonetheless it is demonstrable. Which version of the gene
you have has some control, but there's a "it depends." Which version of
that gene you have has something to do with IQ
if and only if something. Let me tell you what
this FAD2 is involved in. It is involved in
carbohydrate metabolism. Carbohydrates-- carbs,
carbohydrate metabolism-- it codes for some enzyme
that breaks down carbohydrates and no
doubt, some certain kind. OK. What's going on here? You've got two
different versions of a gene that break
down carbohydrates, and you've got one version. And on the average, you'll
have a higher IQ if and only if some environmental
thing was happening. Any guesses what that
environmental thing might be? High carbohydrate diet. High carbohydrate diet, OK. That makes tons of
sense, yes indeed. That winds up being relevant. Remember, you're
now-- not remember, because I didn't
say it already-- but you're looking
at IQ in kids. So you're looking now
at carbohydrate stuff going on early in life. So frame it in that context. So, we've got a start
of an answer here. So what else could be happening? If and only if the
version you have of this gene that's involved
in breaking down carbohydrates, translating into
how you're dealing with the amount of
carbohydrates in your diet, gets you a different IQ
with an if and only if. What would the if
and only if be? Build on that comment there. What else could it be? OK, I'm not even going to look. Somebody shout out an answer. [INAUDIBLE] You're right! OK, Who started? Somebody started
saying something. Who started saying something? OK. Any other ideas? Breast feeding. Who said that? Look at that, they
start pointing. The person doesn't even fess up. Who said that? It's in your notes. You're right. [LAUGHTER] OK. You want to be that way. I'm out of here. OK. So if and only if you read the
damn notes, as it turns out. Yes! Breastfeeding. OK. If and only if you were having
a particular type of diet very early in life
that is extremely rich in the types
of carbohydrates that this enzyme works on. So a very dramatic
if and only if there. I don't know why I
write that stuff. I don't know why I show up here. I should be doing
one or the other and some sort of environment
interaction thing. So yeah, breastfeeding. OK. So one more example, a
very interesting one. And this is one that has at
some point or other, I'd bet, been pertinent to every
single person in this room. Which is, there's another
genetic difference that has something to do with
a certain cognitive aptitude. And this one
everybody knows about. And this one is as follows. If you have-- there's two
different versions of this sort of genetic picture
and they're associated with different
levels of aptitude in this particular
cognitive realm-- this is not a world any more
of one single gene at a time. FAD2, any of these. This instead is talking
about a whole bunch of genes. If you have a Y
chromosome or not. If you are male or female. A genetic trait. And what is the thing
that has been demonstrated most consistently
in the literature forever and ever
and ever in terms of a gender difference in
a cognitive skill in math performance. That has come through endlessly. I've already talked
about it the other week. A small difference in the
average median performance on junior high school
old Johns Hopkins superstar kids taking the SATs. But a big difference in the tail
at the end, high performance, all of that. This comes up in
endless, endless studies. This comes up with kids
at fairly young ages. There is this gender difference
on the average-- yes, we're saying on the average, don't
forget, you say on the average before you even ignore that--
you say on the average, boys are better at
math than girls. Men are better at
math than women. Male plants in the
Gobi desert are better in math than female
plants there. And so we've got a
genetic trait here being a function of whether
or not you get a Y chromosome. But then, a few years ago there
was an astonishingly important paper published in "Science." And what these people did
was look at math performance scores-- of I think it was
480,000 different high school kids. And all over the world. They didn't just
study it in America. They studied it in 40
different countries. And they asked a
very simple question. Which is, are there big
differences in gender quality of life issues in this country? And there is a whole index that
comes out of the World Health UN something or other called
a gender equality index. Which takes into account,
like, if there is-- and if there is, how
dramatic of-- a gender difference in educational
opportunities, in freedom of movement, in
freedom to serve in an elected office, and freedom to vote,
in freedom to choose who you're married to, and obviously
the enormous variability on this planet in terms of that. And what they showed was,
the greater the inadequacies, the greater the difference
in gender treatment in the society across
40 different countries, the bigger the difference
there was in math scores. It's not a function
of gender, it's which society you're
growing up with your gender. What was at the most extreme? Let me make sure I get
the countries right here. Who had the worst profiles
of the 40 countries in terms of the biggest gender
differences in these quality of life measures? Turkey, Tunisia,
and South Korea. Where was the United States? Sitting somewhere
around the middle with most Western
European countries. And which were the countries
on earth which, as a block, had the lowest degree of
different treatment of people in their society
based on gender? The ever handy wonderful
Utopian Scandinavians. So in comes the Scandinavians. And what you show
is, by the time you look at the country on
Earth that has the least gender differentiation of
any, which is Iceland, you notice something different. There still is a
gender difference. In Iceland, girls are better
at math than boys, slightly. A small difference
going on there. But nonetheless, as you go
from the countries in which from day one, girls
transitioning into women are given the most
constraints of freedom of life, that's where you're
going to see the biggest math difference scores. And you look at less and less
of those sort of inequities. And the gender difference
in the math score, by the time it's at the
Scandinavian countries, it's down to zero. And then you get to Iceland,
and it actually reverses. It's got nothing to do
with your Y chromosome. We immediately come
back to the first of those questions from before,
that diagnostic question. You've got a choice. You want to have a sense of,
in this population-- nah, even at the individual level. You've got a choice. You're comparing
two individuals, and you want to
guess which one is better at math than the other. You can either
know their gender, or you could know whether they
grew up in Tunisia or Iceland. Which fact do you want to know? You want to know
about environment. Environment is vastly
more powerful there. Another feature of that
that has been interesting. So you then say
OK, the extremes. OK, so in some societies, the
means are exactly the same. But what about that difference
way out at the highest level of performance. As I mentioned the other
day, in the mid-1980s, looking at the highest first
percentile of math performance on junior high school
kids with their SATs, and there was a 13:1 ratio
of males to females in there. When it was last
studied a few years ago, it's down to a 3:1 ratio. Oh, that's obviously due to
evolution over the last 20 years, because it's
got to be due to genes. That's like saying genes explain
the fact that in 1980s people, like, wore pads
on their shoulders and, like, power sneakers
to work or whatever. And the fact that
nobody does that anymore shows that the gene for wearing
that on your woman executive decor deal has evolved. That's asinine. You do not go from 13:1
to 3:1 with a trait in 20 years in a
highly interbreeding mixed population-- in
other words, humans-- and have this as
a genetic trait. Even function at the
extreme has squat to do with the
genetics of gender. It's, I want to know what
society the person got raised in. I could care less what
their chromosome is. If I want to know
how good they're likely to be at math
compared to the other person, tell me where they grew up. That's the most
important thing to know. Interesting additional thing. The second most reliable
finding in all of measures of cognitive aptitude thing
that has a gender difference is, on the average, girls being
better at verbal performance tasks than boys. Women than men, all
that sort of thing. So what's going on in those
40 different countries there? What you wind up seeing is,
it depends on which country you're in. It depends on that
gender equality index. And what you see is, in
the worst countries there, by these measures, Tunisia-- who
else was it?-- Tunisia, Turkey, and South Korea, what
you see is, yeah, women get better scores
on verbal tests than men. And as you go to the
more equal places, the gender difference increases. In other words, you've
got one of these. The more gender
equal a society is, the less of a sex difference
there is in math capabilities and the bigger the advantage
is for females over males in verbal performance. It's got everything to do
with what society you're in. This gender difference in
this realm of cognitive skills means next to nothing. And if I had any
technological skills-- I kept saying I
was going to bring the figure from this
paper, I'll have it posted-- because
this one figure, it's across all these--
you just look at it and there's the answer. It's got nothing to do with the
genetics of gender difference. OK, so what else would one
want to emphasize here? OK, what else would one
want to emphasize here? Which is, this is a total
mess and totally complicated. OK what have we gotten to so
far in this field of behavior genetics? There's all these
different ancient ways of inferring something. Twins, adoption, there's the
much more modern way, which is finding the actual genes. That's wonderful. That's exciting. Nonetheless, over
and over and over, an environment
gets understudied. Environment is far more
subtle than you think. Environment starts earlier
in life than you think. And when you do it in a very
formal quantitative way, analyzing what genetic
influences are about, you discover that
scientists study things under circumstances where
you constantly underestimate the importance of environment. Blah, blah, etc., etc. But at the end of
the day, don't genes have something to do
with something or other going on with this? And obviously, obviously they
have some very important roles. You have genes that completely
determine aspects of behavior. Single genes that completely
determine your intelligence if you are a housefly. Lots of interesting
studies there showing single gene determination,
a lot of these behaviors. And it doesn't matter sort of
what the economic opportunities are for houseflies in that
country versus humans. What you see there
is yeah, there are realms of behavioral
biology where genes play very, very strong roles. And we will certainly
see some of them in the lectures to come. But with two qualifiers. The first one we
heard the other day. Which is, even when, with all
of these criticisms on board and you've got
every critical tool to slice up anybody
arguing that here's a high heritability important
genetic component, blah, blah, etc. You've completely
cut it to pieces, and there still is
a gene that seems to be doing something important
for trait X. What we heard about the other day is,
nonetheless, think about if there is an indirect genetic
route for getting there. Is there a gene
for extroversion? Or is there a gene
for your height and how people of your
height are treated? Is there a gene for
picking at grubs? Or is there a gene for if you're
really tall, you pick at grubs and people don't
make fun of you. Is all of that business about
indirect genetic effects? So even when you see
a gene for something, you have to begin to ask to
make sure that you are not, in fact, looking at
an indirect effect. Final qualifier. When you look at the
really interesting genes in terms of what they do. When you look at
some of the ones that have the most
to do, say, with what differentiates our genome
from the chimp genome. When you look at things like
juggling your DNA just when you're making new neurons,
what you see over and over is, what human genes are
about most dramatically is coding for ways in
which you have freedom from the effects of genetics. And that is going to be a
theme endlessly in the lectures to come. OK. For more, please visit
us at stanford.edu.