Cookies   I display ads to cover the expenses. See the privacy policy for more information. You can keep or reject the ads.

Video thumbnail
- [Justin] Survival of the Fittest.
For many people,
this phrase is synonymous with evolution.
But we see in nature that sometimes
creatures can act altruistically.
Meaningfully hurting their own survival
and reproduction chances to help others.
In this video, we're gonna built some simulations
to get a better understanding
of which kinds of altruistic strategies work
and which don't.
(soft music)
Alright, let's jump right in with the simulation.
We'll start with the world from the video
on simulating natural selection.
In this world, blob creatures start each day
at the edge of the world.
In the morning, the food appears
and the blobs go out to eat.
The amount of food the blob finds
before running out of energy and returning home
determines whether it dies,
survives till the next day, or reproduces,
passing its genes on to another creature,
except that the genes might mutate.
Now, let's get the creatures
the ability to be altruistic.
Here's how it'll work.
If a blob creature finds two pieces of food
and still has energy left,
it can take one of two routes.
It can look out for itself and its descendants
by deciding to go home early and reproduce,
or it can be altruistic.
Risking its guaranteed offspring
to go and give a piece of food
to another creature who hasn't eaten yet.
And yeah, they regurgitated.
Nature's gross.
At the beginning of our simulation,
half of the creatures will have a copy
of an altruistic gene causing them to be altruistic
every time they get the chance
and the other half of creatures will have copies
of a competing non-altruistic gene.
When we let this simulation run,
what do you think will happen?
Would the selfish creatures take over
or will the altruists triumph through teamwork?
Or maybe they'll stay mostly balanced.
Pause here and make a prediction.
Prediction made?
All right.
(soft music)
Okay, well that was kind of sad.
It turns out that unconditionally sacrificing your offspring
isn't a great long term strategy.
So, how can we give the gene for altruism a better shot?
Well, what if we make acts of kindness
a bit less punishing to the altruistic creatures.
Say, by letting the creatures keep some reproduction chance
when they give their food away.
50% instead of the previous 0%.
So the cost of giving food away
is half of an offspring on average.
Maybe the food was already partially digested.
Again, nature's gross but ickiness aside,
this makes the interaction net positive
instead of just net neutral
which is actually pretty common in the real world.
Okay, so let's restart our simulation
with this lower cost altruism in place.
Now what would you predict?
Hmm.
(light music)
All right, it still doesn't work it seems.
Remember that for a gene to be successful in the long term,
it needs it's copies to keep replicating.
The problem with the gene for purely unconditional altruism
is that it helps copies of competing genes
as much as it helps copies of itself
and it's competitors don't return the favor.
So a successful gene for altruistic behavior
would need to find some way of getting more help to itself
than to it's competitors.
Even if we're making nice creatures,
the gene itself still needs to be selfish.
How could a gene for altruism find a way
to let it's copies coordinate with each other.
One way is to combine two different traits
into the altruism gene.
First, some kind of unique outwardly detectable trait
that can let the gene be recognized
and second, the trait to be altruistic toward creatures
who have that detectable trait.
So let's do that.
Let's add an outwardly detectable trait
to our altruistic creatures.
The classic version of this is green beard
and that's a fun thing to put on the blob,
so let's stick with that.
So the next simulation we'll try,
we'll start out with half creatures
that have the green beard gene
who'll be altruistic toward other creatures
with green beards,
and half creatures without green beards
that will neither help nor get help.
Again, pause to make a prediction.
Are you convinced that the green beards should do well
or might there be another problem?
Let's see.
(light music)
Cool.
I was honestly a little bit worried
before running these simulations
that it still wouldn't work,
but it does.
Maybe you're not that surprised and that's fine
but even if you're not,
this is still a pretty cool moment.
We found at least one kind of gene
that can crack natural selection by causing creatures
to put others before themselves,
even if it's only sometimes.
This is called inclusive fitness.
The fitness includes all the copies of the gene,
not just the ones inside a particular creature.
Don't celebrate too much though
because there's still a problem.
Traits like green beard altruism
aren't actually very common in nature.
There are a few known cases,
for example red fire ant colonies
can have more than one competing queen.
And apparently, the workers can tell which queen
shares her and sets of genes with them,
and then they kill the queens that don't match
and help the queens that do match.
That's cool and everything
but there just aren't very many examples like this.
It turns out to be pretty rare for one gene
to code for two different traits
that happen to work together so nicely.
And even if that does happen,
eventually, mutations could produce multiple genes
that each code for only one of the traits.
So let's set up a simulation to see how it looks
when the traits are on separate genes.
With the traits on separate genes,
they're independent leading to four possible combinations.
The creature can have both, neither,
just green beard or just altruism towards green beards.
Time to make another prediction.
(light music)
Okay, so as I kind of hinted that before that simulation,
the coordination between copies of the altruism gene
is broken
and then the non-altruistic creatures dominate.
But hey, green beards are still cool.
So, we've only gotten one kind of altruism to work so far
and it's a kind that depends on a rare coincidence
and doesn't appear much in nature.
There's got to be something better, right?
Well, in fact there is.
It's known as kin altruism or often kin selection.
Instead of targeting some outwardly detectable
genetically determined trait,
this kind of altruism targets family members,
whatever the traits may be.
So, let's simulate one final version of the altruism gene
that causes creatures to be altruistic
toward their direct parents and direct children.
Now the whole point of this kind of altruism
is that we can't see which creatures have which genes.
So this time, let's hide the graph
and try to predict the results together
while the simulation runs.
The key concern with kin selection
is that even close family members aren't guaranteed
to carry the same gene.
So the altruism gene has to do some gambling.
For any kind of gambling strategy to work
while in the long run,
the cost of playing needs to be lower than the payoff
for a win times the chance that you actually win.
Right?
The average payoff needs to be higher than the cost.
In the context of kin selection,
you'll hear this called "Hamilton's Rule."
Looking at this simulation
and thinking of the altruism gene as the gambler,
the 5% mutation chance
means that there's a 95% chance
that parents and children share the same version
of the altruism gene.
So that's our chance of winning.
The cost of being altruistic as we decided before
is half of an offspring on average
and the benefit to a creature who receives food
is one since that food is converted directly
into offspring.
These numbers aren't exact since both creatures involved
do have other chances to get food
but this should get us pretty close.
And comparing, the expected payout is almost
twice the cost,
so even with the inexact cost in payoff numbers,
it seems pretty clear that the altruism gene
is gonna do well here.
And this is where I realized
that altruism is an illusion
and my heart descended into darkness
only for a little bit though.
Once I dug in, collected some data on what was happening
and found more precise numbers for the cost and benefit,
I figured out what was wrong.
It's that Hamilton's Rule is a lie!
Which I'm sorry to say is gonna require it's own video.
But for now, suffice it to say
that by lowering the cost of the altruistic act,
and cranking up the likelihood of winning
by lowering the mutation chance,
we can find a set-up where a gene for kin selection
tends to flourish.
This is the kind of altruism we see
all over the place in nature
from parents caring for their young
to sterile worker bees helping the queen,
conclusive fitness can be naturally selected.
All right we spent a lot of time in the weeds
in this video,
so before we go, let's not remind ourselves
of the difference between a creature and it's genes.
The genes involved in altruism are still selfish.
The only ones that survive
are the ones that are able to coordinate
their own copies.
But this does not mean the creatures themselves are selfish.
They genuinely care about and make sacrifices
for each other,
whether it's because they're family
or because they just can't resist the look of a green beard.
See you next time.
Hey, thanks for watching.
If you enjoy this video and wanna support more of them,
you can help by subscribing,
sharing with someone else
who you think might also like video,
or if you're so inclined by supporting directly on Patreon.
In any case, thanks again.
(light music)