So, it's been really exciting
to be an astrobiologist nowadays

because astrobiology is starting to
intersect a lot of other fields

that care very deeply about
understanding life -

from the perspective of thinking about
more general principles.

And, one of those fields is
the artificial life field,

which has been exploring
fundamental properties of life

and evolving systems
for a number of decades,

and has made a lot of progress
in computational models

of living processes that are starting to
be able to be used for thinking

about problems relevant to astrobiology.

So, I'm going to just talk about
a little bit of...

where artificial life comes
into providing some new insights

into what biology might be,

and how we might think about life

as a broader class of phenomena
from this astrobiological perspective -

that we want to understand life,
not just on Earth,

but life as it might exist anywhere.

And, one of the things
that I always find really intriguing -

being trained in physics -
is that biology seems to necessitate

a different kind of perspective
on how we should think about

the laws of nature,
how we should think about...

dynamical systems.

And, one of the things that's really
a pretty interesting contrast

between physics and biology is that -
in physics, when we model systems,

we always talk about like
the initial state.

Like, you could talk of the initial state
of the universe

or initial state of the solar system.

And then, you have some fixed
law of motion,

like Newton's law of gravitation,
or we could take Einstein's theory.

And then, we can evolve our system
forward in time

and we can predict the final state.

In principle, we should be able
to do that with...

perfect prediction power if we had,
you know,

big enough supercomputers
and things.

But, there's nothing about the laws
of motion that changes in time.

So, Newton's law doesn't change -

it's just the states that evolve
forward in time.

But, in biology,
we have this really interesting case

where we have an initial state,
like that last universal common ancestor

of life on Earth.

And then, as that system evolves,
it changes,

and there's many possible final states.

So, even though we started
from sort of a...

shared cellular architecture
and early history of life,

because of the interaction of life
and its environment

and the change of information
in those systems,

we've ended up with many, many
different final states for biology and...

So, I'm an organism that descended

from that last universal
common ancestor.

And, so is the bacteria - you know -
on the screen...

the computer screen
that I'm looking at right now.

There's many final states
possible in biology.

And, one of the ways
that we can talk about this

is to even talk about the fact
that in biology

this - the states themselves -

are not the only thing that's evolving,
but the laws are too.

We talk about state dependent rules
or state dependent laws in biology,

and this is the idea -
that, as an organism evolves,

its genetic information is changing.

And, even as it...

expresses that genetic information,

it can actually dictate its own state.

So, biology has its property
of homeostasis

and regulating its own state -
that sort of a state control.

And, this kind of like the rules
are actually controlling this state,

or the constraints
are controlling the state.

So, that seems really fundamentally
different than the situation

that we see in physics
and how we model physical systems

that aren't living.

And the real story there is that -
in biology -

part of the dynamics is the information

and how the information
is changing in time.

And so, in the artificial life field,

people like to try to model
these emergent properties,

or what's going on in biology,

using really simple systems

that allow us to understand
really basic things about life.

One example that is used commonly
in the artificial life field

is cellular automata.

And, one of the reasons
that people really like those

is because they have
really simple local rules,

just like the laws of physics do,

but we see really interesting
global patterns emerging.

And so, for example,

this is an elementary
cellular automata "Rule 150,"

and you can see here
what the pattern is for the rules.

So, if you have three white cells
it maps to one white cell.

If you have three black cells,
it maps to a black cell.

And then, different patterns
of whites and black cells

will either map to a black
or a white cell.

So, it's a very simple rule,

but you'll see that there's this
really regular pattern that emerges

when you run the dynamics
that's not encoded in the rule at all.

So, we talk about that
as an emergent property.

And, life has many emergent properties.

So, one question is -

would those emergent properties of life

be explained by the structure
of the laws of physics alone,

or do we need additional principles,
and do we really need something

like state planet dynamics
or information

or any new principles
that are uniquely biological

to explain how biology has
this sort of many paths

that we observe
through the evolutionary process?

And, that's a great question
for artificial life.

People have thought about that
from many different perspectives

in addition to cellular automata.

But, we'll talk about cellular automata
just a little bit more

because they're a really
nice simple example.

And so, as I said,
this is an example of this idea in physics

of starting with an initial state
of fixed law of motion

and evolving to some final state.

<i>The Game of Life</i> is perhaps
one of the most famous examples

of using this kind of construction
to understand

what emergent properties are
and how complexity can emerge

from simple rules.

So, in <i>The Game of Life,</i>

it's actually a two-dimensional
cellular automata

as opposed to the one-dimensional one
I showed on the previous slide.

And, what you see is,
with <i>The Game of Life,</i>

all kinds of emergent patterns
appearing -

from things that glide
across your screen

that look like little guns
shooting things,

to blinkers,
to all kinds of different structures,

and they can interact
and make more complex structures.

And so, people have actually...
there's sort of a cottage industry

of people studying different emergent
patterns in <i>The Game of Life</i>,

and under what conditions they emerge,

what their computational capacity is.

So, it's been a really great
toy model system to explore this idea

that emergence might...
we might be able to talk about

emergent properties from simple rules.

One of my favorite cellular automata

was actually constructed
by John von Neumann,

and he was really interested in this idea
of self-reproducing machines.

And so, a self-reproducing machine
or automata

would be a system that could make
a complete reproduction of itself.

And, he was actually really inspired
by Alan Turing

and Turing's work
on universal computation.

So, Turing had been interested in
whether you could build a machine

that can compute
any computable function.

And, von Neumann asked the question,
inspired by trying to understand life -

so, this is very early work
in the artificial life field -

could you build a machine

that could build any possible machine
including itself?

And, if it could do that,

then it would be
a self-reproducing machine.

But, it would also be a machine
that could - in principle -

be capable of open-ended evolution,
which is a really important question -

in the artificial life field
related to astrobiology -

about the idea of whether or not evolving
systems can keep evolving forever.

So, if you had a self-reproducing
machine that could build itself,

but not build any arbitrary machine,

it might stall out and not actually be
an evolving system.

And so, von Neumann had this condition
that it should be able to build anything -

so that the space of all possible things
would be completely open,

that it could potentially evolve into,
as long as it could maintain the fact

that it could reproduce itself.

And so, he came up with
a particular architecture

that's necessary for such a machine.

What he basically came up with is that

you need to have some kind
of information content -

the instructions for specifying
the design of the machine.

The machine has to be able to read out
those instructions to build itself,

and then it has to be able to have
something called a "supervisory unit"

that tells it when to copy
the instructions

just as instructions,
rather than reading them out.

So, the instructions have to
have a dual role.

They have to be able to be copied
to be able to reproduce the organism,

but they also have to be read out
to be able to be executed

to construct the organism.

And, the machine doing the construction
is the part that can...

is the machine that can build
any possible machine.

And, he called that
a "universal constructor."

And, there is a direct analogy

with modern cellular architecture
as we understand it,

in that the translation machinery -

ribosomes and all the associated
tRNAs and translation machinery -

could be thought of
as a universal constructor

that can construct any possible proteins.

So, it's not a universal constructor

in the sense that it can make
any possible object,

but it is a universal constructor
in the sense

that it can produce any possible protein.

And, the instruction tape is DNA,

which gets read out by the cell,
but also gets blindly copied

by the cell at other stages
in cellular function.

So, it does seem to be the case

that this abstract idea
from artificial life -

von Neumann's idea of
the self-reproducing automata -

actually maps to some of the function
that we see in biology.

And so, this was a case where
a cellular automata theory

actually predicted some of the logical
architecture of modern organisms.

And so, one of the things that's really
interesting about von Neumann's idea

was that he had envisioned
the possibility of a machine

that could build any possible machine,

which means it could do any possible
transformation on physical matter.

And, most cellular automata
actually can't do that.

So, if you look at cellular automata
and you evolve them forward in time -

in the way that we do,
where we have an initial condition

and the cellular automata rule
is fixed for all time -

not every state transformation
is possible.

So, you can't move from every state
to every other state.

But, there is this idea
that's implicit in von Neumann's theory

for biological evolution -
from this artificial life perspective -

of physical universality,
which is the ability to implement

any transformation on any finite region.

The first physical universal
cellular automata

was actually just realized recently
by Luke Schaeffer...

and he actually demonstrated

that it was possible
to construct such a thing.

And, he did so within this sort of
traditional physics paradigm,

where he started with an initial state
and evolved the system

according to a fixed rule,
and the physical universality

comes about by programming
the initial state.

Now if you're thinking about things
like biological evolution,

then we want to think about
biological systems

in this kind of very abstract way
inspired by artificial life.

And, think about von Neumann's theory

and what it's really telling us
about biology,

one of the things we might hope for
is that biology

would actually be capable of
performing any arbitrary transformation.

And, a good example of that
is modern technology.

So, I think actually modern technology
is a better approximation

to a universal constructor
than an interior of a cell,

in the sense that technology enables
lots of transformations to be possible.

So, for example, we can launch
satellites into space.

And, that's not a transformation
that would be able to be happening -

we wouldn't be...
our planet wouldn't be anti-accreting -

launching things into planetary orbit
without having technology.

So, it does allow transformations
and biology seems to do this in general.

If you think about metabolism...
allows chemical transformations

that seem thermodynamically impossible.

So, if you wanted to talk about
those kind of properties

from this cellular automata perspective,
what you really would like

is to be able to build
a cellular automata

that can perform
any arbitrary transformation

and has this kind of inspiration
from biology.

So, we've been kind of playing around
with those in my group

and this is just sort of an example of...

an artificial life approach
to astrobiology.

Thinking about cellular automata
with state-dependent laws...

So, the idea here is that
we go back to that difference

that we were observing
between physics and biology,

and think about the fact that life
seems to have this property

where the rules and the states
are very tightly coupled.

So... the expression
of my genetic information

determines the state of the cells
in my body.

My mental state determines
something about what I do.

So, if you want to build
systems that do that

you can actually build cellular automata
with state-dependent laws.

And, you see lots of different
rich structures emerging

from these kind of systems,
and they do have this property

of physical universality
that Schaeffer observed,

but from this very different perspective.

And so, these are just some examples
to raise some questions for you

about the kinds of things
that we could be thinking about

from more abstract models of life
in the artificial life field

to get at more general principles
of living systems.

So, we have this idea in mind
that information perhaps

might be this unifying principle of life,
and that really also comes from

the inspiration of
the artificial life field -

because von Neumann was really interested
in this idea of the instruction,

the information content being what's
specifying the design of the machine,

and that the machine could actually
implement those instructions.

And so, in some sense,
what he's talking about

is an algorithmic process
existing in nature

is necessary to have open-ended
evolving systems.

And so, we really don't understand
the basic physical principles of that.

And so, one of the things
I think is incredibly exciting

about working across the field
of artificial life in astrobiology

is that artificial life models have
traditionally been these very abstract,

cellular automata type models,
or there's other things - like <i>Avida</i> -

that are these digital systems
that we program into computers

and we study their properties
and they tell us some things

about emergent properties
or how systems can evolve.

But, astrobiology affords the opportunity
of starting to think about

those kind of things
as chemically embodied systems,

and how do we think about the fact
that the origin of life transition

actually did happen -
these properties emerged

in the natural world from chemistry.

And so, that's really the challenge
for astrobiology moving ahead...

is to take the abstract ideas
from artificial life

and turn them into quantitative science
for astrobiology to build new theories

of what we think life is,
and how the transition

from simple molecules to something
that has the sophisticated logical

and informational architecture
of something

like a von Neumann machine
actually emerged on our planet.

And, what does that actually mean
in the broadest sense?

And, what are the kind of classes
of phenomena

that we might see in the world

that could be inspired
by that understanding...

i.e. can we find aliens?

And so, that's really a great thing
to think about.