07
new behaviors, are not incorporated into our genetic information, and are
08
therefore not passed down to subsequent generations. (There are nuances
09
here, as some environmentally influenced ways that genes are expressed may
10
be heritable, even if the genes themselves don’t change.) Option 2 is a more
11
standard Darwinian explanation. It’s not that previous generations of
12
giraffes wanted to reach higher; it’s just that those that did accrued an ad-
13
vantage that was passed on to their descendants.
14
Then there is option 3, known as “sexual selection.” It is a perfectly plau-
15
sible Darwinian explanation, one that relies on a specific mechanism of
16
selection pressure to achieve the empirical result. Some researchers have
17
suggested that a form of sexual selection is a better explanation than the
18
traditional leafy- treetop story that we tell about the length of giraffe necks.
19
This illustrates one of the difficulties in understanding how evolution actu-
20
ally proceeds in the real world: there may be more than one way to explain
21
the emergence of a single trait.
22
The debate is ongoing. For example, under sexual selection it’s likely that
23
male and female giraffe necks would evolve differently, but the data seem to
24
indicate that they are fairly similar. Option 2 is currently more popular, but
25
new data will continue to impact our credences for each of the different
26
hypotheses.
27
So what about option 4, which avoids any particular evolutionary story-
28
telling? It’s a true statement, but not a useful one in this context. From the
29
poetic- naturalism perspective, natural selection provides a successful way
30
of talking about emergent properties of the biological world. We don’t need
31
to use a vocabulary of evolution and adaptation to correctly describe what
32
happens, but doing so gives us important and useful knowledge.
33
The evolution of life provides a rich source of higher- level phenomena
34
emerging from the fundamental description of reality, including phenom-
35S
ena that have no direct analogue at the deepest level. Because our specific
36N
universe starts in a special state and shows a strong arrow of time, these
292
Big Picture - UK final proofs.indd 292
20/07/2016 10:02:51
E M E R g E n t P u R P O S E
emergent pictures can invoke words like “purpose” and “adaptation,” even
01
though those ideas are nowhere to be found in the underlying mechanistic
02
behavior of reality.
03
04
•
05
A common concern among skeptics of evolution is how it is supposed to
06
lead to the creation of new kinds of things out of the mindless motion of
07
matter. “Purposes” are one obvious example. We say, without apparent em-
08
barrassment, things like “The purpose of the giraffe’s long neck is to help it
09
reach fresh leaves near the treetops.” Another example is “information.”
10
DNA is said to carry genetic information; the optic nerve carries informa-
11
tion from the eye to the brain. Then there is consciousness itself. The con-
12
cern is that these concepts represent a radical break from the mere Laplacian
13
working- out of the laws of physics. How could evolution, which itself is
14
ultimately purely physical, bring these utterly new kinds of things into ex-
15
istence?
16
It’s a natural thing to worry about. The process of evolution is unplanned
17
and unguided. Whether or not genetic information gets passed on to future
18
generations depends only on the conditions of its immediate environment
19
and random chance, not on any future goals. How can an intrinsically pur-
20
poseless process lead to the existence of purposes?
21
But this worry is a little strange, at least in the hands of anyone who ac-
22
cepts that natural selection provides an explanation for more prosaic things
23
like gills and eyeballs. These kinds of organs are “utterly new” in their own
24
way. There is no general principle along the lines of “new kinds of things
25
cannot naturally arise in the course of undirected evolution.” Things like
26
“stars” and “galaxies” come to be in a universe where they formerly didn’t
27
exist. Why not purposes and information?
28
In poetic naturalism, the appearance of “truly new” concepts as one
29
theory emerges from another is the least surprising thing in the world. As
30
time passes and entropy increases, the configuration of matter in the uni-
31
verse takes on different forms, enabling the emergence of different higher-
32
level ways of talking. The appearance of something like “purpose” simply
33
comes down to the question “Is ‘purpose’ a useful concept when developing
34
an effective theory of this part of reality in this particular domain of
S35
applicability?” There may be any number of interesting and challenging
N36
293
Big Picture - UK final proofs.indd 293
20/07/2016 10:02:51
T H E B IG PIC T U R E
01
technical issues to be addressed, but there is no obstacle to the emergence
02
of all kinds of new concepts along the way.
03
•
04
05
Think about Robby the Robot, cleaning up cans from his grid. In the most
06
successful strategies that were artificially generated through many genera-
07
tions of variation and selection, Robby had evolved a technique of not pick-
08
ing up a can on his current square if there were also cans to the east and
09
west. Rather, he would move in one direction— let’s say west— until he ar-
10
rived on a square with a can, but no can on the square just west of his loca-
11
tion. Only then would he double back, picking up all the cans along the way.
12
Why does Robby act in this way? We could simply say, “Those moves are
13
part of the strategy that survives the genetic algorithm process.” That would
14
be the equivalent of answer 4 in the list of giraffe- neck expl
anations above.
15
It’s not wrong, but it’s not very illuminating either. Or we could say, “Robby
16
doesn’t want to forget that there are cans on either side, so he leaves them
17
in place, knowing he will come back and pick them up later.”
18
Is that a sensible way of talking? Robby the Robot doesn’t really want
19
anything. He’s not even a real robot— just a string of ones and zeroes inside
20
some computer memory. Psychologists sometimes speak of the “anthropo-
21
morphic fallacy,” when we attribute human thoughts or emotions to in-
22
animate objects. (My computer gets grumpy if I don’t reboot it every so
23
often.) It may be fun and harmless to speak about Robby as if he has wants,
24
but it’s not really true. Right?
25
Consider the possibility that we have this backward. When we say that
26
Robby the Robot doesn’t really have wants in the same sense that a person
27
does, we are taking the implicit stance that there are things called “wants”
28
that can be correctly attributed to some things in the universe (like human
29
beings) and not to others (like virtual robots). What are these “wants”
30
anyway?
31
The idea that something wants something else is a way of talking that is
32
potentially useful in the right circumstances— a simple idea that summa-
33
rizes a good amount of complex behavior in a convenient way. If we see a
34
monkey climbing a tree, we could describe what’s happening by providing
35S
a list of what the monkey is doing at each moment in time, or for that
36N
294
Big Picture - UK final proofs.indd 294
20/07/2016 10:02:51
E M E R g E n t P u R P O S E
matter we could specify the position and velocity of every atom in the mon-
01
key and the environment at each moment. But it’s immensely easier and
02
more efficient to say, “The monkey wants those bananas that are up in the
03
tree.” The fact that we can say that is a piece of useful knowledge over and
04
above all of those positions and velocities.
05
There is no Platonic idea of a “want” floating out there in the space of
06
ideas that can be properly associated with some kinds of beings and not
07
with others. Rather, there are situations in which it is useful to describe
08
things as somebody wanting something, and other situations in which that
09
is not so useful. These situations can emerge in the natural, undirected evo-
10
lution of matter in the universe. Those wants are as real as things ever get.
11
In the particular case of Robby, it is neither necessary nor especially
12
helpful to characterize his behavior in terms of wants, purposes, or desires.
13
It’s just as easy to simply say what his can- collecting strategy actually is. But
14
the difference between him and a person, as far as the ontological status of
15
“wants” is concerned, is simply a matter of degree. We could imagine a ro-
16
bot with an enormously more complicated programming than little Robby.
17
We might not know much about that specific programming, but perhaps
18
we are able to observe how the robot acts. It may be that the best way of
19
understanding the robot’s behavior is to say, “That robot really wants to
20
pick up those cans.”
21
Under naturalism, there isn’t that much difference between a human
22
being and a robot. We are all just complicated collections of matter moving
23
in patterns, obeying impersonal laws of physics in an environment with an
24
arrow of time. Wants and purposes and desires are the kinds of things that
25
naturally develop along the way.
26
27
•
28
There is a similar story to tell about “information.” It’s worth thinking
29
about, as it will come up again when we start talking about consciousness.
30
If the universe is just a bunch of stuff obeying mechanistic physical rules,
31
how can one thing ever “carry information” about anything else? How can
32
one configuration of atoms be “about” some other configuration?
33
Words like “information” are a useful way of talking about certain
34
things that happen in the universe. We don’t ever need to talk about
S35
N36
295
Big Picture - UK final proofs.indd 295
20/07/2016 10:02:51
T H E B IG PIC T U R E
01
information— we can take the “option 4” viewpoint and just talk about the
02
quantum state of the universe inexorably evolving through time. But the
03
fact that information is an effective way of characterizing certain physical
04
realities is a true and nontrivial insight onto the world.
05
Consider the Voynich manuscript. This is a remarkable and unique
06
book, whose likely provenance has been traced to the early fifteenth cen-
07
tury, possibly from Italy. It is a whimsical volume, replete with fanciful il-
08
lustrations of astronomical and biological subjects. For the most part, the
09
many flora depicted in the illustrations cannot be identified with actual
10
plant species. Most remarkably, the text of the book has proven, to date, to
11
be completely indecipherable. Not only the language but even the apparent
12
alphabet is something that has never been seen before. Statistical analyses
13
of the words and symbols in the writing seem to be compatible with those
14
of ordinary languages, but cryptographers have been stymied in their at-
15
tempts to interpret the text as some kind of code. It may be a very good ci-
16
pher; it may be a unique language that was invented by an individual and
17
then forgotten; or it may be a complete hoax.
18
Does the Voynich manuscript contain information?
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35S
36N
An excerpt of the writing that appears in the Voynich manuscript.
&
nbsp; 296
Big Picture - UK final proofs.indd 296
20/07/2016 10:02:51
E M E R g E n t P u R P O S E
One is tempted to say that it depends on the origin of the book. If it
01
really is a hoax, and the words are some kind of semirandom nonsense, then
02
perhaps it doesn’t contain much information at all. But if it is merely a
03
clever code that will someday be broken, it might contain a great deal—
04
even if that “information” is purely a work of imagination.
05
What if the Voynich manuscript is a code that will never be broken?
06
What if it was originally written with very specific intent, but its meaning
07
has been so well hidden that nobody will ever be able to reveal it? Does it
08
stil contain information? What if the manuscript is placed in a capsule and
09
launched into space, and then the Earth is destroyed by a cataclysmic aster-
10
oid impact, and the book floats through the void for all of eternity. Does it
11
contain information then?
12
We tend to use the word “information” in multiple, often incompatible,
13
ways. In chapter 4 we talked about conservation of information in the fun-
14
damental physical laws. There, what we might call the “microscopic infor-
15
mation” refers to a complete specification of the exact state of a physical
16
system, and is neither created nor destroyed. But often we think of a higher-
17
level macroscopic concept of information, one that can indeed come and
18
go; if a book is burned, the information contained in it is lost to us, even if
19
not to the universe.
20
The macroscopic information contained in a book is relative to the en-
21
vironment in which it is embedded. When we talk about the information
22
contained in the book you are currently reading, what we mean is that these
23
words are correlated with certain ideas that you get upon reading them. You
24
read the word “giraffe,” and the notion of a certain kind of long- necked
25
African ungulate appears in your mind. The same holds for the informa-
26
tion contained in a strand of DNA: it is correlated with the synthesis of
27
certain proteins in the cell. It is this connection with one configuration of
28
matter (a book or a DNA strand) and something else in the universe (the
29
image of a giraffe, or a useful protein molecule) that lets us talk about the
30
existence of information. Without those correlations— if there isn’t, and
The Big Picture Page 50