What Darwin Got Wrong
Update for the paperback edition: Replies to our critics
1. On our book and its reception
The hardcover edition of this book was first published early in 2010. It was intended to
raise two objections to the Theory of Natural Selection (TNS) and to explore their
connections to each other and to familiar questions about evolution. First, we claimed
that TNS is committed to an untenable externalism: Like Skinner, Darwin held that
paradigm explanations of biological (and psychological) structure should invoke relations
between organism and their ecologies. But, whereas Skinner’s externalism was largely
motivated by his methodological commitment to behaviorism, Darwin’s was quite
different; Darwin held that externalism is the price one pays for adaptationism: only an
externalist theory could explain why the features of a creature’s phenotype are so often
well-adapted to the features of its ecology. The explanation on offer is that phenotypes
are shaped by the ecological features to which they are adapted. We suggested, by
contrast, that the appearance of adaptation is in large part illusory. The reason a
creature’s phenotype seems well-adapted to its ecology is that by definition, an
“ecological feature” is one to which the fitness of phenotypic traits is sensitive; and a
“phenotypic trait” is by definition, one that effects a creature’s fitness in relation to its
ecology. We aren’t, of course, the first to suspect that there are vicious circularities
lurking at the heart of TNS. But we have tried to make them explicit, and to document a
variety of recent empirical findings that strongly suggest the crucial role of endogenous
variables in the evolution of phenotypes. About half of our book is devoted to doing so.
The second problem we raised for TNS has, to our knowledge, hardly been noticed
elsewhere in the literature: the tension between its treatment of selection and its treatment
of selection-for. TNS holds, in effect, that though what get selected are kinds of creatures
(kinds of creatures are what flourish, or fail to, in a given ecology), what creatures get
selected-for are certain of their phenotypic traits (viz those phenotypic traits that cause
their fitness.) Problems arise because, unlike selection, selection-for is a paradigmatically
intensional concept: it is perfectly possible that there should be selection-for one, but not
the other, of two coextensive phenotypic traits. The intensionality of selection-for is duly
inherited by a variety of other notions that are interdefined with it, and to which TNS is
committed. These include, in particular, the notion of a phenotypic trait itself (since one
but not the other of coextensive phenotypic traits may be selected-for). This we suggest,
is the logical consideration from which the notorious problems about “arches and
spandrels” eventually arise. We argue that because selection-for is intensional and
selection is not, TNS can’t, even in principle, decide which of its traits is selected for
when a kind of creature is selected. This should hardly be surprising; there is an exactly
parallel situation in cognitive psychology, where the intensionality of the “propositional
attitudes” - beliefs, desires, and the like - offers a prima facie objection to the
naturalizability of “Representational” theories of mind. That there is this previously
widely ignored analogy between the (putative) intensionality of mental processes and the
(putative) intensionality of evolutionary processes is one of the things that make the
present issues philosophically interesting.
1
Our claim is that, given coextensive phenotypic traits, TNS can’t distinguish ones that are
causally active from ones that aren’t. Many of the objections that have been raised
against us seem unable to discriminate this claim from such quite different ones that we
didn’t and don’t endorse, such as: when traits are coextensive, there is no fact of the
matter about which is a cause of fitness; or, when traits are coextensive, there is no way
to tell which of them is a cause of fitness; or when traits are coextensive Science cannot
determine which is a cause of fitness…etc. Such views are, we think, preposterous on the
face of them; we wouldn’t be caught dead holding them. To the contrary, it is precisely
because there is a fact of the matter about which phenotypic traits cause fitness, and
because there is no principled reason why such facts should be inaccessible to empirical
inquiry, that the failure of TNS to explain what distinguishes causally active traits from
mere correlates of causally of active traits, shows that something is seriously wrong with
TNS.
We were, on balance, very pleased the way our book turned out. It seemed to us quite
plausible, in the light of the considerations it raised, that TNS is simply untenable and
that, insofar as current evolutionary theory presupposes it, current evolutionary theory is
due for a thorough reconsideration. We thought of this as a real scientific advance; the
next best thing to finding out what one ought to believe is finding out what one ought not.
We didn’t exactly expect to be awarded a tickertape parade, of course; but we were
looking forward to at least a few warm congratulations. In the event, however, the book
was received very badly. Almost (though not quite) all the reviews were hostile and some
were hysterical. Our arguments and our conclusion were both widely and wildly
misrepresented. Many suspected that we are covert Theists, committed to undermining
the foundations of the Scientific World View (of which they took themselves to be the
anointed custodians). Others reproached us for having opinions on issues that are
proprietary to members of the Guild of Professional Biologists. The blogs, in particular,
were ablaze with anonymous contumely. Well, what did we expect? Hadn’t we heard
there’s a Culture War on?
Some of the objections we’ve seen strike us as too silly to bother refuting. Others deserve
serious replies. The latter should be addressed at length; They will be in future
publications. But there is a number of criticisms that can be replied to succinctly; hence
the present Update. We propose to quote, and rebut, a scattering of short passages from
reviews of our book. Hope springs eternal, so we’re told. We hope, at a minimum, to
clear the ground for more extended discussions. We still believe in the possibility of a
rational, informed, interdisciplinary, consideration of what’s wrong with the conceptual
architecture of TNS.
2. When some biologists (indirectly) agree with us
Several reviewers have suggested that we don’t know enough about biology to criticize a
theory that so many biologists hold dear. The implication is: only someone improperly
educated could say the sort of things we do. But we don’t think our critics are well-
advised to insist on our lack of credentials. For one thing, several of them aren’t
biologists either. For another, it’s a self-defeating line of argument; do they hold that
only theologians are licensed to discuss the existence of God?
2
Everybody makes mistakes; even biologists; even biologists who agree with one
another; even great biologists like Darwin. If you think somebody has made a mistake,
then it’s a good thing for you to say so, so that s/he (or you) can be corrected. Surely that
is common ground among scientists, philosophers, and everybody else who cares about
distinguishing the true from the false. The parochial is the enemy of the true, and should
not be encouraged. But we won’t go on about this; it’s a little embarrassing even to have
to mention it. Instead, we report verbatim some recent passages by fully qualified
evolutionary biologists, each of whom has earned a Ph.D from an accredited institution of
higher learning, and all of whom are explicit in maintaining that neo-Darwinism (the new
synthesis) is gone.
"In the postgenomic era, all major tenets of the modern synthesis have been, if not outright
overturned, replaced by a new and incomparably more complex vision of the key aspects of
evolution. So, not to mince words, the modern synthesis is gone. What comes next? […] a
postmodern state […]. Above all, such a state is characterized by the pluralism of processes and
patterns in evolution that defy straightforward generalization". (our
emphasis)
Eugene
V.
Koonin
(Senior
Investigator,
National
Institutes
of
Health)
(2009
a).
“Evolutionarygenomic studies show that natural selection is only one of the forces that shape
genome evolution and is not quantitatively dominant, whereas nonadaptive processes
are much more prominent than previously suspected”. (our
emphasis)
Koonin,
E.
V.
(2009
b).
“Although 2009 will be marked by a plethora of celebrations on the subject of evolution, most
of the attention is being bestowed on the personalities and historical circumstances
surrounding the theory of natural selection, as if this and its synthesis with genetics in the first
decades of the 20th century marks the culmination of the theory of evolution. It does not.”
…….
“Dogmatic thinking has prevailed all too often in our account, with disastrous consequences
for the progress of the fields of microbiology, molecular biology, and the study of the
evolutionary process. It led to the stagnant and scientifically invalid notion of the prokaryote;
it led to the redefinition of the problem of the gene; and through a slavish adherence to the
modern evolutionary synthesis, it led to a premature declaration of victory in the struggle
to understand the evolutionary process.
……
The study of evolution is poised to cast off a
century of dogma and to become a true science, fully integrated with discoveries that owe
their roots to microbiology and molecular biology. It is time for biology to put its past behind
and begin rethinking the discipline’s future. It can no longer afford to keep the study of
evolution within the narrow confines of the socalled modern evolutionary synthesis.”
(o.e.)
Carl
R.
Woese
(Microbiologist,
University
of
Illinois,
winner
of
the
2000
National
Medal
of
Science)
and
Nigel
Goldenfeld
(Professor
of
Physics
at
the
University
of
Illinois
at
Urbana‐Champaign
and
Head
of
the
Biocomplexity
Group
at
the
University's
Institute
for
Genomic
Biology)
(2009).
“Despite elaborate NeoDarwinist mathematical models that focus on inherited variation in
animals, evidence continues to mount that the branches of “the tree of life” do not just
bifurcate. They do not simply diverge by gradual accumulation of random mutations. Rather
lineages converge, as the result of gene transfers, mergers, fusions, partnerships, anastomoses
and other forms of alliance. The most accurate modern taxonomies recognize that
3
Archaebacteria and Eubacteria have become subkingdoms of the prokaryotes whereas all
nucleated organisms (eukaryotes) evolved symbiogenetically.”
Lynn Margulis (Distinguished University Professor of Geosciences at the University of
Massachusetts, winner of the 1999 Presidential Medal of Science) and Michael J.
Chapman (Marine Biological Laboratory, Woods Hole, MA) (2010),
“There is a growing appreciation among evolutionary biologists that the rate and tempo of
molecular evolution might often be altered at or near the time of speciation, i.e. that speciation is
in some way a special time for genes. Molecular phylogenies frequently reveal increased rates of
genetic evolution associated with speciation and other lines of investigation suggest that various
types of abrupt genomic disruption can play an important role in promoting speciation via
reproductive isolation. These phenomena are in conflict with the gradual view of molecular
evolution that is implicit in much of our thinking about speciation and in the tools of modern
biology. This raises the prospect of studying the molecular evolutionary consequences of
speciation per se and studying the footprint of speciation as an active force in promoting genetic
divergence. …. Speciation might often owe more to ephemeral and essentially arbitrary events
that cause reproductive isolation than to the gradual and regular tug of natural selection that
draws a species into a new niche.” (o.e.)
Chris Venditti (Evolutionary Biologist, The University of Reading UK) and Mark Pagel
(Microbiologist, The University of Reading UK) (2009)
In summary: We have seen how several of the recent discoveries in biology that our book
recounts lead some biologists to explicit non-Darwinian conclusions. Samir Okasha
(2010) pushes them aside saying (correctly) that "they simply concern aspects of biology
about which traditional neo-Darwinism didn't have much to say". But our point about
these biological mechanisms is not that the neo-Darwinists don’t attend to them; but
rather the marginalization of TNS that they suggest. It seems that most of the action may
well be in a different part of town.
3. Replies to critiques from biologists
What follows are brief replies to criticisms that some of our biologist reviewers have
made and that we think are radically wrong-headed; they don’t exhaust the list, but they
are typical.
3.1 Nothing new
A frequent critique we have received is that all the non-selectionist factors and
processes summarized in Part One of our book have been known to evolutionary
biologists for a long time and are all perfectly compatible with the Theory of Natural
Selection (TNS). This is wrong on two counts: First, because we have based that part of
our book mostly on articles published in the last 5 years in specialized biology journals,
and (rightly) presented as innovative by their authors; Second, because it is very hard to
reconcile these discoveries with TNS, as several authors say explicitly (see the quotes
above and more in our book) and almost all of them at least implicitly.
In particular, our critics say that the existence of internal constraints on possible
phenotypic variation is obvious and has been acknowledged to be so for decades, indeed
by Darwin himself. We have doubts about this. Although we are no experts of Darwin’s
publications, those who are say what follows: (see also note 2 to pp. 20-24)
4
“There can be no direction imposed on evolution by factors internal to the organisms, because
the variation upon which selection acts is random in the sense that it is composed of many
different and apparently purposeless modifications of structure. The environment determines
which shall live and reproduce, and which shall die, thus defining the direction in which the
population evolves.”
(Bowler,
2003,
pp.
10–11).
One more qualified quote, by the bio-physicist and bio-mathematician Stuart Kauffman, a
pioneer in the search of physical and self-organizational components of biological
structures and evolution, a scientist highly regarded by Richard Lewontin and the late
Stephen Jay Gould (see Chapter 5):
“A curious, logically unnecessary, but influential feature of Darwin’s thinking was that the
variation within one species which paved the way for emergence of well-marked varieties
constituting two species was an indefinite range. The idea that variations could occur in
virtually any direction, an idea which dominates in Darwin’s work despite attention to
correlations among traits under selection, has had important conceptual consequence. It follows
that selection alone can discriminate which new variants will be found in later generations.
Here is one root of our current idea that selection is the sole source of order in the biological
world”. (Kauffman 1993. Page 6) (emphasis ours)
3.2 Two wrong analogies
We like good analogies, but there are limits. The ones we’re about to quote seem to us
beyond the pale; the kind of far-fetched arguments that responsible scientists should
avoid.
“Thus, the authors argue, there cannot be a universal theory of natural selection, for no general
relationship of phenotype to fitness can be specified. But the same might be said of many other
research programs. For example, the effect of an enzyme is highly context-dependent, so Fodor
and Piattelli-Palmarini presumably would not expect any successful theory in biochemistry”.
Douglas Futuyma (2010, page 692))
The net effect of an enzyme is to catalyze (that is drastically accelerate) a chemical
reaction. This action depends on factors such as temperature, acidity, concentration of the
substrate and of other chemical participants (co-enzymes, inhibitors). The influence of
each of these factors is well understood and separable in principle. Indeed there are
general laws of enzymology, such as the Michaelis-Menten equation of enzyme kinetics.
These processes take place at one well specified level, that of molecular reactions, where
the panorama is totally different from the highly composite one of the genotype to
phenotype relation, where we have multiple levels (from Angstroms to yards), and
multiple kinds of dynamics. In our book we summarize more than a dozen of these
processes; the likelihood of unifying all of them under one theory is negligible. The
analogy with enzymology is, therefore, totally fallacious.
The next one is due to Jerry Coyne:
“Clearly, F&P are confusing our ability to understand how a process operates with whether it
operates. It's like saying that because we don't understand how gravity works, things don't fall.”
… “Our inability to understand all the details [of natural selection] is hardly a reason to claim
5
that natural selection doesn't work.”
(Coyne 2010)
We are not only scientific realists, but scientific hyper-realists. Nothing like the
above ever crossed our minds. We will go back to the analogy with the law of gravity
in a moment, in our reply to Elliott Sober. Let’s concentrate here on just one point.
It’s one thing to lament our failure to understand some or other natural process
which we nevertheless have good reasons to believe occurs. It’s quite another to offer
principled reasons why some or other theory of such a process isn’t viable. Our book
is concerned with the latter in the case of the theory of natural selection. Coyne needs
to rebut these arguments. He doesn’t.
We never said that NS does not operate in the wild because it’s so hard for us
to understand how it works. We say that general explanations based on natural
selection are necessarily based on correlations (between the presence of a trait and
greater reproductive potential), not causes. Detailed, very heterogeneous
explanations of the selection for individual traits, in individual species, in their
particular environments, can sometimes reveal causal factors. There is a radical
difference, on which we insist in our book and in this update. The analogy with
gravity is untenable. Gravity is the cause of the falling of bodies, not a correlation.
3.3 Merging evolution and Natural Selection
In
his
review,
and
in
his
recent
book,
Coyne
regularly
fails
to
distinguish
arguments
about
evolution
and
arguments
about
natural
selection.
For
example,
Coyne
and
Dawkins
both
discuss
at
length
the
circuitous
and
devious
geometry
of
the
laryngeal
nerve
in
mammals,
which
connects
organs
only
a
few
inches
apart, but runs from
the head to the heart, looping around the aorta and then doubling back up to the
neck (Coyne points out that, in the giraffe, this detour involves about fifteen feet of
superfluous nerve).
Then
follows
an
account
of
how
this
oddity
occurred
via
progressive
transformations
from
older
species
of
the
anatomy
of
the
organs,
something
we
have
no
reason
to
question.
Dawkins
and
Coyne
take
such
cases
to
argue
against
evolution
by
“intelligent
design”,
and
so
they
do.
They
are,
however,
thoroughly
irrelevant
to
the
issues
that
our
book
is
concerned
with,
which
is
whether
the
mechanism
of
evolution
is
Natural
Selection.
But
then,
these
data
and
arguments
in
favor
of
the
evolutionary
descent
of
species
are
transmuted
into
data
and
arguments
in
favor
of
the
theory
of
natural
selection.
Questioning
TNS
is
considered
identical
with
questioning
evolution
as
such.
This
conflation
leads
Coyne
to
say:
“Their [our:JF&MPP] claim to have nullified 150 years of science, and one of humanity's
proudest intellectual achievements, with some verbal legerdemain, is not only breathtakingly
arrogant but willfully ignorant of modern biology”.
Enraged
at
having
failed
to
hit
the
target
he
intended,
Coyne
proceeds
to
loose
his
shafts
at
a
venture.
We
repeat:
We
have
no
doubts
about
the
reality
of
evolution,
or,
more
specifically,
about
the
descent
and
radiation
of
species
from
preexisting
ancestors;
and
we
entirely
accept
that
topological
and
functional
transformations
of
internal
organs
6
offer
persuasive
evidence
in
its
favor.
What
we
seriously
doubt
is
the
power
of
natural
selection
to
explain
how
it
happens.
3.4 The argument from the success of artificial selection
Here’s
another
argument
of
Coyne’s:
“If there really were so many constraints on selection, and if development really were so complex
and tightly interconnected that organisms could not respond to natural selection, then why would
artificial selection be so effective at changing animals and plants?”
First of all, we do not say that “organisms could not respond to natural selection”.
What we say is that there are innumerably many different ways of responding,
depending on the phenotype, the species and the environment, defying a unitary
theory. Moreover, to the best of our knowledge, artificial selection has never
managed to produce new species, something that natural selection is supposed to
have done many times. So, even artificial selection is effective only up to a point.
Numerous sub-species have been obtained, by means of repeated selective cross-
breeding, aiming at specific phenotypes (better wool, more milk, stronger muscles
etc.). In our book (page 62 and note 2 page 210) we stress that these desired traits
were invariably accompanied by a number of others (curly tails, floppy ears, piebald
color etc.). These other traits are free riders that were obviously not selected for. The
lesson here is that, in cases of artificial selection, it’s straightforward to decide which
trait was selected for and which one came fortuitously, because we can ask the
human agents involved, or make an educated guess. The burden of our book is that,
on one hand, the distinction between traits that are selected for is essential to
distinguishing causes of fitness from free riders; and, on the other hand, this
distinction can’t be drawn in cases where there isn’t a breeder (including, in
particular, cases of selection in the wild).
3.5 Missing heritability
Coyne makes the following accusations:
“Beyond distorting the scientific literature, F&P make a number of claims that are simply silly. I
mention just one: "The textbook cases of Mendelian inheritance, in spite of their great historical
and didactic importance, are more the exception than the rule." This came as a surprise to me. In
fact, cases of Mendelian inheritance (the random assortment of parental genes into sperm and
eggs) are the rule; if they weren't, genetic counseling would be useless. Statements like this typify
the authors' attitude toward science throughout their book: they seize on some new wrinkle in the
scientific literature, like a rare gene that doesn't behave according to Mendel's rules, and
interpret it as a revolution that nullifies all of mainstream biology. This lack of grounding is often
seen in work by science journalists who make their living touting "revolutionary" new findings,
but it is inexcusable in a supposedly serious book written by academics.”.
We are not surprised that this came as a surprise to Coyne. Indeed genetic counsel is
often (not always, but often) useless, for instance, when well characterized frequent
mutations in over 20 genes explain just 3% or 5% of genetic risk. The case of the
“missing heritability of complex diseases” is not a “wrinkle”, as Coyne would have us
believe. Witness the manifesto by this title published in Nature (October 8 2009, Vol
7
461, pp. 747-753) by 27 leading human geneticists lamenting the situation, and the
following summary by one of the authors, David Goldstein (Richard and Pat Johnson
Distinguished University Professor, Director, Center for Human Genome Variation, Duke
University) in the New England Journal of Medicine on April 23 2009:
“20 gene variants account for 3 percent in the variation of risk susceptibility to type 2
diabetes….If common variants are responsible for most genetic components of type 2 diabetes,
height, and similar traits, then genetics will provide relatively little guidance about the biology of
these conditions, because most genes are “height genes” or “type 2 diabetes genes…News are as
bleak as they could be.”
These are not the irresponsible scientific journalists to whom Coyne compares us. A
quote will say it all. Another of those authors, Leonard Kruglyak (Professor of
Ecology and Evolutionary Biology at Princeton University) in Nature: Vol 456, 6
November 2008, p. 21 says:
“It’s a possibility that there’s something we just don’t fundamentally understand, that it’s so
different from what we’re thinking about that we’re not thinking about it yet”.
Kruglyak refers to the genotype-phenotype relation for complex diseases, but the
same can be said, we think, for complex traits more generally. We suggest that Coyne
absorbs these facts, stops pontificating and pays attention, not to us, but to these
colleagues of his.
Coyne
concludes:
“In the end, F&P's contrarian efforts are all belied by the world of
Richard Dawkinsthe flourishing field of modern evolutionary biology, where natural
selection remains the only explanation for the wondrous adaptive complexity of organisms.”
Please
underline:
“natural selection remains the only explanation” for
later
reference.
3.6 Catching phenotypes
We conclude our replies concerning biology with a critique voiced both by Douglas
Futuyma and Jerry Coyne:
“The ludicrous analogy with which Fodor and Piattelli-Palmarini end: “organisms ‘catch’ their
phenotypes from their ecologies in something like the way that they catch their colds from their
ecologies.” (Futuyma)
“After much demurring, they [i.e us JF&MPP] float the idea that "organisms 'catch' their
phenotypes from their ecologies in something like the way that they catch their colds from their
ecologies." Although this "explanation" links evolution to ecology, it's completely meaningless.
How did ancestral whales catch their flukes and flippers from the water? How did ancestral birds
catch their wings from the air? F&P don't say”. (Coyne)
Actually, we don’t think that whales catch their flukes from the water. This discussion is,
of course, awash in metaphors on both sides, and the thing about metaphors is that if you
don’t treat them with a dollop of subtlety, they are likely to bite you. Darwin’s metaphor
is: “Natural selection is like breeding”. We think it invites failures to notice the difference
between breeding-for (which is intensional) and selection (which is not). Our metaphor
is: “the processes that mediate coming down with a phenotypic trait are like the ones that
8
mediate coming down with a cold”; the point is that both depend on massive dynamic
interactions between a host’s endogenous properties and properties in its environment;
and quite likely the details of such interactions are highly idiosyncratic from case to case.
That’s why nobody in his right mind thinks there could be a general theory of catching
diseases. Why, then do biologists think there could be a general theory of the evolution of
phenotypes?
4 Replies to critiques of the conceptual situation (Part 2 of the book)
4.1 Explanations and definitions
The crucial sentence in Peter Godfrey-Smith’s review of WDGW (London Review of
Books) is:
”if one [but not the other of two linked traits] is causing increased reproductive success, it is
[sic] being selected for, in the sense that matters to evolutionary theory.”
A number of other reviewers have made much the same suggestion, but it won’t do. The
theory of natural selection claims that a trait’s having been selected for causing
reproductive success explains why a creature has it. But then it can’t also claim that “in
the sense that matters” “a trait was selected for” means that it is a cause of reproductive
success. For, if it did mean that, then the theory of natural selection would reduce to a
trait’s being a cause of reproductive success explains its being a cause of reproductive
success which explains nothing (and isn’t true).
This is all old news; because John’s being a bachelor is his being an unmarried
man, John’s being a bachelor doesn’t explain his being an unmarried man. Psychologists
who hoped to defend the “law of effect” by saying that it is true by definition, that
reinforcement alters response strength, made much the same mistake that Godfrey-Smith
does.
Likewise, Elliott Sober says,
“the distinction between selection-for and `free riding’ is nothing other than the distinction
between cause and correlations.”
Later on he says that
“there is selection for trait T in a population if and only if trait T causes organisms to have
reproductive success in the population”.
This, he claims, is a definition of “selection-for”: it’s true by definition that the trait that
is a cause of increased fitness is selected-for but the other is not. However, as we just
saw, that can’t be right. The very heart of TNS is the thesis that, in the paradigm cases,
traits are selected-for because they are causes of fitness; that is, differences of their
effects on fitness explain why some traits are selected-for and others aren’t. But if that’s
so, then the connection between being selected-for and being a cause of fitness can’t be
definitional. The dialectics here precisely parallels arguments that philosophers of mind
offered in ‘50s against the claim that, in paradigm cases, the relation between behavior
9
and mental states is “criterial” (in effect, definitional). If it’s conceptually necessary that
you raise your arm when you want to, then the cause of your raising your arm can’t be
your wanting to raise it. It took fifty years for philosophy to get over this. Must we now
have it yet again? Something really is seriously wrong with the theory of natural
selection, and stipulating that it is true by definition won’t fix it.
4.2 The intensionality of selection-for
Elliott Sober has what seems to us to be a distorted view of the present polemical
situation.
“FP really do maintain that there cannot be natural selection for one but not the other of two
traits that are locally coextensive. However, in Fodor and Sober (2010) Fodor denies that the
book says this.”
What Sober says that the book says is that there can’t be a causal theory of “selection-
for.” But the book doesn’t say what Sober says it does. What it does say is that the
Theory of Natural Selection can’t provide an account of natural selection (because it’s a
causal theory and selecting-for is an intensional relation). So the book proposes a
dilemma: either there is no such thing as natural selection, or, if there is, the Theory of
Natural Selection misdescribes it.
4.3 Can linked properties be distinct in causal role?
Here’s what Ned Block and Philip Kitcher (hereinafter BK) think is one of our
two main errors.
“Their [e.g. our, Fodor and Piattelli-Palmarini’s] specific charge is that, with respect to
correlated traits in organisms - traits that come packaged together - there is no fact of the matter
about which of the correlated traits causes increased reproductive success”.
BK then speculate that we endorse the “very ambitious” claim that when traits are
correlated, there can be no fact of the matter about which trait causes what.
But, of course, we don’t believe, still less make, either of these claims. In fact, we think
that it’s preposterous on the face of it. Indeed, if the causal powers of linked traits can’t
be distinguished, it would not be an argument against the Theory of Natural Selection
that it fails to distinguish them. We therefore spent a whole chapter (Ch. 7) discussing a
number of ways in which the causal roles of confounded variables can be, and routinely
are, assessed. The most obvious of these is J. S. Mill’s “method of differences”: run an
experiment in which one but not the other of the putative causes is suppressed. If you
still get the effect, then it must be the variable you didn’t suppress that’s doing the
causing. People (scientists very definitely included) do this sort of thing all the time, and
with great success. All this is familiar from Phil. 101. Do Block and Kitcher really
believe that, old and battle-weary as we are, could have written a book that gets that
wrong?
The question whether there is a fact of the matter about which variable is the
cause, or about whether this fact of the matter is epistemically accessible, really must not
be confused with whether Natural Selection, as Darwin understands it, is able to
distinguish causes from their local confounds. For reasons the book details, we think it
10
can’t. To repeat: One can work out what caused what in all sorts of ways: use Mill’s
method; or take the system of causes and effects apart and find out what mechanisms
operate inside it; or ask the guy who built it (if somebody did) how it works… and on
and on and on. But Natural Selection can’t do any of these things. It can’t look inside,
and it can’t run experiments, and it can’t contrive theories, and it can’t consult the
intentions of the builder. All natural selection can do is recognize correlations between
phenotypic traits and fitness. And that doesn’t help because, by assumption, if either of
the confounded traits is correlated with fitness, so too is the other, and to the same extent.
Samir Okasha, in his review, commits much the same misreading of our book: He
accuses us of denying the distinction between causes of fitness and free-riders. But our
view is neither that it is impossible to deconfound causes of fitness from free-riders nor
that there is no such distinction. What we do think (and what we do think our book
shows) is that Darwin's theory can't, even in principle, specify a mechanism by which
selection could reliably distinguish causes of fitness from correlates of causes of fitness.
To a first approximation, this is because TNS recognizes only exogenous variables as
selectors, and the only (relevant) fact to which such variables are sensitive, according to
TNS, is the strength of the correlations between phenotypic changes and changes of
fitness. And, of course, correlation doesn't imply causation. Indeed it patently doesn't
imply causation when the correlation in question is identical for both of the candidate
causes; as it is by assumption, in the case where phenotypic traits are linked.
To repeat: It is beside the point that scientists in the laboratory often can
deconfound linked causes; scientists have minds and the process of evolution does not.
Indeed, it is the prima facie connection between intensional states and mental states that
makes the intensionality of “select for” a problem for naturalizing TNS; a point in respect
of which WDGW is vehement.
For a while it bothered us that many of our critics should have so blatantly
misread what we wrote. But we have a theory: It's that the neo-Darwinian community is
so blindly committed to TNS that they allow themselves to reason as follows (implicitly,
to be sure): (1) This book says that TNS can't distinguish causes of fitness from correlates
of causes of fitness. But, it goes without saying that: (2) TNS is certainly true and
everybody knows that it is. So: (3) if the authors claim that TNS can't distinguish causes
from correlates, that must be because they think that there is no such distinction. So (4) I
shall write a review accusing them of thinking that. But if that is indeed how our critics
are reasoning, we protest that it's more than a tad question-begging.
4.4 Laws of evolution
A short summary of the second half of the book might go like this: TNS needs selection-
for to be intensional, but offers no suggestion of how it could be. But, as we remarked
above, if there are laws of evolution (nomologically necessary empirical generalizations
to which evolutionary processes conform) it might be from those that the intensionality of
select-for derives. So it matters to the present question whether there are such laws. The
bad news, according to WDGW, is that there aren’t. This is. Indeed, one of the cases in
which WDGW agrees with what we take to be the consensus view among biologists.
Nobody doubts, of course, that evolution is law-governed; after all, the laws of physics
apply to everything. The present issue is whether there are biological laws of evolution;
that is, laws of evolution that are defined over biological kinds (such as, for example,
11
laws about evolution defined over ecological properties so described and their effects on
fitness so described.) Missing this point has lead to all sorts of confusion including,
notably, the suggestion that if there are no laws of evolution, determinism and/or
mechanism are ipso facto undermined.
Well, Elliott Sober thinks we’re wrong about that. Actually what he says is not
that are such laws, but that we haven’t shown that there aren’t. And indeed we
haven’t. Since the issue is entirely empirical, there’s no question of demonstrative
arguments on either side. There are, however, straws in the wind, and we think they’re
blowing our way.
Here are two reasons for doubting that there are laws of evolution. The first is that
there seem to be no examples of such laws. That is easily explained on the assumption
that, in fact, there are no such laws. The second is that, if there were laws of evolution,
they would have to be horrendously complicated. A long tradition of modeling evolution
has indentified at least the following factors, among others: effective population size,
density-dependent selection, drift with or without selection, migration, gene flow and
horizontal transmission, the diffusion of neutral mutations, mutational bias, biased gene
conversion, differentials in fertility, sexual selection, variable sex ratios, the overlap of
fertile generations, the fixation of deleterious alleles, phenotypic plasticity, and various
kinds of epistasis (gene-gene interactions). Sober says (rightly) that complexity isn’t, in
and of itself, an argument against the putative laws. But the kind of complexity that laws
of evolution would require is, we think, without precedent in the other sciences. First of
all, laws of evolution would have to take into consideration interactions at vastly
heterogeneous levels: molecule to molecule, gene to gene, gene to cell, cell to cell,
developmental module to developmental module, tissue to tissue, organism to organisms
of the same species, organism to organisms of different species, and all these to the local
ecology. The heterogeneity concerns both sheer size (from Angstroms to miles) and the
conceptualization of the relevant kinds. His failure to understand this is part and parcel of
Sober’s mishandling of one of his own examples:
“The gravitational force now acting on the earth depends on the mass of the sun, the moon, and
of everything else. It does not follow that there are no laws of gravity, only that the laws need to
have numerous placeholders…. The fact that an effect has numerous complexly interacting
causes does not show that there are no laws about this complex cause/effect relation”.
Well, of course there are laws of gravity; principally that the gravitational force between
objects varies directly with their total mass and inversely with the square of their
distance. Notice, however, that this law is quite simple; in particular, it has no ‘place
holders’ for the sun, the moon, the Earth or anything else except the masses and distances
of the objects involved. That’s why the law of gravity would be unaffected even if there
weren’t the sun, the moon, or the earth.
What goes on when explanations appeal to laws is something like this: there are
variables for relevant properties of things that fall under the laws; and there are
specifications of the “initial conditions” in some domain to which the laws apply.
Neither the moon nor its mass gets mentioned by the laws of gravity; but both do get
mentioned in specifying the conditions that obtain when the theory of gravity is used to
predict the gravitational force between (eg.) the moon and the earth. In consequence, the
laws of gravity have very many fewer “place holders” than there are things in the
12
universe to which they apply. We won’t argue for this view; but please take our word for
it that a lot depends on getting it straight.
So now the question arises whether this picture is plausible for the (putative)
evolutionary laws of trait fixation. We think it pretty clearly isn’t; not, however, because
there are very many creatures to which the laws would have to apply, and very many
environmental features with which such creatures may interact. Rather, it’s because of
the awesome heterogeneity of levels and kinds we have mentioned, and of the ways in
which interactions of creatures with their environment depend on what kind of creature
it is and what kind of environment it is interacting with. As we saw two paragraphs back,
laws don’t need place-holders for each thing that falls under them, but they do need
placeholders for each kind of thing that falls under them.
To make the point slightly differently, there are typically many kinds of creatures
that can share an environment, and many kinds of environments that creatures can share.
(We’re told that more than ten thousand species share Central Park). That being so, the
putative laws that determine fitness as a function of such interactions would have to be
complicated in precisely the way that the laws of gravity are not: They would need “place
holders” for each of the kind of creatures that they apply to and for each kind of
environment that the creatures can interact with. And, to repeat, though the number of
things a law applies to doesn’t determine how many placeholders it needs, how many
kinds of things it applies to does. Given all that, could there be such laws about how
creature/environment interactions determine fitness? In principle, sure there could. But
are there such laws? We think the probability is asymptotically close to nil. The kind of
complexity that does tell against a putative law is the kind that proliferates kinds beyond
necessity.
There are other things Sober’s review says that we think are wrong; for example,
we think it’s wrong about whether truths about individual events support counterfactuals
(except for the dreary counterfactual that if exactly the same thing were to happen again,
all else being equal, exactly the same effects would ensue.) But, for present purposes,
we’re content to leave it here.
4.4 TNS versus sufficient reason
David Papineau, in his review says:
“If Fodor and Piattelli-Palmarini are right, polar bears don’t have white fur because it confers
advantages in the Arctic; we don’t have eyes because they help us to see; and in general there is
no tendency for natural selection to preserve adaptive traits”.
Could we really be denying that the reason polar bears are white is that being white hides
them in the snow? No. Part of the story about why polar bears are white is surely that
there were many causal chains in which white polar bears got missed by their predators
(and/or were able to sneak up on their prey) more regularly than polar bears that were less
white. On our view, tracing such causal chains is what natural history does for a living.
But a theory of Fs doesn’t consist of an enumeration of causal chains in which Fs are
involved. A theory of Fs is an account of what Fs have in common as such. Accordingly,
a theory of trait evolution is an account of what instances of trait evolution have in
common as such. (Notice, in passing, that “as such” is intensional). So what does TNS
say about what instances of trait evolution have in common as such? What, for example,
13
does it say about what the evolution of four chambered hearts in mammals, and of long
necks in giraffes, and of web spinning in spiders and of bipedal gait in us have in
common qua instances of trait evolution? Just this: In every such case there has to be
something about the creatures (or about their ecology, or both) such that those of the
creatures that were F flourished more than otherwise similar creatures that were not F.
Well of course there has to be. That follows just from the “principle of sufficient reason”
according to which if something is F, there must be something that caused it to be F; and,
of course, whatever the “something” is, it has to be either internal to the organism or
external to the organism. There’s no place else that it could be. On our view there is no
theory of evolution. All there is, is natural history.
Speaking of the adaptive function of the eye (as Papineau urges us to do) a
species of jellyfish (the cubozoan jellyfish, Tridpedalia cystophora discovered in the
waters near Puerto Rico) has 24 globular eyes in 6 groups of 4 (called rhopalia), very
similar to our vertebrate eyes, but no brain to collect the images, no optic nerve, and the
lenses can only form images behind the retina. No adaptive explanation is in sight,
though the genetic and developmental mechanisms responsible for this feast of structure
without function are well understood.
4.5 On mathematical models
Samir Okasha and other reviewers hope to vindicate TNS by appealing to the "paradigm"
(sic) explanatory power of mathematical models of natural selection. We are fully aware
of the long and illustrious tradition of mathematical theory of natural selection and, more
generally, of evolution; from the Hardy-Weinberg law of equilibrium between allele
frequencies (1908)) to the works of Ronald Fisher, J. B. S Haldane and Sewall Wright
(1924-1937) to George R. Price’s theorem (1970, 1972) all the way to the present day
(for thorough expositions see Provine 1971/2001 and Rice 2004). However, as a leading
historian of mathematical evolutionary theories says:
“They [Fisher, Haldane, Wright, Hogben, Chetverikov and other mathematical modelists] all
disagreed, often intensely, with each other about actual processes of evolution in nature, even
when their models were mathematically equivalent.” (William B. Provine 1988, p. 56) (our
emphasis)
Several critiques of the plausibility of many such models have been raised by qualified
biologists including, just to name a few, Carl Woese, Andre’Ariew and Richard
Lewontin, Richard Michod and even Massimo Pigliucci, who is by no means in
sympathy with our view of TNS. In particular, Carl Woese, in a recent interview with
Marc Buchanan for the “New Scientist”, says:
"Biology built up a facade of mathematics around the juxtaposition of Mendelian genetics with
Darwinism, and as a result it neglected to study the most important problem in science - the
nature of the evolutionary process.”
(Buchanan 2010)
And it is again beside the point that scientists are quite often successful in constructing
models of such phenomena as the evolution of sex ratios in a population; or of how actual
foraging strategies approximate ideal foraging strategies; etc. The point is that such
14
models aren’t causal explanations; they don’t do - they don’t even purport to do - what so
many proponents of TNS claim that it does: explicate the causal mechanism of evolution.
The most strenuous defenders of the modern synthesis state explicitly that, although
causal inference is desirable, mathematically, all that is required is correlation. In general,
mathematical models can only be as good as the idealizations on which they are based. In
the words of a leading expert and author of a comprehensive technical treatise:
“It is in the nature of model building that our models often hinge on assumptions that we know
are not exactly true. What is interesting about [two such] assumptions – monomorphic
populations in which variant strategies appear one at a time and populations that respond
quickly to environmental changes – is that they are contradictory. A population cannot quickly
evolve to a new equilibrium unless it has a substantial amount of heritable variation. If evolution
always had to wait for a new variant to arise by mutation, it would be a very slow process,
especially if each new mutation differed from the previous state by only a small amount. Thus,
when one of these assumptions is a good approximation, the other one ceases to be.”
(Sean H. Rice, 2004, Page 289)
Mathematical model building can make explicit the consequences of certain idealizations,
but it doesn’t even purport to reveal the causal mechanisms that sustain the phenomena;
whereas our worry about TNS is that no causal mechanism could do what it claims that
the process of selection-for does.
Conclusion
We continue to believe that there’s a lot that Darwin Got Wrong. We continue to believe
that the issues implied by the externalism of his account of selection, and by his failure to
notice the intensionality of selection-for, are in need of thorough and careful
consideration. Thus far, the critical responses to our attempts have not been edifying;
mostly a howl of reflexive Darwinism, with very little attention paid either to the
structure of the arguments or to their repercussions. But we’re told that hope springs
eternal. Our hope, at a minimum, is to have cleared the ground for calmer and much
more responsible polemics. We still believe in the possibility of a rational,
interdisciplinary, discussion of the empirical warrant and the conceptual architecture of
TNS. But we must admit that we don’t believe in it now as much as we did a year ago.
References for this Update
Block, N. and P. Kitcher (2010). "Misunderstanding Darwin” Boston Review 35(2): 29.
Buchanan M. (2010), “Horizontal and vertical: the evolution of evolution”, in New
Scientist, 2744: 34-37
Coyne, J. A. (2010). “The improbability pump”. The Nation, (May 10th ).
Futuyma, D. J. (2010). "Two critics without a clue." Science 328 (7 May ): 692-693.
Godfrey-Smith, P. (2010). It Got Eaten. [Review of the book What Darwin Got
Wrong.] London Review of Books, 32 (13), 29-30. Retrievable from
15
http://www.lrb.co.uk/v32/n13/peter-godfrey-smith/it-got-eaten.
Goldstein, D. B. (2009). "Common Genetic Variation and Human Traits." New England
Journal of Medicine 360 (17 April, 23): 1696-1698.
Kauffman, S. A. (1993). The Origins of Order: Self-Organization and Selection in
Evolution, Oxford University Press.
Koonin, E. V. (2009 a) "The "Origin" at 150: is a new evolutionary synthesis in sight?"
Trends in Genetics 25 (11): 473-475.
Koonin, E. V. (2009 b). "Darwinian evolution in the light of genomics." Nucleic Acids
Research 37 (4): 1011-1034.
Manolio, T. A., F. S. Collins, N. J. Cox, D. B. Goldstein, L. A. Hindorff, D. J. Hunter, M.
I. McCarthy, E. M. Ramos, L. R. Cardon, A. Chakravarti, J. H. Cho, A. E.
Guttmacher, A. Kong, L. Kruglyak, E. Mardis, C. N. Rotimi, M. Slatkin, D.
Valle, A. S. Whittemore, M. Boehnke, A. G. Clark, E. E. Eichler, G. Gibson, J. L.
Haines, T. F. C. Mackay, S. A. McCarroll and P. M. Visscher (2009). "Finding
the missing heritability of complex diseases." Nature 461 (8 October): 747-753.
Margulis, L. and L. J. Chapman (2010) Kingdoms & Domains: An Illustrated Guide to
the Phyla of Life on Earth, Elsevier: San Diego, London
Okasha, S. (2010): “Review of Fodor and Piattelli-Palmarini’s What Darwin Got
Wrong.” Times Literary Supplement, (March 26th)
Papineau, D. (2010): “Review of Fodor and Piattelli-Palmarini’s What Darwin Got
Wrong.”Prospect168:83-84.
http://www.kcl.ac.uk/content/1/c6/04/17/80/FodorDarwinrevProspect.doc.
Provine, W. B. (1988). “Progress in evolution and meaning in life”. In Evolutionary
Progress. M. H. Nitecki, Ed. Chicago, IL, University of Chicago Press: 49-74.
Provine, W. B. (1971 (republished 2001)). The Origins of Theoretical Population
Genetics. Chicago, IL, University of Chicago Press.
Rice, S. H. (2004). Evolutionary Theory: Mathematical and Conceptual Foundations.
Sunderland, MA, Sinauer.
Venditti, C. and M. Pagel (2009) “Speciation as an active force in promoting genetic
evolution” Trends in Ecology and Evolution 25 (1): 14-20
Woese,
C.
R.
and
N.
Goldenfeld
(2009).
"How
the
Microbial
World
Saved
Evolution
from
the
Scylla
of
Molecular
Biology
and
the
Charybdis
of
the
Modern
Synthesis."
Microbiology
and
Molecular
Biology
Reviews
73 (1
March)):
14‐
21.
16
