In Defence
of Selfish Genes
Richard Dawkins
I
have been taken aback by the inexplicable hostility of Mary
Midgley’s assault.[1] Some colleagues
have advised me that such transparent spite is best ignored, but
others warn that the venomous tone of her article may conceal the
errors in its content. Indeed, we are in danger of assuming that
nobody would dare to be so rude without taking the elementary precaution
of being right in what she said. We may even bend over backwards
to concede some of her points, simply in order to appear fair-minded
when we deplore the way she made them. I deplore bad manners as
strongly as anyone, but more importantly I shall show that Midgley
has no good point to make. She seems not to understand biology or
the way biologists use language. No doubt my ignorance would
be just as obvious if I rushed headlong into her field of
expertise, but I would then adopt a more diffident tone. As it is
we are both in my corner, and it is hard for me not to regard the
gloves as off. I will try to make my reply constructive, in the
hope that it may interest those who have not read Midgley’s article,
as well as those who have. Unattributed quotations with page numbers
will all be taken from her article. Since it was my book, The
Selfish Gene (Oxford: Oxford University Press, 1976), which
stimulated her attack, it will also be necessary for me to quote
from it. I shall divide my reply into eight sections.
Definitional
Misunderstanding
‘[Dawkins’]
central point is that the emotional nature of man is exclusively
self-interested, and he argues this by claiming that all emotional
nature is so. Since the emotional nature of animals clearly is not
exclusively self-interested, nor based on any long-term calculation
at all, he resorts to arguing from speculations about the emotional
nature of genes…' (p. 439). Midgley raises the art of misunderstanding
to dizzy heights. My central point had no connection with what she
alleges. I am not even very directly interested in man, or at least
not in his emotional nature. My book is about the evolution of life,
not the ethics of one particular, rather aberrant, species.
I
shall return to this misunderstanding of me, but for the moment
let me concentrate on her more serious misunderstanding of the definitional
conventions of the whole science of ‘sociobiology’, a science of
which she aspires to be a serious scholar.[2]
When biologists talk about ‘selfishness or ‘altruism’ we are
emphatically not talking about emotional nature, whether of human
beings, other animals, or genes. We do not even mean the words in
a metaphorical sense. We define altruism and selfishness
in purely behaviouristic ways: ‘An entity… is said to be altruistic
if it behaves in such a way as to increase another such entity’s
welfare at the expense of its own. Selfish behaviour has exactly
the opposite effect. "Welfare" is defined as "chances
of survival", even if the effect on actual life and death prospects
is . . small . . . It is important to realize that the above definitions
of altruism and selfishness are behavioural, not subjective.
I am not concerned here with the psychology of motives . . . that
is not what this hook is about. My definition is concerned only
with whether the effect of an act is to lower or raise the
survival prospects of the presumed altruist and the presumed beneficiary’
(The Selfish Gene, pp. 4-5).
It
follows from such a behaviouristic definition of altruism and selfishness
that ‘calculation’, whether long-term or not, is irrelevant, as
is ‘emotional nature’. I assume that an oak tree has no emotions
and cannot calculate, yet I might describe an oak tree as altruistic
if it grew fewer leaves than its physiological optimum, thereby
sparing neighbouring saplings harmful overshadowing. A biologist
would be interested in calculating the genetic and other conditions
which would be necessary for such ‘altruism’ to be favoured by natural
selection: for instance, it might be favoured if the saplings were
close relatives of the tree. Philosophers may object that this kind
of definition loses most of the spirit of what is ordinarily meant
by altruism, but philosophers, of all people, know that words may
be redefined in special ways for technical purposes. In effect I
am saying: ‘Provided I define selfishness in a particular way an
oak tree, or a gene, may legitimately be described as selfish’.
Now a philosopher could reasonably say: ‘I don’t like your definition,
but given that you adopt it I can see what you mean when you call
a gene selfish’. But no reasonable philosopher would say: ‘I don’t
like your definition, therefore I shall interpret your statement
as though you were using my definition of selfishness; by
my definition your concept of the selfish gene is nonsense, therefore
it is nonsense’. This is, in effect, what Midgley has done:
‘Genes cannot be selfish or unselfish, any more than atoms can be
jealous, elephants abstract or biscuits teleological’ (p. 439).
Why didn’t she add to this witty little list, for the benefit of
quantum physicists, that fundamental particles cannot have charm?
If
I spoke of a ‘selfish elephant’ I would have to be very careful
to state, over and over again, whether I meant the word in its subjective
or its behaviouristic sense. This is because a good case might be
made that elephants subjectively experience emotions akin to our
own selfishness. No sensible case can be made that genes do, and
I therefore might have thought myself safe from misunderstanding.
To make doubly sure, I still went to the trouble of emphasizing
that my definition was behaviouristic. The many laymen who have
read my book seem to have had little trouble in grasping this simple
matter of definition.
Did
Midgley, perhaps, just overlook my definition? One cannot, after
all, be expected to read every single word of a book whose author
one wishes to insult. But in the present case no such excuse can
be made. ‘My’ definition is not private to me. It is essentially
the same kind of definition as is used by all modern biologists
who write about social behaviour in animals, and Midgley is supposed
to know about these things. Actually I think it is arguable that
we ethologists (‘sociobiologists’) have overdone our insistence
on objective, behaviouristic definitions of words like ‘hunger’,
‘fear’ and ‘selfishness’. Maybe one day we will all come round to
the minority view of Donald Griffin (The Question of Animal Awareness,
New York: Rockefeller University Press, 1976) that the present
anti-subjective bias of ethological language constitutes ‘an obsolete
straitjacket’. But for the time being, whether we like it or not,
it just is the case that biologists use these words in a special,
restricted sense. A philosopher who wishes to understand biologists
must, therefore, learn this basic feature of biological language,
particularly a philosopher who aspires to write about biology. The
imagination reels at what a mind labouring under Midgley’s definitional
misconception must make of almost any of the modern literature on
animal behaviour.
Egoism
To
Midgley it evidently follows from her misunderstanding of my words
that I am advocating an egoistic view of human ethics, or at least
that I ‘would like to be an egoist’ (p. 446). But even if, to grant
the inconceivable, I really was saying that genes had a selfish
‘emotional nature’ (p. 439), it would not follow that I thought
human beings had one too. And even if I did think human beings
were fundamentally selfish, it would not follow that I welcomed
the idea. In fact, of course, to the extent that I am interested
in human ethics (a rather small extent), I disapprove of egoism.
To the extent that I know about human psychology (again, a rather
small extent), I doubt if our emotional nature is, as a matter of
fact, fundamentally selfish. And I of course do not think genes
have emotional natures at all.
Let
me try to say again what I do think. The facts of ethology certainly
deny individual egoism as a rule in nature. Every ethologist knows
this, and examples abound in my book. how, then, is the Darwinian
to explain individual altruistic behaviour in animals? ‘Group selection’
is one possible answer: a species, or other group, may selfishly
survive at the expense of rival groups if the individuals within
it behave altruistically towards each other. But unfortunately,
except under very special conditions, biologists now agree that
group selection cannot work in nature. There is no authoritative
support for the once fashionable habit of explaining animal adaptations,
altruistic behaviour among them, as ‘for the good of the species’.
Midgley, incidentally, has this old biology A-level reflex well
developed, as when she says ‘What is maladaptive . . . damages the
species’s chances of surviving’ (Beast and Man, p. 149),
and .... there is a problem about evolution, which runs "Can
a species survive if each member of it sometimes does things which
do not (in fact) pay him?" (op. cit., p.117). The contemporary
biologist would say that whether or not a species survives is, though
doubtless an interesting question, nothing to do with Darwinian
selection. Darwinian selection does not choose among species.
What,
then, does it choose among? The favoured answer is ‘individuals’.
In a sense this is correct, but only if we put it very carefully;
what matters is not differential survival of individuals,
but differential inclusive genetic fitness of individuals. The fitness
of an individual (again, this is a special technical usage, different
from everyday usage) means its success in getting copies of its
genes represented in future generations. Fitness is a difficult
quantity to calculate and a difficult concept to understand (see,
for instance, Midgley’s own misunderstanding of it in Beast and
Man, pp. 138-140). My suggestion is that we can lessen the risk
of misunderstanding if we shift our attention from the organism
as agent, to the gene itself. Inclusive fitness is, I have only
half facetiously pointed out, ‘that property of an individual organism
which will appear to be maximized when what is really being maximized
is gene survival’.[3] We may say, with
the majority of modern specialists, that maternal care is favoured
by natural selection because of its beneficial effects on the inclusive
fitness of the mothers concerned. Or, we may say what is essentially
the same thing in terms of the selfish gene: genes that make mothers
care for their young are likely to survive in the bodies of the
infants cared for; genes that make mothers neglect their infants
are likely to end up in dead infant bodies; therefore the gene pool
becomes full of genes that induce maternal care; this is why we
see maternal care in nature.
In
effect, what I have done is to reject ‘the selfish group’ as an
explanation of individual altruism, to say ‘the selfish individual’
is a better, but more complex and easily misunderstood, alternative,
and to offer ‘the selfish gene’ as a simple, correct alternative.
The details are by no means simple, however, and my book is a working
out, in various ways, of the complications and implications of this
fundamental principle, that individual behaviour, altruistic or
selfish, is best interpreted as a manifestation of selfishness at
the gene level.
To
illustrate the kind of argument I was making, I used an analogy:
‘If we were told that a man had lived a long and prosperous life
in the world of Chicago gangsters, we would be entitled to make
some guesses as to the sort of man he was. We might expect that
he would have qualities such as toughness, a quick trigger finger,
and the ability to attract loyal friends
Like
successful Chicago gangsters, our genes have survived, in some cases
for millions of years, in a highly competitive world. This entitles
us to expect certain qualities in our genes’ (The Selfish Gene,
p. 2). If anybody had suggested to me that it was possible to
misread that passage as saying that people are essentially Chicago
gangsters I would have laughed. Yet this superhuman feat of misunderstanding
is exactly what Midgley manages to achieve, ‘. . . telling people
that they are essentially Chicago gangsters is not just false
and confused, but monstrously irresponsible’ (p. 455). I ask Midgley
to look again at my words, take a few deep breaths and read them
calmly and quietly. See the role of my Chicago gangster analogy.
The point was that knowledge about the kind of world in which a
man has prospered tells you something about that man. It had nothing
to do with the particular qualities of Chicago gangsters.
I could just as well have used the analogy of a man who had risen
to the top of the Church of England, or been elected to the Athenaeum.
In any case it was not people but genes that were the subject
of my analogy.
Reciprocal
Altruism
Midgley’s
misunderstanding of the theory of reciprocal altruism is a special
case of her more general muddle, already alluded to, about animals
‘calculating’. The evolutionary theory of reciprocal altruism, largely
due to R. L. Trivers, was the subject of J
L. Mackie’s paper in this journal which was the immediate
stimulus for Midgley’s attack. Briefly, Trivers suggested that the
principle of doing favours in the ‘expectation’ of their possibly
being returned later, which we understand at the level of
conscious calculation, can be made to work in an evolutionary model
without assuming conscious calculation. The appropriate mathematics
is the theory of games, as I illustrated in my simple explanatory
model of three ‘strategies’ called ‘cheat’, ‘sucker’, and ‘grudger’
(The Selfish Gene, pp. 197-201). Now Midgley appears to think
that reciprocal altruism can only work in animals that can ‘calculate’.
She quotes E.O. Wilson’s surprising statement that ‘Human behaviour
abounds with reciprocal altruism consistent with genetic theory,
but animal behaviour seems to be almost devoid of it’
(Midgley’s italics, not in original, not acknowledged). Midgley
goes on: ‘[Wilson] accounts for this (as I do) by the lack of calculation
in animals, but seems not to see that, since these "animals"
are the subjects we are dealing with for almost the whole of evolution,
any "genetic theory" inconsistent with their capacities will
have to be revised’ (p. 444).
I
would have been surprised if Wilson had really invoked ‘the lack
of calculation in animals’, and indeed, as far as I can see, he
does not. What he does suggest is that '…in animals relationships
are not sufficiently enduring, or memories of personal behavior
reliable enough, to permit the highly personal contracts associated
with the more human forms of reciprocal altruism’ (Sociobiology,
p. 120). I think Wilson underestimates the power of animals
to recognize and remember each other, but, be that as it may, he
is talking about memory, which is quite different from Midgley’s
‘calculation’. More importantly, far from the theory of reciprocal
altruism needing calculation, it doesn’t even need memory, at least
in the ordinary sense of the word. All that is required is some
functional equivalent of a memory of past favours. It does
not have to be a real memory residing in the nervous system. This
is, indeed, the novelty of Trivers’ contribution, since any fool
can see that the principle of reciprocation will work in a species
that is capable of remembering past favours and calculating debts.
Midgley might have realized this if, instead of relying on her admittedly
slightly misleading secondary source, she had gone back to the primary
source (R. L. Trivers, ‘The Evolution of Reciprocal Altruism’, Quarterly
Review of Biology 46 (1971), 35 - 57).
She
might even have got the point from The Selfish Gene (pp.
201-202): ‘Trivers discusses the remarkable symbiosis of the cleaner-fish.
Some fifty species, including small fish and shrimps, are known
to make their living by picking parasites off the surface of larger
fish of other species. The large fish obviously benefit from being
cleaned, and the cleaners get a good supply of food... In many cases
the large fish open their mouths and allow cleaners right inside
to pick their teeth, and then to swim out through the gills which
they also clean. One might expect that a large fish would craftily
wait until he had been thoroughly cleaned, and then gobble up the
cleaner. Yet instead he usually lets the cleaner swim off unmolested.
This is a considerable feat of altruism because in many cases the
cleaner is of the same size as the large fish’s normal prey . .
. Each cleaner has his own territory, and large fish have been seen
queueing up for attention like customers at a barber’s shop’ (not
a real barber’s shop with scissors and electric clippers, I suppose
I now have to add). ‘It is probably this site-tenacity which makes
possible the evolution of delayed reciprocal-altruism in this case.
The benefit to a large fish of being able to return repeatedly to
the same "barber’s shop", rather than continually searching
for a new one, must outweigh the cost of refraining from eating
the cleaner.’
The
important point is that neither calculation nor memory of past favours
need be invoked. Site-tenacity on the part of both kinds
of fish is sufficient. The site-tenacity, which, by the way, is
a commonplace of fish ethology, acts as a kind of equivalent of
a memory. In Darwinian terms we say that, given site-tenacity
by both cleaners and cleaned fish, natural selection favours merciful
behaviour by large fish towards their cleaners. Calculations of
probable future benefit are done by the biologist, not by the fish
(they might be done by the fish, but that is incidental).
The fish simply do things which have consequences in given conditions,
and natural selection judges them by those consequences.
The
idea of animals behaving as if calculating odds without really
doing so is fundamental to an understanding of the whole of sociobiology:
‘Just as we may use a slide rule without appreciating that we are,
in effect, using logarithms, so an animal may be pre-programmed
in such a way that it behaves as if it had made a complicated
calculation . . . This is not so difficult to imagine as it appears.
When a man throws a ball high in the air and catches it again, he
behaves as if he had solved a set of differential equations in predicting
the trajectory of the ball. He may neither know nor care what a
differential equation is, but this does not affect his skill with
the ball’ (The Selfish Gene, pp. 103-104; see also my reply
to Marshall Sablins: misunderstanding number 3 in R. Dawkins, ‘Twelve
Misunderstandings of Kin Selection’, Zeitschrift fur Tierpsychologie
51 (1979), 184-200).
There
are other odd things in Midgley’s section on reciprocal altruism.
For instance she devotes a paragraph to a trenchant and forceful
advocacy of the obviously undisputed proposition that ‘The main
source and focus of altruistic behaviour in animals is the care
of the young, which in most species will certainly never be repaid’
(p. 440, my italics). Who is supposed to be surprised? Not me,
I am relieved to note, since reciprocation occupies a very small
part of my book and kin-selected parental care rather a large one.
Midgley’s target in this case is J. L. Mackie (‘The Law of the Jungle’,
Philosophy 53 (October 1978)), but her shot is aimed not
at his main point (which she seems to have overlooked), but at his
little aside about Nietzsche.
Before
explaining why I think Mackie’s paper may be an important contribution
to biology, I cannot leave the subject of parental care without
calling attention to the following, from Midgley: ‘This persistent
difficulty in reducing parents to the egoist pattern is just the
kind of thing which makes Dawkins’s typical readers - people with
vaguely egoist leanings about individual human psychology - willing
to follow him in losing touch with the observed facts of motivation
altogether and taking off for the empyrean with the Gene’ (pp. 443-444).
It is one thing to insult the author of a book, but how dare
Midgley pontificate about its ‘typical readers’? I don’t think
I have had the pleasure of meeting any readers of Mrs Midgley’s
book, but no doubt they vary and would resent prejudiced generalizations
about their ‘leanings’ and ill-informed slurs against their critical
faculties.
Mackie’s
Contribution
Midgley’s
emotional reaction to a few words and phrases used by Mackie seems
to have blinded her to the potentially important suggestion he was
making. I shall explain this, since Mackie himself did not follow
his train of thought to its conclusion. Group selection is the hypothetical
process whereby natural selection chooses among whole groups of
organisms, as opposed to choosing among individuals (see J. Maynard
Smith, ‘Group Selection’, Quarterly Review of Biology 31,
277-283). As I have explained, it is widely agreed to be an unworkable
theory, but if it did work it would be important since it could
explain altruistic behaviour: groups containing altruistic individuals
are less likely to go extinct than groups of selfish individuals.
Mathematical models by Maynard Smith himself and others have shown
that the theoretical objections to group selection would largely
vanish if we were allowed to assume the existence of high genetic
variance among groups compared to within-group variance. This is
a technical way of saying that there has to be a tendency for fellow
group members to share more genes with each other than they share
with random members of the population at large. This assumption
will clearly be met if genetic relatives go about in family groups,
but then we are dealing with the well-understood phenomenon of ‘kin
selection’, not group selection at all. Mackie’s contribution, though
he does not put it like this, is to have offered us a new mechanism
whereby the variance-differential necessary for group selection
could be maintained. The argument is as follows.
My
game-theoretic analysis of ‘cheats, suckers and grudgers’ led to
two alternative stable solutions. A population dominated by cheats
would not be invaded (evolutionarily speaking) by suckers or grudgers.
If, however, a population chanced to acquire more than a critical
frequency of grudgers, natural selection would suddenly start favouring
grudgers, until they became a runaway majority. The concept of a
bistable system is a slightly subtle one, and it is not surprising
that Midgley misunderstood it in her summary: ‘Dawkins concludes
that Cheats and Grudgers would exterminate Suckers, and Grudgers
might well do best of all’ (p. 440). The whole point is not that
grudgers might do better or worse than cheats, but rather that whichever
of the two happened to attain more than a critical frequency
in the population would, by virtue of that fact, do better
than the other. For the present argument, the important consequence
is that such a bistable system is a recipe for high between-group
variance: some populations would stabilize at the grudger equilibrium;
others would stabilize at the cheat equilibrium. Populations with
intermediate relative frequencies would be inherently unstable,
and natural selection at the individual level would push them to
one extreme or the other. Selection within groups would thus see
to it that variance between groups was high. Mackie’s argument is
that group selection would now have a real chance to work, differentially
extinguishing groups of cheats at the expense of groups of grudgers
(reciprocal altruists). It is too early to say, yet, whether formal
mathematical models will uphold this possibility, but if they do,
Mackie’s paper in Philosophy will have to be seen as a useful
contribution to biology. I should add that a brief similar suggestion
has been made independently by M. J. Wade (‘A Critical Review of
the Models of Group Selection’, Quarterly Review of Biology 53
(1978)).
Models
Midgley
describes my model of cheats, suckers and grudgers as an ‘absurdly
abstract and genetically quite impossible situation’ (p. 440), and
as a ‘grossly simplified and distorted scheme’ (p. 441). But of
course it is abstract, simplified and distorted. This is what
models are, and that is what gives them their usefulness. It is
the very property which made my model useful to Mackie and which
stimulated his useful contribution. Models do not aspire to mimic
reality faithfully. If they did, they would not be models, they
would be reality. In physics, for instance, it is sometimes convenient
to imagine a body - it may even be described as a train - travelling
at nearly the speed of light past an observer, who sees the passengers
hideously foreshortened. Only a pedant would point out that trains
can’t go that fast, and that in any case the observer wouldn’t have
time to see the passengers. If a philosopher made such an objection
against the writings of a particular physicist, we could justly
conclude that he or she did not understand the first thing about
physics, since all physicists make use of such simplified models.
Yet this is almost exactly the nature of Midgley’s objection to
my ‘grudger/sucker/cheat’ model. If she had objected that it was
a bad model I would have listened sympathetically, but that is not
what she did. She appears not to have understood that it was a model
at all.
In
the present state of evolutionary biology, the preferred models
embody various kinds of deliberate simplification, and one of the
most fashionable of these deliberate simplifications is the ‘one
gene one strategy’ model that worries Midgley so much. I am only
one of many biologists for whom it is a convenient weapon in our
theoretical armoury. Others who frequently wield it include J. Maynard
Smith and E.O. Wilson, to name two biologists whom Midgley singles
out for special praise in her article. It is ironic that she should
compare my ‘gene-selection’ treatment of the paradox of sex, to
its disadvantage, with a passage from John Maynard Smith’s rightly
praised The Evolution of Sex (Cambridge University Press,
1978). Like nearly all Maynard Smith’s works, this book abounds
in simplified models of exactly the kind Midgley castigates. If
she had read beyond the Preface to page 113, Midgley would have
found Maynard Smith specifically endorsing gene-selection models
of sexuality, invoking in his support the very passage from The
Selfish Gene which Midgley describes as a ‘danger’.
If
a philosopher attacked modern evolutionary biology as a whole for
its reliance on over-simple models, again we would have to listen.
But a philosopher who intemperately attacks one particular biologist
for doing exactly what most of his professional colleagues do, and
have done for decades, displays fundamental and profound ignorance
of the methods of biology. It may be that we shall eventually find
today’s ‘one gene one strategy’ models too simple to be useful.
The intuition of professionals varies here. My own hunch, for what
it is worth, it that most of the major principles of present day
‘strategy’ models will survive future injections of genetic complexity,
while the quantitative details of their predictions will not. We
must patiently wait and see.
Genes
‘There
is nothing empirical about Dawkins. Critics have repeatedly pointed
out that his notions of genetics are unworkable’ (p. 439). No critic
is named. The footnote refers only to a 1978 paper of mine.[3]
Midgley says that in this paper I have ‘eventually’ made an
‘attempt to answer some of these criticisms’. In fact I made no
such attempt, because no such criticisms were known to me. If Midgley
will cite the ‘repeated’ criticisms I will read them with attention
and, if appropriate, reply to them.
My
notions of genetics are actually much more conventional than Midgley
thinks. She herself would have a great deal of trouble with the
concept of the gene, as it is ordinarily used by geneticists: ‘For
selection to work as [Dawkins] suggests by direct competition between
individual genes, the whole of behaviour would have to be divisible
into units of action inherited separately and each governed by a
single gene . . . To convince us that this is so, Dawkins brings
up once more the case of Rothenbuhler’s Hygienic Bees, creatures
which have been appearing in suspicious isolation as a stage army
in all such arguments for some time. Actually, not only does the
bees’ case stand alone, but it is certainly not proven. To show
that even the simple behaviour it involves is really governed by
only two genes would take something like seventy generations of
outbreeding experiments to ensure that the effects described are
not due to the close linkage of genes at a whole series of adjacent
loci, and even this would not show that these genes affected nothing
else’ (p. 449). There are so many muddles interwoven here, it is
hard to know where to start unravelling.
Probably
the first point to make is that whenever a geneticist speaks of
a gene ‘for’ such and such a characteristic, say brown eyes, he
never means that this gene affects nothing else, nor that it is
the only gene contributing to the brown pigmentation. Most genes
have many distantly ramified and apparently unconnected effects.
A vast number of genes are necessary for the development of eyes
and their pigment. When a geneticist talks about a single gene effect,
he is always talking about a difference between individuals.
A gene ‘for brown eyes’ is not a gene that, alone and unaided, manufactures
brown pigment. It is a gene that, when compared with its alleles
(alternatives at the same chromosomal locus), in a normal environment,
is responsible for the difference in eye colour between individuals
possessing the gene and individuals not possessing the gene. The
statement ‘G1 is a gene for phenotypic characteristic
P1’ is always a shorthand. It always implies the existence, or potential
existence, of at least one alternative gene 2, and at least one
alternative characteristic P2. It also implies a normal developmental
environment, including the presence of the other genes which are
common in the gene pool as a whole, and therefore likely to be in
the same body. If all individuals had two copies of the gene ‘for’
brown eyes and if no other eye colour ever occurred, the ‘gene for
brown eyes’ would strictly be a meaningless concept. It can only
be defined by reference to at least one potential alternative. Of
course any gene exists physically in the sense of being a length
of DNA; but it is only properly called a gene ‘for X', if there
is at least one alternative gene at the same chromosomal locus,
which leads to not X.
It
follows that there is no clear limit to the complexity of the ‘X’
which we may substitute in the phrase ‘a gene for X’. Reading, for
example, is a learned skill of immense and subtle complexity. A
gene for reading would, to naive common sense, be an absurd notion.
Yet, if we follow genetic terminological convention to its logical
conclusion, all that would be necessary in order to establish the
existence of a gene for reading is the existence of a gene for not
reading. If a gene G2 could be found which infallibly
caused in its possessors the particular brain lesion necessary to
induce specific dyslexia, it would follow that G1, the gene which
all the rest of us have in double dose at that chromosomal locus,
would by definition have to be called a gene for reading. Imagine
a tribe in which almost everybody had G2 and therefore
could not learn to read. Now the rare possessors of G1 would be
the sole literates and, provided adequate educational opportunities
were available to all, reading behaviour would be inherited according
to the elementary laws of Mendelian genetics. Dyslexia would not,
of course, be the only describable effect of such a gene. All genes
are fundamentally ‘genes for making proteins’, but it is a routine
convenience in genetics to accept other labels such as ‘gene for
brown eyes’. Which of the intricately ramified consequences of the
fundamental protein effect we choose to use as a label is simply
a matter of convenience. The hypothetical ‘gene for dyslexia’ would
almost certainly have other psychological or perceptual effects,
but in our world where reading is so important dyslexia might well
be its most salient effect, and the dyslexia label would therefore
be convenient. The same gene, in a Pleistocene environment, might
earn a different label, say ‘gene for being unable to read animal
footprints’. Similarly, a gene for total blindness would obviously
prevent reading, but it would not be convenient to label it by this
property since other effects of total blindness would be more noticeable.
The normal alternative to a gene for total blindness could sensibly
be called a gene for seeing, but not a gene for reading. This is,
of course, a hypothetical example. I know of no evidence of a gene
for dyslexia. My only point is that the complexity, per se, of
a behaviour pattern such as reading is irrelevant to the plausibility
of there being a single gene ‘for’ that behaviour pattern. To summarize
the reason for this, it is that differences between behaviour
patterns can have unitary and simple causes, even if the behaviour
patterns themselves are highly complex.
It
is no part of my world view that the whole of behaviour must be
‘divisible into units of action inherited separately and each governed
by a single gene’. Since Midgley is not the only person to have
had trouble in grasping this point, let me use an analogy which
others seem to have found helpful. The genetic code is not a blueprint
for assembling a body from a set of bits; it is more like a recipe
for baking one from a set of ingredients. If we follow a particular
recipe, word for word, in a cookery book, what finally emerges from
the oven is a cake. We cannot now break the cake into its component
crumbs and say: this crumb corresponds to the first word in the
recipe; this crumb corresponds to the second word in the recipe,
etc. With minor exceptions such as the cherry on top, there is no
one-to-one mapping from words of recipe to ‘bits’ of cake. The whole
recipe maps on to the whole cake. But suppose we change one word
in the recipe; what now emerges from the oven is a different cake,
different through its whole substance. If we have 100 cakes baked
according to the first version of the recipe and 100 cakes baked
according to the second version of the recipe, it will be possible
to say: although there is no one-word-one-crumb mapping from recipe
to either cake, it is true that a one word difference between these
two recipes is solely responsible for the only consistent differences
between this set of 100 cakes and that set of 100 cakes.
To
repeat, then, geneticists are not concerned with ‘one gene one bit-of-animal’
mapping. They are concerned with ‘one gene-difference one animal-difference’
mapping. And just as geneticists are concerned with inter-individual
differences, so is natural selection. Natural selection can be said
to choose individuals versus rival individuals, but it is only if
the responsible differences between the individuals are due to genes
that natural selection can have any evolutionary consequences. For
instance, if selection favours fleetness of foot within a preyed-upon
species, but individual differences in fleetness of foot are entirely
non-genetic in origin, no evolutionary change will result from the
selection: fast runners will come to predominate among the survivors
of each generation, but they will not pass their fleetness of foot
on to the next generation, so no evolution will be seen. It follows
that if we believe that X is a Darwinian adaptation, we are committing
ourselves to the belief that, in the past anyway, there must have
been at least one gene ‘for’ X. And Midgley’s implication that the
hygienic honey bee is the only known example of a gene effect on
behaviour (it isn’t, of course; it is just the most spectacular),
and that even it may be suspect, is tantamount to a disavowal of
the entire principle of the evolution of behavioural adaptation
by natural selection!
We
now come to the allegedly important distinction between a single
gene and a linked series of adjacent genes, and the statement that
it would take ‘something like seventy generations of outbreeding
experiments’ to demonstrate a single gene effect as opposed to a
close linkage effect. I hope nobody was impressed by the spurious
impression of scientific precision conveyed by that ‘seventy generations’.
Why seventy, not seven hundred or seven thousand? No magic number
of outbreeding experiments can settle the issue, because it is a
non-issue, or, more precisely, because ‘the gene’, as I use the
term, is an asymptotic, not an all or none, concept. If a series
of adjacent genes is so closely linked that it takes n generations
of breeding experiments to separate them, then for practical purposes
we can treat them as one gene (‘supergene’ it is sometimes called),
provided n is large in relation to the time span we are interested
in. And the time span we are interested in here is the evolutionary
time span. If we are examining a particular behaviour pattern as
a possible Darwinian adaptation, we will be content to regard it
as controlled by a single gene provided natural selection, too,
‘regards’ it as controlled by a single gene - that is, provided
the risk of the supergene’s being split into its component sub-genes
is small compared to the risk of its being eliminated by the natural
selection pressures we are investigating.
It
is admittedly true that ‘the gene’ is an asymptotic rather than
an all or none concept only if defined in a particular way. A molecular
biologist might define it so that it became an all or none concept.
But I am not a molecular biologist, and I made my definition very
clear: ‘A gene is defined as any portion of chromosomal material
which potentially lasts for enough generations to serve as a unit
of natural selection’ (The Selfish Gene, p. 30). Midgley
quotes this definition, expressing surprise that I got it from George
Williams (whom she rightly admires), and adding, as though it were
an objection, that I ‘might be talking about any section of the
DNA’ (p. 451).[4]
That
is my point. I am not searching for an ideal, indivisible, atom-like
unit. I am searching for a chunk of chromosomal material which,
in practice, behaves as a unit for long enough to be naturally selected
at the expense of another such fuzzy unit. I agree that there are
difficulties in this way of looking at evolution, but I believe
I have shown them to be less great than the difficulties inherent
in any other way that has been suggested. The individual organism
is a fuzzy unit too (think of vegetatively propagating plants),
yet it is current orthodoxy that ‘the individual is the unit of
selection’. The group of individuals is even more fuzzy, and it
is partly for this reason that it is no longer regarded as a significant
unit of selection. The truth is that there is no hard atomic unit
of natural selection, but I believe my ‘fuzzy gene’ or ‘replicator’
is the most convenient approximation.
Once
again, philosophers should be particularly sympathetic towards special-purpose
re-definitions of words, but actually the present case hardly deserves
to be called re-definition at all. There never has been a generally
agreed definition of the gene. Pre-molecular usage, in practice,
amounted to the gene of the Williams definition, although in principle
it was thought of as an indivisible ‘bead’ on a chromosomal string.
In the 1950s, molecular biology showed that there were no atomistic
beads, and Seymour Benzer [5] suggested
that ‘the gene’ should be replaced by three terms: the muton was
the minimum unit of mutational change; the recon was the minimum
unit of recombination; and the cistron was defined in a way that
was strictly applicable only to micro-organisms, but for practical
purposes it amounted to the unit of protein synthesis. Which of
the three gene definitions one used was to depend on one’s purposes.
But Benzer’s purposes were all molecular. For the student of adaptation
in whole organisms yet another unit, which I shall call the ‘optimon’,
is required. The optimon is that unit to which we refer when we
speak of a Darwinian adaptation as being ‘for the good of’ something.
Williams, in effect, defined the gene as equivalent to what I am
calling the optimon. In The Selfish Gene I followed him,
but I have since suggested substituting the more general term replicator,
since ‘gene’ gives rise to confusion (and how!). This whole
area of units of genetic function and units of adaptive benefit
is fraught with important difficulties, but the alleged difficulties
manufactured by Midgley are not among them. I do not claim that
my essay on replicator selection [3] solves
all the problems, but I think that it, and the paper of the philosopher
David Hull [6] that follows it, are
honest attempts to face up to the difficulties, and that some progress
is being made.
Midgley
(p. 454) quotes with approval Stephen Jay Gould’s courteously expressed
criticism: ‘No matter how much power Dawkins wishes to assign to
genes, there is one thing that he cannot give them - direct visibility
to natural selection. Selection simply cannot see genes and pick
among them directly. It must use bodies as an intermediary . . .‘[7]
I find it impossible to imagine what it would even mean to
say that genes were directly visible to natural selection. Of course
they have to use bodies as an intermediary. That is why my book
is mostly about the behaviour of individual bodies (see especially
Chapter 4 for a discussion of the role of bodies as machines programmed
to preserve genes, like computers programmed to win games of chess).
Natural selection favours genes (replicators) versus their alleles
by virtue of those genes’ effects on bodies. But it is not the bodies
that survive; they reproduce their genes and die. Only genes survive,
in the form of information copies of themselves (why, by the way,
does Midgley think the perfectly obvious fact that ‘a gene cannot
perpetuate itself but only likenesses of itself’ (p. 446)
is such a ‘crashing’ disaster for my case? It is one of the very
linch-pins of my case!). Evolution consists in the differential
copying success of genes relative to their alleles. The genes which
exist in the world are, obviously, the genes whose replicas in previous
generations were successful in getting themselves copied. Such success
is achieved by means of influence on the development of bodies.
Bodies, therefore, tend to have what it takes to propagate genes,
and may properly be regarded as engines of gene propagation - ’survival
machines’.
Sociobiology
Midgley’s
malice at times becomes positively catty, as, for instance, when
she gratuitously remarks that my ‘pages are virgin of originality…'
(p. 444), my material having all been drawn from ‘evolutionists
such as W. D. Hamilton, Edward O. Wilson, and John Maynard Smith
who are not directly interested in individual psychology at all’.
In another place she quotes a sentence from Wilson’s Sociobiology
(Harvard University Press, 1975; ironically the sentence is
the very one on reciprocal altruism, which, as I showed above, Midgley
so pathetically misunderstood). She then adds: ‘Dawkins . .. ignores
Wilson’s reasoning here, as he does most other things that do not
suit him’ (p. 444). I did not ‘ignore’ Wilson’s reasoning: at the
time of writing (1975) I, together with most other people, had not
had an opportunity of seeing Wilson’s book. After completing my
book in essentially its final form I obtained a copy of Sociobiology,
and managed to slip into my final draft a brief mention of it
(a criticism of Wilson’s treatment of the theory of kin selection:
I prophesied that he would muddle people, and p. 140 of Midgley’s
Beast and Man shows my forecast to have been amply fulfilled).
This was the only change Sociobiology caused in my entire
text. Only after The Selfish Gene had gone to press did I
read Wilson’s excellent work from cover to cover, and even then
(early 1976) I must have been one of the first people in Britain
to do so. Any claim that I was influenced by Wilson is simply false.
The claim that I drew material from Hamilton and Maynard Smith is,
of course, true. I am proud of it, and acknowledged my debt to them,
and to George Williams and Robert Trivers. Like E. O. Wilson, I
was trying primarily to synthesize and interpret our field (it wasn’t
called sociobiology then), and only incidentally trying to break
new ground (although I think both Wilson and I would be disappointed
if we were thought to have broken no new ground). Both Wilson and
I would have been sadly remiss if we had not given great prominence
to Hamilton’s ideas on kinship and other topics. In my opinion [8]
Wilson was rather remiss in virtually ignoring Maynard
Smith’s game-theoretic concept of the evolutionarily stable strategy.
As for the statement that Wilson is ‘not directly interested in
individual psychology at all’, hollow laughter seems the only appropriate
response. Whatever does Midgley think the ballyhoo, the political
demonstrations, the ‘Sociobiology Study Group of Science for the
People’ are all about? If anyone remains in doubt, I recommend Wilson’s
On Human Nature (Harvard University Press, 1978).
Concluding
Remarks
If
the reader discerns in my reply signs of what appears to be undue
rancour, I beg him or her to scan a few random sentences of Midgley’s
paper and judge the provocation. It is not an invited book review,
remember, but a voluntarily contributed article. Her concluding
footnote would be hard to match, in reputable journals, for its
patronizing condescension toward a fellow academic (a fellow academic,
moreover, who is a professional in the field under discussion, a
field in which the critic herself is most charitably described as
trying hard): ‘Up till now, I have not attended to Dawkins, thinking
it unnecessary to break a butterfly upon a wheel. But Mr Mackie’s
article is not the only indication I have lately met of serious
attention being paid to his fantasies’ (p. 458). Incidentally, when
Midgley says she has not ‘attended to’ me before, this is not strictly
accurate. In Beast and Man (e.g. p.131) will be found criticisms
of the concept of the selfish gene’, but it is an orphaned concept,
named but without a responsible author. Her readers were served
up with the criticism, without being trusted with the information
that ‘the selfish gene’ being criticized is, in fact, a real book,
with an author, a date, and a publisher—a book that they might go
away and judge for themselves against her criticism. Worse, in her
Introduction (p. xxii), the concept of ‘the selfish gene’ is solemnly
attributed to Edward Wilson, a fact which probably annoys him even
more than it annoys me (he tells me he finds my ideas reductionist).
What, in the circumstances, are we to make of her publisher’s claim
on the dustjacket that Midgley’s comments on ‘Wilson’s concept of
"the selfish gene are the most serious and sustained criticism
of Wilson yet published’?
Let
me not end on a negative note. Midgley has a lot to say about metaphor,
and I can end constructively by explaining why it was unnecessary
for her to say it. She thought that I would defend my selfish genes
by claiming that they were intended only as a metaphor, and assumed
that I was speaking metaphorically when I wrote, ‘We are survival
machines-robot vehicles blindly programmed to preserve the selfish
molecules known as genes. This is a truth which still fills me with
astonishment’ (The Selfish Gene, p. ix). But that was no
metaphor. I believe it is the literal truth, provided certain key
words are defined in the particular ways favoured by biologists.
Of course it is a hard truth to swallow at first gulp. As Dr Christopher
Evans has remarked, ‘This horrendous concept - the total prostitution
of all animal life, including Man and all his airs and graces, to
the blind purposiveness of these minute virus-like substances -
is so desperately at odds with almost every other view that Man
has of himself, that Dawkins’ book has received a bleak reception
in many quarters. Nevertheless his argument is virtually irrefutable’
(The Mighty Micro, London: Gollancz, 1979, 171). For my part,
what has gratified me is that the anticipated bleak reception has,
in the event, been confined to so few quarters, and such
unpersuasive ones.[9]
New College, Oxford
References
1 M. Midgley, ‘Gene-juggling’,
Philosophy 54 (October 1979).
2 She recommends her own book (M. Midgley, Beast
and Man, Hassocks: Harvester Press, 1979) ‘For a fuller discussion
of sociobiological ideas…'.
3 R. Dawkins, ‘Replicator Selection and the Extended
Phenotype’, Zeitschrift fur Tierpsychologie 47, 61-76.
4 It is hard to resist a flourish as I quote almost
exactly the same words from a recent, forward-looking review by
Francis Crick, architect (with J. D. Watson) of the modern molecular
concept of the gene: ‘The theory of the "selfish gene"
will have to be extended to any stretch of DNA’ (F. H. C. Crick,
‘Split Genes and RNA Splicing’, Science 204 (1979), 270).
Crick’s point is elaborated in two further molecular biological
papers whose titles betray no coy reticence about applying the word
‘selfish’ to DNA molecules! (L. E. Orgel and F. H. C. Crick, ‘Selfish
DNA: the Ultimate Parasite’, Nature 284 (1980); W. F. Doolittle
and C. Sapienza, ‘Selfish Genes, the Phenotype Paradigm and Genome
Evolution’, Na lure 284 (1980). As for my definition of the
gene, its derivation from Williams is not word for word, but I have
conveyed the clear message of pp. 22-25 of his Adaptation and
Natural Selection (New Jersey: Princeton University Press, 1966).
My definition is a rendering, for laymen, of two technical sentences
from these pages of Williams: ‘I use the term gene to mean "that
which segregates and recombines with appreciable frequency"
'; and ‘a gene could be defined as any hereditary information for
which there is a favorable or unfavorable selection bias equal to
several or many times its rate of endogenous change’.
5 S. Benzer, ‘The Elementary Units of Heredity’,
The Chemical Basis of Heredity, W. D. McElroy and B. Glass
(eds) (Baltimore: Johns Hopkins, 1957).
6 D. L. Hull, ‘The Units of Evolution: a Metaphysical
Essay’, Studies in the Concept of Evolution, U. J. Jensen
and R. Harré (eds) (Hassocks: The Harvester Press, in press).
In view of her spirited remark that I should either learn to do
metaphysics or retreat out of sight altogether, Midgley might be
amused at the following from Hull’s manuscript: ‘Although he is
likely to be shocked, if not offended, at being told so, Dawkins
(1976, 1978) has made an important contribution to the metaphysics
of evolution in his explication of "replicators". Like
Monsieur Jourdain, who was astonished to discover that he had been
speaking prose all his life, Dawkins may well be equally surprised
to discover that he has committed an act of metaphysics.’
7 S. J. Gould, ‘Caring Groups and Selfish Genes’,
Natural History 86 (December 1977). Gould is a well-known
palaeontologist who would probably be surprised at Midgley’s description
of him as ‘a geneticist’ (Beast and Man, 66). Midgley, in
turn, might be surprised at some of the things Gould has written
elsewhere, for instance: ‘Natural selection dictates that organisms
act in their own self-interest. They know nothing of such abstract
concepts as "the good of the species". They "struggle"
continuously to increase the representation of their genes at the
expense of their fellows. And that, for all its baldness, is all
there is to it; we have discovered no higher principle in nature’
(S. J. Gould, Ever Since Darwin (London: Burnett Books, 1978),
261).
8 In Hamilton’s opinion too, as is clear from his
reviews of both our books (and by the way, nobody in the world is
better qualified to review either of them): W. D. Hamilton, review
of The Selfish Gene (Science 196, 1977, 757-759); W. D. Hamilton,
review of Sociobiology (Journal of Animal Ecology 46, 1977,
975-983).
9 Some of the more constructive arguments in this
paper are developed further in my forthcoming book, The Extended
Phenotype (Oxford: W. H. Freeman & Co., 1982).
Home
| Articles | Philosophy
| Think |
Supplements | Events | Centres
| Links | Join
| About | Contact
Webmaster:
Dr Jeremy Stangroom
|