Inkwell: Authors and Artists

    
spineless? :-)

I don't think the distinction should be pushed too hard. It was a way
to get into the arguments about sociobiology and to make them more
comprehensible. It's not a grand classification of all the thinkers
that have ever been.
  

    

Ok, I confess, I'm minimally clued about this stuff, though generally
interested in evolution and science.  Having read neither of these thinkers
in the original, I was sorta hoping for a thumbnail description of the work
each has done, and of their divergences, so there's something to hurl those
funny metaphors about metaphors at.  Then we can tell what sticks, and
what slithers down.  Anybody willing to do or point to a very simple
summary?
  

    <hidden>
  

    
<53> is hidden to avoid deflecting the topic back into the discussion of
memes.
  

    
"What sticks and what slithers down" is nice.
Here's the thumbnail. Evolutionary biology is a huge subject. There
are two ways to approach it: to find out what actually happens, and to
find the mathematical rules that make everything happen. Obviously, the
ultimate theory encompasses both but, like I said, it's a huge subject
and everyone has to start somewhere.

Dawkins and his friends start from the mathematics. They begin with
population genetics -- the study of how gene frequencies alter in ideal
populations -- and from game theory. This approach made a huge strides
in the 1960s due to the mathematical advances made by W.H. (Bill)
Hamilton, who is now professor of zoology at Oxford, and an American
journalist named George Price, who was so horrified by Hamilton's
discoveries that he checked them, reformulated them even more clearly,
had some sort of a breakdown, and killed himself. The work of Hamilton
and Price is part of what Dawkins popularised in the Selfish Gene.
Specifically, they showed, mathematically, how a gene for "altruistic"
behaviour could spread through a population at the expense of a gene
for "selfish" behaviour. This is one form 

There were two further strands in the Selfish Gene. The first was
ethology, which is based on the insight that animal behaviour is as
much shaped by natural selection as animal bodies. It can be analysed
in terms of the advantage it gives to animal genes. When this snalysis
is combined with game theory (Price, again worked on this, with John
Maynard Smith in Britain. Robert Trivers in the US) and the whole lot
worked out using computers, you can get very sophisticated and
wonderful models of gene flow and animal behaviour. 

But all these theories take for granted what "genes" actually are.
They are dependent on a definition of "gene" that has nothing to do
with DNA. A gene, for Dawkins, is essentially a unit of heritable
material which could just as well be made of green jello as DNA
provided green jellos had the right characteristics. Nor is he very
interested in the processes by which genes come to have effects on the
outside world. It is enough for his theories that they should do so.
The blanks can be filled in later. 

This kind of bold analysis has been enormously successful. But it is
possible to carry it too far. The point is that "gene" defined as
"something hereditary that makes a difference" is all you need for
mathematical analysis. But it may be a completely misleading way to
thing of "gene" as "particular bit of DNA that makes a particular
difference in the world". For when you get down to the level of
molecular analysis a single gene consist of lots of discontiguous bits
of DNA and the sort of traits we're interested in at a higher level can
be produced by the interaction of large numbers of genes. 

These are the sort of objections that Gould and his friends made (cont
next message)


 reason for Price's breakdown, as I understand it, was that 
  

    
(this is going to be more of a hangnail than a thumbnail. Still, the
two sides agree about so much that it is impossible to be clear, fair,
and short).

Gould started off as a paleontologist (Dawkins was an ethologist,
under Nico Tinbergen). So, broadly speaking, he starts from practice
and moves towards theory, rather than the other way round. To the
theoreticions like Maynard Smith, he never gets there ("Gould is a man
whose ideas are too confused to be worth bothering with"). 

His original scientific theory was "Punctuated Equilibrium", which is
normally understood as a thoery about how species emerge. I think it
actuallya theory about why they persist. The fossil record shows that
speciation can happen very quickly in geological time. In a way it is
more puzzling that some things have persisted for hundreds of millions
of years than that other species have been really short-lived. So
Gould, with Niles Eldredge, proposed that the natural condition of
species was to persist, and that speciation happened normally in a
quick burst. Like thinking about genes instead of organisms, this was a
way of reinterpreting existing data. Like that, too, it was widely
misunderstood. The bastard child of Gould is the X files, just as the
bastard child of Dawkins is having "genes for" liking Elizabethan lute
music.

Gould, Eldredge, and Lewontin, all maintained that if the simple
mathematical equations of the populaiton biologists were right, you
would expect the natural world to be in a state of constant flux, and
species to be constantly changing. Dawkins and his friends replied that
they were not nearly so simple-minded as they were accused of being,
and that even if punk eek were true, it proved nothing. Creationists
announced that the whole thing proved that evolution was a mess.  

Against the Creationists the Dawkinsians maintain that natural
selection is necessary and sufficient to explain all the apparently
designed complexity in the world. the Gouldians agree with this. But
they thing there is a hell of a lot more undesigned complexity around
than the Dawkinsians are willing to exist. 

They also argue for technical and philosphical reasons, that you
cannot read off actual human behaviour and psychology from the
mathematics describing the behaviour of idealised genes. You will come
to an especially nasty end if you try to do this by using the language
of "selfishness". There's some dispute about whether Dawkins actually
meant to do that in the Selfish Gene. I think he did it, but without
intending to.

Our own dear <sumac> has a very elegant example of the Gouldian attack
on "selfishness" in Salon at the moment.
  

    
When will *The Darwin Wars: How Stupid Genes Became Selfish *The Darwin
Gods* come out in the States?
  

    
Gail, thanks so much for asking (in #52) the question I've been meaning to.
And Andrew, thanks for the explanation.
  

    
I hope it made sense. the trouble is that by the time you have written
an entire book on the subject, you can't see the trees for the
brambles. 

As for when the book comes out in the states -- when I find a
publisher. I don't know. The curious can always order it form
amazon.co.uk
  

    
I like that summary.
  

    
Me too.

I have just read a book by Lee Dugatkin, in which he says that
these arguments map to Locke & Hobbes, to liberal & conservative.
He defines liberal as the belief that people are fundamentally good
and conservative as the belief that people are neither good nor
bad, but learn to be these things (and this is where he locates
himself).  He doesn't even address the belief that people are
fundamentally bad/selfish.
  

    
The mapping to Hobbes is spot on.

I'm less sure about Locke and the Gouldians, though there is certainly a
lot there. I did press Hamillton about all this the first time I
interviewed him: he would go no further than that sociobiology proved that
communism could never have worked. In 1995 this was not a sensational
insight. But it may be as far as you can go. Mary Midgley would say that
sociobiology also proves that completely atomistic individualism can't
work either. I don't think that's disputed, though it is a matter of
emphasis.
  

    
Yes, the matter of emphasis is so important.  It's as if people were
at daggers drawn over whether the glass is half full or half empty.

In the nature-nurture argument, for example, I think just about
everyone now thinks that humans are produced by some of each, but
just as everyone is shaking hands and agreeing, one party says, "Of
course nature and nurture are both involved -- but nature is more
important" and the other party shrieks with rage "Nurture is more
important!" and the common ground is lost sight of.
  

    
Yet in another perspective, differences of emphasis are _tremendously_
important. Like differences in style, they go to the heart of our
individuality. We don't expect to find them among scientists, because
science aspires to a god-like impartiality that transcends individuals.
And in a sense it gets there. But as soon as you leave the equations
behind, you're back with emphasis and style.
  

    
Except that paleontology never really gets to the equations. It's a science
without experiments in the physicist's sense.
  

    
If it isn't quantified, is it science?

Conversely, do the numbers sometimes get in the way of 
what they originally described?
  

    
The heart of science is the self-corrective process of replicating research,
not numbers in and of themselves.  It seems to me that many of the people I
meet in the process of doing professional skepticism imagine that science is
about taking the mystery out of everything, or reducing humans to
predictable globs of program code.  And yet that would be much more true if
astrology worked, since saying that you could actually work out a person's
life and personality from a chart of the positions of the planets when they
were born is much more prescriptive and lacking in mystery than any
understanding science has to offer of how humans work.

the Skeptic is about to publish an article by Rupert Sheldrake claiming that
physicists and other hard scientists should be required to conduct
experiments to see if double-blind testing would be important in their field
as they claim it is in testing for ESP (or as is required for medical
testing.

wg
  

    
I've heard it said that the language of science is mathematics, but that's
more true of some sciences than of others, and I don't buy it as a
generality.  Mathematics is a set of useful tools, but the language of
science is reason.

If Gould never arrives at theory, as Andrew suggested some from the other
camp have charged, maybe it's because he's practicing a descriptive brand
of science, long on detail and short on reducibility.  He's proposed a
theory which seems to explain his data.  That his theory isn't founded in
the manipulation of constructs (idealized genes) matters little or not at
all.

The formal theorists have done such a fine job of illuminating the
behavior of a gradually evolving system composed of specifics-shorn gene
constructs, that we can compare their conclusions with observables and
conclude that the material world is otherwise.

Would that be a fair summation?
  

    
I think that's fair. Mind you, I don't think that the observed
features suggesting punk eek contradict neoDarwinian genetics, and
neither does anyone else. But they suggest some of the layers and
layers of complexity and constraints which hold species in existence. 

Wendy's point about astrology is absolutely right, and goes to the
heart of why so many technically educated people are fundamentalists.
It's precisely because they expect science _should_ work like
astrology. 

ONe of the important questions to ask the more simple-minded
science-vs-religion types is just how much of a scientific education is
needed to innoculate you against fundamentalism: three years? Five
years? a degree? a doctorate? Then ask how many of the people who made
the "Islamic bomb" in Pakistan had doctorates in physics, the hardest
science of all. 

If anything, I think a scientific education makes you more vulnerable
to fundamentalism, and to the belief that there is a user's manual for
the world, if only we can track down the manufacturer. 

There are historical reasons for this. The confidence that such a
manual existed, derived from theology, was one of the great stimuli to
the rise of Western science. There's an important sense in which, if
the inquisition had not existed, no one would have taken any notice of
Galileo. 
  

    
I should mention at this point that I am about to disappear for a
week: It's half-term in the mayfly season, so any parent who takes
their responsibilities seriously must teach their daughter to fly fish.

I had hoped this would coincide with an outburst of Islamic hostility
towards me for a story I am doing about the Koran. But it's publication
has been postponed so I'll be back in England in time to  be burned in
effigy in person, though I hope this won't prove necessary.
  

    
I hope you have a great fishing trip, Andrew: the stars look good for
that.
  

    
Go Fish, Andrew.


re: science. I think of science as a methodology for processing
inquiries. 
  

    
Go Fish, indeed.

And for your return, I invoke the name of Thomas Kuhn.
  

    
Hello, coming late to this conversation, on the suggestion of (jmara).  I
have read most of the foregoing and I just would first like to say that I
find Andrew's observations on the sociobiology debate quite insightful and
interesting.  I would like to add a few small comments of my own.  Note 
that I am not a biologist but I have had a keen interest in this and other
closely related fields for many years.

I have to say that I find myself neither in the Dawkins nor the Gould camp,
exclusively.  The main problem I have with Dawkins is that he appears to
ignore the complexity which surrounds what one might call a transition between
different levels of organization; what Gregory Bateson (a man whose work is
often ignored in these sorts of debates but who should really be required
reading for anyone in neuroscience, genetics, or cognitive science) would
call logical levels (he clearly borrows this from Russell though really it
is more a metaphorical use of the term as Russell used it).  Clearly genes can
and do operate in a relatively clean mathematical fashion at the gene level of
organization, and there are feedback loops which cross every level of
organization from the biochemical all the way up to the scale of individual
organisms and, further, to the level of societies and the global ecosystem as
a whole.  However, the transition between even two "adjacent" levels of
organization can present many surprises; if this were not the case, it would
be possible to write a mathematical equation that could predict human behavior
from first principles.

I am reminded of an example I like to use, from a conference on computational
neurobiology that I attended several years ago.  One researcher had built
a mathematical model of the nervous system of a very simple worm which has
a C-bending reflex; essentially a reflex which causes the worm to bend
away in a C shape when stimulated on its side.  What is interesting is that he
measured the connection strengths between the nerve cells and the muscles in a
live worm and found many "non-intuitive" connections; for example, a sensory
neuron on the right side would be expected to contract the muscles on the
right side, and relax the muscles on the left; however, they found that
in many cases, it was precisely the opposite.  There were a mix of
connections which, as a whole, created a smooth C-bending reflex movement;
but taken in isolation, the connections were seemingly random.

What's more interesting is that they created a mathematical model of the
neurons and their connections and they used a statistical training algorithm
to get the model to produce precisely the same C-bending reflex outputs that
the biological network produced.  When they went in to look at the trained
connection strengths, they got the same pattern: sometimes the connection
was what you would "intuitively" expect; sometimes the opposite.

Now, I am certainly not claiming that Dawkins would find this result
surprising; in fact, just the opposite, he also argued that altruistic
genes could thrive even in a purely Darwinian context.  However, my point is
that once you pass even one level of complexity, the potential for massive
computational complexity that can transpire in that level transition makes
it very difficult to ascribe direct causal relationships between features
found at one level (a neuron connection in this case) and another level (the
overall bending reflex).  Sometimes, clearly, one can do this, but it is just
as likely that you have to consider extremely complex interactions between
thousands of genes in order to predict their outcome.  Once this becomes
a possibility level-crossing associations become much more suspect.

On the other hand, I would have to say that I agree with the general
premise that we have, in some sense, no self; this is not to say that we
do not appear to have a self, nor is it to say that we do not act as though
we have a self.  But I would argue that the "Buddhist" view that strictly
speaking we have no self, is essentially correct.  However, because of my
general distrust of assertions which cross levels, this does not lead me to
throw away all large-scale metaphor as meaningless.  Precisely the
opposite: notions of personal responsibility, individuality, and so forth, are
not meaningless because these notions live not at the fundamental level of
analysis in which it becomes possible to say we have no self, but rather they
are abstract approximations which refer to phenomena at the same level of
organization as large-scale fuzzy concepts like "self"; so to the extent that
it is meaningful to use the word "self" to refer to anything at all, it is
meaningful to use those terms as well.

Metaphor is very important precisely because of the computational barrier
which exists at every organizational level-crossing.  In some important sense
we cannot extend a reductionistic description arbitrarily to larger and
larger scales, not because there is some magical "self" that "really exists"
at the higher levels, but because the complexity introduced is far too great
to allow such a facile overuse of an otherwise credible reductionist theory.
This is in fact a matter of principle; one cannot even in principle make
some computations, and thus one cannot have a perfect reductionist description
which operates at all levels of description.  At best one can hope for
a perfect description at *one* level of description, with metaphor filling
in approximately everywhere else.  This is not to say reductionism cannot
poke through the levels; of course it can in many isolated instances.  But
not everywhere, not in general, not completely.  While I am certain Dawkins
et al would agree with this, I think they have certainly far less appreciation
of how fundamental and formidable a barrier computational complexity really
is.

Apologies for the length.  I look forward to Andrew's return.
  

    
Applicable term: emergent properties.