ITALO CALVINO was an acclaimed novelist, essayist and journalist. In the
course of his career (he died in 1985), he moved from militant communism and
neorealism to postmodernism and fantasy. With this profile, why would his work
appeal to the scientifically inclined? Be patient: this intriguing book bridges
the gap between science and the arts brilliantly.

In Why Read the Classics? Calvino’s curiosity leads him on an odyssey
in space as well as time, as he scrutinises texts both literary and nonliterary
in search of works he sees as classic, according to his own quirky criteria. So
to Calvino a book makes the grade if well-read people will admit only to
rereading it rather than encountering it for the first time. Amusingly cynical,
this, but he goes on to say that a classic will also reward every rereading as
if it were a first reading, and that, paradoxically, even a first reading gives
a sense of familiarity yet novelty.

These criteria are also part of a repeating, nested pattern to be found in
the works he discusses. In Galileo’s work he traces the divide between the
pursuit of knowledge through an indirect reading of the book of nature through
reading the books of Aristotle, and a direct reading through empirical inquiry.
Aristotle’s works are classics because they remind us of the world and what we
know of it, and can stir our imagination when we inquire into it, even if they
are not nature itself. Galileo reminds us of this, and his books are classics
too.

Calvino celebrates a wide range of great thinkers in these provocative
essays. Here are writers from the ancient world, the Renaissance and recent
times, and from the old and new worlds. Many, like Galileo, will be familiar
(Dickens, Conrad, Defoe, Homer, Hemingway, Tolstoy, Twain), but some will not
(Raymond Queneau, Cesare Pavese and Eugenio Montale, and Ariosto, and the
mathematician Girolamo Cardano).

Galileo’s books also offer, as do many of the other nonliterary works
examined by Calvino, the view that to read the book of nature we need a
language, and that language is mathematics. His essays reveal a broad
understanding of how the inability of most people to understand mathematics
presents a problem. This is particularly true in the study of human affairs, as
Calvino reveals in his examination of, among others, Gianmaria Ortes, an
18th-century Venetian cleric. In his book A Calculation of the Pleasures and
Pains of Human Life, Ortes offers views on the causes of human conduct not
so distant from those of modern biologists. He also had a wry sense of
self-parody in the recursive application of his own views to his own work. This
and the other essays are a reminder to us that “rereading” the classics can
amuse as well as reward.

Given Eco’s success in writing for a wider public, most notably with The
Name of the Rose and Foucault’s Pendulum, you might expect an
entertaining and informative set of essays. Sadly, no. The individual chapters
are repetitive in argument and content, and examples from the history of
linguistic inquiry are often far too obscure to allow anything more than a
display of the fact that the author knows them. You’re left wondering what all
the fuss is about. The most glaring example of this tendency is that great play
is made of the fact that good ideas and discoveries have follow ed from the
pursuit of quite daft ones.

Overall, the book feels like it’s been thrown together from a motley
collection of lectures and talks. They may have been fine with a dramatic
delivery on an appropriate occasion. It is hard not to think that this
collection would not have made it into print without the author’s substantial
reputation. But the book certainly does him no favours. Any New
ĐÓ°ÉÔ­´´ reader with ÂŁ12.99 to spare and an interest in serendipity
would do much better by spending the money on time connected to
http://serendip.brynmawr.edu/

Goodbye, Descartes by Keith Devlin, Wiley, ÂŁ17.99,
ISBN 0 471 14216 6

MEANING and reason have long been identified as the central features of human
intelligence. But their analysis has always been problematic. Progress towards a
satisfactory understanding of them has been meagre, whether one looks at the
soft underbelly of the human sciences in sociology and anthropology, through the
increasingly firm areas of psychology and linguistics, or even the hard shell of
philosophy, logic and certain branches of mathematics.

Progress has always seemed to have been greatest when reason has been
distilled off from meaning and language through the construction of idealised
models of human rationality, but the success has always been more apparent than
real. The most recent and telling instance of this has been the field of
artificial intelligence (AI). Despite having the powerful medium of the computer
in which to form and explore its ideas, it has singularly failed to deliver on
the prophecies and promises of its founders that machine-based general
intelligence was within our grasp.

The failure of a scientific project can have many causes, but the
mathematician Keith Devlin argues that this failure should give us pause for
thought. He argues that the intellectual tradition in which AI and its
precursors is placed is fundamentally flawed, and that we need new approaches to
meaning and reason in language in which the insights and analyses of the softer
sciences are combined with rigorous mathematical analysis of pattern and system
in a new “soft mathematics”. Devlin’s Goodbye, Descartes is a manifesto
for change to those new ways. What is interesting in his argument is the highly
accessible historical overview he provides of human attempts to characterise,
clarify and extend human reason.

His history begins in ancient Greece with the two schools of logic created by
the Stoics and Aristotle. For Devlin, these schools share one important feature:
both analysed argument structure independently of content. They established the
precedent of separating reason from meaning, and laid the basis for the
difficulties he perceives today. At the time, a link remained between the use of
language in speaking and understanding. But the development of logic over the
next two thousand years by writers and philosophers from Cicero and Apuelius
through William of Ockham to Gottfried Leibniz and Leonhard Euler severed that
link.

A major constraint on the speed of development in this area has been the lack
of a suitable way of symbolising propositions and arguments. The Stoics
presented their abstract formulations in lengthy and confusing sentences in
which the nature and precision of the abstraction is easily lost. Reading
Devlin’s examples of these provokes the thought that doing logic as a Stoic must
have been like doing long division with Roman numerals.

All of this changed with the work of George Boole in the mid-19th century. A
mathematician—today he would probably be considered a cognitive
scientist—he provided a symbolic algebra for representing what he saw as
the laws of thought. This algebra allowed the structure of an argument to be
represented independently of any specific meanings in the thoughts that underlay
it. The result was a golden age in mathematical logic. Problems of meaning were
pushed aside. It was also an era in which the influence of René
Descartes, in promoting the reduction of science to mathematics, was
profound.

These moves have been hugely beneficial for the study of the natural world,
but Devlin argues that their influence on the analysis of mental life has led us
to a dead end. The inheritors of Boole and Descartes, most importantly Noam
Chomsky in linguistics and Alan Turing in computational theory, are doomed to
failure, he says. Meaning and context are crucial to the use of language, and in
the use of language in communication lies the human intellect.

Completing this stage of his argument is not easy, and arguably not possible
given our current state of knowledge. Many telling features of human abilities,
however, support Devlin’s view. Not least among these is the curious fact that
humans are so bad at conforming to the exacting standards of models of sound
reason. This is true of both logic and probability. The success of lotteries is
clear evidence of human inability to handle probability on a grand scale, but
even when the numbers are small we often fail miserably—as demonstrated by
the Monty Hall problem. “This example is notorious,” says Devlin, “for leading
otherwise sane and rational citizens to heights of frenzied argument.”

You are a game show contestant who must choose to open one of three
doors—A, B or C. Behind two lies a banana, behind the third, £10
000. You win what is behind your chosen door. You pick a door, A, but before you
open it, Monty (the host, who knows what is behind each door) opens one of the
two remaining doors (C) to reveal a banana. He offers you the chance to change
your choice from A to B, on payment of £10. Most people refuse—even
though B has twice the chance of having the ÂŁ10 000 pounds behind it. They
find the idea that it would double their chances deeply counterintuitive. Many
find it hard to understand the reasoning even when it is explained.

A central theme in Devlin’s argument is that meaning in language is
impossible to decontextualise, and that the meaningful elements in the language
can subvert completely proposed ideal argument structures. He provides many
examples, and the experience of psychology over the past century provides many
more.

For example, Peter Wason entertained many with his verbal illusion in which
meaning and intent completely subvert apparent formal structure. If we take a
pair of sentences such as “No head injury is too small to be ignored” and “No
missile is too small to be banned”, we find that whichever way we analyse the
structure of one, it provides a counterintuitive result for the structure of the
other. For example, characterising the first sentence as “no head injury is to
be ignored” accords with our intuitions as to the meaning of the first sentence,
but the analogous treatment of the second produces the opposite of what we take
it to mean, that is, “no missile is to be banned”.

Pointing out failings is always easier than making sound proposals for a new
beginning. Devlin’s manifesto is, therefore, unsurprisingly sketchy when it
turns to what should replace the Cartesian tradition. What it does offer is an
interesting and eclectic mix of sociology, linguistics, psychology and
philosophy.

This is the sociology of individual lives and how we understand one another
through what we say and do in particular contexts. He draws on other disciplines
largely where they have moved meaning and context to centre-stage in their
analyses. This work has proved to be of great applied value in the design of all
sorts of knowledge-based devices. As Devlin rightly observes, the work to build
a “soft mathematics” to support it has barely begun.

In presenting all of this, especially the history, Devlin necessarily uses a
broad brush, but he does not sacrifice the significant themes. He writes with
great clarity, and while in parts it is not an easy text, it often deals with
very tough material which he makes easier to understand than many other
writers.

Finally, if anyone is still puzzled by the claim that switching doors in the
Monty Hall problem can double your chance of winning, here is the reasoning.
There is a 1/3 chance of the ÂŁ10 000 being behind door A, and a 2/3 chance
of it being behind B and C together. If you know there is a zero chance of it
being behind C, then there must be a 2/3 chance of the money being behind B.
Thus, switching doubles your chances. If this brief explanation does not
convince you, then buy the book. You will not be disappointed.

To pass on their genes to subsequent generations, there must be some reproductive advantage to make big brains worth the cost. In the branch of the primate family tree that leads to modern humans, this advantage has been such that brain size has more than doubled in the past 2 million years to a volume of some 1500 cubic centimetres. This steady growth in brain size made possible a steady growth in intelligence and an ever-increasing mastery of the world.

But if size mattered, argues Derek Bickerton in Language and Human Behaviour, then whatever happened to the even larger-brained Neanderthals? And why was the gradually increasing brain size of our ancestors not matched by improvements in the tools they used? His answer is that size isn’t everything. What does matter is how the brain is structured. The structural change that made possible human language as we know it was a crucial evolutionary event.

Before this change human groups spoke a protolanguage that linked items in the world to words. But the words could be linked to one another only in simple chains that did not allow the expression of complex propositions – for example, cause-and-effect relationships. Such propositions became possible only when a grammar emerged to organise words into sentences that enabled their representation.

This proposition-making capacity offered huge benefits such as an extensive modelling of the world in “off-line” thinking that was not tied to immediate action. Planning became possible once humans could contemplate the effects of alternatives, picking options most likely to be successful. The analytic, conscious mind was born.

Accounts of the evolution of human intelligence, language and consciousness are breathtaking in scope but hard to support empirically: we cannot test Neanderthal linguistic competence. Consequently, claims often rest on characterising rival theories as implausible, and on suggesting that the favoured theory is inevitable given what we know of modern human linguistic and intellectual capacities.

Bickerton’s account is no exception, but he does draw interestingly on studies of language breakdown, and his own earlier work on pidgins and creoles, to support his principal contention that there was a sharp break between the protolanguage of the early hominids and the language available to the earliest Homo sapiens.

A pidgin occurs when two groups which do not share a common tongue come into contact. In its primal form, it is a collection of words that may be juxtaposed, but which do not create propositions in their juxtaposition. A pidgin is a protolanguage. It is like the utterances of a child in the earliest stages of language development or the sign-sequences of one of the many apes that have been taught a sign language.

A creole, by contrast, develops in the mind of a child learning a language for the first time who is exposed to a pidgin. That child will hypothesise a syntax, then use it to produce novel constructions. The child is midwife and parent at the birth of a language.

Bickerton argues that there is no intermediate form. In the ontogeny of a single language, a creole, we see the phylogeny of human language and intelligence. His claims may not be uncontentious but this book sheds light on a fascinating issue.