Home Page || Publication List || Public Bookmarks || The Calvin Bookshelf (8 August 97)
William H. Calvin
This page is at http://WilliamCalvin.com/1990s/1997DeaconBkRevNYT.htm

copyright ©1997
William H. Calvin and
The New York Times Book Review

The Co-evolution of Language and the Brain

By Terrence W. Deacon.

Illustrated. 527 pp. $29.95 New York: W. W. Norton & Company, 1997


William H. Calvin is a theoretical neurophysiologist at the University of Washington, the author of "How Brains Think" and "The Cerebral Code."

In our evolutionary ascent from an ape-like ancestor, we gained our most prized possession, the mental abilities that underlie language. We're still trying to figure out what language is (from monkey cries to structured syntax), how it works (the short-term processes in the brain that construct and deconstruct utterances), and why it evolved (the Darwinian processes that bootstrapped it over the long run).

That's what Terrence W. Deacon's book, "The Symbolic Species," is about. His first section is on symbols and language, the next tackles the brain's language specializations, and the last addresses the coevolution of language and the human brain, ending up with Darwinian views of consciousness. It's a work of enormous breadth, likely to pleasantly surprise both general readers and experts.

Deacon's first illustration shows four cartoon views of the theories used to explain language: the common-sense view, B. F. Skinner's associations, Noam Chomsky's innate grammatical knowledge, and the "mentalese" instincts that Steven Pinker describes. Though I am agnostic on what terms to use, I have no doubt that human infants come with an enormous "acquisitiveness" for discovering patterns -- and this helps them to zero in on any human dialect.

Newborns detect typical speech sounds; pretty soon, atypical sounds are forced into a standard category that the infant has created. A six-month-old Japanese infant can still detect the English /L/ and /R/ sounds. But after another six months of exposure to the Japanese phoneme that lies between them (and not to English phonemes), the infant standardizes on the Japanese phoneme and hears /L/ and /R/ as the same old thing. In English, infants may need 44 categories; Polynesian is easier, with only 11 phonemes to detect.

And infants start picking up the common bundles of phonemes: throughout the preschool years, they average six new words a day.

Between 18 and 36 months of age, children analyze the longer patterns of words they hear -- and they discover some rules. In English, they'll figure out the add -s rule for plurals and add -ed for past tense. They're soon fitting one sentence inside another: I think I saw him leave to go home has four verbs and "This is the house that Jack built" carries nested embedding to extremes.

Three major steps in the first three years is only warming up. Judging from how demanding they become about bedtime stories, they've detected the far-longer patterns involving many sentences, such as Aristotle's rule that all good stories have a distinct beginning, middle, and end. This pattern-hunger doesn't require even average intelligence, nor is it limited to speech. Deaf children can do the same four levels of pattern discovery by observing body movements -- but only if they are immersed in a fluent sign-language environment, and equally early in the first few years of life.

Nothing like this obsession with extracting hidden patterns is seen in other animals (after a recent lecture, someone asked me what a child's first thought might be about -- and I replied, "Eureka! Another rule!"). Some of the pattern-hungry children even go on to become scientists, trying to guess the larger obscure patterns that tell us how the world works and where we came from.


Many types of scientists have been attracted by the problem of how fancy language got started, how it was fitted into an ape brain. But in our search for pattern, we've often been like the blind men examining the elephant.

Chomsky, for example, is famous for declaring humans have a "language organ" but he doesn't seem to be interested in the evolution, anthropology, or the neurology of it. You'd think it was some sort of supercharger, tacked on to a conventional engine by a hopeful monster mutation -- and that raises skeptical eyebrows. As Deacon says, "innate Universal Grammar is a cure that is more drastic than the disease. It makes sweeping assumptions about brains and evolution that are no less credible than the claim that children are super-intelligent learners."

Since the Chomskians aren't interested in explaining how human language ability originated, many of us assume that any real piecing together of the big story of language evolution is going to come from people acquainted with more than one leg of the elephant, those who can combine the perspectives of the cognitive neurosciences, anthropology, primatology, linguistics, and of developmental and evolutionary biology.

And that describes Terry Deacon pretty well. He's known as a biological anthropologist at Boston University and a neuroscientist at Harvard Medical School, accustomed to grounding his thinking in known anatomy and physiology. As The Symbolic Species proves, he's at home all across the relevant territory. If you've been reading Derek Bickerton, Daniel Dennett, Merlin Donald, Ray Jackendoff, Melvin Konner, Philip Lieberman, Steven Pinker, or Sue Savage-Rumbaugh, then you are surely going to want to read Deacon's lovely book.


Evolutionary explanations for language and brain enlargement are often offered up in an overly simplistic way that gives Darwinian endeavors a bad name. Deacon neatly avoids this by showing how flexibility during life (learning and creativity) eventually helps to reward genetic variations leading in a similar functional direction. This form-follows-function principle has been known for a century but it's still poorly appreciated. Since Deacon's introduction of the Baldwin effect is correct but unintuitive, let me illustrate it here via recipes.

Anyone who has ever asked for a copy of "that wonderful recipe" ought to realize that the donor has long since stopped consulting the recipe card: she just improvises from memory and, over the years, has improved the cake (or whatever) considerably beyond what would result from faithfully following her written recipe.

Still, her point-of-departure version of the recipe is what gets copied, usually with some unintentional mutations (thanks to spots and smears on the old recipe card, you confuse teaspoons with tablespoons, or transcribe the oven temperature incorrectly). Most errors are bad news -- but no one asks for copies of them, only those next-generation recipes whose errors came closer to the master chef's actual practice. This dropout creates a slow convergence in copying errors toward written recipes with a combination of ingredients, amounts, times, temperatures, and assembly procedures that -- with some common-sense tweaking -- will satisfy "good taste."

The Baldwin effect allows unrecorded tweaking from flexible behavior to secondarily drag along relevant genes ("recipe items") in the long run; it's Darwinian but at one remove. Thus relevant gene combinations "fill in" behind the behavioral advance. Deacon makes good use of this behavior-first-anatomy-follows principle in addressing toolmaking by the larger-brained Homo species that evolved from the australopithecines with ape-sized brains: "The Baldwinian perspective suggests... that the first stone tools were manufactured by australopithecines, and that the transition to Homo was in part a consequence rather than the cause.... The large brains, stone tools, reduction in dentition, better opposability of thumb and fingers, and more complete bipedality found in post-australopithecine hominids are the physical echoes of a threshold already crossed [in behavior].... Another way to look at this is to say that many of the physical traits that distinguish modern human bodies and brains were ultimately caused by ideas shared down the generations."


Deacon next discusses what aspects of language could have become established via good old Darwinian means during hominid evolution, after first being carried along culturally. He promptly makes the usual mistake: being overly specific. "If symbolic communication did not arise due to a `hopeful monster' mutation of the brain, it must have been selected for." But selection favoring language need not be via the success of language per se. For example, the brain has circuitry shared between planning complex hand-and-arm movements and complex language. Variants that promote hammering and throwing success may also aid language ability -- and vice versa. This bundling sometimes provides a genuine "free lunch."

Deacon's deep knowledge of anthropology is evident as he analyzes the stage setting of 2.5 million years ago, particularly the reproductive strategies involving hunting and the provisioning of mates and offspring: "Symbolic culture was a response to a reproductive problem that only symbols could solve: the imperative of representing a social contract." This involves exclusive mating arrangements: "Though philandery, cuckoldry, and desertion are common... in other species, adultery is more than this. It involves betrayal, and there can be no betrayal without [implicit promises]."

And it involves "But you owe me!" The considerable benefits of reciprocal altruism require that freeloading be minimized, which requires ways of categorizing debts. As the linguist Derek Bickerton has recently noted, the nonlanguage task of remembering who owes what to whom sets you up for understanding structured sentences. They carry over into linguistic argument structure (those word categories involving actors, recipients, beneficiaries, and so forth), which provide major clues to understanding a story-like sentence about who did what to whom.

In his research lab, Deacon transplants embryonic brain tissue to study evolutionary and developmental brain differences, and tries to develop new cell replacement therapies for brain damage. You can see this perspective in the way he frames sociobiological questions. He avoids those constructs ("Language Acquisition Device, or LAD") so often created by the cognitive cognesetti when trying to define their way into the unknown (in a different context, Deacon quotes Goethe: "When an idea is wanting, a word can always be found to take its place"). This grounding in biological ideas gives Deacon's well-focussed book an entirely different flavor than many language origins books, with a lot to chew on. Some will find this threatening, others refreshing.

PLEASE NOTE: This draft differs in minor ways from the slightly shorter published version. If you plan to quote anything, see the published version at http://www.nytimes.com/books.

My old family recipe teaching example for
the Baldwin Effect is from my 1990 book,
The Ascent of Mind (Bantam, now out of print
in English but just out in German translation).

Home Page || Publication List || Public Bookmarks || The Calvin Bookshelf