Behold the Talking Chimp

SMALL HINTS, LARGE CHANGES:Courtesy of Rick Effland; Design, Erica P. JohnsonAlthough the genetic differences are small, as illustrated in the above stretch of FOXP2, and the neural differences still largely unknown, there is a world of difference between the mind of a chimp and the mind of a human.From our common ancestor with chimpanzees, it took only six million years, give or take, to develop the ability to speak. And, as we now know, the vast majority of our genetic material has been inheri

Gary Marcus
Jun 6, 2004

Courtesy of Rick Effland; Design, Erica P. Johnson

Although the genetic differences are small, as illustrated in the above stretch of FOXP2, and the neural differences still largely unknown, there is a world of difference between the mind of a chimp and the mind of a human.

From our common ancestor with chimpanzees, it took only six million years, give or take, to develop the ability to speak. And, as we now know, the vast majority of our genetic material has been inherited unchanged. Language, and whatever else separates us from chimpanzees, has its origins in alterations to no more than about 1.5% of the nucleotides in the genome,1 a pretty neat trick, when you consider how handy talking can be.

How did evolution pull it off? Some important clues have already come in, such as a recent study showing that there has been an...


At the moment, the most obvious obstacle is technological. Although scientists have become quite proficient in working with genes and in studying the brain, we still have a long way to go. We can measure the activities of genes by the thousands, but when it comes to neurons, we're still limited to measuring a handful at a time (with so-called single-cell recording techniques) or measuring the gross activity of hundreds of thousands of neurons that make up a "voxel" in a typical magnetic resonance imaging (MRI) scan. Whatever makes the difference between human and chimp may fall between these two extremes. The individual neurons of a human brain aren't likely to be much different from the individual neurons of a chimpanzee brain, yet thus far MRI-based comparisons of human and chimp have turned up relatively little. Instead, the most important differences may lie at a "circuit" level that we do not yet have tools to observe.

But even if we had the right tools for measuring the brain, we still might not know what to look for. Genetics has advanced as much as it has in the last half-century in no small part because Mendel was able to characterize abstractly what molecular biologists should look for. Alas, neuroscience may not yet have had its Mendel. Although we know that the brain is an information processor, much remains to be discovered about the nature of that information and how it is processed. And it is hard to see how we can fully understand which genetic changes were crucial for the evolution of language until we have a firmer grasp on what human-specific neural changes were crucial for the language's evolution.


Courtesy of Athena Vouloumanos

Still, with all these caveats in place, I think there is great reason for optimism. It has long been clear that resolving the mystery of language will require cooperation among researchers steeped in a broad range of disciplines, from genetics and neuroscience to psychology, anthropology, and linguistics. Now, for the first time, there are real signs of that happening. Brain imagers are beginning to collaborate with professional linguists, psychologists are collaborating with geneticists, and so forth. Such studies have already overturned the simplistic story found in many textbooks (Broca's Area for grammar, Wernicke's Area for meaning) and stimulated a great deal of new research, aimed at understanding language as the product of a far more complex network that spans the brain.

Although it is too early to tell for sure what we will discover, my research into the interface between language, genetics, and neurosciences leads me to predict that this flurry of activity will converge on two fundamental truths.


First, the biggest differences may stem from remarkably subtle changes. For instance, a big part of what underlies language is the ability to represent what linguists call hierarchical structure, small units coming together to form larger units that in turn form still larger units: an article (the) and a noun (girl) coming together to form a noun phrase; a verb (loves) and the noun phrase coming together to form a verb phrase (loves the girl); another noun phrase combining with the verb phrase to form a sentence (The sailor loves the girl); and that sentence combining with other bits to form still further sentences (Mr. Lynch knows that the sailor loves the girl).

In a computer, such an ability can be added quite cheaply. As artificial intelligence pioneer John McCarthy showed in 1956, when he invented the computer programming language LISP, all that is required is a new convention for organizing things in memory, not any new hardware. Analogous changes to the organization of human memory could, all by themselves, have profoundly altered the nature of human communication systems.45

Of course, it is unlikely that language emerged in a single evolutionary step. Other crucial innovations may have included a new system for rapidly acquiring names for things, ultimately generalized into a system that could acquire words for actions, abstract ideas, and the like, and modifications to nonlinguistic cognitive systems, such as our systems for understanding the goals and actions of other people.


Second, although human language is qualitatively different from any other communication system found on the planet, I suspect that we will discover that the neural circuitry of language is, to a remarkably large extent, built out of preexisting parts. The French Nobel laureate François Jacob once described evolution as a tinkerer, who makes use of whatever available bits and pieces might already be around. This is obviously true when one compares, say, the wing of a bird, the flipper of a dolphin, and the arm of a human being: Adaptations are often variations on a theme. My prediction is that as we come to understand the neural and cognitive basis of language, we will see that the same strategy of reusing preexisting parts applies in mental systems.

For example, Broca's Area, a region of the brain that has long been thought to play an essential role in language, may have evolved from neural areas in ancestral species that were used for temporal sequencing or motor control.6 Similarly, FOXP2, the only gene that has thus far been definitely tied to language,7 is not unique to humans. It is found in other species, from songbirds and mice to chimpanzees, and appears to be expressed in similar parts of neural tissue.8 Yet the human and chimpanzee versions of FOXP2 differ by just two amino acid positions, underscoring the extent to which the genes involved in language are likely to be tinkerings with preexisting machinery rather than wholesale innovations.

If there is a lesson to be learned, it is this: If we are ever to understand ourselves fully, we will first need to understand from whence we came.

Gary Marcus, associate professor of psychology at New York University, studies the roots of human language acquisition and builds computer models of the development of brain and cognition. He is author of The Birth of the Mind: How A Tiny Number of Genes Creates the Complexities of Human Thought (Basic Books, 2004).

He can be contacted at

This article was written with the support of the NIH, the Human Frontier Science Program, and the able editorial assistance of Athena Vouloumanos.

Interested in reading more?

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?