Advancing Basic Science for Humanity
Babies, Birds and Words
Human infants learn to talk and starlings learn to sing, but just how different are the two?
The answer on one level is obvious. Songbirds never learn to talk, at least not in the way even poorly educated humans can. Nor do other animals, such as monkeys and dolphins, that appear bright and expressive but never seem able to make up more than the simplest of sentences, if even that. In its complexity and power to convey ideas and emotions, the human gift of language does seem unique in all nature.
But deep down, at its roots in the cells and circuitry of the brain, that gift may not be exclusively human after all. It’s superior, yes, but is it truly one-of-a-kind? This is a much-disputed question among scientists, and it’s central to the research mission of the Kavli Institute for Brain and Mind (KIBM) at the University of California San Diego.
The Mystery of Human Language
KIBM seeks to explain the links between the brain and all the experiences and behaviors we collectively label the “mind.” Of all human traits, language may be the most mysterious -- and the most difficult to link to traits in other species. But KIBM scientists such as Jeff Elman, the Institute’s co-director, think the animals can tell us quite a lot about how we learn to talk, even if they’re not such great talkers themselves.
Using different “warbles” and “rattles” from the natural song of a male European starling (Left), KIBM researcher Tim Gentner and his coauthors created artificial songs that followed two different patterning rules or grammars. Starlings then were taught to distinguish between sets of grammars and songs using classic reinforcement techniques. They were rewarded with food for pecking at a button (Diagram, Right) when they heard a song from the context-free set and for refraining when they heard one from the finite-state set. Starling photo courtesy of Daniel Balekaitis. Apparatus diagram courtesy of T.Q. Gentner.
Elman, a UCSD professor of Cognitive Science, believes that the capacity for language evolved in a process that began long before the first recognizably human beings walked the earth. It didn’t “spring de novo in some genetic big bang,” he says. Given that, he says, the brain’s basic linguistic tools should also be present in related species: “One of things we firmly believe is that whatever drives language – the mechanisms that make it possible – have to be present in our close kissing cousins.”
This approach joins neurobiologists, such as KIBM’s other co-director Nicholas Spitzer, and cognitive scientists such as Elman in studies of brain development and learning across species lines and at all phases of life (Spitzer specializes in the electrical activity of the brain in embryos). Their research extends to not-so-close “kissing cousins,” such as songbirds. One notable study at UCSD, published in 2006, focused on the ability of starlings to learn their songs and, significantly, to grasp a key principle of grammar.
A research team led by Assistant Psychology Professor Tim Gentner found that starlings could be trained to recognize sound patterns based on recursion – the embedding of sentences within sentences. Human language uses recursion to fold two or more sentences into one, as in “The man chased the dog who chased the cat” (another way of saying “The man chased the dog” and “The dog chased the cat”). For the starlings, the “sentences” were simple sound sequences, but the birds appeared to know which sequences followed the rules of embedding and which did not. (Gentner is continuing his research into the learning process of birds with the help of a $30,000 Innovative Research Award from KIBM in the 2007-08 academic year).
Elman says this finding was crucial because it countered the hypothesis, put forward by a number of linguists including Noam Chomsky, that recursion is what makes human language unique. If the birds do it (and who knows, maybe the bees too), the divide between human and animal expression is not fundamental, and the differences are matters of degree more than kind. If true, this is good news for researchers, because it enables them to learn a great deal about the human brain and human learning by studying other species.
A functional magnetic resonance image (fMRI) shows different areas of the brain that are active in language acquisition at different ages. Areas colored in blue are the more active regions in 3-year-olds (and adults) as they listen to speech. Areas colored in orange are more active in toddlers (age 1-2).
This evidence for a cross-species common ground also reinforces a particular way of thinking about the brain – as a highly adaptable organ that performs its remarkable feats mainly though learned processes rather than innate structures. If the raw material of the human brain is more or less like that of other animals (with one big difference: There’s far more of it), then the human mastery of language must come mainly from the way in which our brains are trained by experience, starting in the womb.
The brain certainly has the flexibility to be a good student. Imaging shows, for instance, that the regions most associated with language in adults, such as Broca’s area and Wernecke’s area, do not play the most important role in learning of language during infancy. In the first two years, Elman says, the frontal regions of the brain are much more important to this task, perhaps because they are much more developed in humans and because they are involved in social interactions. Researchers have found that damage to the brain to a child before or just after birth does not have the devastating impact it would have on an adult. Elman and his colleagues have been studying such children for about 30 years as they grow into adulthood. By the ages of 6 to 8, he says, children with early brain lesions “function in the normal range of behavior. Some are now in college.”
The brain’s ability to adapt, adjust and even relocate functions such as language makes it more sturdy and reliable than the typical computer, which can be thrown into catastrophic failure by a line of malicious code. In Elman’s view, the brain is more like a vast assembly of parallel computers working in concert. He is widely known as a leader in the development of this “neural network” model in his three decades of research at UCSD and KIBM. As he and other KIBM scientists shed more light on the mind through such studies, they also show the way to knowledge that could be put to use in medicine and education. What they are learning is, in part, reassuring: The brain is resilient. It also carries a warning against neglecting the intellectual development of the very young. “It is becoming clear that the first two years of life are critical,” says Elman, and that depriving them of normal language exposure can do damage that is more lasting than the effect of brain lesions. But there’s a positive side, he adds: “If you can catch the kids early, you can do a lot of good.”
© 2008 The Kavli Foundation