Funded Grants

The evolution of language: Perceptual and computational constraints

The popular writer John Horgan famously declared a few years ago that science had come to an end. There are no more hard solvable problems left. I am pleased to report that after combing through Horgan's treatise, there is at least one scientific problem for which he provides no satisfying insights. It is both hard and solvable: the mystery of language evolution. This problem is hard because it demands an understanding of how brain, mind, and behavior evolved in the service of building a complex system of internal computation and external expression in the form of sounds that carry ideas. Due to advances over the past few, years, this problem is also solvable, but we need a new perspective.

What is the problem of language evolution? As Darwin articulated over 100 years ago, to address problems of evolutionary origins we must explore the design features of the trait in question, the extent to which they are shared with other species, and the selection pressures that led to the origins expression of the trait. Language evolution poses a problem on several levels. First, our direct ancestors are no longer around, and the fossil record is mute with respect to many interesting aspects of language including its neural architecture, the interface between computation and perception, and the role of peripheral organs in carrying inner thought to outward expression. Second, studies of nonhuman animal communication provide few strong cases of homology or analogy with language, thereby suggesting that the interesting evolutionary events occurred after the divergence point with our last common ancestor, some 5-6 million years, ago. Third, studies involving the use of signed or artificial languages by human-reared apes have met with mixed success, especially with respect to the infinite expressive power of human language; whereas apes can learn 400 or so symbols, this pales in relationship to the 60,000 words that a typical high school graduate knows. This claim is not meant to diminish what apes have learned about such symbolic systems, but rather, to point out that the mechanism underlying symbol acquisition and use is likely to be quite different. Fourth, many if not most researchers focusing on this problem have centered their efforts on the communicative aspects of language, as opposed to the computational and perceptual mechanisms that it relies on, and that may have evolved for reasons other than communication. The aim of this proposal is to explore to what extent human language relies on uniquely evolved specializations, focusing on three components: neurobiology, computation, and perception. Specifically, we ask whether each of these three components evolved uniquely for language, or whether they evolved for more general functions prior to the emergence of language. We argue that only comparative studies of humans and animals can address this problem.

Why is the time ripe for such an approach? Over the past five years, my lab, in collaboration with linguists, cognitive psychologists, and neuroscientists, has developed new techniques that allow us to address these age old problems from a different perspective. Whereas dozens of studies have explored similarities and differences between humans and animals with respect to language processing, almost all have been based on techniques that involve extensive training. Our approach, in contrast, involves no training and directly parallels studies that have been conducted with human infants. As a result, we are not only capable of looking at the kinds of mechanisms that are spontaneously recruited, but in addition, can more directly compare the results obtained with human infants.

Our research program has been designed to test the following hypothesis: The language faculty relies on a suite of uniquely human computational mechanisms that interfere with evolutionarily ancient, and non-language-specific perceptual mechanisms. Consequently, the goal of research on language evolution is to uncover the computational mechanisms that animals lack, and the extent to which the constraints on language acquisition (over evolution and in development) rely on the emergence of new computational resources as well as their interface with non-language-specific mechanisms. This research program relies an three different techniques. Behaviorally, we implement habituation-discrimination methods that are the mainstay of studies of human language acquisition research; these techniques work beautifully with our target monkey species: tamarins and marmosets. Neurobiologically, we will explore the neural substrates involved during perceptual and computational tasks using fMRI with an awake monkey preparation. All of the tasks involve passive listening, and previous work with this technique has yielded highly reliable results with respect to passive olfactory discrimination. Lastly, we explore the role of experience by taking advantage of the breeding biology of our subjects: both species naturally give birth to twins (dizygotic). Our goal is to raise the twins in different auditory environments to understand how linguistic experience shapes perceptual distinctions. Theoretically, we use these three approaches to further explore the claim that perceptual mechanisms are shared across human and nonhuman primates, relying on general auditory mechanisms as opposed to mechanisms that evolved uniquely for speech perception. Similarly, we build on previous results suggesting that the kinds of statistical learning mechanisms that humans implement during language processing are highly limited in nonhuman primates, thereby revealing why animals have not evolved a system of communication like ours.

Descartes famously proclaimed that our existence as individuals depends upon our ability to cogitate, reflect, and ponder. Language provides us with a medium for thinking about who we are, and for sharing our thoughts with others. When Gregor Samsa was transformed into a beetle in Kafka's Metamorphosis, he certainly lost the expression of language, but neither his sense of self nor his internal language. By understanding how language evolved, we will illuminate one of the most fascinating historical passages of all, from a true beetle without a sense of self to a monkey or ape with a rich mental life, but little to no capacity to express it, to a human adult who can translate the inner language of the mind into an external expression, whether it is about a baseball score, a romantic moment, or the latest findings in science.