Language ability as exaptation
An interesting article in Babel's Dawn: Building a New Brain from Old Parts; I retake it here. The article argues a gradualistic and selective emergence of language (something I agree with) and dismisses the role of Gouldian exaptation and spandrels in the process—which I don't quite agree with. See my comment below, and see the original article for additional links. The question of exaptation and of unforeseeable and catastrophic turns in evolution is a key one in Gould's evolutionary theory, and I highly appreciate Gould's awareness of the role of retrospection and hindsight in evolutionary theory and his ability to discern the potential for narrative fallacies in our narratives of evolution. Here goes then Bolles's paper which I would argue is fully anti-Chomskian but only half anti-Gouldian:
Edmund Blair Bolles (Babel's Dawn): Building a New Brain from Old Parts:
Spandrels are famous as inevitable byproducts of arches and as metaphors of evolutionary theory; although sometimes spandrels can leave the scholar with more to say than just, "It's a spandrel."
If language is unique to humans, there must be something unique about our brains. But how does a brain acquire something unique? That question is one of the real mysteries of speech origins and only now is a satisfactory solution beginning to take shape. Unsatisfactory answers have been in rich supply, so it is encouraging to see beyond them at last.
One idea has been that language is the natural byproduct of having a very big brain. Stephen Jay Gould was a proponent of this idea, calling such byproducts 'spandrels.' It was a metaphor taken from architecture. If you put an arched doorway into a building, you will get some curved wall space. The spandrel comes with the territory and is to be explained by referring to the arched door, not to any decorative details the builder may have added to the spandrel. The spandrel explanation laughs at the mysteries of language. Language is as it is because the language spandrel is what it is. The mysteries endure, hope of understanding dissolves.
A second idea is that a sudden mutation gave us a language faculty. The most popular lone-mutation hypothesis has been that in a leap we were given the capacity to manipulate symbols recursively. After March's conference in Barcelona the recursion idea seems mortally wounded (see: Recursion Can Be A 'Side Effect' and Words are More Human than Syntax), but the search for a single mutation that delivers us to language goes on.
These solutions have been tried because the standard evolutionary appeal to gradual change seems on its face unable to account for something as unique as language. Nevertheless, this long-dismissed account appears to be correct. The March and April issues of Nature Neuroscience included two papers and a news story about brain evolution. The news account concluded, "[the papers] suggest that the neural circuitry in humans evolved gradually from primate precursors.... The human language circuits did not appear de novo through a chance mutation or as a 'spandrel' of increased brain size... but instead have their basis in modified versions of neural structures shared by related species."
The articles in Nature Neuroscience were by James K. Rilling et al., “The evolution of arcurate fasciculus revealed with comparative DII” (abstract here), Christopher I. Petkov et al., “A voice region in the monkey brain,” (abstract here), and Asif A Ghazanfar , “Language evolution: neural differences that make a difference” (report here; see p. 384 for quotation cited above).
The papers examined the evolution of specialized areas and the evolution of connections that integrate specialized areas into complex functions. A familiar specialization in speech, for example, is voice recognition. Besides knowing what a person is saying; we can tell who is saying it. Petkov’s team found that monkeys (macaques) have a specialized area of their brain devoted to identifying the voices of other macaques, just as humans have a specialized area for identifying human voices. Thus, voice recognition seems to be older than language, but the macaque’s system is in a different region of the brain from the human system. Voice recognition does not seem to be a spandrel, because it has emerged in different anatomical contexts, nor does it look like the product of a single mutation. Instead, it appears to be a complex function with a long, evolutionary history.
The same holds true for the Rilling team’s examination of the emergence of a connection that supports language. Its paper examined a tract that links the front and rear portions of the brain, the “arcuate fasciculus,” best known for connecting the two most celebrated language regions in the brain, Broca’s Area and Wernicke’s Area. Despite its importance to speech, Rilling's team identified homologs of the arcuate fasciculus in macaques and chimpanzees. The macaque pathway is simpler and connects fewer points than the chimpanzee's, and the chimpanzee pathway is simpler and connects fewer points than the human's. Thus the pathway has exactly the kind of gradual history expected by the most orthodox evolutionists.
But if spandrels and lone leaps are out, what of the objections to gradualism that inspired those explanations in the first place? The most persistent objection is that we do not see little bits of language in other species. Macaques and chimpanzees do not talk at all. Something unique must have happened.
A theme that has appeared a few times on this blog concerns the integration of existing brain function. Last February work on autism suggested that joint attention arises from the union of two distinct forms of attention, one reflexive and one intentional. (See: How the Brain Supports Conversation) The findings reported by Rilling’s team indicate that integration of function has been evolving along one pathway for a very long time. Friedmann Pulvermüller’s report in Barcelona (see: Brain Circuitry Challenges Linguistic Models) also paid close attention to integration of function.
The importance of integration can be seen just by observing what happens when the arcuate fasciciculus is damaged. The separate language areas continue working properly, but coordinating the functions becomes a challenge. Speakers are able to understand what they hear, but unable to repeat what was said. So you can say, “Pass the salt,” and the listener can pass the salt, but the listener cannot say, “You said ‘pass the salt.’” Damage to this pathway also gives readers a problem reading aloud, and when sufferers misspeak they can recognize the error but have a difficult time correcting it.
Gradualism and integration suggests that we will understand the neurological role in speech origins by looking for the enhancement and coordination of previously existing functions. Last week's post on tool making shows what I mean (see: Toolmaking and Speech) We began by using existing sensorimotor skills and then coordinated them in new ways and complicated them. Emerging from that process came something quite new: interest in neutral topics.
I like this idea. Here is a metaphor: Early people had fire and they made it with a little bow. They also had darts or spears. They created a new weapon,the bow and arrow, that was very complicated compared to a spear. But they did not create it from scratch. They made the bow bigger and the spear smaller and then tweaked them a bit. All the stages of creating the bow and arrow were useful; the steps were each short etc. This is good evolution narrative with no big jumps and no useless bits being kept around too long.
I also like the idea of the tool making skill at the start of language because it is tracing back to a starting point in the 'hand' rather than the 'voice'. (That is just my bias showing.)
Posted by: JanetK | June 16, 2008 at 04:19 AM
(and my own comment: ) Hm, I'm not sure that spandrels will go away as easily as that. When you say at the end of your paper "Emerging from that process came something quite new: interest in neutral topics" - isn't there a whiff of the spandrel there? Discussion of spandrels should be complemented with a discussion of exaptation (also Gould's term) and surely there must be a role for exaptation when a detour through something "not crucial" (i.e. neutral topics) actually becomes a shortcut due to the exponential increase in coordinated action enabled by language - something which comes as an "unintended" consequence of these developments, but certainly feeds back on them and enables further evolution along that line. I don't think gradualism and spandrels are as irreconcilable as they seem to be at first sight.
Posted by: JoseAngel | June 16, 2008 at 07:01 AM
BLOGGER: The crucial difference between spandrels and gradualism is selection. With spandrels it is the door that is selected, not the spandrel. With gradualism, you get what you select.
—yeah, you do get what you select, but the point is that you get many things you don't select as well—you get spandrels. And exaptative uses both of spandrels and of selected traits. And they do have effects... such as ideology's feedback on evolution, allowed by language. (JAGL).