It has to be admitted that on the whole we do not understand how the brain works. Nevertheless, some crucial elements seem to emerge. One is that development of the normal brain is enormously plastic, even though the power of genetic factors is obvious. One classic example is that in the same brain areas of identical twins the two hemispheres of the same individual resemble each other more closely than the same hemispheres in the two people (Changeux, 1983).
Another insight is that a tremendous amount of variation and selection is going on during brain ontogenesis. This is a Darwinian-type process, no doubt. As the psychologist William James recognised a long time ago, natural selection of heritable variation is the only known force that can lead to adaptations, so let us apply it to brain ontogenesis and problem-solving as well (James thought that even learning is a result of selection of variation within the brain). There are several expositions that all regard the brain, one way or the other, as a 'Darwin machine' (Calvin and Bickerton, 2000). Here I stick to the formulation by Changeux (1983), because we think this is the most relevant to the language problem. According to this view, the functional microanatomy of the adult cortex is the result of the vastly surplus initial stock of synapses and their selective elimination according to functional criteria (performance).
We have just learnt in Section 3 that a very large part of the human brain can process linguistic information, including syntactical operations. This means that there is no fixed macro-anatomical structure that is exclusively dedicated to language, but some functional micro-anatomical structure must be appropriate, otherwise it could not sustain language. This further suggests that there is some statistical connectivity feature of a large part of the human brain that renders is suitable for linguistic processing. From the selectionist perspective there are three options: the initial variation in synaptic connectivity is novel; the means of selection on functional criteria is novel; or both. Maybe both component processes are different in the relevant human brain areas, and I do not dare to speculate about their relative importance.
This idea must be seen in close connection to the one presented by Rapoport (1990) about the co-evolution of brain and cognition (within a population of humans). The traditional view is the so-called bottom-up mechanism: that a genetic change of some neural structure is subjected to selection and, based on its performance, it either does spread or it does not. There is, however, a so-called top-down mechanism, which could have more significantly contributed to the evolution of human cognitive skills, including, especially we argue, language. The crucial idea is as follows:
• Due to the plasticity in brain development, enhanced demands on a certain brain region lead to less synaptic pruning (a known mechanism).
• Less synaptic pruning is assumed to lead to more elaborate (and more adaptive) performance.
• Any genetic change contributing to the growth of the brain area thus affected will be favoured by natural selection.
There are two important connections that must be pointed out. First (observed by Rapoport himself), the top-down mechanism is a more detailed exposition of the late Allan Wilson's idea (Wyles et al., 1983). Thus an larger brain, due to its more complex performance, alters the selective environment (in social animals composed of conspecifics to a great extent), which selects for an even larger brain, and so on. Second, and perhaps more important, this mechanism is also a neat example of a Baldwin effect (or genetic assimilation), when 'learning guides evolution'. As Deacon (1997) pointed out, it is trickier to apply the idea of genetic assimilation to language than usually thought. The reason for this is that the performed behaviour must be sufficiently long lasting and uniform in the population. It is thus hard to imagine how specific grammatical rules, for example, could have been genetically assimilated. This point is well taken, but here we speak of a different thing: the genetic assimilation of a general processing mechanism that is performed by virtue of the connectivity of the underlying neural structures.
Our claim is that the most important, and largely novel, faculty selected for was the ability of the networks to process syntactical operations on symbols that are part of a semantically interwoven network. The specific hypothesis is that linguistically competent areas of the human brain have a statistical connectivity pattern that renders them especially suitable for syntactical operations. In conclusion, we think:
• The origin of human language required genetic changes in the mechanism of the epigenesis in large parts of the brain.
• This change affected statistical connectivity patterns and dynamical development of the neural networks involved.
• Due to the selectionist plasticity of brain epigenesis, co-evolution of language and the brain resulted in the genetic assimilation of syntactical processing ability as such.
An intriguing possible example of gene-culture co-evolution has recently been raised by Bufill and Carbonell (2004). They call attention to a number of facts. First, human brain size did not increase in the last 150,000 years, and it did even decrease somewhat in the last 35,000 years. Second, a new allele of the gene for apolipoprotein E originated sometime between 220,000 and 150,000 years ago. This allele improves synaptic repair (Teter et al., 2002). The original form entails a greater risk of Alzheimer disease and a more rapid, age-related decline in general (Raber et al., 2000). More importantly, ApoE4 impairs hippocampal plasticity and interferes with environmental stimulation of synaptogenesis and memory in transgenic mice (Levi et al., 2003). Interestingly, the ancestral allele decreases fertility in men (Gerdes et al., 1996). The facts taken together indicate, but do not prove, a role in enhanced synaptogenesis in a period when syntactically complex language is thought to have originated. More evidence like this would be welcome in the future, since one such case can at best be suggestive.
Various people (e.g. Premack, 2004) have called attention to the fact that besides language, efficient teaching (which differs from learning), imitation, and a developed theory of mind are also uniquely human resources. I also stress the trait of human cooperation (Maynard Smith and Szathmary, 1995), which is remarkable because we can cooperate even in large non-kin groups. My proposal is that these traits are not by accident together. They form an adaptive suite, and presumably they have co-evolved in the last 5 million years in a synergistic fashion. The relevant image is a co-evolutionary wheel (Fig. 1): evolution along any of the radial
spokes presumably gave a mileage to all the other capacities, even if the focus of selection may have changed spokes several times.
This hypothesis is testable; and there is evidence in its favour already. Take the case of autism, for example. Affected people have a problem with the theory of mind, communication, and they can be seriously challenged in the strictly linguistic domain as well (Fisher and Marcus, 2005). The prediction is that there will be several to many genes found, that will have pleiotropic effects on more than one spoke of the wheel in Fig. 1.
Was this article helpful?