Christiansen & Chater (C&C) make an excellent argument, both against the plausibility of an innate language-specific Universal Grammar (UG) and for the likelihood of the cultural evolution of language in response to selective pressures associated with use and learnability. In terms of challenging the concept of an innate UG, one of their strongest arguments concerns the implausibility of the independent evolution of highly similar UGs within isolated populations, and also the improbability that a common ancestor UG could be maintained in its original form should a previously unified population split and disperse.
C&C's argument that apparent linguistic universals are therefore attributable to the cultural evolution of language, as opposed to biological evolution of an innately specified UG, hence depends on a notion of convergent cultural evolution. In biological evolution, convergent evolution refers to a process by which species that are only distantly related independently evolve analogous adaptations in response to similar environmental pressures. Convergent cultural evolution therefore refers to situations in which different populations independently develop similar socially transmitted behaviours despite different ancestral histories. In C&C's account, similarities in languages will arise due to similarities in human brains, thereby imposing similar selective pressures on the process of cultural evolution. In other accounts (e.g., Deacon Reference Deacon1997; Reference Deacon, Christiansen and Kirby2003a; Reference Deacon, Weber and Depew2003b), similarities may also emerge as a result of functional constraints associated with communication that are, in a sense, external to brains (Deacon [Reference Deacon, Christiansen and Kirby2003a], for example, has suggested that many of the features common to human languages might also be found in any symbolic communication system, such as one used by an alien species from another planet). On either of these accounts, however, it is essentially convergent cultural evolution which is producing the observed uniformities in behaviour across populations.
In order to fully develop and support such views, it would be useful to understand more about how such cultural convergence is likely to operate. Although there are many cases that can be, with hindsight, attributed to convergent cultural evolution (e.g., the independent development of writing systems [Mesoudi et al. Reference Mesoudi, Whiten and Laland2004]; or cross-cultural similarities in colour terms [Deacon Reference Deacon1997]), ideally we would have access to complete datasets to track the relevant behavioural histories, and would be able to control likely sources of variation. Mesoudi (e.g., Mesoudi Reference Mesoudi2007; Mesoudi & O'Brien Reference Mesoudi and O'Brien2008) has argued that, for just such reasons, experimental approaches are particularly valuable in the study of culture.
In our experiments (e.g., Caldwell & Millen Reference Caldwell and Millen2008) we have been making use of “microsociety” (e.g., Baum et al. 2004), or “microculture” (e.g., Jacobs & Campbell 1961), methods in order to allow us to study cultural phenomena under controlled laboratory conditions over short time periods. Generational succession is simulated within small test groups of participants by periodically removing the longest-standing member of the group and replacing them with a naive individual. While in the test group, participants are required to complete simple tasks using everyday materials, with clear goals. In one of our procedures, participants were instructed to build a paper aeroplane which flew as far as possible, and in the other, they were instructed to construct a tower from spaghetti and modelling clay which was as tall as possible. We have shown that, in both cases, information accumulated within the groups such that later generations produced designs which were more successful than earlier ones.
However, more relevant from the point of view of the target article is that we also have evidence of convergent cultural evolution. For each of our tasks we ran ten chains each of ten participants, and different chains had no contact with one another (simulating isolated cultures). Photographs of the objects produced by each participant were rated for similarity. Although there was clear evidence of cultural variation (designs were rated as being more similar to others from the same chain, compared with those from different chains), there was also evidence of convergence, since designs further down the chains were more similar to one another compared with those at the start of the chains. So, the tenth designs from each of the ten chains were rated as being more similar to one another, compared with the first designs from each of the chains. Thus, we have been able to demonstrate experimentally that over generations of cultural transmission, behaviours can become increasingly similar when shaped by similar selection pressures (in our case, the goal measures).
Our experiments are admittedly highly simplified models of cultural transmission, and the goals involved in our tasks are very far removed from the communicative functions performed by language. C&C also draw particular attention to the fact that language evolution may be a special case in terms of cultural evolution, in that users are not consciously designing their behaviours (although see Mesoudi Reference Mesoudi2008 for a different view). All the same, we believe our results are illustrative of general phenomena which we suspect to be at work across a variety of contexts involving cultural transmission. Importantly, our experiments clearly show that complex adaptive behaviours can arise through cultural transmission, and that these behaviours, shaped by similar selection pressures, can show effects of convergence across isolated populations.
