The role of working memory in the cultural evolution of linguistic structure

Human language gives its users the ability to combine and recombine signals in endlessly novel ways, to express meanings that have never been expressed or even thought of before. An obvious question is how our communication system evolved to be so uniquely complex in the animal kingdom. But while language is a defining human trait, its origins cannot be studied directly: it leaves behind no fossils, and there are no naturally-occurring proxies in other species.

In recent years, researchers have found ways to simulate language evolution with human participants and computer models. An experimental paradigm has emerged which uses iterated learning, "a process in which an individual acquires a behaviour by observing a similar behaviour in another individual who acquired it in the same way" (Kirby et al. 2008). This work has found that humans have cognitive biases which mean that structure inevitably emerges when a new language needs to be both learnable by future generations and capable of expressing as much meaning as possible.

Mechanisms external to language itself may explain why this structure - which we typically call 'grammar' - evolved in all human languages. One such mechanism is working memory, the 'blackboard' which stores information for just long enough to process it in conscious attention.

Working memory can only handle a limited amount of information before it is exhausted and the information decays, as anyone who has tried to remember a group's drinks order while waiting to be served at a bar can confirm. This project will test whether the limitations of working memory could have put pressure on language to become more structured as it evolved culturally.

I am not the first to hypothesise that working memory could have constituted a selective pressure for the emergence of complex grammar. In particular, Coolidge (2012) proposed that as language evolves, grammatical structure is selected for as a means of bypassing the limits of working memory capacity. But as intuitively appealing as this theory is, it has never been substantiated through experimental evidence. Previous work has also left vague whether this 'selection' is for a biological adaptation for syntax, or a process of cultural evolution whereby memory imposes a transmission bottleneck that promotes structure through iterated learning.

The focus of this project is therefore to use the experimental paradigm outlined above to test whether linguistic structure evolves more quickly through cultural transmission - or provides more of an advantage in successfully communicating information - when the load on working memory is greater.

Getting to the bottom of the factors that drive language evolution is central to our understanding of what makes us human. And to truly understand the origins of language, it is not enough simply to know how it evolved chronologically. A more fundamental question is why it evolved along the lines that it did; in other words, once the earliest stages of this new communication system got off the ground, what pressures in hominins' environment selected for particular characteristics of that communication system or made particular characteristics adaptive? This project seeks to advance that line of enquiry.

There are also potential implications for the treatment of specific acquired memory deficits if greater structure is shown to place a lower load on working memory. There are conflicting views over whether linguistic structure is domain-specific or indicative of more general cognitive abilities, but at the least the findings could be of use to speech and language therapy.