Sunday, 16 September 2018

If only this skull could still talk ...

If only this skull
could still talk
I am always happy to answer questions about how my research relates to a particular research paper and in a discussion on the FutureLearn course "A Question of Time" Paul asked about "On the antiquity of language: the reinterpretation of Neandertal linguistic capacities and its consequences" by Dan Dediu and Stephen C. Levinson. This paper concludes that Neanderthals and Denisovans may have had very similar language capacities to us - they too had complex tool making technologies. Paul asked:
When did the conditions arise to allow more rapid evolution of information (technical, such as for tools but also hunting & food gathering & processing techniques, cultural info etc.) ? After all, we possess much the same genes as our ancestors did 60,000 years ago, but our world is significantly different due to evolution of information.
Dediu and Levinson's paper looks in detail at the relevant literature about the discover of human fossils, evidence for tool making , etc. while my approach starts in a very different way by looking at how a network of neurons might evolve into an intelligent brain. The immediately relevant parts of my model are as follows:

  • The starting point is evolutionary economics. An animal has to spend time and energy (a) gathering and eating food, (b) breeding, (c) escaping predators, (d) growing a brain and learning how to use it. 
  • Large brains need a significant amount of energy to work compared with other organs so have to work hard to justify their existance.
  • Because everything an animal’s brain learns is lost when the animal dies, no species will evolve a brain bigger than it can use in a lifetime, because the time and energy used filling an over-sized brain with information that is never used could be better spent eating, breeding and escaping. 
  • Trial and error learning is time consuming (basically it involves doing something repeatedly until it is routinely done correctly) so this puts an effective limit on brain size and intelligence. There are various strategies but primates (and some other mammals) have bigger and more effective brains than average because they evolved with a long life span, and the parents spend a lot of time supervising a comparatively small number of infants. However the “cap” on how much can be learnt economically in a lifetime still limits the size of the brain.
  • Humans (and even some of the other apes) tend to differ from animals because they frequently use tools.
  • In evolutionary terms inventing a tool means nothing unless each generation can pass the tool making skills to the next generation. Tools which improve survival alter the evolutionary economics and so tool making could justify the development of a slightly bigger brain. 
  • About 5 million years simple tool making by early hominins probably helped them survive and they started to evolve slightly bigger brains. And with a slightly bigger brain they could make slightly better tools. The process was limited by the speed of genetic evolution and was very slow – and it is therefore not surprising that in the Old Stone Age hand axes were made in the same way for about a million years. 
  • It is reasonable to assume that family groups had some form of language for communicating with other members of the group. This can clearly be seen today in family groups of species such as bonobos, wolves, elephants and orcas. 
  • Say 500,000 years ago it is reasonable to assume that some kinds of meaningful communications could take place between closely related hominin groups – such as the Neanderthals, Denisovans and Sapiens forebears. This can happen in other species and recently a solitary dolphin has been observed to learn "porpoise" - See "Solitary Dolphin clicks with Porpoise Companions in the Clyde."
  • Now language is a tool, and at some stage it started to be used as a tool to teach tool-making in a way that is more efficient than trial and error. Evolutionary economics means that it is now possible, with the same brain, to learn about more tools – or to learn to build better tools. 
  • But language is itself a tool – so if language makes it possible to build better tools it makes it possible to build a better language – which in turn makes it possible to teach how to make even better tools. (Still happens. Old fogeys like me tend to grumble because the younger generation is always changing the language!)
  • What we have is a significant tipping point. Generation by generation language becomes slightly more effective, so that we can make better tools and a better language  so in the next generation things get even better ... and better ... In fact what we have is what is an auto-catalysed reaction which has led to an explosive growth in our tool-making capacity in modern times with no need to an increase in brain size.
  • In fact there are several possible tipping points, each of which accelerates the process. Language allows information to be passed from generation to generation so that everything a brain has learnt in a lifetime is not lost when the brain dies. Tools such as writing and now computers allows the storage of vast quantities of information. Language also makes it easier to use other humans as tools so we now use other humans  to grow and deliver our food needs, build shelters, help us travel large distances, and allow us to live longer.
One of the problems with this model is knowing when the process started. In a demonstration of the autocatalysed chemical reactions, such as that between nitric acid and alcohol nothing seems to happen for several minutes, then the liquid takes a faint brown tinge, which rapidly darkens and then there is an explosive reaction, the actual timings depending on the strength of the nitric acid. A similar process in the development of language could also work on a "slow, slow, quick, quicker, bang!"basis but on a totally different timescale It might start with very small, almost imperceptible, changes over a million years or more (the old Stone Age?), a shorter period with small but detectable changes (the Middle Stone Age), a faster change with obvious changes (New Stone Age), and the really explosive growth starts with the coming of civilization and the invention of writing. The fact that the first "evidence" for our increasing intelligence took place in Africa and involved our own species may be a pure accident of history, no more significant than the role of Western Europeans, and particularly the British, in the Industrial Evolution. Once the "Isn't language such a powerful tool" genie was out of the bottle it would have been rapidly adopted by all human species that came aware of it, just as modern technology now embraces the whole world.

So back to the paper by Dediu and Levinson. They have produced a good descriptive summary of the evidence suggesting how language gradually appeared, with a detailed review of the relevant literature, and as such their model and my model fit well together although we have approach the subject from very different angles. In particular we both consider that the foundations of language started long before the apparent explosion in the ability of humans to make tools.

The most significant difference is that their paper says nothing about the underlying mechanism which allowed the brain to "transform" in this way without any apparent significant genetic change. Such omissions are very common in accounts of modern research relating to many different aspect of the brain, For instance a typical example appears in this months Scientific American, where a news item on "Brain Bar Codes" starts with the words. "Neuroscientists know a lot about how individual neurons operate but remarkably little about how large numbers of them work together."

My approach is based on a direct assault of this area of uncertainty in brain research. I start with research carried out some years ago on a very unconventional computer language CODIL - which was designed to provide friendly human-computer interaction. A recent reassessment suggests that CODIL can be interpreted in term of a neural network and could also be a simple model of human-human interaction - perhaps mimicking early human natural language. As a result I am exploring the way that a CODIL-like model can explain how the animal brain could morph from a pattern-recognition systems, via a set processing system, to a rule-based programming system almost certainly capable  of supporting natural language and human intelligence. 

So back to Paul's question about when language became significant. , and when did the earliest forms of human natural language originate? The key point is a "chicken and egg" type problem. The neural net model suggests that animal brains do not live long enough to learn enough to become very intelligent - while serious intelligence needs information to be shared between many different brains over generations. This suggests that you can't have advanced intelligence without language - and you can't have language without intelligence! But is the a tipping point where a species could develop a brain that was just capable of supporting a vert  very simple language?
If you could inject the seed of a very simple language into the system it could grow - very slowly at first - but at an accelerating rate culminating in an explosive tipping point. Dating the start of this process could be very difficult - but what is more relevant is the point when information shared between individuals became a more important factor than the genetic expansion of the brain. At the moment I am still thinking about it - and need more information, but it is almost certainly significant that over the last 100,000 years or so the brain of our species has apparently become "much more intelligent" but has actually stopped getting bigger - and may even be getting fractionally smaller. This could suggest that the tipping point when language became the dominant feature in the evolution of human intelligence was perhaps one or two hundred thousand years ago, but that it's precursor had been an significant feature for a million or more years before that.

No comments:

Post a Comment