Thursday, 28 June 2012

Observations on Heyes' paper “The Evolution of Human Cognition”

Cecilia Heyes' paper “New Thinking: the evolution of human cognition” has just been published in Philosophical Transactions of the Royal Society and is available online. It is the introductory paper to a special issue where the other papers are behind a pay wall.

It is over 20 years since I abandoned my research on CODIL and took early retirement,effectively on health grounds, and just over a year since I decided to come online to see what relevant developments had happened during the intervening years. While my research was oriented to providing a flexible information tool to help people handle poorly structured problems many aspects of the design involved trying to understand how people processed information. As such I was thrilled to discover Cecilia's review as a means of getting back up to speed.

Her dismissal of the approach which she refers to as “Evolutionary Psychology” reminds me very much of some of the disagreements I had with the “Artificial Intelligentsia” in the 1970s and 80s. Then the establishment view (and unfortunately the establishment effectively controlled the research purse strings) was that the way forward to understanding intelligence was to study mathematically well defined tasks such as playing chess and solving formal logical problems of little practical relevance to the rather messy real world we actually live in. It is clear from the review that many cognitive psychologists have been following the stored program computer philosophy of trying to build specific models for specific tasks. I was therefore delighted that this approach is being rejected. The idea that the brain consists of general purpose units which can do anything fits in well with what I was trying to say some thirty years ago – when I was suggesting that the starting point for modelling the way humans process information should be the flexible way they could handle open end, fuzzy and otherwise ill-defined problems, and not detailed studies of well-defined tasks.

Following on from a discussion of brain structures I came to an interesting section which covers recent studies on human evolution and the behaviour of animal. I felt things were going in the right direction when I read:

Whiten & Erdal's pre-Pleistocene perspective focuses on the comparison of humans with chimpanzees. They identify five major components of the 'human socio-cognitive niche', five dimensions on which humans excel - cooperation, egalitarianism, theory of mind, language and culture - and in each case they review evidence that the behaviour/cognitive competence was present to some degree in the common ancestor of humans and chimpanzees. For example, chimpanzees cooperate when hunting and mounting raiding parties on other troops; show signs of egalitarianism when sharing meat and forming coalitions that thwart dominant males; appear to be able to attribute perceptions and goals, if not beliefs and desires, to others; and, in addition to having an extensive repertoire of communicative gestures, chimpanzees use vocalizations in a flexible, context-dependent way to signal information about food and social roles. Whiten & Erdal note that there is a 'yawning gulf in the cultural achievements of chimpanzees and humans', but even in this domain they find signs of continuity. Field studies have yielded reports of more than 40 chimpanzee traditions-involving food processing, tool use and vari­ous social behaviours - and many of the social learning processes found in humans are also present in other animals, including chimpanzees.

My approach to thinking about biological evolution is to assume that there is nothing new under the sun, as everything is no more than a slight increment on what has gone before, and that human thinking mechanisms were little more than greatly amplified versions of the mechanisms of our ape relatives, For the same reason I was pleased to see that in the abstract of Cecilia's paper Grist and Mills: on the cultural origins of cultural learning she writes:

I find that recent empirical work in comparative psychology, developmental psychology and cognitive neuroscience provides surprisingly little evidence of genetic adaptation, and ample evidence of cultural adaptation.

However as I went further into the main paper I found that while there was a lot of very interesting science I felt that often the wrong questions were being asked. This is typified by the words from the start of the the section headed What? Domain-general Developmental Mechanisms:

Many articles in this theme issue ... present a very different view. They suggest that humans are born with extraordinarily powerful cognative-developmental mechanisms. ... The genetically inherited cognitive-developmental mechanisms use computational processes that are also present in other animals, but they are uniquely powerful in their range, capacity and flexibility.

I cannot possibly agree with such a view. Of course there are exceptional aspects of human behaviour – but to suggest that we are born with extraordinarily powerful and unique mechanisms run counter to the way biological evolution works. Nearly all changes are made in small increments – and these changes are often related to comparative rates of development of various parts of the body rather than fundamental changes in plan. The real problem in the above statement is that there seems to be no clear model of how the brain processes information at the general cognitive level and as humans are not impartial observers it is very attractive to think that the “missing links” in the brain model are “extraordinarily powerful and unique mechanisms”.

I discussed the fact that there were missing links in brain research a year ago today in the opening brainstorm The Black Hole in Brain Research, If you try and stand back from the detailed research and get an overview of brain research the situation is rather as if some aliens had obtained a number of working personal computers – each running different applications. One machine is dissected to try and understand the wiring, and monitors are installed in others to see which circuits are electrically active as the applications run. Other alien scientists study individual application in great depeth – and end up learning a lot about the applications – but gain virtually no insight into how the tasks the computers are doing are related to the hardware. In such circumstance it would be very easy for the aliens to assume that the central processor of the computer contains “extraordinarily powerful and unique mechanisms” when it is actually has a quite small set of operations which manipulate numbers in a way which superficially appear to have little to do with the tasks the computers are running. However we know that the real power of the computer lies not in the hardware (different computers can have different instruction sets) but in the way the patterns of “0”s and “1”s in memory are linked together. Superficially there can't be anything much simpler that an array of binary digits – which is why the aliens failed to understand the role of software.

So back to the brain and the evolution of human cognition. If you are talking about any form of evolutionary model you need to have a clear starting point and a clear finishing point so that you can assess what has changed, and consider the evolutionary pressures that brought about that change. This is easy enough with an organ like the hand, where one can examine both living and fossil examples.

To properly understand the evolution of human cognition you also need to have an understanding of the mechanisms within the brain (both animal and human) so you can measure the evolutionary changes that have taken place, and hopefully predict the likely reasons for the change. Unfortunately it is effectively impossible to directly read out the contents of a living brain of an animal to work out what it understands, and there is no way we can ever get fossil brainwaves from our long extinct forebears. But unless you have a clear idea of the evolutionary starting point it is totally inappropriate to make strong statements such as the finishing point includes “extraordinarily powerful and unique mechanisms” especially in the light of the statement quoted earlier that there is “surprisingly little evidence of genetic adaptation”.

If we cannot experimentally determine the starting point it could be possible to look at models based on the minimal amount of information an animal will need to remember past events and use the remembered information to make simple predictions about the present. Given such a model it should then be possible to ask how it would adapt to various evolutionary pressures which lead towards what we know about human cognition. If the process is successful we have a possible evolutionary pathway and we can compare the start point and end point mechanisms. If the approach is unsuccessful it should at least give us clues as to how we might build a better starting model.

In fact I have been exploring one possible model in this blog, for instance in brainstorms such as An Evolutionary Model of the Brains Internal Language and Step Outside the box to understand the Evolution of Intelligence. I am well aware that the initial model has limitations, which is why I welcome contributions to the “brainstorms” on this blog, but the model appears to explain some features of in human mental development over the last few million years, and suggests a tripping point clear, where there could be a small but significant change in the underlying mechanism. I will be preparing a further brainstorm on this topic sometime in July, which will be looking further than previous posts into the relationship between the brain's biological mechanisms and cultural knowledge.

No comments:

Post a Comment