Has archaeology become a pseudo-science?
A stone 'totem pole' found at Gobekli Tepe: image courtesy of Travel the Unknown
When I started out on this journey I never thought I would need to explain or justify the scientific method to any academic, let alone those members of an academic discipline ending in ‘ology’. But over the last few years, and particularly in recent weeks, it has become clear to me that there are, indeed, some academic disciplines, even some ostensibly scientific ‘ologies’ that have diverged so far from the scientific path that they are behaving more like religions.
Since publishing our ‘cave art’ paper, I have received the views of a few archaeologists who, in my opinion, are behaving irrationally. Our cave art paper is clearly a perfectly normal scientific paper. We have a zodiacal hypothesis deduced from observations at sites like Gobekli Tepe and Catalhoyuk, and we perform a perfectly sensible and fair scientific test of our hypothesis using the best data available from Western European Palaeolithic cave art. And yet when I claim that our hypothesis is essentially proven, in a scientific sense, by the strength of the observed correlation between our zodiacal hypothesis and established radiocarbon dates, I am met with derision from some archaeologists. There have been claims that we are ‘lucky’, or have treated the data unfairly in some ill-defined sense, or that the hypothesis itself is somehow an ‘embarrassment’, or that the scientific method is simply not valid for studying ancient symbols. I can only conclude from this that some archaeologists, and in my present situation it seems to be a majority, although this might be a false perspective, have lost track of why their discipline is an ‘ology’. In other words, they appear to have lost sight, diverged or abandoned the scientific method. To me, this is completely amazing.
It is perfectly reasonable to react to our discovery of an ancient zodiac with surprise - amazement even. It is perfectly reasonable also to scrutinize our methods, i.e. our handling of the data and statistics. But until a flaw in our methodology is found, it is not reasonable to dispute our conclusions. Here, I will try to explain why this is so – essentially, I am describing the scientific method.
First, I should explain my background so that you can have at least some faith in what I am saying. I have a PhD in theoretical physics from the University of Bristol, a leading UK university with an excellent reputation in physics internationally. I then studied as a post-doctoral researcher in the department of Chemistry at Imperial College London (acknowledged as one of the World’s top universities) for over 5 years before taking academic positions teaching chemical engineering at the Universities of Strathclyde and then Edinburgh in Scotland. Edinburgh is also acknowledged as a world-leading university, frequently within the world’s ‘top 20’. I am also a Fellow of the Royal Society of Chemistry, which recognises my contribution to the chemical sciences in the UK.
With this background, and with 25 years of experience doing scientific research, I have a reasonable grasp of the scientific method, or how science works, and why it is the best approach to discovering new knowledge. The aim of science, in a nutshell, is to find the most likely explanation for observations. That’s all. Clearly, when approaching any problem we must consider a range of alternative explanations. But the difference between science and non-science, is that science searches for the explanation that is most likely to be correct. To determine the ‘likelihood’ of competing explanations requires the use of statistics, which is just a branch of logic.
So, the scientific method is founded on the statistical analysis of data, or measurements. It enables the rational choice between different explanations of the same data. This is the reason science has been so spectacularly successful over the last few centuries in advancing knowledge – it allows us to discard some explanations as being almost certainly false or accept others as being almost certainly correct, or at least as being the best explanation from among those available. The scientific method is essentially responsible for our modern way of living. We could not have got to our present level of sophistication using non-scientific methods. There is no dispute about this.
Now, any discipline wishing to label itself as a science, and therefore label its practitioners as scientists, must use this scientific method. It is a corruption to call oneself a ‘scientist’, and yet perform research that does not involve the statistical analysis of measurements. So where does this leave other disciplines, like history, that take a different approach?
Clearly, there are some problems, often faced by those in the humanities, that are just too difficult to take a purely scientific approach. Sometimes the problem is difficult to even define precisely, or measurements are either impractical or impossible. This is fine. It means that other approaches to this kind of knowledge must be pursued, as there is no other option. In these cases, it typically means that reasoned debate takes place, assumptions are made, and conclusions are stated. But rarely can these conclusions be made with much certainty. In effect, all we have is the comparison between different explanations of the same observations, but with little to no confidence in which explanation is most likely to be correct. This is not a criticism of non-scientific disciplines. It is simply a fact that not all questions have quantifiable answers.
However, it will sometimes happen that problems that were once thought to be so difficult that a scientific approach cannot be taken, are, in fact, much simpler than previously believed. This is what has just happened with our cave art research. Until just a few weeks ago it was generally believed among archaeologists, as I understand it, that Palaeolithic cave art could not be ‘decoded’, and that its meaning would always remain enigmatic and subjective. Our work, using the scientific method, shows this is wrong and that there is actually a perfectly simple way it can be understood based on our zodiacal hypothesis.
This should be seen as progress, as it opens the door to an entirely new research field – the evolution of an ancient zodiac. It means we now have an alternative method for dating ancient artefacts and a better understanding of ancient ways of thinking. Of course, we are now able to discard other subjective interpretations of this cave art (that are inconsistent with ours) as being almost certainly wrong. This might cause some friction with practitioners of non-scientific approaches to interpretation of this art, as it means they have almost certainly been ‘barking up the wrong tree’, and this is difficult for some to accept. However, non-scientific disciplines should always ensure their debates among alternative explanations respect the scientific facts. Sometimes, as has just happened, radical new facts are discovered that essentially render old non-scientific debates obsolete. This is the nature of paradigm change.
One of the cornerstones of the scientific method, the one we used in our cave art paper, is known as ‘hypothesis testing’. It is simply a rational way of choosing between alternative explanations for observations. Typically, it works like this; the predictions of an idea, theory, hypothesis or other statement of knowledge are compared statistically with actual ‘real’ measurements. Essentially, we look for a ‘correlation’ between predicted and real data sets. We then try to estimate the probability that that this correlation could have occurred by pure chance. If this probability is very small, then it is rational to conclude that chance had nothing to do with it, and instead the hypothesis, theory etc is a good explanation for the observed correlation. It doesn’t prove the explanation is correct in a mathematical sense, but it does provide confidence that the hypothesis might, in fact, be correct.
Of course, there are many problems with this approach, because it is undertaken by humans with all our frailties. We need to agree on the measurement process and the statistical analysis of the resulting data, and this is not always straightforward. Moreover, if we wish to make strong claims about the truth of the hypothesis, then we also need to be sure there is no reasonable alternative explanation (other than pure chance) for the observed correlation, and we need to agree on a level of confidence beyond which it is reasonable to make such strong conclusions.
In the case of our cave art paper, there are no real difficulties like this. The measured and predicted data sets which we compare are straightforwardly obtained with sufficient accuracy and precision to be useful, and their statistical analysis is particularly simple. There are also no other reasonable explanations for the observed correlation between them, which is extremely strong. So, we can rationally claim our zodiacal hypothesis is almost certainly true. In fact, the statistics are so good that we are entitled, I think, to claim it has been ‘scientifically proven’.
And yet, this research appears to be unpopular among the academic archaeology discipline. Why is this? For some archaeologists it is as though there will never be sufficient evidence to take this zodiacal hypothesis seriously. They essentially are rejecting the scientific method, if only for this specific issue. This is irrational. What might they be thinking? Why are they so sceptical?
This is an important issue. Are prehistoric symbols really impervious to the scientific method? Are there really problems for which such scepticism is rational? This issue, of confidence levels and scepticism is tricky.
In my career as a scientist, I approach every new research paper I read with scepticism. But this is only temporary. The aim of reading the paper is to dispel this scepticism to arrive at a scientific viewpoint, i.e. a judgement as to how likely the ideas being investigated are to be true. Scepticism is not a permanent state of mind – eventually I need to take a view which depends on the strength of the evidence presented.
Scrutiny of the methodology is key. Only if the methodology is flawed might my scepticism remain. But once I am satisfied that the methodology is fine, or at least sufficient for the issue investigated, then I’ll take a view on the conclusions which depends on the strength of the results. A problem we have in communicating the strength of the results is a lack of sufficient words in the English language to convey such a wide range of possible outcomes. For example, if the work shows only a very weak correlation, say a chance of 1 in 10 of being wrong, the hypothesis ‘might’ be correct. A stronger result, at the level of 1 in 100 of being wrong say, is ‘probably’ correct, while it might be said to be ‘very likely’ correct at 1 in 1,000, and so on.
But what words should we choose to convey confidence levels of 1 in 10 thousand, or 1 in 1 million, or 1 in 1 billion? And at what confidence level is it okay to proceed as though the hypothesis had been proven? I have seen an archaeology paper presented at a conference that argued that a confidence level of 1 in 10,000 of being wrong is sufficient to claim a discovery. For particle physicists the level is around 1 in 2.5 million. This does not mean that this level is considered sufficient to claim ‘absolute truth’, i.e. mathematical proof, of a hypothesis. But it is sufficient to allow publication of a result as though a discovery had been made, i.e. to make very strong claims indeed.
Intuitively, it seems a confidence level of around 1 in 1 million of being wrong is considered sufficient to be convincing for most people, i.e. to claim the hypothesis is ‘almost certainly’ correct. You have about the same probability of throwing 8 sixes in a row on a dice as this. It’s a very demanding level of confidence. If this actually happened in a game of dice, you’d be entitled to suspect foul play and examine the dice for a bias. The strength of our zodiacal hypothesis based on the correlation in Palaeolithic art is around several hundred to several thousand times stronger than this. That’s an insane level of confidence. Of course, it’s rational to scrutinize our methodology. But until a flaw or alternative explanation is found, it really is completely irrational to dismiss this level of confidence as just ‘lucky’, or to remain sceptical. To date, no flaw or alternative explanation has been found.
Indeed, when we factor in the probability that the Vulture Stone, i.e. Pillar 43, at Gobekli Tepe also appears to support the zodiacal hypothesis at a level of around 1 in 100 million, it is safe, in my view, to claim the hypothesis is scientifically proven. At the combined level of 1 in 53 quadrillion of being wrong, scepticism is no longer an option. Only the delusional would deny our hypothesis at this level of confidence. Yes, there is a degree of subjectivity in this value now due to the pattern matching method we used for Pillar 43, but this is trivial compared to this overwhelming result.
And yet, some archaeologists insist with complete conviction that it is me who is delusional, and that my work is pseudo-archaeological. This leaves me stumped. I cannot understand their way of thinking. It is alien to me. I have to conclude that if my work is pseudo-archaeological, then orthodox archaeology is pseudo-scientific. In other words, it has become a belief system, or religion, which suppresses rational thought. Only the irrational could continue to deny our hypothesis as nonsense with such strong conviction.
Essentially, what I am saying is this. The correlation we have discovered in Paleolithic art and Gobekli Tepe is as real as the moon, and demands an explanation. What explanation is there other than ours?