My latest Ockham’s Razor talk for Australian ABC Radio National. How some of us can’t recognize facial patterns, but how most of us fall for imaginary patterns in the real world, and what we can do about it.

Here is the link to the podcast:

http://www.abc.net.au/radionational/programs/ockhamsrazor/seeing-patterns-(even-when-they-aren%E2%80%99t-there)/8421130

And here is the original unedited transcript. Enjoy!

I’ve known Robyn Williams for twenty-odd years. We have a common mission of trying to persuade people that science, and scientific thinking, are really important for our future. I recently found that we also have another problem in common. We both suffer from prosopagnosia, otherwise known as face blindness. Robyn and I both have the greatest difficulty in recognizing and remembering faces.

We were at a meeting together recently in Sydney’s Government House. We hadn’t seen each other for a while, but I was able to identify Robyn because he had been announced as the chairman of a session. I walked over to say hello, and he looked at me blankly. Only after he bent down to read my name badge did he realize who I was and started chatting.

I suppose it’s lucky that we weren’t chimpanzees, because chimpanzees don’t recognize each other’s faces, but distinguish individuals by the patterns on their buttocks.

Prosopagnosia can be very embarrassing. I have walked past my own brother in the street without recognizing him. Worse still, I walked straight past my wife of just a few weeks when she was waiting to meet me at an airport. Only after I had walked past her did I think to myself “I’m sure I’ve seen that hat.”

At least I didn’t mistake my wife for the hat, as a patient of the psychologist Oliver Sacks once did. Robyn reminded me that Sacks himself suffered from prosopagnosia, to the extent that he couldn’t even recognize his own face. Once in a restaurant he started to groom his beard, using the restaurant window as a mirror. Slowly it dawned on him that the reflection was not joining in the grooming, but that he was in fact looking at real person on the other side of the window, who was staring back at him with some puzzlement.

Those of us who suffer from prosopagnosia are unable to recognize facial patterns. An opposite condition is apophenia, which is seeing patterns where none really exist.

Apophenia can be a very serious neurological condition, but we are all inclined to look for patterns in events, and there are two reasons why we are prone to see patterns even when they aren’t there. One is evolutionary. The other is mathematical.

The evolutionary reason for our tendency to believe falsehoods is a matter of whether it is more costly to mistakenly accept a falsehood (psychologists call this a Type I error) or to mistakenly reject a truth (psychologists call this a Type II error). Say you were a primitive person walking through high grass in an area where tigers were known to be present. You notice a rustle in the grass. It could be the wind, but it might also indicate the presence of a menacing carnivore. According to the evolutionary argument, those who survived to pass on their genes were the cautious ones who bet on the carnivore and took appropriate defensive measures, even though most of the time it might have been the wind.

Evolutionary biologists Kevin Foster and Hanna Kokko argue that “as long as the cost of type II errors – that is, of mistakenly rejecting a truth – is high enough, natural selection can favour strategies that frequently make type I errors and generate superstitions.” In other words, as they demonstrate with mathematical rigour, superstitions are adaptive.

Mathematical rigour also comes in to the second reason why we tend to see patterns when they are not there, and also why we believe that, because we have perceived a pattern, then the pattern must have some meaning.

Maybe it does; maybe it doesn’t. In science, finding a pattern in things is just a start. Our real job is to investigate whether the pattern is real and, if so, whether it has meaning; that is, whether there is a causative link that is generating the pattern, or whether the correlations that we observe are no more than random.

Say, for example, that you have a map of Devon and Cornwall, and plot the positions of the hundred or more standing stones that have been there since ancient times. Out of these, a dozen or so form a very accurate straight line stretching for over a hundred miles. Intuition screams that these must have been purposely placed along this line, but in this case intuition is misleading. In fact, as an astronomer friend showed me, you can take the same number of random stars from a map of the sky, and find a dozen or so that also lie very accurately on a straight line.

Why should this be? It comes down to some very sophisticated mathematics that was discovered in the 1920s by one of my heroes – a young Cambridge mathematician called Frank Plumpton Ramsey, who showed vividly how even the most abstruse mathematics can be very relevant to real life. Using advanced mathematics, he worked out just how much of its income a nation should realistically invest in its future – advice that many of today’s governments are ignoring, to the peril of us all.

The discovery that concerns us here, though, is called Ramsey’s theorem. It shows that, as a group of objects gets larger and larger, connections and patterns must inevitably emerge. If we are in a group of six people at a party, for example, there must inevitably be at least three of us who are linked through knowing each other, or three who are linked in the negative sense that they do not know each other.

Ramsey’s theorem really comes into its own as the numbers get larger. Its significance for the real world is that we can always find a chain of negative or positive correlations between people or events if the group is large enough. Unless we have separate evidence for an underlying cause, however, it is wise to be wary. Ramsey’s theorem shows that such chains of connection in large masses of data are quite likely to be statistical artifacts; no more than a matter of chance.

One example is astrology, which goes back at least to the ancient civilizations of Mesopotamia (the Babylonians, the Assyrians and the Sumerians) some 5000 years ago. The peoples of these civilizations believed that there must be a link between unusual heavenly and earthly events. So they recorded such events in a set of books that is now known as the omen series. Each page contained two columns – one of unusual heavenly events like eclipses, the other of earthly events with which they were thought to be correlated.

When an eclipse, an occultation, or an unusual arrangement of stars occurred, they would look through their lists to see whether the event might have been linked with the occurrence of a good crop, the death of a king, or even the collapse of an empire in the past. According to Ramsey’s theorem, they were bound to find some connections within such a large mass of data.

The ancient astrologers could be forgiven for their ignorance of Ramsey’s theorem. The same cannot be said for some of today’s highly paid prophets of society and the market place, who are also seeking and finding correlations and patterns among large masses of data, often apparently ignorant of the fact that they are bound to find such connections.

Compare the situation with global warming. Here too there is a pattern – a slow but persistent increase in atmospheric carbon dioxide concentrations, accompanied by an increase in global temperatures. But is this just a statistical artefact? Those who desperately don’t want to believe in it would say so. But here we have an underlying mechanism – we know why increasing carbon dioxide concentrations can cause a temperature increase. We also have postulated mechanisms that can be translated into computer models. When all of the different models point to the same conclusion, we would be foolish indeed to dismiss them. Mind you, some people are foolish indeed, especially when their financial interests are at risk. Just think of the desperation with which cigarette manufacturers in the 1960s and 1970s dismissed the mounting evidence that smoking causes lung cancer, despite the fact that scientists had identified the biological mechanisms involved.

Some perceived patterns have real meaning. Others exist only in our minds. How can we tell the difference, especially when our evolutionary history drives us to believe that the very existence of perceived patterns and connections implies that they must have some meaning, although mathematics suggests otherwise.

What we need is a baloney detection mechanism. This is what science provides, because science works on one simple principle – check your beliefs against reality, as Robyn did when he bent down to read my name tag. We can’t all be scientists, but we can all use this simple scientific principle. We should be suspicious of any claim that relies solely on apparent patterns for its veracity. The reality lies below, and if those who make the claim can’t show why it should be so, we should be very suspicious indeed.

IMAGE: Source unknown, but would like to credit