Daily Archives: May 20, 2009

Cheesus, It’s Just a Type 1 Error

cheesus2Dan and Sara Bell have seen Jesus again. Once more, he has appeared in a convenience snack. In this case, the deity appears in the form of a “Cheeto” – a rather disgusting corn-based munchie from the US that sticks to the roof of your mouth and clogs the gaps in your teeth. We covered this tendency of seeing the divine in an earlier blog on pareidolia where Jesus turned up on the backside of a dog. No doubt, Dan and Sara will try their luck on eBay where other examples of divine apparitions in snacks such as cheese toasties and pizzas have sold for silly money. 

This nonsense bring me to Michael Shermer’s piece in this month’s Scientific American about what he calls “patternicity” – the human tendency to find meaningful patterns in meaningless noise. (He also gives, “SuperSense” a good plug in the article so I am hoping this mention will drive some of his readers over here). Patternicity is the consequence of a brain that automatically seeks out structure in the environment, looking for significant events. This tendency is particularly strong in the case of detecting people and faces as our brains readily interpret all manner of random configurations as evidence for others. As the Scottish philosopher Hume said, “We find human faces in the moon, armies in the clouds.”

The psychologist Stewart Guthrie has argued that this is a perceptual bias that means we are trip-wired to detecting the presence of potential enemies even when the evidence is so weak. It is better to assume that someone is hiding in the bushes rather than ignore them. So if you see a face in the bushes when there is none, this is know as a Type 1 error because you have inferred the presence of something which is not really there. On the other hand, if you ignore the face in the bushes when there really is an enemy hiding in there then that would be a Type 2 error. The evolutionary argument is that it is better to make a Type 1 error than a Type 2 error because the consequences of ignoring the evidence of a potential enemy are much greater than assuming that someone is there.

However, I don’t think you necessarily need an evolutionary argument based on potential threat for the person bias. All the evidence suggests that newborns (human and monkey) have built-in mechanisms for paying extra attention to faces so we are supersensitive to any face-like pattern to begin with. We even have special face processing areas in the visual parts of our brain. So having a perceptual bias could arise from a variety of different advantages, not necessarily enemy-detecting.

Shermer’s discussion of Type 1 and Type 2 is very relevant to one of my arguments developed in SuperSense – namely that individual propensity to supernatural belief is supported by their interpretation of ambiguous evidence. You can test this by presenting people with computer tasks where they have to detect a faint pattern that may or may not be present among random noise.

Individuals who score highly on scales that measure supernatural belief are also more likely to make Type 1 errors compared to those who score low on such measures who make Type 2 mistakes. So we vary in our susceptibility in detecting evidence and how we interpret it. If you already have a strong belief that there are significant patterns out there, then you will more readily find evidence for it. Beautiful theory, isn’t it. I can easily see all the evidence to support it! Or that might be my SuperSense at work.

12 Comments

Filed under book publicity, In the News, Research