In a paper in Science from October 2008, Jennifer Whitson and Adam Galinsky report that placing people in situations where they lack control increases the false perception of patterns because of a need to impose structure on even random events.
This study is very interesting because it helps us understand why we develop superstitions and the like, which are based on false pattern recognition. It does not, however, speculate on why some of those superstitions take hold and last (e.g., buildings without thirteenth floors) and some do not (e.g., my efforts to get my tee ball team to wear pink socks after a 3-4, 4RBI game and a laundry accident).
I, however, am not above some wild speculation. The defense of superstitions, quack medical treatments, etc. frequently goes like this: A medical treatment works or it does not work. If it works, people who use the treatment are more likely to live, people who don’t are more likely to die and the treatment keeps getting used. If it does not work, people who use the treatment are more likely to die, people who don’t are more likely to live and the treatment stops getting used. That makes intuitive sense. It sounds a lot like selection, and we like selection.The Chinese have been using acupuncture for thousands of years, supposedly. It must be safe and effective. People have been eating a Mediterranean diet for thousands of years. It must be good for you. We’ve all heard these arguments before and probably made some of them ourselves. Let’s forget for a moment that acupuncture may not be ancient or Chinese or that Italians only started eating tomatoes in the last 300 years.
Seems intuitive, right? Chances are, however, that our intuition is just flat out wrong. Why? Evolutionary theory says we are wrong.
The principles of evolutionary theory can be applied to any system that consists of discrete, varying, replicating units. In fact, most of those evolutionary principles were developed before we knew that genes were made of DNA. In this sense, ideas are like genes. They are relatively discrete. They have different versions. They can replicate. A brief illustration of this point follows.
Ideas are discrete
Idea 1: What is your favorite ice cream flavor?
Idea 2: What is your favorite color?
Ideas have different varieties
Idea 1.1: I like mint chocolate chip.
Idea 1.2: I like vanilla.
Ideas can replicate
Person A: I can’t decide if I prefer mint chocolate chip or vanilla.
Person B: I like mint chocolate chip.
Person A: B is right. Mint chocolate chip is better than vanilla.
So, maybe the superstitions that last are selected for? Often, when people (including scientists) think evolution, they think selection, but they forget about the three other forces that drive evolution: drift, mutation, and migration. Selection and drift are related, as they both act to reduce variation. We are going to be interested in the interplay between selection and drift.
Drift is the effect of random chance on populations. Because real populations are not infinite, not all individual random effects are balanced out by random effects in the opposite direction. In real populations, random events affect outcomes. As the population size decreases, drift becomes more important and selection becomes less efficient (i.e., larger selective benefits are needed to overcome the effects of drift). Eventually, drift could lead to the fixation (i.e., everybody has the same version of the gene or idea) of a gene or idea with no fitness benefit or, even, a negative fitness effect.
There is a range of fitness effects within which the gene or idea variant acts like it has no effect on fitness, even if it has a positive or negative fitness effect in individuals (expressed as the coefficient of selection – how much proportionally less fit one version is compared to the more fit version). This range is dependent on the effective population size (the number of randomly mating individuals needed to explain the population genetics of a group) according to the equation s=1/4Ne, where “s” is the selection coefficient and “Ne” is the effective population size.
The absolute population size for humans is approximately 6 billion. The genetic effective population size is estimated at 10,000. That means that selection dominates for selection coefficients larger than 0.000025 (2.5 excess deaths per 100,000 individuals). What is the effective population size for ideas? With the advent of the internet, one might expect that the effective population size for ideas would be large; but, most information originally comes from wire services or text books. We can make plausible arguments for large or small effective population sizes for ideas.
Let’s try an example to estimate a limit to the effective population size for ideas. The by-products of coal burning power plants are estimated (let’s use a number I heard once) to kill 30,000 Americans (total population ~300 million) each year. Although 30,000 sounds like a lot, it is only 1 extra death per 10,000 people (s=0.0001). When we rearrange our equation s=1/4Ne to Ne=1/4s, we get an upper-bound for the effective population size for ideas of 2500 (it could be smaller). Is this number reasonable when compared to an absolute population size of 300 million? It is a 5 order of magnitude difference, but not all 300 million of us are freely exchanging ideas with anyone and everyone else. The intellectual equivalent of free sex with the entire population is reserved for a few wire services, television channels, and text books. Maybe 2500 is reasonable.
I wonder how small the effective population size for ideas was in the Europe that used blood letting? Could we reasonably be talking about effective population sizes for ideas of less than 100 people? That is fertile ground for bad ideas. Dark Ages anyone?
This article was the subject of my “Speaking Up” radio interview on Sketpically Speaking (airing 27 August 2010) and was originally posted at Science 2.0 in November 2008.