Friday, November 23, 2012

Bad Argument: Illusion of Cause

Ice Cream... causes polio?
(Photo credit: Wikipedia)
Our brain seeks to associate cause and effect, so we can recognize the pattern in the future and react accordingly. However, very often, our brain makes the wrong choice, by associating the effect with the wrong cause, or finding a cause where there is none. That is known to psychologists as "illusion of cause", and was popularized in the book 'The Invisible Gorilla".

Illusion of cause has many forms, many have long Latin names that I shall not attempt to repeat. I have even covered a few of them before as bad arguments. In no particular order (and probably incomplete list):
  • Correlation is not causation [Read more]
  • Spotting patterns when it's merely random distribution  [Read more]
  • Two things happened consecutively are not necessarily cause and effect
  • Attribution to God
Today the discussion will be on the two topics that weren't covered before. 

Two things happened consecutively are not necessarily cause and effect

When two things happened "together" (close in time and maybe space), the mind often assumed that one caused the other. However, there many factors that could affect the time sequences. 

For example, serious illness and depression often go together. So what caused the other? The generally accepted explanation is the serious illness caused the depression, but some "gurus" are known to claim the opposite, that it is depression that allow your body to be weakened and susceptible to serious illness. 

Another example of correlation problem is disease. Incubation period of a disease, for example, and different people's response to the disease, could allow a person who contracted the disease later to either never show symptoms (i.e. be a carrier only) or show symptoms late and only mild symptoms, while others who contracted the disease early, and had full-blown symptoms. Thus, discovery of the disease in one location first does NOT prove he's patient zero, the source. 

In murder mysteries, as the investigation expands, more and more bodies are discovered, but it is important to plot their times of death, not times of discovery. Investigators cannot be allowed to mistake the sequence of discovery as the sequence of the crime in progress. That would screw up their timeline. 

But the point is simple: just because the two events happen close to each other in time does NOT necessarily mean they caused each other, and then you can't tell what's cause and what's effect. 

Here is another example... and one that's often used to illustrate this point...

In the turn of the 20th century, polio is a major epidemic. Kids are left crippled and no one could figure out the reason. They only know it's really prevalent in the summer time. Here's Freakonomics co-authors explaining "cause and effect" fallacy... 

In 1940, Dr. Benjamin Sandler wrote a paper linking sugar consumption (think ice cream) and polio. No, this is not a joke. The paper is still available today. Don't want polio? Stop eating ice cream, the doc said. Ice cream consumption curve is same shape as polio epidemic curve. Government was seriously considering banning ice cream to curb polio. 

Eventually we figured out the virus that causes polio, and vaccine was introduced and epidemic was eradicated, and ice cream survived and we now still have Ben and Jerry's

Just because both items (ice cream consumption and polio instances) peak together doesn't mean they are causing each other! 

Now consider this: what if ice cream was banned, and polio instances went down? Now that will be evidence that ice cream really caused polio, wouldn't it? But there's no science behind it either. It's still a correlation. 

Yet every day you see the equivalent: we did "this", something happened, so "this" must have worked. 

Attribute to God

When there's nothing to attribute the cause to, one fallback is God, whatever name you call Him.

Christians say "It is God's Will" or "God's Hand Moves in Mysterious Ways"

Muslims say "Inshallah" (If Allah wills it), though it's usually used in a different context.

But that's just a "process of elimination"... It can't be anything else you know, so it must be God.

Note the caveat: anything else YOU KNOW. Maybe you don't know enough to explain it or describe it.

Imagine for a moment, that you are an 19th century peasant, barely educated, just enough to farm and raise livestock. Now imagine one day, you found this object in your field (it's actually a helicopter). Can you describe it using what you *should* know? You can't.  You'll say it is a shiny metal object with odd portrusions with a long tail and dropping branches, about size of 10 cows, resting on what appears to be skis, but made of metal. It is clearly not man-made (at least, not men contemporary to 19th century), so it must be made by God! (or the Devil!)

It's all a matter of context, not necessarily cause. 

Enhanced by Zemanta

No comments:

Post a Comment