Friday, March 28, 2008

E2.0: John Allen Paulos: Probability and Error

This is the 27th in a new series of weekend posts taken from the presentations at the Salk Institute’s “Beyond Belief: Enlightenment 2.0.”. I have placed an index of essays in this series in an introductory post, Enlightenment 2.0: Introduction.

The main purpose of education should be to turn children into adults who are capable of making intelligent decisions. We know, through a great deal of research, that there are certain types of errors that people too commonly make that derail their attempts to make plans or evaluate policy. So, we should have a mandatory class somewhere in our education system where children learn what these common mistakes are, why they are mistakes, and how to avoid them.

The next presenter on Enlightenment 2.0, John Allen Paulos, is a mathematician who went through a number of mistakes that people generally make concerning mathematics and probability. He focused specifically on mistakes that people make in arguments for the existence of God. Yet, these are general mistakes that generally interfere in an agent’s ability to make successful plans and evaluate policy.

The Odds of Tokens and Types

For example, there is the mistake of confusing the probability that some token event will occur versus the possibility that some type of event will occur. For example, a person is dealt 13 cards from a deck of cards. Whatever set of cards he gets, the odds of getting that particular combination of cards is extremely small. Yet, he was going to get some combination of cards – every one of which has the same minuscule chance of being the combination he actually receives.

Intelligent design proponents like to use the argument that the odds of a particular combination of events happening is so remote that there must be an intelligent designer at work to make it happen. Yet, this is like arguing that receiving a particular combination of cards is so remote that an intelligent dealer must have purposefully picked out those cards.

In another example of the same type of reasoning, I would like the reader to consider the odds that he or she would actually be here, in the universe, today given the amazing set of events that had to have happened in the past. Let’s just go back 2000 years. The chance meetings, births, deaths, pregnancies (planned and unplanned), the events that determined a person’s personality that determined with whom or whether they had sex with – the odds of my being here today are vanishingly small. I cannot even imagine how small those odds were. And this is just going back 2000 years. If we go back 200 million years, the odds become even smaller.

Yet, here I am.

If I were to follow the reasoning of the intelligent design theorists, the minuscule odds of my being here implies that there must have been an intelligence at work engineering everything that happened up to the point at which I was born, just to overcome those amazingly small odds. Everything else that happened in history – the wars, the plagues, the conquests, slavery, travel, everything – were merely side effects of this master plan to bring us into existence. Otherwise, the odds being what they were, we would not be here.

That’s absurd. Given the way things were 2000 years ago, it was certain that somebody would exist today, just as the bridge player was going to receive some combination of cards. The fact that I am one of those people – or that the hand in question was one of those hands – is not all that remarkable.

Conjunction Bias

Which option is more likely? Is it more likely that A will happen, or that A and B will happen? For example, if you were rolling a die, which is more likely? (1) that the first roll will be a ‘6’, or (B) that the first roll will be a ‘6’ and the second row will be a ‘6’? The first has to be more likely than the second. That much is obvious.

However, there is a set of experiments that show that if a story is told a particular way, a substantial portion of the people will say that ‘A and B’ is more likely than A alone. These experiments involve a story about a person (e.g., Linda). After hearing the story, the listener is asked, “Which is more common? (A) Linda became a college professor, or (B) Linda became a college professor and a political activist? Under certain telling of the story, people (mistakenly) choose B.

But option B is absurd. B can, at best, be equal to A.

Confirmation Bias

Paulos mentioned confirmation bias quickly, because it is a widely understood and recognized bias. However, in his brief mention he linked confirmation bias to the Gulf War – the bias that allowed an administration to see evidence it wanted to see and dismiss evidence suggesting conclusions it did not like.

This bias has cost the United States 4,000 lives, will ultimately cost over a trillion dollars, has killed hundreds of thousands of people in Iraq and maimed countless more, did huge damage to the country in terms of a generation of children who will see no school or training to do any sophisticated work other than the manufacture of bombs.

All of this could have been avoided, perhaps, if our education system had decided to devote a few resources to classes that taught the American population about the traps that people get into and how to avoid them. The people would have had a better understanding of the types of pitfalls that might get us into an unnecessary expense of lives and money. The Bush Administration itself might have had a few more people who understood how to read evidence and to protect it from the chance that they were misreading it. Or (what is more likely) the Bush Administration would have continued to push its favorite interpretation on the American people who would have been better equipped at detecting their deception.

If the school system is supposed to be providing us with information that would help us make sound decisions, and information on the type of epistemic mistakes we are likely to make will help protect us from harm, then it follows that these premises that we should be giving our children knowledge about the types of mistakes we are likely to make.

So, there certainly must be room in the educational system for a class on cognitive errors. We can pay for it out of the future lives and monies saved by not going along with the next avoidable bloody war.

No comments: