Book Review: Thinking Fast and Slow Part 3
An uncomfortable realization of human nature which came to light as part of an experiment carried out by Richard Nesbitt and Eugene Borgia at the University of Michigan. The experiment starts by telling students about a previous study, whereby participants in an experiment were led to individual booths and invited to speak over an intercom about their personal lives and problems. There was a 2 minute time limit per person. Only one microphone was active at any one time. There was 6 participants in each group and one was a plant (fake) participant. This plant participant spoke first (with a preprepared and planned script given to them by the experiment conductors). They describe issues adjusting to a new city and embarrassment that they are prone to seizures especially when stressed. All other participants then had a turn. When the microphone again was with the plant participant, the plant faked a seizure coming on, asking for help, stuttering their words and stuttering out they thought they were going to die. The microphone then cuts out.
Out of the 3 groups, with 15 genuine participants, only 4 out of the 15 got out of their booth.
5 more got out but would have been well after the seizure victim more than likely choked.
The experiment was to prove that individuals feel relieved of responsibility when they know others have heard the same request for help.
There is some potent feature (perhaps the diffusion of responsibility) that induces normal decent people to behave in surprisingly unhelpful ways in these circumstances.
2. Nesbitt and Borgia wanted to find out, could the students belief about human nature which had just been proven to them, change?They showed students two videos, short and bland of two seemingly nice, normal, decent people. They described their hobbies and plans for the future etc. After watching a video students are asked how quickly that person had come to the aid of the stricken stranger.
Looking at the base theory set by the original experiment. We’ve been told only 4 out of 15 people rushed to help at the first request. The probability on this basis, of a participant being immediately helpful to the stricken stranger is 27%.
This should lead one to believe any unspecified participant likely did not rush to help.
Baysean logic requires adjustment to any new relevant information.
In this study the videos were purposefully designed to be uninformative so there was no reason to suggest that that these individuals would be either more or less helpful. There were two groups in this second experiment.
Group 1 were told only about the parameters of the first experiment but not the results. This group predicted that both individuals in the second experiment would rush to help.
Group 2 knew the parameters of the first experiment and the results. They also predicted that both individuals in the second experiment would rush to help. They knew the probability was 27% yet remained convinced that the people they saw in the video would rush to help.
3. Finally, they took a new group of students and showed them the parameters of the first experiment but not the results and also told them the new two individuals they had just seen had not helped the stricken stranger and asked them to guess the results/outcomes.
The students guesses were extremely accurate.
The study shows that people who are taught surprising statistical facts about human nature may be shocked and surprised, but this does not mean that their understanding of the world has really changed.
But when they were surprised by individual cases - two nice people who hadn’t helped - they immediately made the generalization and inferred that helping is more difficult then they had thought.
4. There is a deep gap in our thinking of statistics and our thinking about individual cases.
Even compelling causal statistics will not change our long held beliefs rooted in personal experience.
You are more likely to learn something by finding surprises in your own behavior, than by hearing cold hard facts about people.
5. Daniel Kahneman has two “favorite” equations:
Success + Talent = Luck
Great Success = A little more talent + A lot of luck
6. Regression to the mean
The book outlines an important principle of skill training: rewards for improved performance work better than punishment of mistakes.
Say for example the author was explaining this concept to a group of rally drivers and their trainers.
The trainer disagrees with this concept of rewards over punishment for improved performance because they insist that in their experience, rally drivers that pull off an extremely difficult maneuver in training, when they are praised by their trainers, the next time they try the same move they usually do worse.
In contrast, poor executed maneuvers where trainers scream at the rally driver to try harder and do better, often do better next time.
Daniel Kahneman describes this as a joyous moment of insight for him, because while the trainer was astute and not wrong, the inference he had drawn from reward and punishment was wrong!
Instead, what he had described was regression to the mean.
The rally driver who cleanly executed was probably more lucky than usual in that maneuver, which would explain why the next time he tried it he was more likely to deteriorate. Similarly, the rally driver whose performance was bad was more than likely going to improve on the next attempt!
7. This idea is seen when the author applies to a 2 day golf tournament.
The scenario assumes that on both days the average score of the competitors was par 72.
A player who did very well on the first day closed with a score of 66.
A immediate inference is that the golfer is more talented than the average participant in the tournament.
The expectation that the talented golfer will retain the same level of talent on the second day, and so will the others. This doesn’t however, account for luck on the second (or any) day.
Your best guess is that it will be an average level of luck, neither good nor bad and more than likely not a repeat of their luck on Day 1. This is the most one can say and we expect the difference between the two golfers to shrink on the second day.
The golfer who did well on Day 1 is likely to be successful on Day 2 as well but less than on the first because the unusual luck they enjoyed on Day 2 is unlikely to hold.
The golfer that did poorly on Day 1 will probably be below average on Day 2 but will more than likely improve because their probable steak of bad luck on Day 1 will more than likely not continue.
Best predicted performance on Day 2 is closer to the average than on the evidence of which it is based (Day 1). This is regression to the mean.
The broader notion is that regression occurs when the correlation between two measures is less than perfect.
8.Whenever the correlation between two scores is imperfect, there will be regression to the mean.
9. Lets look at this interesting statement that the book outlines:
“Highly intelligent women tend to marry men who are less intelligent than they are.”
Most people even with experience with statistics will look for a causal link to this statement things like highly intelligent women want to avoid the competition of highly intelligent men.
10. Next, look at this statement:
“The correlation between the intelligence of spouses is less than perfect.”
Its the same statement! However one is drastically more interesting and conversation starting than the other. Nevertheless they are algebraically equivalent.
If the correlation of between the intelligence of spouses is less than perfect, and on average men and women differ in intelligence, then it is a mathematical inevitability that highly intelligent women will be married to husbands who are on average less intelligent than them (and vice versa of course).
Again we see how our minds are strongly biased to causal explanations rather than mere statistics.