“In 1954, I was drafted as a second lieutenant, and after an eventful year as a platoon leader I was transferred to the Psychology branch of the Israel Defense Forces. There, one of my occasional duties was to participate in the assessment of candidates for officer training. We used methods that had been developed by the British Army in the Second World War. One test involved a leaderless group challenge, in which eight candidates, with all insignia of rank removed and only numbers to identify them, were asked to lift a telephone pole from the ground and were then led to an obstacle, such as a 2.5-meter wall, where they were told to get to the other side of the wall without the pole touching either the ground or the wall, and without any of them touching the wall. If one of these things happened, they had to declare it and start again. Two of us would watch the exercise, which often took half an hour or more. We were looking for manifestations of the candidates' characters, and we saw plenty: true leaders, loyal followers, empty boasters, wimps - there were all kinds. Under the stress of the event, we felt, the soldiers' true nature would reveal itself, and we would be able to tell who would be a good leader and who would not. But the trouble was that, in fact, we could not tell. Every month or so we had a "statistics day," during which we would get feedback from the officer-training school, indicating the accuracy of our ratings of candidates' potential. The story was always the same: our ability to predict performance at the school was negligible. But the next day, there would be another batch of candidates to be taken to the obstacle field, where we would face them with the wall and see their true natures revealed. I was so impressed by the complete lack of connection between the statistical information and the compelling experience of insight that I coined a term for it: "the illusion of validity." Almost twenty years later, this term made it into the technical literature (Kahneman and Tversky, 1973). It was the first cognitive illusion I discovered.
Closely related to the illusion of validity was another feature of our discussions about the candidates we observed: our willingness to make extreme predictions about their future performance on the basis of a small sample of behavior. In fact, the issue of willingness did not arise, because we did not really distinguish predictions from observations. The soldier who took over when the group was in trouble and led the team over the wall was a leader at that moment, and if we asked ourselves how he would perform in officer-training, or on the battlefield, the best bet was simply that he would be as good a leader then as he was now. Any other prediction seemed inconsistent with the evidence. As I understood clearly only when I taught statistics some years later, the idea that predictions should be less extreme than the information on which they are based is deeply counterintuitive.
The theme of intuitive prediction came up again, when I was given the major assignment for my service in the Unit: to develop a method for interviewing all combat-unit recruits, in order to screen the unfit and help allocate soldiers to specific duties. An interviewing system was already in place, administered by a small cadre of interviewers, mostly young women, themselves recent graduates from good high schools, who had been selected for their outstanding performance in psychometric tests and for their interest in psychology. The interviewers were instructed to form a general impression of a recruit and then to provide some global ratings of how well the recruit was expected to perform in a combat unit. Here again, the statistics of validity were dismal. The interviewers' ratings did not predict with substantial accuracy any of the criteria in which we were interested.
My assignment involved two tasks: first, to figure out whether there were personality dimensions that mattered more in some combat jobs than in others, and then to develop interviewing guidelines that would identify those dimensions. To perform the first task, I visited units of infantry, artillery, armor, and others, and collected global evaluations of the performance of the soldiers in each unit, as well as ratings on several personality dimensions. It was a hopeless task, but I didn't realize that then. Instead, spending weeks and months on complex analyses using a manual Monroe calculator with a rather iffy handle, I invented a statistical technique for the analysis of multi-attribute heteroscedastic data, which I used to produce a complex description of the psychological requirements of the various units. I was capitalizing on chance, but the technique had enough charm for one of my graduate-school teachers, the eminent personnel psychologist Edwin Ghiselli, to write it up in what became my first published article. This was the beginning of a lifelong interest in the statistics of prediction and description.

I had devised personality profiles for a criterion measure, and now I needed to propose a predictive interview. The year was 1955, just after the publication of "Clinical versus statistical prediction" (Meehl, 1954), Paul Meehl's classic book in which he showed that clinical prediction was consistently inferior to actuarial prediction. Someone must have given me the book to read, and it certainly had a big effect on me. I developed a structured interview schedule with a set of questions about various aspects of civilian life, which the interviewers were to use to generate ratings about six different aspects of personality (including, I remember, such things as "masculine pride" and "sense of obligation"). Soon I had a near-mutiny on my hands. The cadre of interviewers, who had taken pride in the exercise of their clinical skills, felt that they were being reduced to unthinking robots, and my confident declarations -"Just make sure that you are reliable, and leave validity to me"-did not satisfy them. So I gave in. I told them that after completing "my" six ratings as instructed, they were free to exercise their clinical judgment by generating a global evaluation of the recruit's potential in any way they pleased. A few months later, we obtained our first validity data, using ratings of the recruits' performance as a criterion. Validity was much higher than it had been. My recollection is that we achieved correlations of close to .30, in contrast to about .10 with the previous methods. The most instructive finding was that the interviewers' global evaluation, produced at the end of a structured interview, was by far the most predictive of all the ratings they made. Trying to be reliable had made them valid. The puzzles with which I struggled at that time were the seed of the paper on the psychology of intuitive prediction that Amos Tversky and I published much later.
The interview system has remained in use, with little modification, for many decades. And if it appears odd that a twenty-one-year-old lieutenant would be asked to set up an interviewing system for an army, one should remember that the state of Israel and its institutions were only seven years old at the time, that improvisation was the norm, and that professionalism did not exist. My immediate supervisor was a man with brilliant analytical skills, who had trained in chemistry but was entirely self-taught in statistics and psychology. And with a B.A. in the appropriate field, I was the best-trained professional psychologist in the military.” http://www.nobelprize.org/nobel_prizes/economic-sciences/laureates/2002/kahneman-bio.html I've always been under the understanding that the trade-off for better performance of punishment is a decrease of physical health of the person, creating a shorter long-term benefit. Punishment induces stress, stress is accompanied by a prolonged increase of cortisol in the body. Effects of prolonged cortisol levels are. * Anxiety * Depression * Digestive problems * Heart disease * Sleep problems * Weight gain * Memory and concentration impairment
On the preconscious level, we pay more attention to negative stimuli than to positive stimuli.
Negative information is processed more thoroughly than positive information. This can be demonstrated even on the level of neural activity.
In terms of impression formation, negative information by far outweighs positive information (telling one lie can make you a “liar” forever).
Bad memories are engraved deeper in our brains and can be retrieved more easily.
Losing a certain amount of money feels worse than winning the same amount of money feels good. Basically, that´s what Kahneman and Tversky got their Nobel prize for in economics in 2002.
Bad events in our lives have a stronger and longer-lasting effect than good events. This is nicely demonstrated by the fact that we do have word for the consequences of very very bad events (trauma), but there´s no corresponding term for the positive side of the emotional continuum.
Negative feedback has a stronger and longer-lasting effect on us than positive feedback.
Therefore, we put a lot more emphasis on avoiding negative information pertaining to ourselves than focusing on integrating positive information.
In close relationships, one bad event can ruin everything. Yet, a lot of positive events cannot save a relationship “forever”.
Bad parenting has a stronger negative effect on the development of the children than good parenting has on positive development.
This list could go on forever. And: there´s hardly any exception to be found.

http://www.carlsonmba.umn.edu/Assets/71516.pdf http://mappalicious.com/tag/daniel-kahneman/ J. A. Gray