Heuristics and Biases: The Psychology of Intuitive Judgment 1st Edition

In 1982, Daniel Kahneman and Amos Tversky edited a volume, "Judgment under Uncertainty." This served as a culmination of their and others' research, bringing together in one volume a large number of reports on how humans make decisions under conditions of uncertainty. In short, they contended that under such conditions, people tend to use heuristics or decision-making shortcuts. This can lead to suboptimal decision-making.

Since, much research has built upon the earlier works. Indeed, there are now two streams in the research on heuristics--one fairly optimistic, exemplified by works of scholars such as Gerd Gigerenzer, and the other more pessimistic, exemplified by this particular volume, edited by Gilovich, Griffin, and Kahneman.

The introduction sets the stage for the myriad essays making up this book. The editors note in the Preface that (page xv): "The core idea of the heuristics and biases program is that judgment under uncertainty is often based on a limited small number of simplifying heuristics rather than more formal and extensive algorithmic processing. These heuristics typically yield accurate judgments but can lead to systematic error." The Introduction itself provides an historical overview of this line of work and notes some of the critiques of this body of research.

The individual essays themselves note some of the basic heuristics (or decision-making shortcuts). To illustrate: representativeness. Here, one takes a small number of cases and generalizes from these. E.g., oh, I knew a couple college basketball players and they were pretty dumb. Hence, one then generalizes and concludes that all basketball players are not so smart. In short, one generalizes from a poor sample. This is one of the roots of stereotyping, which can lead to all manner of mischief.

What is at stake with the study of heuristics and biases? These raise real questions about the common assumption that humans behave rationally, using something like a cost-benefit calculus to make decisions. This has profound implications. Much policy is based on people behaving rationally. If that assumption is wrong, then government decisions based on a flawed view of humans' decision-making isn't likely to have the desired effects.

Part Two explores new theoretical directions. One of the pluses of this volume is that it includes works by those who see heuristics as positive. For instance, an essay by Gigerenzer and colleagues makes the point that heuristics may do better as a source of decision-making than even statistical predictions.

Part Three looks at real world applications, from "the hot hand in basketball" to an evaluation of clinical judgments to political decisions.

In short, this volume covers a lot of territory. The work is not meant for Joe Six Pack. It is written by academics and may be a bit dense for some readers. But there is much at stake with the research program described in this volume. I think that many people would find the struggle to understand the arguments here as worthwhile.

I highly recommend this work.

I based a whole module of my PhD in management on this book. The heuristics and biases described in this book can be turned into power conversations in management to affect decisions in real world situations. With my understanding of these concepts, I can now easily detect them in management discussions, which allows me the opportunity to clarify, amplify, and simplify decision making situations.

This collection of articles has its origin in the work of one of the editors (Daniel Kahneman) and Amos Tversky (now deceased) in the 1970's. The first article in the book gives an introduction to this work and a brief historical survey. This work, along with current developments, is extremely important, for it sheds light on the differences (if any) between "intuitive judgment" and judgment that is based on more quantitative, mathematical, or algorithmic reasoning. If human judgment in uncertain environments is based on a limited number of simplifying heuristics, and not on extensive algorithmic processing, this would be very important for someone who is attempting to implement or simulate human reasoning in a machine. Economics, finance, and political decision-making are other areas that need a more accurate view of human judgment. Indeed, the "rational agent" assumption in classical economics, wherein the person makes choices by assessing the probability of each possible outcome and then assigning a utility to each, is considered to be fundamental, even axiomatic. It is therefore of great interest to examine challenges to this assumption.

In order to test the rational agent assumption, experiments must be conducted to test whether indeed the human assessment of likelihood and risk does indeed conform to the laws of probability. The data obtained in these experiments must then be judged as to whether it can be used to decide between the rational agent model and models of human judgment that are based on "intuition" (however vaguely or mystically this latter term is defined).

The authors of the first article in this book discuss some of the work on these questions, in particular the research that involved comparing expert clinical prediction with actuarial methods. The latter were found to perform better than the former. Even more interesting is that the clinician's assessments of their abilities were very far from what the record of success actually indicated. Some research has also indicated that intuitive judgments of likelihood do not correspond to what is obtained by Bayesian reasoning patterns.

These results, as the authors discuss, motivated performance models that were not based on the assumption of full rationality, but rather on what is called `bounded rationality.' The developers of this model felt that the processing limitations of the human brain dictated that humans must choose very limited heuristics when engaged in decision-making.

Also of great interest, and discussed in another article in the book, is the human ability to engage in affective forecasting. The latter involves the making of decisions based on the predictions of the emotional consequences of future events. The authors study the accuracy of affective forecasting and the accompanying notion of `durability bias.' The latter notion arises when individuals attempt to estimate how long particular feelings will last, and this estimation seems to be considerably longer than what actually occurs. The authors discuss some of the reasons for the durability bias in affective forecasting. One of these is ordinary misconstrual, where events are thought to be more powerful than what are actually realized, resulting in the overestimation of the duration of the affective responses to these events. Another regards the difficulty in forecasting affective reactions to events about which much is known. In addition, the authors point to "defensive pessimism" as to another of the reasons for inaccurate affective forecasting. This allows for mental preparation for the consequences of an event, and for positive feelings when the affective duration is smaller than what had been predicted. The main emphasis of the authors' article though is much more interesting than these explanations, for it involves the notion of a `psychological immune system.' Quoting the research of many psychologists, and arguing in analogy to the ordinary biological immune system, the authors view this system as one that protects the individual from an "overdose of gloom." Further, the functioning of the psychological immune system is optimized when it is not brought into the conscious focus of the individual. This `immune neglect' however has as a consequence the durability bias, in that if an individual fails to recognize her negative affect will decrease and be subjected to psychological mechanisms that assist greatly in this diminution, then she will tend to overestimate the time duration of her emotional reactions. The authors discuss empirical studies of durability bias in their article, and discuss some of the consequences of their studies. One of these concerns the possibility that humans could be mistaken about their own internal experiences. This is a very troubling possibility, but the authors give many references that purport to support it. This research shows that not only can people be completely mistaken about their feelings toward an object, but that their actual behaviors is better evidence of their internal states than what they report verbally.

Another interesting article in the book concerns the topic of automated choice heuristics. This area has arisen as a reaction to the idea that human choice can be predicted using theoretical models of optimal choice. Instead, one must identify the heuristics the people use to simplify their choices. These heuristics are used to restrict or compress the amount of information that is processed by the human brain and also to deal with the complexity in which this information is assimilated. There are many different theories of choice heuristics, and some of these are discussed in the article. Some of these theories involve heuristics that are "deliberate", i.e. involve the elimination of aspects and slower cognitive processes, and some involve heuristics that are "automatic" and judgmental, i.e. that arise from cognitive processes that are rapid and not controllable. Judgmental heuristics is also referred to as `System 1' heuristics in the article, whereas deliberate heuristics is referred to as `System 2' heuristics. The authors give a very interesting overview of automated choice heuristics, involving choices that are based on immediate affective evaluation, and choices that are using the option that is first thought of. All of these discussions, as are all the others in the book, are extremely important.