Heuristics can be ecologically rational Heuristics: The Foundations of Adaptive Behavior edited by Gerd Gigerenzer, Ralph Hertwig and Thorsten Pachur. Oxford University Press, 2011. $99.95 (hbk) (844 pp.) ISBN 13: 9780199744282 Hal R. Arkes Department of Psychology, Ohio State University, 240N Lazenby Hall, 1827 Neil Avenue, Columbus, OH 43210-1234, USA Unlike many edited ‘greatest hits’ volumes, this book’s chapters explore exactly one viewpoint: the fast-and-frugal heuristics research program of the Adaptive Behavior and Cognition (ABC) Research Group at the Max Planck Institute for Human De- velopment in Berlin. The studies emanat- ing from this research program suggest that very simple cognitive heuristics com- prise highly accurate inference strategies. These primitive heuristics enable humans to perform cog- nitive tasks both with great speed and with a truncated information search – hence the name of the research pro- gram ‘fast-and-frugal’. If you want to become familiar with this important research and the seminal publications that comprise it, there is no better sourcebook to own. The seminal article in this research program is the 1996 Psychological Review article by Gigerenzer and Goldstein [1], which comprises one of the earliest chapters in this volume. In this article and in the immediately subsequent publications, the ABC Group provided evidence for several counterintuitive findings. First, humans who use very simple algorithms that ignore much of the available infor- mation can nevertheless make inferences more accurately than far more sophisticated techniques such as multiple regression. Second, the success of the fast-and-frugal heur- istics appears to violate the well-known speed–accuracy trade-off, according to which heightened accuracy can be achieved only by moderating one’s processing speed. Giger- enzer and colleagues showed that the simple fast-and- frugal algorithms can be fairly accurate despite their speed. Their swiftness is attributed to the fact that they do not process much of the available information in the environment. Third, because the algorithms do not process much of the available information, their surprisingly high performance on inference tasks necessarily violates some of the tenets of rationality. Readers of this volume who are not familiar with the ABC research agenda may be surprised as they confront some of the remarkable conclusions in the various chap- ters. For example, according to the recognition heuristic, people may make highly accurate inferences based on their mere recognition of only one of the available options. The use of such a heuristic implies that people who can recog- nize all of the options will not be able to utilize this effective strategy to make choices, because its use requires that some alternatives not be recognized. As a result, persons who know less, that is who recognize fewer options, may perform better than those who recognize more options. This is the paradoxical ‘less-is-more effect’. The primitive recognition heuristic has been shown to perform remark- ably well in such diverse tasks as predicting the outcome of the Wimbledon tennis tournament and gauging the rela- tive size of cities. How can the recognition heuristic and other simple algo- rithms do so well in competition with more sophisticated strategies? Beginning with Chapter 1 by Gigerenzer and Brighton, several of the articles make the point that al- though more complicated strategies with multiple param- eters may fit prior data fairly well, such models capture noise in the data. Thus they do less well in predicting new outcomes when cross-validation is assessed. The simple heuristics may not fit the prior data as well as the complex strategies, but they perform just as well or even better on cross-validation. In other words, they suffer less loss in going from postdiction to prediction. Martignon and Hoffrage summarize this best in their chapter: ‘In prediction a robust strategy rather than an optimal one is required’ (p. 270). One of the intellectual pleasures of this book is the expansion of the fast-and-frugal heuristic program to a wide variety of domains. For example, authors of the latter chapters demonstrate the applicability of heuristics to setting bail, diagnosing cardiac problems, catching Fris- bees, and foiling burglars. In addition to choosing articles that demonstrate the applicability of heuristics, the editors have included articles that promote consideration of the deeper implications of this research program. The landmark research of Tversky and Kahneman [2] characterized heuristics as quick-and-dirty strategies that usually provide a good approximation to the correct answer but occasionally lead decision makers astray. The fast-and-frugal research program casts heuristics in a different light: heuristics are viewed as ecologically rational adaptations to the requirements of the environment. Giger- enzer and colleagues use Simon’s [3] metaphor of a pair of scissors whose two blades represent, respectively, the struc- ture of the task environment and the computational capa- bilities of the actor. In some situations a particular computational heuristic is most appropriate for a person to function effectively. In another setting a different heuris- tic is needed for the scissors to ‘cut’ properly. This viewpoint renders moot the distinction between the rational and the psychological; heuristics are not seen as subrational but instead are viewed as adaptations that evolution has de- vised to fit various information-processing demands of the environment. The evolution of the fast-and-frugal research program itself is covered in some of the chapters, many of whose authors are erstwhile or current skeptics of the program. Book Review Corresponding author: Arkes, H.R. (arkes.1@osu.edu). 260