PSYCHOLOGY How Do Scientists Think? David Lagnado ave you ever wondered what goes on second half of the book, Feist charts this proinside scientists’ heads when they gression with originality and insight. His formulate a grand theory? Or when speculations on the origins of scientific they decide what hypothesis to test? How thinking are particularly impressive and draw does this differ from the munwell on recent cognitive psyThe Psychology of dane reasoning involved when chology. He identifies several Science and the Origins you explain why your car core components of thought— of the Scientific Mind won’t start or choose a birthobservation, categorization, day present for a relative? pattern recognition, hypotheby Gregory J. Feist More generally, do scientists sis testing, and causal thinkYale University Press, New use the same cognitive meching—and argues that these Haven, CT, 2006. 336 pp. $38, anisms available to us all were progressively augmented £25. ISBN 0-300-11074-X. (supplemented with formal, as scientific thinking passed conceptual, and technological from the preverbal stage tools)? Or does scientific thinking require through to the explicit research we have more specialized cognitive abilities, avail- today. Critical developments along the way able to only a talented few? included explanatory thinking (greatly aided If you are interested in such questions, by the advent of language), measurement, then Gregory Feist’s The Psychology of mathematics, and finally the hallmark of Science and the Origins of the Scientific modern science, the experimental method. Mind is the book to read. As the title sugThis account is well argued and innovagests, Feist (a psychologist at the University tive, but more could be made of the dynamic of California, Davis) argues the case for a interplay between the key components. For new discipline of “psychology of science” example, both observation and categorization and explores the evolutionary and historical are hypothesis-driven (1) and can be influroots of scientific thinking. The first half of enced by prior causal thinking (2). This imthe book gives a brief history of three domi- plies that these components co-develop rather nant areas in which science itself has been than arise in an incremental fashion. Further the object of study (history, philosophy, and support for such co-development is prosociology of science) and reviews a wealth vided by the recent of research implicitly engaged in the psy- emphasis in cognitive chology of science. This research is divided neuroscience on actionalong traditional lines (biological, develop- based representations mental, cognitive, personality, and social (3). Thus it appears psychologies), and Feist makes a convincing that our internal modcase for their inclusion in the new discipline. els of the world are However, his survey lacks an overarching heavily shaped by the framework and reads more as an assortment demands of effective from subordinate disciplines. (The desired action. Indeed, “motor unification is not helped by the traditional cognition” could be divisions already in place.) If we are envi- added as a key comsioning a new discipline, now is a great time ponent in the preto rethink the classic taxonomy—if not to verbal stage of scienreplace it, at least to give it a sound and log- tific thought. ical explanation. Notably absent from the book are any What of the origins and precursors of sci- discussions of the formal or normative modentific thought? How did we move from pre- els that scientists (or everyday reasoners) literate hunter-gatherers who eat their meat ought to use and how these models relate to raw to sophisticated reasoners with a taste descriptive models of scientific reasoning. for relativity theory and fine cuisine? In the Although it is common to distinguish how people actually reason (descriptive) from how an ideally rational person would reason The reviewer is in the Department of Psychology, Uni(normative), both play crucial roles in versity College London, Gower Street, London WC1E 6BT, current psychological research. Normative UK. E-mail: d.lagnado@ucl.ac.uk H 1390 8 SEPTEMBER 2006 VOL 313 SCIENCE Published by AAAS models serve both as standards against which to appraise human performance and as a framework for understanding cognition (4, 5). For example, there is a growing movement in cognitive psychology and neuroscience that advances a Bayesian perspective on the mind (6). Indeed, one of the appeals of causal maps (which are discussed by Feist in his chapter on cognitive psychology) is that they are formally well defined and normative (7). The question of whether people use fully fledged causal graphs (and Bayesian methods), or instead use simplifying heuristics that approximate these norms, is contentious. But there is little doubt that formal models are critical to the development of cognitive models. Moreover, the psychology of science has a special stake in these issues, because the status of normative models is itself keenly debated in current philosophy of science. Another topic of concern is Feist’s attempt to prescribe guidelines for recognizing scientific talent (and its consequences for education and selection policies). He makes much of correlational studies that allow predictions of scientific achievement from intelligence and personality tests and demographics. Such an emphasis is worrying for two reasons: First, there are wellknown problems with using correlational studies as a basis for policy interventions. Correlation does not imply causation, and these studies may include all kinds of confounding factors. Second, even if the predictors are valid precursors for the prototypical scientist, would we really want to risk excluding less stereotypical thinkers? Einstein would have fared pretty poorly in terms of early college achievements. Lastly, there is a hint of paradox in introducing a new discipline to bridge the gap between related disciplines. Once the new discipline is established (complete with specialized conferences and journals), it runs the risk of reducing rather than increasing cross-disciplinary talk. There are now three independent groups that need to share information rather than two, so new bridges must be built, and so on. In the case of the psychology of science, this is not just a theoretical worry. The subdisciplines of psychology already suffer a lack of integration and cross-fertilization; adding another discipline (however much its content spans the divide) might simply add to the problem. www.sciencemag.org CREDIT: DAVID RIDLEY/IMAGES.COM/CORBIS BOOKS ET AL. BOOKS ETAL. The barbarism of specialization looms anew. In spite of these worries, The Psychology of Science and the Origins of the Scientific Mind succeeds on many levels. Feist pulls together a vast range of psychological research with clarity and insight, and he advances an intriguing framework for the cognitive origins of scientific thinking. The book makes a strong case for an integrated study of the psychology of science. about the workings of nature Big Bang. Or it might with interested nonscientists. be an in-depth examinaWe should therefore weltion of new and exciting come books like Marcus developments in a parChown’s The Quantum Zoo: ticular discipline. Or, A Tourist’s Guide to the Neverfor that matter, it might ending Universe. Chown’s work just involve bringing a is an admirable attempt to storyteller’s eye and a delve into the mysteries of gift for narrative to illuthese two great theories, quanminate a forbidding comtum mechanics and relativity, plex of ideas. References and express them in terms Unfortunately, The 1. R. L. Gregory, Eye and Brain: The Psychology of Seeing that an intellectually curious Quantum Zoo isn’t really (Oxford Univ. Press, Oxford, ed. 5, 1998). nonexpert can understand. And distinguished in any of 2. M. Jeannerod, Motor Cognition: What Actions Tell to the Self (Oxford Univ. Press, Oxford, 2006). for the most part the book those ways. Chown is a 3. S. A. Sloman, Causal Models: How People Think About the succeeds. Chown (a science fine explainer, but he World and Its Alternatives (Oxford Univ. Press, Oxford, writer who trained as a physidoesn’t take us over any 2005). cist) has a pleasant writing ground that others haven’t 4. J. R. Anderson, The Adaptive Character of Thought (Erlbaum, Hillsdale, NJ, 1990). style and a facility with simtrod before. For exam5. D. Marr, Vision: A Computational Investigation into the ple metaphors and analogies ple, after a good explaHuman Representation and Processing of Visual that helps bring difficult con- Tunneling site. Proton tunelling nation of bosons and Information (Freeman, San Francisco, 1982). 6. N. Chater, J. B. Tenenbaum, A. Yuille, Trends Cognit. Sci. cepts into sharp focus. allows hydrogen fusion in the Sun to fermions takes us up to 10, 287 (2006). The book is divided into occur “even at the ultralow tempera- the connection between 7. J. Pearl, Causality: Models, Reasoning, and Inference two sections: “Small Things” ture of 15 million degrees.” spin and statistics, Chown (Cambridge Univ. Press, Cambridge, 2000). and “Big Things.” In the forsimply admits that this 10.1126/1132467 mer, as you might guess, he covers the quan- “brings us to the end of what can easily be tum world, explaining the crucial ideas of conveyed without opaque mathematics.” PHYSICS superposition and interference, and braving Later, after foreshadowing about how superdifficult topics such as the uncertainty prin- fluid helium can do strange things like crawl ciple, entanglement, and the collapse of the up the sides of a container, the book never wave function. Chown moves easily from actually explains why that happens. Dark historical examples such as Young’s double- energy is not explained any more deeply slit experiment and Rutherford’s scattering than “the repulsive force of empty space.” to modern issues such as quantum comput- After whetting our appetites for more subSean M. Carroll ers and teleportation. In the second section, stantive explanations, we are left feeling a here’s no reason why devoted to relativity, he little unsatisfied. everyone shouldn’t underswiftly covers the basics The primary shortcoming of the book The Quantum Zoo stand the basics of quanof spacetime and relativ- seems to be the lack of some specific point tum mechanics and relativity. ity, gravitation, and cosmo- to the project. The subtitle, A Tourist’s Guide A Tourist’s Guide to the These two cornerstones of 20thlogy. The appropriate hot to the Neverending Universe, gives an indiNeverending Universe century physics have become a topics are mentioned, if cation of the unfocused nature of the text. Marcus Chown basis for our deepest understandbriefly: black holes, string I suspect that Chown could have written an Joseph Henry Press ing of reality, as well as of great theory, inflation, and dark interesting and useful book about quantum (National Academies practical importance to familiar energy. The brevity of the mechanics, starting with the basics and Press), Washington, DC, technologies from lasers to the text is not a shortcoming; going into some detail about modern devel2006. 212 pp. $24.95, C$27.95. global positioning system. And, not every popular book opments in atomic and molecular physics, ISBN 0-309-09622-7. despite their reputations for needs to be a massive and quantum information theory, and quantum being somewhat abstruse and comprehensive tome. The computation. Or, alternatively, an interestinaccessible, the basic points of each theory popular audience at which The Quantum Zoo ing and useful book about relativity, concan be stated simply enough that an inter- is aimed should learn a lot from reading the centrating on some specific aspect such as ested person with no technical background book and enjoy themselves in the process. gravitational waves, black holes, or dark in physics should be able to understand And yet, there is a sense in which the energy. Instead, Chown’s book is compethem. At a time when science seems both book is a disappointment. There are other tent but uninspiring, a somewhat superfimore central than ever and more removed books out there, after all, that deal with the cial look at the foundational theories of from our everyday world, it is certainly topics of quantum mechanics and relativity. modern physics. The explanations are worth the effort to share what we’ve learned To stand out from the crowd, any new entry clear, and the interested reader will be able should have something distinct to offer. It to learn quite a lot. But there is not quite might be the unique insight of a true master any reason to choose The Quantum Zoo The reviewer is at the Physics Department, California of the field, as we find when Richard from among the other titles on the popularInstitute of Technology 452-48, 1200 E. California Blvd, Feynman writes about quantum electrody- science shelf. Pasadena, CA 91125, USA. Web site: http://preposternamics or George Gamow writes about the ousuniverse.com 10.1126/1130369 The Universe, Too Quickly Toured CREDIT: PHOTOS.COM T www.sciencemag.org SCIENCE VOL 313 Published by AAAS 8 SEPTEMBER 2006 1391