Journal of Behavioral Decision Making, Vol. 7, 293-297 (1994) Book Reviews ANCHORED NARRATIVES: T H E PSYCHOLOGY O F CRIMINAL EVIDENCE, W. A. Wagenaar, P. J. van Koppen, and H. F. M. Crombag, New York: St Martin’s Press and Hertfordshire: Harvester Wheatsheaf. 1993. (Based on DUBIEUZE ZAKEN: D E PSYCHOLOGIE VAN STRAFRECHTELIJK BEWIJS, H. F. M. Crombag, P. J. van Koppen, and W. A. Wagenaar, Amsterdam: Contact, 1992.) This is an excellent book. It tells the reader a great deal about the actual practice of legal decision making and introduces an original and convincing psychological theory of the trial judge’s decision-making processes. It describes several useful conceptual tools that are well known to judgment and decision-making experts and applies them in a comprehensible manner to concrete legal decision-making situations. Also, it is a good model of how to write a scientific analysis of legal decision making that really connects to actual practice, the best to appear since Kalven and Zeisel’s classic on The American Jury. The anchored narrative model of judicial reasoning is simple: ‘First an assessment is made of the plausibility of the prosecution’s account of what happened and why, and next it is considered whether this narrative account can be anchored by way of evidence to common-sense beliefs which are generally accepted as true most of the time’ (p. 10). On the surface this does not sound like a surprising account of how judges reason. But what is remarkable is the extent to which the outcome of the process is dominated by the narrative; as the authors put it, ‘. . . a good story is better than half of the proof (p. 44). Wagenaar, van Koppen, and Crombag (hereafter ‘the authors’) even conclude that a major obstacle to effective defense in criminal trials is the trial judge’s proclivity to construct good narratives (implying guilt) even when the prosecution’s narrative is incomplete or weak. The power of coherent narratives in legal decision making has been noted by several other theorists and practitioners. Probably the strongest case for its role in juror decision making in American trials has been made by Pennington and Hastie (summarized in Pennington and Hastie, 1991 and 1993a). However, although the basic message about the psychology of decision processes is the same, Anchored Narratives goes beyond earlier contributions in many important directions. First, Anchored Narratives focuses on the behavior of trial judges. There is not much research on trial judges; there are scattered studies of their personalities and attitudes and the relationship of these dispositional variables to their verdicts and there are a few studies of sentencing and bail-setting judgments, but not much 0 1994 by John Wiley & Sons, Ltd. else. Anchored Narratives is the best book ever written from a behavioral science perspective on the behavior of trial judges. The message that judges (like the rest of us) are suckers for a good story is an important corrective to the myth that judges are rational thinkers who are better than ordinary people because they are good a t keeping track of chains of inference and always use ‘safe anchors’ for important conclusions. Second, the authors skillfully develop a method of analyzing decisions in ‘anomalous cases’ that provides an exemplary model for exploratory research in this domain. Incidentally, it is also a good way to attract attention from legal professionals and the lay public (see the authors’ prefatory comments on the initial reception of this book in the Netherlands). The authors identified a corpus of 35 cases in which appellate court reviews and information from attorneys indicated that the trial court verdict was ‘a “dubious” conviction of the defendant’ (p. 11). They carefully studied the records from these cases, especially the justifications for verdicts provided by judges and tribunals in the original trials (required under many conditions according to Dutch law), and extracted information about the decision processes and their possible inputs. One of the great pleasures of reading this book was derived from becoming familiar with the multitude of concrete examples from the 35 anomalous cases (a commentary on the famous ‘Ivan the Terrible’ trial of John Demjanjuk in Israel is a bonus). Third, the authors describe several valuable analytic tools from the judgment and decision-making field in addition to their anchored narratives model. The introductions were clear enough that readers from outside of the source field will be able to grasp the essentials and see their applicability to legal practice. We especially liked the frequent use of Signal Detection Theory-inspired summaries of the conceptual relationships between states of the world (e.g. guilty versus innocent defendant) and judgments (e.g. convict versus acquit). This format allowed the authors to discuss base rate and decision criterion effects in a conceptually clear framework and to quickly analyze the problems caused by troublesome forms of evidence such as polygraph test results, anatomically correct doll tests, and confessions. Fourth, the most original portion of the contribution was the discussion of inferential chains and anchors. The theory will surely be expanded and revised, but the effort to specify the character of the inference chains, the point at which reasoning does (or should) stop, and the nature of ‘safe’ and ‘unsafe’ anchors that are used frequently in legal reasoning is an excellent beginning (other researchers developing the theme of 294 Journal of Behavioral Decision Making Vol. 3, Iss. No. 4 narrative reasoning have not had much to say about on the large multidisciplinary literature studying text this critical aspect of the process; cf. Pennington and comprehension and discourse analysis), and the most Hastie, 1993b). Probably the most intellectually chal- complete account of the factors that affect confidence lenging portion of the book was the discussion of the in a decision based on narrative reasoning. In comparimanner in which entire narrative structures are son, the authors’ more general claim that a story has anchored (‘The Anchoring Construction as a Whole’, ‘a readily identifiable central action; and a context (setpp. 73-81). Here the authors show that compensatory ting) that provides an easy and natural explanation of reasoning (strong argument-anchor chains can com- why the actors behaved in the way they did’ (p. 36) pensate for weak or missing chains) is characteristic seems a bit abstract. This abstractness may also contriof judges’ reasoning, although legal doctrine usually bute to the incompleteness of the authors’ discussion proscribes compensatory proof and demands com- of the relation between the story and general elements pletely satisfied conjunctive support for verdicts. The of legal definitions of crimes (identity, a c ‘ m reus, mens authors also connect the discussion of proof construc- rea) and their discussion of ‘Which story parts need tion to the implications of variations in decision criteria to be anchored?’ (pp. 62ff.). Of course, this may be and in base rates of guilty or innocent defendants. This a matter of style. Our own habit in research is to aim was very good stuff and we hope that it will instigate for detailed, perhaps overly detailed, hypotheses about further analyses by the authors and others interested mental representations - we find commitment to in the theory and practice of legal decision making. detailed hypotheses really moves our research forward. (More analysis from the authors’ good starting point Of course, it may be that premature precision is an is needed: for example, the disconnection (when base error. If so, it is definitely not an error that can be rates vary) between posterior odds ratios (like Black- attributed to the authors of Anchored Narratives. stone’s famous rule, ‘It is better that ten guilty persons When we asked American trial judges to think aloud escape than one innocent suffer’ rule) and outcome uti- while making decisions, we found their thoughts were lities is noted, but the full implications of this relation- saturated with previously encountered cases. In other ship are not worked out; cf. Connolly, 1987; DeKay, words, American judges d o a lot of ‘case-based reason1994). ing’ using analogies to old cases, that they retrieve from Another related contribution is the useful discussion their long-term memories, to decide new cases. ‘Oh, of biases in hypothesis-testing habits, established in this case reminds me of the Miller case last year, and laboratory research, and their manifestation in the he was guilty of manslaughter . . .’ However, there is selection of evidence during pre-trial procedures. The almost no reference to case-based reasoning in the authdescriptions of the general strategies of ‘trawling’ for ors’ analysis of Dutch judges’ decisions. Perhaps the evidence and suspect-driven searches by the police are Dutch judges (like the Americans) also rely on caseilluminating. Again, the concrete discussions of con- based reasoning, but traces of the case-based process fessions, eyewitness identification, expert witnesses were just not visible through the ‘window’ provided including psychiatrists, and other unreliable witnesses by their post-decision written justifications (in a system are especially useful to practitioners and academics that does not use precedent cases as a basis for justificabecause they include so many details from the corpus tion). Or, much more interesting, this may be the basis of anomalous cases. for a fascinating hypothesis that merits further comThis book is excellent, but we can make a few parative research: ‘Code Law’ (Dutch) judges do not remarks about ideas that should be developed more reason in a case-based manner, but ‘Common Law’ fully in the research that we hope will be stimulated (American) judges rely heavily on case-based analogical by its publication. There are now several scholars inter- reasoning. ested in the narrative character of legal decision making One of the most original contributions of this book (e.g., the authors; Pennington and Hastie; Allen, 1991; is the discussion of the relationship between a narrative Amsterdam and Hertz, 1992; Bennett and Feldman, and its support. The image of inference chains is useful 1981; Jackson, 1988; Twining, 1988; and others). How- and the authors’ discussion of when the chain stops ever, the concept of a well-formed story in legal contexts and decision maker says ‘I’m anchored’ is uniquely oriis still murky; there are many open questions. What ginal (in psychology; but see treatments by e.g. Hartare its parts? How is it constructed? How are inferences man, 1986; Wigmore, 1937; and many others in about completeness, coherence, and coverage of the adjacent fields). The catalogue of ‘safe’ and ‘unsafe’ evidence made? The authors advance the analysis of anchors is also doubly valuable because, as in so many the concept of a story over their predecessors, especially parts of this book, the authors provide many concrete with reference to how the story is linked to evidence examples of actual practice and instances from their and role of ‘anchors’ that warrant (in Toulmin’s terms) 35 anomalous cases. In the conclusion of the book, the links in the inference chains. But, as the authors ten ‘Universal Rules of Evidence’ are presented; most note (p. 36), Pennington and Hastie have proposed the of these rules (numbers 3-6, 8, and 9) are admonitions most detailed description of a story structure (based to improve practice when choosing and connecting to Book Reviews anchors. However, the discussion did not make contact with the substantial literature on the inference chains and belief support in legal analysis (e.g. Anderson and Twining, 1991), police investigations (Binder and Bergman, l984), and decision analysis (Edwards, 1991; Edwards et al., 1990).We were especially disappointed that David Schum’s analysis of investigation and proof was not cited and compared to the authors’ approach (Schum, 1986, 1994; Schum and Tillers, 1991); in our view Schum is one of Wigmore’s most brilliant heirs. Perhaps this is an oversight caused by the chasm between American and Continental literatures and because many of the best sources that were not cited are recent publications. Much of the controversy created in the Netherlands by the publication of this research was probably caused by the form of the authors’ own case for their conclusion that judges’ verdicts are often poorly anchored and subverted by good stories. The basic logic supporting this conclusion is indirect. The authors d o not look at the judges’ written justifications and find a lot of storytelling. (Pennington and Hastie and others have found this kind of direct evidence for reliance on stories in jurors’ reports of their decisions.) Rather, starting from a deliberately selected sample of cases with ‘dubious verdicts’ and weak written justifications, the conclusion that ‘Proof by Narrative’ describes the judges’ decision process rests on an indirect argument by elimination. Thejudges d o not appear to be reaching their conclusions via strong chains of logical inferences from sound evidence and valid warrants (anchors), therefore they must be doing something else. Since the authors’ alternative to ‘logic’ is ‘Proof of Narrative’, they conclude that ‘good stories’ must be the major influence on the judges’ decisions. A critic might argue that the authors are exhibiting some of the bad habits of reasoning that they themselves decry in triai judges’ behavior. The authors select an admittedly small and unrepresentative sample of cases (evidence). Their basis for concluding ‘Proof by Narrative’ is indirect and not directly revealed in affirmative evidence. They construct many explanatory interpretations of the actions and thoughts of various actors in crime and courtroom events. Also, they d o not present a complete catalogue of alternative hypotheses about judges’ reasoning processes and systematically summarize confirming and disconfirming evidence for each hypothesis. However, this critical evaluation of the volume is not warranted. First, the argument is much stronger than the gloss we have presented above. Great care is taken to summarize the anomalous case-selection process and to locate the selected cases in a comprehensive map of the population of cases flowing through the Dutch criminal justice system. The most prominent alternative hypotheses are presented and evaluated (although not in a comparative format). The authors are very honest about their comparative inferences 295 from the ‘hard evidence’ that is in hand and they frequently qualify their most interpretive conclusions with the label ‘speculative’. Finally, they treat their argument as a strongprimafacie case for the anchored narratives model, but they acknowledge that proof awaits the outcome of additional research. Early in the volume the authors note that ‘As psychologists we claim to study universal principles of reasoning’ (p. 15). But there is an inevitable tension between the desire to say something that is practically useful in actual courts at a particular moment in history and in a particular legal culture and the goal of saying something general enough to attain the status of a universal principle of human psychology. In our view, the authors have done the right thing, in directing their conclusions at current practices in inquisitional and adversarial law courts. Perhaps because they are a truly interdisciplinary team of psychological and legal talents, they have done a better j o b of making these connections than almost everyone working on the behavioral science and law enterprise today. Anchored Narratives will be understood by intelligent readers from the law and from the science sides of the field and it will improve practice and theory applied to the processing of evidence in legal decision making. If there is a truly universal message about human nature in the book, it is an echo of the most elementary principle of ‘cognitivism’ (from philosophy and psychology). People are designed to find out what is going on around them, the way they d o this is to create models of the important parts of their world inside their heads. In legal settings, these cognitive models take the form of narrative stories summarizing what the decision makers believe happened in significant events such as crimes, quarrels, and disputes. The surprising conclusion is that the thinkers’ cognitive models take on a life of their own, often independent of their anchors via evidence to firm beliefs, and that ‘A good story is better than half of the proof‘. ACKNOWLEDGEMENTS The authors want to thank William Pizzi (Professor of Law, University of Colorado) and Robert J. Hallisey (Superior Court Judge, retired, Massachusetts) for useful advice about judicial decision making. NSF Grants numbered SES 91221 54 and SES 9 1 13479 provided partial support for the authors of this review. REIDHASTIEand NANCY PENNINCTON Psy chologji Department University of Colorado Boulder, Colorado 80309-0344 USA REFERENCES Allen, R. J. ‘The nature of juridical proof‘, Cardo:o Law?Review, 13 (1991), 373-422. 296 Journal of Behavioral Decision Making Vol. 3. Iss. No. 4 Amsterdam, A. G. and Hertz, R. ‘An analysis ofclosing arguments to a jury’, New York Law School Law Review, 37 (1992), 55-98. Anderson, R. J. and Twining, W. Analysis of Evidence: How to do things with facts based on Wigm0re.s ‘Science of Judicial Proof’, Boston: Little, Brown, 1991. Bennett, W. L. and Feldman, M. S. Reconstructing Reality in the Courtroom: Justice andjudgment in American culture, New Brunswick, NJ: Rutgers University Press, 1981. Binder, D. and Bergman, P. Fact Investigation: From hypothesis toproof, St Paul, MN: West, 1984. Connolly, T. ‘Decision theory, reasonable doubt, and the utility of erroneous acquitals’, Law and Human Behavior, 11 (1987), 101-1 12. DeKay, M. L. How Sure is Sure Enough? Outcome Evaluations and Action Thresholds in Binary Decisions, unpublished doctoral dissertation, University of Colorado, Boulder, CO, 1994. Edwards, W. ‘Influence diagrams, Bayesian imperialism, and the Collins’ case: An appeal to reason’, Cardozo Law Review, 13 (1991), 1025-1079. Edwards, W., Schum, D. and Winkler, R. ‘Murder and (of?) the likelihood principle’, Journal of Behavioral Decision Making, 13 (1990), 75-89. Harman, G. Change in Mew. Cambridge, MA: MIT Press, 1986. Jackson, B. S. Law, Fact. and Narrative Coherence, Liverpool: Deborah Charles, 1988. Kalven, H., Jr and Zeisel, H. The American Jury, Boston: Little, Brown, 1966. Pennington, N. and Hastie, R. ‘A cognitive theory of juror decision making: The Story Model’, Cardox Law Review, 13 (1991), 519-557. Pennington, N. and Hastie, R. ‘A theory of explanation-based decision making’, in Klein. G., Orasanu, J., Calderwood, R. and Zsambok, C. E. (eds), Decision Making in Action: Models and methods, pp. 188-204, Norwood, NJ: Ablex, 1993a. Pennington, N. and Hastie, R . ‘Reasoning in explanation-based decision making’, Cognition, 49 (1993b), 123-163. Schum, D. A. ‘Probability and the processes of discovery, proof, and choice’, Boston University Law Review, 66 (l986), 825-876. Schum, D. A. Evidential Foundations of Probabilistic Reasoning, New York: Wiley, 1994. Schum, D. A. and Tillers, P. ‘Marshalling evidence for adversarial litigation’, Curdozo Law Review, 13 (1991), 657-704. Twining, W. Rethinking Evidence, Oxford: Basil Blackwell, 1990. Wigmore, J. H, The Science of Judicial Prooj? As given by logic, psychology, and general experience and illustrated in judicial trials (3rd edition), Boston: Little, Brown, 1937. INSIDE T H E JUROR: THE PSYCHOLOGY OF JUROR DECISION MAKING, R. Hastie (ed.), Cambridge: Cambridge University Press, 1993, 277 pp. ISBN Ck521-41988-3 (hc). that has a major impact on the life of the accused, and that is based on incomplete, unreliable, probabilistic, or even untrue information, is the same for lay people and professionals. Our own work on reasoning processes in professional judges (cf. Wagenaar et al. 1993) could have been described in one of the sections even without the reader realizing that the topic had shifted. What, then, is the characteristic property of this decision process, that remains so much the same, despite the extensive training and experience of the professional judge? An important distinction is made by Lola Lopes in her commentary in Chapter 13. She explains that the decision process can be modelled as a ‘meter model’ or as a ‘story model’. In the meter model information is used to adjust some interior reading of the perceived guiltiness of the accused. In the story model the decision maker attempts to construct a story about what happened; the match between the constructed story and the indicted behavior of the accused leads to a judgment of guilt. Lopes argues that both models are right, and I urge the reader to consult her argumentation. Personally, I favour the story model over the meter model, This book has two parts: Part I contains an introduction by the editor, and seven chapters on how jury decision making is modelled by decision theorists. Part I 1 contains six commentaries by outstanding researchers in the field. The commentaries add the sort of discussion that is urgently needed in this area, and thus lift the book to the level that makes it a very good buy. At the same time these commentaries make it a ridiculous undertaking to write a critical review, because the work of the commentators cannot be improved. The only additional contribution that I can make stems from the fact that my country, The Netherlands, has no juries, but courts of professional judges. Perhaps that is the reason why the Journal of Behavioral Decision Making asked my opinion. After reading the introductory chapter by the editor, I realized that juries and professional judges are much alike. The logical impossibility to draw a conclusion