www. NYLJ.com White-Collar Crime Volume 257—NO. 62 Monday, April 3, 2017 Combatting Hindsight In White-Collar Criminal Investigations

20/20 hindsight can sometimes mean 20 to life.

By Nicolas Bourtin

n the otherwise forgettable Star Wars prequel “The Phantom Menace,” Jedi I Knight Qui-Gon Jinn instructs a young Anakin Skywalker to “always remember: your focus determines your reality.” Qui-Gon’s insight is an important one, but you need not be a Jedi to recognize this basic principle of human : You tend to find evidence of whatever it is you already believe, whether it’s proof that your boss undervalues your work, or evidence that a job candidate who looks right for the job also possesses the necessary skills. Shu tt er st oc k This effect—what psychologists call “”—is just as strong when we look backwards in time. Thucydides, the great historian of the Peloponnesian War, observed that “peo- of the phenomenon known as “hind- and, as I’m going to discuss, in white- ple make their recollections fit with their sight bias.”1 Decades of psychological collar investigations most of all. Those suffering.” In other words, our perception research have proven the universal ten- of us who practice in this area quickly of the past similarly bends to conform to dency not only to look for evidence to learn that white-collar criminal investiga- what we “know” in our hearts to be true. confirm a conclusion you have already tions are often heavily outcome-driven, This tendency to find confirmation for reached, but also to greatly overesti- leaving them especially vulnerable to the facts already “known” is at the heart mate how foreseeable an outcome was distortions of fairness and rationality that once you know that the outcome has hindsight bias can produce. The problem taken place.2 is not insoluble, but solving it requires a Nicolas Bourtin is a litigation partner at Sullivan & Cromwell and deputy managing partner of the firm’s The effects of hindsight bias are particu- broader awareness of hindsight bias, a criminal defense and investigations group. larly significant in criminal investigations greater understanding of the depth and Monday, April 3, 2017

dimensions of the issue among the white- institutions that dealt with Madoff felt and developed rules—or “de-biasing” collar community, and consideration of inevitable. methods, in the jargon of psycholo- the range of potential solutions. gists—to mitigate the effects of hindsight Hindsight Bias In Legal Settings bias.10 Federal judges and legislators have The Problem of Hindsight Numerous empirical studies have docu- designed pleading requirements for secu- In White-Collar Crime mented the effects of hindsight bias on the rities fraud class actions to deter suits Hindsight bias infects white-collar inves- judgment of both experts and laypeople in premised solely on “fraud by hindsight,” tigations so meaningfully for two principal situations that involve medical diagnoses, in Judge Henry Friendly’s famous phrase.11 reasons. The first is that in financial fraud, political strategy, and more.3 Its impact Various legal rules aim to de-bias hindsight more often than not, guilt or innocence in judgments of legal culpability by both assessments of a defendant’s level of care. turns not on what a defendant did, but and judges has been shown to be One method is to restrict the information on the defendant’s mental state in doing equally pervasive.4 Typical “mock ” available to the factfinder—for example, it: with what intent did he or she make studies ask subjects to evaluate the rea- the inadmissibility of subsequent remedial a statement in a securities filing, push sonableness of a defendant’s decision.5 measures evidence in negligence cases.12 for a particular accounting treatment, Evaluators are randomly divided into a Another is to relax the standard of review approve a payment to a third party, or “foresight group,” which is given all the or defer to professional norms—for exam- perform some other job function that now information available to the defendant ple, the business judgment rule applicable appears suspect? That intent must usu- at the time of the decision, and a “hind- to corporate officers.13 ally be inferred from an ambiguous factual sight group,” which is additionally told the Nevertheless, hindsight bias has proven record, further amplifying the subjective outcome.6 These tests have consistently extremely difficult to combat. Studies have nature of the fact finder’s job. The result is found little success with jury instructions that innocent mistakes, poor judgment, or that merely warn about its influence.14 In even negligence can look like intentional The effects of hindsight bias are the flood precautions study, for example, conduct when viewed through the mag- instructions cautioning that “hindsight nifying lens of hindsight. particularly significant in crimi- vision is always 20/20” and urging jurors Even worse, the degree to which a nal investigations and in white- to “think of all the ways in which the event particular event is judged to have been collar investigations most of all. in question may have happened differently improper after the fact and the notoriety or not at all” failed to eliminate hindsight it generates have an outsized impact on shown that the hindsight group—the bias.15 On the other hand, more elaborate what inferences criminal investigators group playing the role of typical investi- de-biasing procedures than those typically draw about the mental state of the persons gators, judges, and juries—will view the used in court—requests for jurors to list involved. As discussed below, empirical outcome as more likely and the defendant alternative explanations along with spe- studies have shown that the more nega- as more culpable than the foresight group. cific evidence supporting each counter- tive the outcome, the more pronounced And the more negative the outcome, the factual—have proven more successful.16 the hindsight bias. Real life examples of more pronounced the bias.7 One exam- this phenomenon are not hard to come ple involved the choice of whether to Addressing Hindsight Bias by. When Bernie Madoff’s 48-year-old Wall undertake expensive precautions against In White-Collar Practice Street firm was revealed to have been a the threat of a flood. It found that three Given that white-collar crime displays Ponzi scheme of historic proportions, law quarters of the foresight group determined virtually every feature shown empirically enforcement agencies immediately set out that flooding was too unlikely to justify to intensify or entrench hindsight bias, it to prove that the banks and investment extra precautions, while a majority of is hardly surprising that this area of the advisors who had worked with Madoff the hindsight group instead estimated a law suffers from the problem. As noted, knew what he was up to. This, in spite of ­significantly higher probability of flood- white-collar criminal and regulatory the fact that government agencies had ing and found negligence in the failure to enforcement depends chiefly on after- themselves failed to detect Madoff’s fraud, take precautions.8 Empirical analyses of the-fact judgments of foreseeability and despite having received complaints and decisions by actual juries and judges show state of mind.17 Major white-collar investi- warnings years earlier. Given the magni- similar results.9 gations often follow the kinds of extreme tude of the fraud and the media In certain fields, such as civil litigation, outcomes—a company’s collapse, a Ponzi it drew, liability on the part of financial lawmakers have recognized the problem scheme’s unraveling, a full-blown ­financial Monday, April 3, 2017

crisis—that heavily bias estimates of the emotional and professional price paid To its credit, the DOJ has acted recently likelihood. An exaggerated view of an by compliance officers put through the to address unconscious racial, ethic, gen- outcome’s foreseeability, coupled with “near death experience” of such investi- der and other among all of its law an understandable motivation to assign gations is high. enforcement agents and prosecutors. blame retroactively, makes it difficult even Given the prevalence of hindsight bias It should do the same with respect to for experienced, well-meaning investiga- in white-collar criminal enforcement, the cognitive biases like hindsight bias. DOJ tors to credit alternative explanations for law’s failure to adapt is regrettable. Any prosecutors and law enforcement agents a defendant’s actions or omissions, such strategy to combat , psychol- should be trained on how to de-bias their as innocent blind spots in judgment or ogists have found, requires not only aware- evaluation of evidence—for example, by misplaced optimism. ness of the bias, but also awareness of its engaging in formal exercises to create The human impact of this effect on well- magnitude and direction, a motivation to and test counterfactuals and alternative meaning professionals is real. To take one correct it, and some means of correcting hypotheses, including by employing “Dev- all-too-common example, every bank anti- it.18 And although some areas of law have il’s Advocate” teams to challenge prosecu- money laundering officer suffers sleepless shown a capacity to address hindsight torial theories. nights worrying that one of the bank’s cus- bias—for example, through evidentiary Notwithstanding the difficulties, legal tomers will turn out to be at the center rules and judicial instructions in simple rules and practitioners in the area of of the next big scandal. Major banks have negligence cases, or pleading and burden- white-collar criminal enforcement can millions of customers, and anti-money shifting rules in corporate governance and and should adapt to reduce hindsight laundering officers charged with detect- securities litigation—white-collar criminal bias—not simply to ensure that white- ing and preventing terrorist financing and law lags behind in tackling or even recog- collar criminal enforcement is fair and money laundering must sift through mil- nizing these vulnerabilities. Neither the rational, but also to ensure that it achieves lions of daily transactions to try to find rules of criminal procedure or evidence, its critical function of deterrence. If hind- evidence of criminal activity. In view of nor criminal statutes themselves, take into sight bias undermines the law’s ability to the enormousness of the data haystack, account the pernicious effect of hindsight distinguish lawful conduct from wrong- the law requires that a bank’s internal con- bias in financial fraud cases. As noted doing, it weakens the deterrent value of trols be “risk based” and not foolproof. But above, a simple reminder of 20/20 hind- compliance defenses, for example in the compliance officers know that if a terror- sight in jury instructions is not effective. anti-corruption and anti-money launder- ist attack, large-scale narcotics bust, or Given this reality, the responsibility falls ing contexts.19 Hindsight bias does not massive fraud reveals connections to one to defense attorneys to posit alternative only undercut the law’s effectiveness in of their bank’s customers, they will find theories and to urge prosecutors, judges, conveying legal risk. It also fundamentally themselves under the intense spotlight and juries to view the facts at the time the hampers our ability to understand and of criminal, regulatory, and sometimes events occurred, rather than through the learn from past experience. Social scien- political investigations. And they know lens of hindsight. tists have amply demonstrated that even that those connections, no matter how Compounding the problem is the infre- the most experienced, objective, and well- difficult to detect at the time, will appear quency with which white-collar investiga- meaning decision makers will systemati- glaring in the harsh light of hindsight. tions, particularly corporate investigations, cally overestimate the foreseeability of Worst of all for the compliance officer ever make it in front of a neutral fact-finder. bad outcomes once they have occurred. is the nightmare scenario where a now- As has been much documented and dis- If the law fails to correct for this error, then notorious customer did raise suspicions in cussed elsewhere, most corporate criminal legal decision makers have little chance of real time, suspicions that were allayed by resolutions are out-of-court settlements, encouraging rational risk-mitigation and explanations that now appear implausible where the investigators themselves serve ensuring the fair administration of justice. or pretextual to investigators viewing the as the arbiters of whether any crime was facts retrospectively. Innocent contempo- committed. The overwhelming majority of ••••••••••••••••••••••••••••• raneous communications concerning the law enforcement agents and prosecutors The author thanks former Sullivan & Cromwell risk posed by a customer can falsely sug- act in good faith and want their investi- summer associate Laura Savarese for her invaluable research and contributions to this article, and Jim Ger- gest indifference or willful blindness when gations and charging decisions to be fair. aghty of National Review for the inspiration for the read later by investigators possessing the But without understanding and correcting Star Wars quote. full knowledge of the customer’s criminal for the unconscious cognitive biases to 1. There’s a reason why hindsight bias also goes more colloquially by the name “Monday morning quarterback- activity. And even though criminal charges which we all fall victim, they necessarily ing”: When Super Bowl LI reached its improbable conclu- in such circumstances are thankfully rare, suffer from their own critical blind spots. sion, millions of football fans—including those who had Monday, April 3, 2017

turned the game off in the third quarter because it was 8. See Kamin & Rachlinski, supra note 5, at 98-99. that only intensive personalized feedback reduced the clear that New England had no hope of coming back— 9. See, e.g., Galen V. Bodenhausen, “Second-Guessing bias); Merrie Jo Stallard & Debra L. Worthington, “Reduc- quickly took to criticizing Atlanta’s decision to continue the Jury: Stereotypic and Hindsight Biases in Perceptions ing the Hindsight Bias Utilizing Attorney Closing Argu- to throw the ball late in the fourth quarter instead of run- of Court Cases,” 20 J. APPLIED SOC. PSYCHOL. 1112 (1990) ments,” 22 L. & HUM. BEHAV. 671 (1998) (in a study of ning the ball and settling for a field goal. (finding that subjects considered the evidence in a crimi- jury determinations of director liability, finding some 2. For seminal accounts of the hindsight bias, see nal trial more incriminating when they were told the defen- success in reducing hindsight bias through the defense’s Baruch Fischhoff, “Hindsight ≠ Foresight: The Effect of dant was found guilty and less incriminating when they closing argument and little success through jury instruc- Outcome Knowledge on Judgment Under Uncertainty,” were told the defendant was found not guilty, as compared tions warning against the use of hindsight). J. EXPERIMENTAL PSYCHOL.: HUM. PERCEPTION & to subjects given only the evidence and not the outcome); 15. Kamin & Rachlinski, supra note 5, at 97, 99. PERFORMANCE 288 (1975), and & Daniel Mitu Gulati, Jeffrey J. Rachlinski, & Donald C. Langevoort, 16. See, e.g., Hal R. Arkes et al., “Eliminating the Hind- Kahneman, “Judgment Under Uncertainty: Heuristics and “Fraud by Hindsight,” 98 NW. U. L. REV. 773 (2004) (study- sight Bias,” 73 J. APPLIED PSCYHOL. 305 (1988) (finding Biases,” 185 SCIENCE 1124 (1974). Psychologists often ing judges’ strategies to correct for hindsight bias in secu- that where foresight and hindsight subjects had to esti- distinguish between different types of hindsight effects: rities fraud cases); Chris Guthrie, Jeffrey J. Rachlinski, & mate the probability of different medical diagnoses, hind- first, adjusting the perceived probability of an event after Andrew J. Wistrich, “Inside the Judicial Mind,” 86 COR- sight bias was lower for hindsight subjects asked to state learning it has occurred; second, failing to realize that NELL L. REV. 777 (2001) (finding federal magistrate judges reasons why different possible diagnoses might be cor- learning the outcome has altered one’s perception, and susceptible to hindsight bias, though to a lesser extent rect); Martin F. Davies, “Reduction of the Hindsight Bias so mistakenly believing that one’s ex ante and ex post than lay decision makers); Hastie, Schkade, & Payne, supra by Restoration of Foresight Perspective: Effectiveness of judgments would be similar (sometimes referred to as note 2, at 96-108 (finding substantial hindsight bias in jury Foresight and Hindsight-Retrieval Strategies,” the “I knew it all along” effect); and third, concluding that determinations of recklessness and liability for punitive 40 ORG. BEHAV. & HUM. DECISION PROCESSES 50 (1987) another person would likewise have made equivalent ex damages); Reid Hastie & W. Kip Viscusi, “What Juries Can’t (finding reduction of hindsight bias where subjects listed ante and ex post judgments. See Reid Hastie, David A. Do Well: The Jury’s Performance as a Risk Manager,” 40 supporting facts for various potential outcomes). But Schkade, & John W. Payne, “Looking Backward in Punitive ARIZ. L. REV. 901 (1998) (concluding that hindsight bias in see Lawrence J. Sanna, Norbert Schwarz, & Shevaun L. Judgments: 20-20 Vision?,” in PUNITIVE DAMAGES: HOW punitive damages determinations was more pronounced Stocker, “When Backfires: Accessible Content JURIES DECIDE 96, 107 (Cass R. Sunstein et al. eds., 2002). for juries than experienced judges); Marianne M. Jen- and Accessibility Experiences in Debiasing Hindsight,” 3. For a recent review of research on hindsight bias nings et al., “Causality as an Influence on Hindsight Bias: 28 J. EXPERIMENTAL PSYCHOL. 497 (2002) (finding that generally, see Neal J. Roese & Kathleen D. Vohs, “Hind- An Empirical Examination of Judges’ Evaluation of Profes- attempts to reduce hindsight bias by considering alterna- sight Bias,” 7 PERSPECTIVES ON PSYCHOL. SCI. 411 sional Audit Judgment,” 21 J. OF ACCT. & PUB. POL’Y 143 tive outcomes exacerbated the bias and attributing this (2012) (noting that over 800 scholarly papers had been (1998) (observing hindsight bias in judges’ evaluations of result to the perceived difficulty of listing ­counterfactuals). published on hindsight bias as of 2012). For examples auditing opinions). 17. See Samuel W. Buell, “Is the White Collar Offender of studies on hindsight bias in varied settings, see Hal 10. See Rachlinski, supra note 4, at 602-24 (explaining Privileged?,” 63 DUKE L.J. 823, 841-46 (describing char- R. Arkes et al., “The Hindsight Bias Among Physicians how the law has adapted to accommodate hindsight acteristics of white-collar criminal offenses); Donald C. Weighing the Likelihood of Diagnoses,” 66 J. APPLIED bias). Langevoort, SELLING HOPE, SELLING RISK: CORPORA- PSCYHOL. 252 (1981) (finding hindsight bias in doctors’ 11. See Denny v. Barber, 576 F.2d 465, 470 (2d Cir. 1978) TIONS, WALL STREET, AND THE DILEMMAS OF INVES- medical diagnoses); Baruch Fischhoff & Ruth Beyth, “‘I (affirming the dismissal of a securities class action where TOR PROTECTION 43-44 (2016) (discussing hindsight knew it would happen’—Remembered Probabilities of the lead plaintiff had “simply seized upon disclosures problems specific to white-collar crime and securities Once-Future Things,” 13 ORG. BEHAV. & HUM. PERFOR- made in later annual reports and alleged that they should enforcement). MANCE 1 (1975) (studying hindsight bias in Americans’ have been made in earlier ones”); Rachlinski, supra note 18. See Timothy D. Wilson & Nancy Brekke, “Mental predictions of the outcome of President Richard Nixon’s 4, at 616-17 (analyzing heightened pleading requirements Contamination and Mental Correction: Unwanted Influ- visit to China); MISSION COMMAND CTR., OF EXCEL- in §10(b) cases as a response to hindsight problems); ences on Judgments and Evaluations,” 116 PSYCHOL. LENCE, U.S. ARMY, COGNITIVE BIASES AND DECISION Gulati, Rachlinski, & Langevoort, supra note 9, at 775 BULL. 117, 119 (1994). MAKING: A LITERATURE REVIEW AND DISCUSSION OF (finding that federal courts cited concerns with hindsight 19. For a similar point regarding securities regulation, IMPLICATIONS FOR THE U.S. ARMY (2015), available at in nearly a third of all published securities class action see Gulati, Rachlinski, & Langevoort, supra note 9, at 774 http://usacac.army.mil/sites/default/files/publications/ opinions). (“The hindsight bias … creates a considerable obstacle HDCDTF_WhitePaper_Cognitive%20Biases%20and%20 12. See Rachlinski, supra note 4, at 617-18; Dan M. to the fundamental task in securities regulation of sorting Decision%20Making_Final_2015_01_09_0.pdf (collecting Kahan, “The Economics—Conventional, Behavioral, and fraud from mistake.”). research on hindsight bias in military decision making Political—of ‘Subsequent Remedial Measures’ Evidence,” and strategies to counteract it). 110 COLUM. L. REV. 1616 (2010) (analyzing arguments for 4. For reviews of research by psychologists, behavioral Federal Rule of Evidence 407 as a means of combating economists, and legal scholars on hindsight bias and hindsight bias in tort cases). de-biasing strategies within the legal system, see Erin 13. See Rachlinski, supra note 4, at 619-23 (on the M. Harley, “Hindsight Bias in Legal Decision Making,” 25 business judgment rule); see also id. at 608-13 (on legal SOC. COGNITION (SPECIAL ISSUE: THE HINDSIGHT BIAS) defenses based on compliance with regulation or custom, 48 (2007); Jeffrey J. Rachlinski, “A Positive Psychological in cases involving commercial or medical decisions). Theory of Judging in Hindsight,” 65 U. CHI. L. REV. 571 Delaware courts have specifically cited the problems (1998); Neal J. Roese, “Twisted Pair: Counterfactual Think- of hindsight bias and judicial competence as rationales ing and the Hindsight Bias,” in BLACKWELL HANDBOOK for applying the business judgment rule in cases alleging OF JUDGMENT AND DECISION MAKING 258 (Derek J. failures of director oversight. See In re Citigroup Deriv. Koehler & Nigel Harvey eds., 2004); Cass R. Sunstein & Litig., 964 A.2d 106, 124 (Del. Ch. 2009) (“[The doctrine] Christine Jolls, “Debiasing Through Law” (John M. Olin follows from the inadequacy of the Court, due in part to Program in Law and Econ., Working Paper No. 225, 2004). a concept known as hindsight bias, to properly evaluate 5. See Kim A. Kamin & Jeffrey J. Rachlinski, “Ex Post whether corporate decision-makers made a ‘right’ or ≠ Ex Ante: Determining Liability in Hindsight,” 19 L. & ‘wrong’ decision.”); Stone v. Ritter, 911 A.2d 362, 373 (Del. HUM. BEHAV. 89 (1995); see generally Harley, supra note 2006) (“With the benefit of hindsight, the plaintiffs’ com- 4 (reviewing studies). plaint seeks to equate a bad outcome with bad faith.”). 6. See Harley, supra note 4. 14. See, e.g., Baruch Fischhoff, “Debiasing,” in JUDG- 7. See id. at 51 (“The severity of a negative outcome MENT UNDER UNCERTAINTY: HEURISTICS AND BIASES can have dramatic effects on the size of hindsight bias, 422 (, Paul Slovic, & Amos Tversky eds., Reprinted with permission from the April 3, 2017 edition of the NEW YORK LAW JOURNAL © 2017 ALM Media Properties, LLC. All rights reserved. Further with larger bias resulting from more severe negative out- 1982) (reviewing various strategies to reduce hindsight duplication without permission is prohibited. For information, contact 877-257-3382 comes.”). bias, including warnings about its effects, and finding or [email protected]. # 070-04-17-01