Improving prosecutorial decision making: some lessons of cognitive science.

• Author: Burke, Alafair S.

TABLE OF CONTENTS INTRODUCTION I. COGNITIVE BIAS: FOUR EXAMPLES OF IMPERFECT DECISION MAKING A. Confirmation Bias B. Selective Information Processing C. Belief Perseverance D. The Avoidance of Cognitive Dissonance II. THE ETHICAL PROSECUTOR AND COGNITIVE BIAS A. Investigation and Charging Decisions B. Sticky Presumptions of Guilt C. The Disclosure of Exculpatory Evidence D. Stickier Presumptions of Guilt Postconviction III. IMPROVING PROSECUTORIAL DECISION MAKING A. Improving the Initial Theory of Guilt B. Educating Prosecutors and Future Prosecutors C. The Practice of "Switching Sides" D. Second Opinions and Committee Input E. Prosecutorial Involvement in Innocence Projects F. Repairing Brady CONCLUSION INTRODUCTION

Earl Washington Jr., a black mentally retarded farmhand, spent seventeen years in prison--nearly ten of them on death row--for the rape and murder of a white woman before DNA tests linked another man to the crimes. (1) The evidence originally implicating him was questionable. The victim provided little identifying information, indicating before her death only that a black man had attacked her. (2) Washington was convicted based largely on his own confession, even though he simultaneously provided factually inconsistent confessions to four other crimes and did not know the location of the crime scene, whether other people were present, or even the victim's race without the assistance of leading questioning from police. (3) When defense counsel sought postconviction relief based on the discovery of another man's DNA on a blanket linked to the crime, prosecutors resisted. (4) Even after Washington was pardoned because of exonerating evidence, prosecutors insisted that he remained a viable suspect. (5)

Earl Washington's case raises questions about the discretionary decisions his prosecutors made from the moment they received the case. Why did they look past apparent problems with the confession? Why did they resist the evidentiary testing that ultimately exonerated an innocent man? Why would they still not concede innocence after his exoneration? Traditionally, commentators have clothed the study of prosecutorial decision making in the rhetoric of fault, attributing normatively inappropriate outcomes to bad prosecutorial intentions and widespread prosecutorial misconduct. (6) From this perspective, Earl Washington's conviction, and his prosecutors' refusal to concede his innocence even after a gubernatorial pardon, result from prosecutorial overzealousness, (7) a culture that emphasizes winning, (8) the absence of "moral courage," (9) and the failure of prosecutors to act as neutral advocates of justice. (10)

This focus upon incentives, priorities, and values as potential taints upon the exercise of prosecutorial discretion reveals an implicit but important assumption about prosecutors: they are rational, utility-maximizing decision makers. Prosecutors choose to overcharge defendants, withhold exculpatory evidence, and turn a blind eye to claims of innocence; therefore, the traditional inference goes, they must value obtaining and maintaining convictions over "doing justice." (11) To ensure that prosecutors do not rationally opt for misconduct to maximize their conviction rates, the fault-based literature recommends reform through changes to the prosecutorial cost-benefit analysis. Common strategies include more stringent ethical rules, (12) increased disciplinary proceedings and sanctions against prosecutors, (13) and professional (14) and financial (15) rewards based on factors other than just obtaining convictions.

Consider, however, a different explanation for the failure of prosecutors always to make just decisions. Perhaps prosecutors sometimes fail to make decisions that rationally further justice, not because they fail to value justice, but because they are, in fact, irrational. They are irrational because they are human, and all human decision makers share a common set of information-processing tendencies that depart from perfect rationality. A compelling body of cognitive research demonstrates that people systematically hold a set of cognitive biases, rendering them neither perfectly rational information processors, nor wholly random or irrational decision makers. (16) Drawing on the cognitive literature, the growing literature of behavioral law and economics explores the limitations of cost-benefit rationality, challenging the assumption of traditional economists that people are perfect wealth maximizers. (17) From both the cognitive and behavioral economics literature emerges a theory of bounded rationality that seeks to explain how cognitive biases and limitations in our cognitive abilities distort perfect information processing in nonrandom, predictable ways. (18)

Others have suggested how cognitive bias and bounded rationality can affect juries, (19) judges, (20) the regulation of risk, (21) federal rulemaking, (22) corporate disclosures, (23) contract law, (24) consumer choice, (25) employment discrimination, (26) and group deliberations. (27) This Article seeks to explain how cognitive bias can affect the exercise of prosecutorial discretion. Viewing cases like Earl Washington's through a lens of human cognition, rather than fault, colors not only the description of the problem, but also the recommended solutions. When the underlying problem is human irrationality, rather than the malicious intentions of a single prosecutor or the indifference of a prosecutorial culture, the result is a far more complicated story about the criminal justice system. If prosecutors fail to achieve justice not because they are bad, but because they are human, what hope is there for change?

This Article explores the potential that even "virtuous," (28) "conscientious," (29) and "prudent" (30) prosecutors fall prey to cognitive failures. Part I summarizes four related aspects of cognitive bias that can affect theory formation and maintenance. Part II explores how these cognitive phenomena might adversely affect the exercise of prosecutorial discretion. Finally, Part III proposes a series of reforms that might improve the quality of prosecutorial decision making, despite the limitations of human cognition.

1. COGNITIVE BIAS: FOUR EXAMPLES OF IMPERFECT DECISION MAKING

   Decades of empirical research demonstrate that people's beliefs are both imperfect and resistant to change. Once people form theories, they fail to adjust the strength of their beliefs when confronted with evidence that challenges the accuracy of those theories. (31) Indeed, theory maintenance will often hold even when people learn that the evidence that originally justified the theory is inaccurate. (32) At the same time that people fail to consider information that disconfirms a theory, they tend both to seek out and to overvalue information that confirms it. (33)

   This Article explores four related but separate aspects of cognitive bias that can contribute to imperfect theory formation and maintenance: confirmation bias, selective information processing, belief perseverance, and the avoidance of cognitive dissonance. Confirmation bias is the tendency to seek to confirm, rather than disconfirm, any hypothesis under study. (34) Selective information processing causes people to overvalue information that is consistent with their preexisting theories and to undervalue information that challenges those theories. (35) Belief perseverance refers to the human tendency to continue to adhere to a theory, even after the evidence underlying the theory is disproved. (36) Finally, the desire to avoid cognitive dissonance can cause people to adjust their beliefs to maintain existing self-perceptions. (37) This Part summarizes the empirical literature regarding each of these cognitive phenomena.

   1. Confirmation Bias

      When testing a hypothesis's validity, people tend to favor information that confirms their theory over disconfirming information. (38) Good evidence suggests that this information-seeking bias results because people tend to recognize the relevance of confirming evidence more than disconfirming evidence. (39) This is true even when an effort to disconfirm is an essential step towards confirmation of the hypothesis under test.

      For example, in a classic study of confirmation bias in hypothesis testing, Peter Wason presented subjects with four cards and told them that each card contained a letter on one side and a number on the other. (40) The revealed sides of the four cards displayed one vowel, one consonant, one even number, and one odd number. (41) Subjects were then asked which of the four cards needed to be turned over to test the following rule: If a card has a vowel on one side, then it has an even number on the other side. (42)

      Proper scientific method requires that researchers seek to disprove their working hypotheses. (43) Accordingly, the rational test of Wason's "if vowel, then even number" card test is to turn over the vowel and odd number cards. The vowel card provides a relevant test because discovery of an odd number on its backside would disprove the tested rule. The odd number card provides an equally relevant test because the discovery of a vowel on the backside of the odd number card would disconfirm the proposed "if vowel, then even number" rule in the same way as any odd number on the other side of the vowel card.

      However, to test the rule that all vowel cards had even numbers on the other side, subjects overwhelmingly selected from their four choices either just the vowel card, or the vowel and the even number cards. (44) They failed to select the odd number card that was necessary to test the rule properly. (45) Moreover, subjects who chose the even number card erred still further because that card's backside offered no probative value to the rule at hand. (46) The subjects' choice of cards demonstrated that the subjects failed to recognize the importance of disconfirming evidence and instead sought information...