The hidden structure of fact-finding.

• Author: Spottswood, Mark
• Position: Abstract through II. A Primer on Dual-Process Cognition B. The Associative Structure of System I Unconscious Thinking, p. 131-166

ABSTRACT

This Article offers a new account of legal fact-finding based on the dual-process framework in cognitive psychology. This line of research suggests that our brains possess two radically different ways of thinking. "System 1" cognition is unconscious, fast, and associative, while "System 2" involves effortful, conscious reasoning. Drawing on these insights, I describe the ways that unconscious processing and conscious reflection interact when jurors hear and decide cases. Most existing evidential models offer useful insights about the ways that juries use relevant information in deciding cases but fail to account for situations in which their decisions are likely to be affected by irrelevant stimuli. The dual-process approach, by contrast, is able to explain both probative and prejudicial influences on decision making. As a demonstration, I use the dual-process framework to explain the surprising result in People v. Rivera, a case in which a jury convicted a man of rape and murder despite the admission of exonerating DNA evidence. This result, I suggest, was not the product of an unusually lazy or unreasonable jury but rather illustrates the way that our ordinary cognitive processes can lead us to endorse quite unreasonable results if primed using certain common prosecutorial strategies.

After elaborating the dual-process model in a descriptive form, I then consider some of its normative implications. Many leading evidence scholars have argued that verdicts resting on "pure" or "naked" statistical evidence are problematic. Although the dual-process model of fact-finding is descriptive rather than normative, it nevertheless provides surprising insight into this debate by showing that our intuitive discomfort with verdicts that are based on purely statistical data may arise from the failure of such evidence to speak in terms that our unconscious, intuitive System 1 can process reliably. In such circumstances, intuitions about outcomes should be treated with caution. Thus, what unites the seemingly disparate examples of the Rivera trial and the naked statistical evidence debate is that, in both contexts, it feels right to do wrong.

CONTENTS INTRODUCTION I. EXISTING MODELS OF FACT-FINDING A. The Multiple Uses of Evidence Models B. A Brief Taxonomy of Fact-Finding Models 1. The Bayesian Model 2. Bayesianism's Rivals: Stories, Comparisons, and Coherence C. What the Existing Discussion Has Taken for Granted II. A PRIMER ON DUAL-PROCESS COGNITION A. System 1 Is a Hidden, Unconscious Information Processor B. The Associative Structure of System I Unconscious Thinking C. System 2 Processes Information in a Conscious, Effortful Manner D. The Complex Interplay of Dual-Process Reasoning III. A DUAL-PROCESS ACCOUNT OF FACT-FINDING A. System 1 Cognition at Trial: The Background Associative Model B. The Impact of Party Appearance and Identification on the Background Associative Model C. System 1 Cognition: Adding Case-Specific Units to the Associative Network in Response to Trial Evidence D. The Role of System 2 Reasoning in Shaping the Rivera Verdict IV. PROFESSOR COHEN'S GATECRASHERS, OR: HOW I LEARNED TO STOP WORRYING AND LOVE STATISTICAL EVIDENCE CONCLUSION INTRODUCTION

For quite some time, evidence scholars have debated which framework among several competing alternatives--best describes the process by which judges and juries evaluate evidence and make verdict decisions. (1) This discussion has centered on two rival accounts of fact-finding. The "Bayesian" family of inferential models seeks to account for evidential analysis in probabilistic terms, with each new item of evidence contributing to changes in jurors' estimations of how likely it is that disputed facts are true. (2) A competing group of theorists asserts that fact-finding involves the construction and comparison of stories, or narratives, and the fact finder chooses the narrative that best explains the given pattern of trial evidence, without any separate consideration of individual probabilities. (3)

In this Article, I reflect on the shortcomings of both the Bayesian and story-comparison frameworks as descriptions of how jurors determine facts in the messy environment of real-world trials. Both approaches fail to account adequately for the unconscious aspects of fact-finding. (4) And, unfortunately, we cannot fully comprehend how real jurors are likely to process evidence unless we account for prejudice and associative intuition in addition to more rational styles of inference. (5)

To see the shortcomings of purely rationalist models, consider People v. Rivera, (6) a case in which an Illinois jury convicted a man for raping and murdering a child despite compelling forensic DNA evidence that a different, unidentified person committed the crime. (7) Rivera arose from the death of eleven-year-old Holly Staker in 1992. (8) Almost twenty years after her murder, all Illinois jury was called to decide, for the third time, (9) whether Juan Rivera committed the crime. (10) The facts of the case were gruesome: the killer had kicked in the door of the apartment where Holly worked as a babysitter, raped her both vaginally and anally, and then stabbed and strangled her until she was dead. (11) And the prosecution had an excellent card to play: when questioned by the police, Rivera had confessed to the crime. (12) In addition to this confession, the State offered evidence showing that Rivera had tried to concoct a fictitious alibi for the night of the crime and that other jail inmates had also heard him confess to the killing. (13) Based on the prosecution's evidence alone, it would seem like an easy case for conviction.

Once the defense had its say, however, the picture looked strikingly different. According to new and uncontradicted expert testimony, DNA samples taken from the semen recovered during Holly's autopsy did not match Rivera's DNA profile. (14) What is more, Rivera was on electronic home monitoring when the crime was committed, and there were no records suggesting that he had left his home that night. (15) To explain Rivera's confession, the defense offered evidence that he had a low IQ that bordered mental retardation; that, during a break between police interviews, jail medical personnel had witnessed him behaving in ways suggesting psychosis; that his story had changed in response to leading questions by the police; and that his initial confession was inconsistent with the facts of the actual crime. (16)

In other words, despite the vivid evidence connecting Rivera to the crime, he was probably--indeed, almost certainly not the man who raped and killed Holly Staker. The jury in Rivera's case, having heard all of this evidence, was instructed that Rivera was presumed innocent and should be convicted only if the jury was convinced of his guilt "beyond a reasonable doubt." (17) After four days of deliberations, they concluded that this standard had been met, finding him guilty. (18) The result was decried in the media and then eventually reversed on the ground that the evidence had been insufficient to permit any reasonable jury to convict. (19)

It is hard to know what to make of the jurors' inability to see reasonable doubt in this pattern of evidence. Although the case was disturbing, some might wish to construe this as an isolated event with little overall import. (20) the criminal jury, after all, is selected in a quasi-random fashion from members of the general public, and every so often that process might select a highly unreasonable group of people by a mere quirk of statistics. Seen in this light, the case could be viewed as the random result of a deck stacked against an innocent defendant, in which an overzealous prosecutor (21) and a judge who had already made up his mind that Rivera was guilty (22) allowed a weak case to go before a jury that was either unusually lazy or biased, or both. But I believe this case deserves closer study. The jury's decision in Rivera can be explained in terms of well-studied features of unconscious cognition that we all share. (23) This explanation, in turn, makes the trial both more understandable and more disturbing, because it implies that other juries, even if well motivated, might have reached the same perplexing result if exposed to a similar pattern of evidence. (24)

To provide an account of fact-finding that can explain cases like Rivera, we must depart significantly from the rationalist tradition of evidential modeling, in search of a framework that can accommodate prejudicial inferences as well as reasonable inferences. If we limit ourselves to rational inferences about probabilities and coherent narratives founded on the admitted evidence, we will be at a loss to explain why twelve ordinary people could have thought that it was not just probable, but beyond any reasonable doubt, that Rivera was Holly's killer. What we need, instead, is a similarly robust way to analyze and describe the way that the jury's views about what inferences were and were not reasonable were shaped, in significant part, by irrelevant or prejudicial factors that operated largely outside the jury's awareness.

Although evidence scholars have spent a substantial amount of time chronicling particular ways that we might fall prey to unconscious biases in the courtroom, (25) few scholars have incorporated these scattered insights into the broader accounts of inference at trial. (26) We lack, in other words, a framework in which we can examine inferential prejudice in the organized and systematic way that the Bayesian and story-comparison accounts bring to the analysis of inferential probity.

In this Article, I develop a systemic account of the ways that unconscious cognitive processes interact with the more familiar aspects of judge and jury decision making. (27) In so doing, I will draw heavily on the dual-process tradition in cognitive psychology. This dual-process framework, which cognitive scientists developed to...