The cognitive psychology of circumstantial evidence.

• Author: Heller, Kevin Jon

Empirical research indicates that jurors routinely undervalue circumstantial evidence (DNA, fingerprints, and the like) and overvalue direct evidence (eyewitness identifications and confessions) when making verdict choices, even though false-conviction statistics indicate that the former is normally more probative and more reliable than the latter. The traditional explanation of this paradox, based on the probability-threshold model of jury decision-making, is that jurors simply do not understand circumstantial evidence and thus routinely underestimate its effect on the objective probability of the defendant's guilt. That may be true in some situations, but it fails to account for what is known in cognitive psychology as the Wells Effect: the puzzling fact that jurors are likely to acquit in a circumstantial case even when they know the objective probability of the defendant's guilt is sufficient to convict. This Article attempts to explain why jurors find circumstantial evidence so psychologically troubling. It begins by using a variety of psychological research into judgment and decision-making--Kahneman & Tversky's simulation heuristic in particular--to argue that jurors decide whether to acquit in a criminal case not through mechanical probability calculations, but on the basis of their ability to imagine a scenario in which the defendant is factually innocent. The Article then examines the basic epistemological differences between direct and circumstantial evidence and shows how those differences normally make it easier for jurors to imagine a factually exculpatory scenario in a circumstantial case. Finally, the Article concludes by discussing how an ease-of-simulation model of jury decision-making improves our understanding of why false verdicts occur.

TABLE OF CONTENTS INTRODUCTION I. THE PROBLEM DESCRIBED A. Jurors' Misevaluation of Direct and Circumstantial Evidence 1. Direct Evidence 2. Circumstantial Evidence B. The Greater Reliability of Circumstantial Evidence 1. Error Rates 2. False-Conviction Statistics II. THE TRADITIONAL EXPLANATION AND THE WELLS EFFECT A. The Traditional Explanation B. The Wells Effect III. THE EASE-OF-SIMULATION HYPOTHESIS IV. THE SIMULATION HEURISTIC AND THE EASE-OF-SIMULATION MODEL A. The Simulation Heuristic B. The Simulation Heuristic in the Legal Context 1. Use of the Heuristic 2. Toward a Simulation-Based Model of Decision-Making V. THE EASE-OF-SIMULATION MODEL APPLIED A. The Differences between Direct and Circumstantial Evidence 1. Representational vs. Abstract 2. Narrative vs. Rhetorical 3. Univocal vs. Polyvocal 4. Unconditional vs. Probabilistic B. Imagining the Factually Inculpatory Scenario 1. Event Representation 2. Structural Coherence 3. Vividness C. Priming the Imagination of a Factually Exculpatory Scenario 1. The Need for Priming 2. When Priming Occurs 3. The Certainty Effect 4. What About Unreliability? 5. The Reliability of Circumstantial Evidence D. Imagining the Factually Exculpatory Scenario 1. Recognizing Evidence of Innocence 2. Creating an Exculpatory Storyline 3. Imagining the Factually Exculpatory Scenario 4. "Undoing" the Factually Inculpatory Scenario VI. THE PARADOX OF CIRCUMSTANTIAL EVIDENCE REVISITED A. "Ideal-Typical" Direct Cases B. "Ideal-Typical" Circumstantial Cases C. "Mixed" Cases D. False Verdicts VII. CONCLUDING THOUGHTS INTRODUCTION

"He[, the Judge,] says to them: Perhaps ye saw him running after his fellow into a ruin, ye pursued him, and found him sword in hand with blood dripping from it, whilst the murdered man was writhing [in agony]: If this is what ye saw, ye saw nothing." (1)

Classical Jewish law was profoundly skeptical of circumstantial evidence. Such evidence was per se inadmissible in a criminal case; to convict, direct evidence of the defendant's guilt--specifically, the testimony of two witnesses who saw the defendant commit the crime--was required. (2) The rationale for the rule, according to Talmudic scholars, was the need to protect the innocent from unjust conviction: because of its probabilistic nature, not even the strongest circumstantial evidence could completely prove the defendant's guilt. (3) Maimonides explained as follows:

Do not let this puzzle you, or think the law unjust. For among events which are within the bounds of possibility, some are very probable and others highly improbable, and still others are in between the two.... If we do not give judgment even on the basis of a very strong presumption, the worst that can happen is that the sinner will be acquitted; but if we punish on the strength of presumptions and suppositions, it may be that one day we shall put to death an innocent person; and it is better and more satisfactory to acquit a thousand guilty persons than to put a single innocent man to death.... (4) Modern Anglo-American law, of course, does not view circumstantial evidence as inferior to direct evidence. Wigmore says that "it is out of the question to make a general assertion ascribing greater weight to one class or to the other." (5) The Supreme Court agreed in Holland v. United States, holding that direct and circumstantial evidence are "intrinsically no different." (6) The Court acknowledged that circumstantial evidence can lead to an incorrect result, but noted that in "both instances, a jury is asked to weigh the chances that the evidence correctly points to guilt against the possibility of inaccuracy or ambiguous inference. In both, the jury must use its experience with people and events in weighing the probabilities." (7)

Some scholars, in fact, have argued that circumstantial evidence is superior to direct evidence. Edmund Burke was of that opinion, (8) as was William Paley, who famously explained that a "concurrence of well-authenticated circumstances composes a stronger ground of assurance, than positive testimony, unconfirmed by circumstances, usually affords. Circumstances cannot lie." (9)

Error rates and false-conviction statistics support Burke and Paley. Both indicate that direct evidence--false confessions and mistaken or perjured eyewitness identifications--is much less reliable than circumstantial evidence. For example, studies have shown that eyewitness identifications are mistaken more than 58% of the time, (10) whereas less than 1% of DNA matches turn out to be erroneous. (11) Similarly, although 68% of the false convictions that Bedau and Radelet identified in their famous study resulted from problems with direct evidence, only 9% resulted from problems with circumstantial evidence. (12) So if we take seriously Maimonides' assertion that it is better "to acquit a thousand guilty persons than to put a single innocent man to death," direct evidence would seem to be a far better candidate for categorical exclusion than circumstantial evidence. Circumstances may sometimes lie, but witnesses lie far more often.

Empirical research into jury decision-making, however, indicates that jurors agree with Maimonides. That research, discussed in detail below, has consistently found that jurors dramatically undervalue circumstantial evidence and just as dramatically overvalue direct evidence. One study of direct evidence, for example, found that jurors overestimated the accuracy of eyewitness identifications by more than 500%. (13) Conversely, the "most surprising finding" of a study of circumstantial forensic evidence "was how easily people can be persuaded to give no weight" to such evidence. (14)

There is, in short, an unsettling paradox concerning the use of circumstantial evidence in criminal trials: although it is far less likely to lead to a false conviction than direct evidence, jurors are so reluctant to use it to convict that a circumstantial case often results in the opposite problem, a false acquittal. (15) Indeed, it is only a slight exaggeration to say that the Talmudic prohibition on circumstantial evidence lives on in the minds of modern-day jurors via the shibboleth--regularly disseminated by television and the movies--that "it's all circumstantial." (16)

The question is--why?

The traditional explanation is that jurors simply do not understand the importance of circumstantial evidence--DNA matches, fingerprint comparisons, and the like--and thus underestimate its effect on the objective probability of the defendant's guilt. (17) In one study of blood evidence, for example, sixty-six percent of mock jurors misunderstood the evidence in a way that led them to significantly underestimate its probative value. (18)

That explanation, however, is not the whole story. On the contrary, a series of sophisticated mock-jury studies has found that jurors are likely to acquit in a circumstantial case even when they know the objective probability of the defendant's guilt is sufficient to convict. (19) That phenomenon, known as the Wells Effect, (20) is irreconcilable with the traditional explanation, because it indicates that jurors' undervaluation of circumstantial evidence is psychological, not cognitive. (21)

To understand jurors' reluctance to convict in circumstantial cases, then, we need to explain why jurors find circumstantial evidence so psychologically troubling. Such an explanation might allow us to understand not only their undervaluation of circumstantial evidence, but their overvaluation of direct evidence as well.

The object of this Article is thus twofold: (1) to explain why jurors are likely to acquit in a circumstantial case even when they know that the evidence is objectively sufficient to convict; and (2) to explain why jurors are likely to convict in a direct case even when there is reason to believe that the evidence may be unreliable. The Article's central thesis is that both phenomena result from the same cognitive mechanism--the fact that jurors decide whether to acquit not through mechanical probability calculations, but on the basis of their ability to imagine a scenario in which the defendant is factually innocent. I will argue...