Psychology’s Reasonable Doubts

IN DOUBT: THE PSYCHOLOGY OF THE CRIMINAL JUSTICE PROCESS. By Dan Simon. Cambridge, Mass.: Harvard University Press. 2012. Pp. 405. $45.00

I.

The criminal justice process does not work as well as we think it does. Set aside the overworked public defenders who cannot provide adequate counsel[1] and the inordinate reliance on plea-bargaining.[2] Set aside improper forensic treatment[3] and unreliable informants.[4] Set aside the massive discretion mandatory minimums give to prosecutors[5] and the accusation that the criminal justice system has replaced Jim Crow laws in the subjugation of African-Americans.[6] Look only at jury trials with competent defense counsel, honest police investigators, and fair prosecutors. Even under these idealized conditions, the fact remains: The criminal justice process does not work as well as we think it does.

Dan Simon’s In Doubt: The Psychology of the Criminal Justice Process details the damage. While acknowledging that the true rate of false convictions is unknowable, Simon notes that the error rate in capital homicides is approximately 3-4%, and the overall rate of false convictions “most likely considerably higher.”[7] Even these numbers underestimate the defects of criminal justice processes because they include “easy cases,” which police solve without any investigative effort by detective units and make up the majority of solved crimes.[8] In “difficult cases,” when the investigative resources of the criminal justice system are brought to bear to overcome evidentiary shortfalls, the false conviction rate is higher than the base rate, already over 5%. In Doubt turns to social psychology to understand why the criminal justice system so often goes awry.

Although his data in the introduction suggest a coming jeremiad against the criminal justice system, Simon’s goals are more modest. He expressly rejects a “fundamental institutional redesign” in place of recommendations that are “practical, feasible, and readily implementable in the short or medium term.”[9] He moves through stages of the criminal justice process, identifying the ways that cognitive biases derail the quest for accuracy and ending each chapter with a list of recommendations to minimize errors. In the investigative stage, the “adversarial pull” drives investigators to become too confident of suspects’ guilt; eliminating the incentives for clearing crimes could reduce institutional—and therefore internal—pressures.[10] Detectives can unknowingly encourage victims to select the suspect out of lineup procedures; “double blind” procedures eliminate this potential error.[11] Interrogators often coerce suspects into false confessions; convincing them that the confidence they place in their ability to interpret physical cues for truthfulness is exaggerated would reduce this risk.

As Simon proceeds through issue areas at all stages of the criminal justice process, certain psychological phenomena reappear as causes of error. The “coherence effect” explains people’s bidirectional process of reasoning; just as facts guide conclusions, conclusions radiate backwards and reshape facts.[12] Thus when juries learn of an inculpating piece of evidence that suggests the outcome of guilt, they retroactively find an alibi more suspect or an ambiguous composite to more resemble the suspect.[13] The effect causes investigators to perceive the evidence as more supportive of their conclusions than it really is, encouraging police to investigate innocent suspects. Months later, the jury may experience the same effect in fitting disparate pieces of evidence into a coherent whole of guilt.

The fallibility of memory also emerges as an issue in several different contexts. Memories are neither as complete nor as accurate as we imagine. Despite common belief,[14] memory is not like a video camera. Witnesses cannot know in advance which details of an event are going to be crucial in a criminal trial and simply do not remember most of them. What they do remember is not necessarily accurate. People systematically underestimate distances, overestimate duration, and misestimate speeds.[15] Most relevant for Simon, they often misremember criminal events and are unable to accurately identify their perpetrators, despite conventional wisdom that these memories are extremely probative at criminal trials.

Simon presents to the reader a criminal justice system riddled with scores of small inadequacies that all too often accumulate in false convictions. His tone, however, is not that of a firebrand, but a cautious reformer, methodically identifying problems and suggesting fixes.

If anything, Simon treads too carefully in avoiding sweeping conclusions or impugning actors’ motivations. While documenting the “adversarial pull,” detailing how incentive systems push investigators towards the inclination to infer guilt, Simon inserts the caveat that, “There is reason to believe that most law enforcement personnel in most police departments withstand the adversarial pressures, and conduct thorough and fair investigations.”[16] Simon fails to provide the “reason to believe,” however, and gives no citation.[17] His caveat is implausible; effects on motivation cannot simply be “withstood,” and alter behavior in subtle and sometimes imperceptible ways. While investigations—perhaps even most investigations—may still be generally fair, they will be less fair in the presence of distorting effects on motivation. The phrasing is also oddly imprecise for an academic work. “[M]ost law enforcement personnel in most police departments” could refer to nearly any number of officers, depending on how many officers comprise the first “most” and the size of the departments compromising the second.

Simon next writes, “The adversarial pull, however, is likely to wreak havoc in investigations conducted under intense pressures and performed by those who lack a disciplined professional temperament.”[18] While “temperament” and “time pressures” likely affect the magnitude of motivational distortions, imagining a fork where the “adversarial pull” either “wreaks havoc” or is “withstood” entirely oversimplifies motivation’s effects.

Simon’s own motivations likely explain why he—who is so careful not to overreach throughout the book—becomes careless with concepts and phrasing here. The information he presents in the book is radical, yet he insists that poorly designed institutions and subtle psychological phenomena explain many of these errors, not the mendacity of the police or prosecutors. In making the case for good faith actors and technocratic fixes, he sometimes goes too far.

Simon’s general level of caution makes his occasional condemnations more powerful. He reserves his ire—or as close as he comes to it—for practices that plainly do more harm than good and should be abolished entirely. In-court identifications, when a witness identifies the defendant as present in the courtroom during trial, are hollow formalities, as even “a tourist who happened to stumble into the courtroom” could identify the person sitting at the defense table. The practice is useless, Simon claims, but psychologically manipulative, drawing on the emotional response to the performance. Simon finds the practice too worthless “to warrant empirical investigation,” and finds “puzzling” legal scholars’ indifference to the tactic’s continuance. He recommends not merely reforming the practice but abolishing it.

He reserves a similar fate for composite images of suspects. While the police construct composites with a feature-based process, incrementally adding discrete facial features, memory for faces is holistic. The action of creating the composite can even contaminate the witness’s original memory of the perpetrator, replacing it with the composite itself.

II.

Simon’s recommendations are granular, going out mostly to police departments and trial judges. He rarely mentions legal doctrine; when he does, he often buries a resigned sentence in the middle of a paragraph or a footnote about the Supreme Court’s permissiveness, detachment, or naïveté.[19] While on the book’s penultimate page he pines generally for a Due Process Clause jurisprudence that demands accuracy, he does not address the constitutional doctrine concerning the more specific concerns he raises throughout the book.

Consider Manson v. Brathwaite, when the Supreme Court examined whether the Due Process Clause demanded the exclusion of unnecessary and suggestive eyewitness identification procedures.[20] It determined the linchpin of admission must be reliability and laid out the factors courts should use to consider it: “[T]he opportunity of the witness to view the criminal at the time of the crime, the witness’ degree of attention, the accuracy of his prior description of the criminal, the level of certainty demonstrated at the confrontation, and the time between the crime and the confrontation.”

The Court sets out the prongs for a legal test, but the reliability of eyewitness testimony is as much a scientific question as a legal one. As Simon notes, there have been over 2,000 studies on eyewitness testimony,[21] and these should dictate “reliability,” not prongs the Court laid out in 1977. Some of the Court’s amateur psychology is quite good—as Simon notes, memories fade quickly and the time between the crime and the confrontation is crucial. Some of it is not. While the “witness’ degree of attention” might intuitively seem to be relevant to preserving memory, high levels of witness stress actually impair identification accuracy.[22]

While Simon does mention Brathwaite, he does so well after Chapter Three (entitled “Eyewitness Identification of Perpetrators”), in a footnote, and only to mention that the Court should never have admitted the identification from the case.[23] There are several reasons why Simon may be uninterested in the constitutional doctrines that govern the phenomena he deems inaccurate. First, the Court’s insistence on process over accuracy may be so endemic that Simon believes it is unable to regulate smaller questions of the constitutionality of reliability without a broader philosophical shift. Second, police investigations are opaque; much of the information courts need to regulate investigators currently goes unrecorded. Lastly, collecting reliable evidence initially is more important than evaluating evidence later, as the presence of faulty evidence induces other faulty evidence through “tunnel vision” of the investigators and the “coherence effect” on all involved.[24]

Two of In Doubt’s main themes, however, suggest the potential for constitutional regulation of flawed processes. First, Simon’s primary recommendation through much of the book is electronic recording of police activity, from interrogations, to identification lineups, to all witness statements. Jurisdictions have begun to adopt electronic recording,[25] and the trend will likely spread. As more information becomes available, courts will be able to better regulate the investigative process.

Second, Simon notes those present in the courtroom are no more accurate in determining factors like the trustworthiness of witnesses than those examining the cold record.[26] The deference that appellate courts give to factfinders is thus largely unearned. If taken seriously, the suggestion that appellate courts should be more aggressive in examining the factual record could also lead to closer regulation of legal processes.

III.

In Doubt is a careful and methodical book, one that resists the temptation to rail against the criminal justice system and instead slowly eviscerates it. It sets aside the egalitarian concern that the system works unequally for the technocratic concern that the system does not work at all. Simon writes in his concluding line that, pending reform, “[W]e have no choice but to question the evidence we use and to live in doubt as to who we set free and who we punish.”[27] But we do have a choice. We can, like the Supreme Court, continue to indulge the fiction that jury trials as currently existing are unparalleled factfinders, or we can begin the hard work of piecing together a system that delivers acceptable outcomes. If we choose the latter, In Doubt is a good place to start.