10 Chap. L. Rev. 623 (No PDF)

Chapman Law Review

Spring 2007



Mark A. Godsey 

Copyright (c) 2007 Chapman Law Review; Mark A. Godsey


The advent of post-conviction DNA testing in the past twenty years has spawned an “Innocence Revolution,” in which hundreds of Americans imprisoned or on death row for serious crimes like murder and rape have been conclusively proven innocent and released.  DNA testing gives our criminal justice system something like a crystal ball–it allows us to look back in time with absolute clarity to determine how and where things went wrong in ways we never could before.

It was not uncommon in earlier decades to find judicial opinions espousing the view that wrongful convictions simply do not occur in America. Now, however, we know better. From studying the cases of wrongful convictions, we now know that eyewitness identification is not nearly as reliable as once believed.  We also know that hundreds of innocent people have been convicted through the use of “junk science,” such as bite mark analysis or microscopic hair comparison, which DNA testing has proven to be wildly inaccurate. 

Most importantly for my purposes here, we now know that suspects will confess all too often to crimes that they did not commit, leading to wrongful convictions.  Of all the things that DNA has taught us about our criminal justice system, this phenomenon is perhaps the hardest to accept. The idea that a suspect would falsely confess to a crime that he did not commit seems counterintuitive and nonsensical. Psychiatrists and social scientists are examining the reasons why false confessions occur, and we have a long way to go before we will have a complete answer. At any rate, DNA testing has proven that false confessions by innocent people occur more often than most would have ever imagined.

For centuries, constitutional confession law in the United States has been concerned with the reliability and accuracy of confessions. Prior to admitting the confession into evidence, a court in this country had to examine whether the confession at issue was accurate and trustworthy. In the 1986 case of Colorado v. Connelly,  however, the United States Supreme Court suddenly subverted reliability as a factor when considering the admissibility of a confession. Then, ironically, shortly after Connelly was decided, DNA testing started to reveal why we should be very concerned about the reliability of confessions after all. Indeed, within a decade of the Connelly decision, the false confession problem in this country had been laid bare.

In this essay, I attempt to shed light on this legal irony. Part I briefly traces the role of the reliability factor in constitutional confession law from its origins through today. Part II briefly explores our recent discovery of the false confession problem through DNA exonerations. In the Conclusion, I offer some thoughts on how this problem should be resolved.

I. The Reliability Factor: A Historical Examination

Interestingly, courts in the eighteenth and nineteenth centuries were much more aware of and concerned with the problem of unreliable confessions than their modern counterparts. Anyone today who is concerned about innocent people being convicted by the use of false confessions would be envious of the protective doctrines in place during confession law’s infancy. Indeed, prior to the ratification of the Bill of Rights, the common law in England was fast developing a rule of evidence that was highly suspicious of confessions. The rule made it quite easy for a defendant to have his confession thrown out if the interrogator made any sort of promise or threat of the type quite common in modern interrogations.  The reason for this delicacy was the fear of false confessions–that any sort of inducement might cause an innocent person to utter an untruth.  As the seminal 1783 case of The King v. Warickshall  noted:

Confessions are received in evidence, or rejected as inadmissible, under a consideration whether they are or are not intitled [sic] to credit. A free and voluntary confession is deserving of the highest credit, because it is presumed to flow from the strongest sense of guilt, . . . but a confession forced from mind by the flattery of hope, or by the torture of fear, comes in so questionable a shape . . . that no credit ought to be given to it; and therefore it is rejected.

In Warickshall and the American cases that followed its lead in the nineteenth century, even slight pressures–such as those which would be considered the norm in modern interrogation rooms–were deemed impermissible.  Many of these cases involved hints of leniency by the interrogators. In the 1845 American case of State v. Bostick,  for example, a promise by the interrogator–that the suspect would receive leniency and would be “sent home” if she confessed–rendered the resulting confession inadmissible. The court stated:

[W]here promises of favor or threats are used, the great danger is, that the confession, whether verbal or written, may be untrue; proceeding, not from a sense of guilt, but from the influence of hope or fear. In such cases, the confession is rejected. Therefore, a confession obtained by temporal inducement, by threat, or by a promise of hope or favor, having some reference to the party’s escape from the charge, held out by a person in authority, is inadmissible. 

In 1897, the Court in Bram v. United States  incorporated this common law rule of evidence, presently called the “voluntariness rule,” into constitutional confession law by making it the test for admissibility under the Fifth Amendment’s Self-Incrimination Clause.  Bram further illustrates the highly suspicious attitude of courts toward confessions during this era. In Bram, the defendant confessed after the interrogator made a mild and subtle suggestion of leniency.  The Court focused on the following statement by the interrogating detective: “If you had an accomplice, you should say so, and not have the blame of this horrible crime on your own shoulders.”  The Court explained:

[H]ow could the weight of the whole crime be removed from the shoulders of the prisoner as a consequence of his speaking, unless benefit as to the crime and its punishment was to arise from his speaking? . . . Thus viewed, the weight to be removed by speaking naturally imported a suggestion of some benefit as to the crime and its punishment arising from making a statement.

[The detective], in substance, therefore, called upon the prisoner to disclose his accomplice, and might well have been understood as holding out an encouragement that by so doing he might at least obtain a mitigation of the punishment for the crime which otherwise would assuredly follow. 

Acknowledging that the inducement offered by the detective was slight, the Court noted, “the law cannot measure the force of the influence used, or decide upon its effect upon the mind of the prisoner, and therefore excludes the declaration if any degree of influence has been exerted.” 

Flashing forward to the mid-twentieth century, one finds reliability as an important, though not quite integral, factor in determining the admissibility of confessions.  The fifty-year period that is marked by Brown v. Mississippi  in 1936 at the front end, and Colorado v. Connelly  in 1986 at the back end, saw the reliability factor thrive at times before its final demise in Connelly.

Indeed, in the 1936 Brown decision, the reliability factor that originated in the common law voluntariness cases was still going strong. In throwing out the confessions of several African-American defendants who had confessed after being tortured by a sheriff and local mob in Mississippi, the Court focused on the unreliability of confessions obtained in such a manner.  The court held that convicting the defendants solely on such unreliable evidence violated due process as it amounted to a “mere pretense of a trial.” 

In the middle part of the twentieth century, reliability remained an important factor in confession admissibility, but it was no longer the sole driving force. The reliability of a confession was merely one of many factors that a court considered when deciding whether to admit or reject a confession under the governing due process voluntariness test.  Nevertheless, during this era, courts could exclude a confession under due process based solely on the ground that it was unreliable. 

In 1986, however, the Court handed down its decision in Colorado v. Connelly.  In that case, a mentally ill man had walked up to a police officer in Denver and confessed to a murder.  A state-employed psychiatrist testified that the defendant had confessed while suffering from psychotic delusions in which God told him to confess or commit suicide. (The state apparently did not contest the issue of mental illness). Made under such circumstances, the confession was quite unreliable. Indeed, the police were unable to corroborate that there had been an unsolved murder during the month and year in which the defendant claimed he had committed his crime.

Despite such concerns about the confession’s reliability, the Court found no constitutional barriers to its admission into evidence.  Turning 200 years of confession law on its head, the Court stated: “A statement rendered by one in the condition of respondent might be proved to be quite unreliable, but this is a matter to be governed by the evidentiary laws of the forum, and not by the Due Process Clause of the Fourteenth Amendment.” 

With this language, the Court greatly undermined the role of reliability in the equation of confession admissibility. It may be possible to limit this case on the grounds that, because the officer in question did nothing to induce Connelly’s confession, there was no state action. Nevertheless, Connelly made clear for the first time that unreliability is not a stand-alone basis on which a court can exclude a confession on constitutional grounds. 

In sum, the period from Warickshall in 1783 through Bram in 1897 reflects the high-water mark in the criminal justice system’s concern for false confessions. Through the middle part of the twentieth century, reliability remained a chief concern. In 1985, the Court subverted this long-standing policy, holding that a confession’s unreliability could not be the sole reason for excluding a confession under the Constitution.

II. The Recent Discovery of the False Confession Problem

After Connelly was decided in 1986, Barry Scheck and his network of Innocence Projects across the country began using DNA testing to exonerate scores of inmates convicted of serious crimes.  To date, more than 300 individuals have been exonerated in this Innocence Revolution.  Because DNA was either not left at the scene by the perpetrator, not collected by the police, or not properly preserved by the authorities in the vast majority of cases, the number of exonerations to date represents a small fraction of the number of innocent people wrongfully convicted.

The DNA exonerations to date have revealed something particularly remarkable for those who study confessions. In 25% of the cases of wrongful conviction, the innocent person falsely confessed.  Before DNA, there were few avenues to verify a false confession. Simply put, the inmate would claim that he had confessed falsely, but most people, including in many cases his lawyer, would not believe him. Now, scores of case studies involving actual false confessions exist, and they raise troubling questions.

The problem of false confessions has become so pervasive that it frequently appears in the mainstream media. It has been covered by nearly every television show dealing with criminal justice. O, The Oprah Magazine (of all sources), recently explored this phenomenon in a glossy full-color feature. 

While perhaps the most famous case of false confessions is the Central Park jogger case,  pop fiction writer John Grisham’s recent work and first piece of non-fiction, The Innocent Man, may do the most to bring public awareness to this phenomenon. In a riveting fifteen-page section in the middle of the book, Grisham puts the reader in the shoes of the suspect during a five-hour interrogation. He describes the pressures, emotions and thoughts that lead to a false confession. In the end, the suspect, Tommy Ward, who was later sent to death row on the basis of his false confession, decides to “play along” with the police and confess for two reasons: First, he is exhausted and worn down by the relentless questioning and pressure of the tag-team interrogation.  Second, he has an undying belief in the criminal justice system.  He believes that no innocent person could be convicted in this country. He naively believes that if he tells a story, the questioning will end and the truth will quickly be discovered.  Ward rationalizes: “Tomorrow, or the next day, the cops would realize that the story did not check out. They would talk to Karl, and he would tell the truth. They would find Odell Titsworth, and he would laugh at them. Play along. Good police work will find the truth.”  The perceived low risk that the false confession would stand the test of time, coupled with the overwhelming desire to say anything to end the barrage of intense questioning and pressure, gave way and resulted in a false confession. While I cannot do justice to Grisham’s work in this short essay, anyone who struggles to understand how a rational and sane person could falsely confess to a murder should read the true story of The Innocent Man.

It may take years for social scientists and psychiatrists to fully understand why and how often false confessions occur. In the meantime, University of San Francisco Law Professor Richard Leo and others have produced a rich series of scholarly articles examining the root causes of this problem.  For now, however, we know that false confessions are a problem, we know that many innocent people have been convicted as a result of false confessions, and we know that, because of Colorado v. Connelly, courts have no constitutional grounds on which to deal with this problem.


Many have offered ways to deal with the problem of false confessions. Videotaping interrogations is an obvious improvement, as it allows a judge and jury to accurately see what went on in the interrogation room.  Several of the documented false confessions were videotaped, and watching a replay of the interrogation allows one to see why the suspect falsely confessed in a way that could never be understood without such documentation. Richard Leo and others have suggested a corroboration rule and new reliability test based on a multitude of factors. 

I endorse all of these approaches. My point in this essay, however, is less ambitious than attempting to solve the problem. I seek merely to bring to light a sad irony in the development of our confession law. The Supreme Court killed off the “reliability policy” as a factor in constitutional confession law in 1986, and then, suddenly, the problem of unreliable confessions surfaced with a vengeance. Today we have empirical proof that false confessions occur more often than we have ever imagined, yet our courts are substantially less concerned with the problem than they were 200 years ago.

I have previously written that the Self-Incrimination Clause should not be concerned with reliability; I stand by that statement today.  The Court in Connelly erred, however, when it held that the Due Process Clause is not concerned with the admission of unreliable confessions. The Connelly Court offered no legal basis for this holding, and ignored or distorted history to reach its conclusion. Indeed, confession law and the Due Process Clause have been steeped in concerns about reliability for more than 200 years, and rightfully so. No case is better than Brown v. Mississippi for illustrating how due process is offended by the introduction of a false or perjured confession that results in a “pretense of a trial.”

As we move forward into the next century of confession law, it is my hope that we learn from our recent discovery of the false confession problem, recognize our mistake in Connelly, and heed the lessons of courts from a bygone era.

Copyright © 2007 by Mark A. Godsey

Professor of Law, University of Cincinnati College of Law; Faculty Director, Lois and Richard Rosenthal Institute for Justice/Ohio Innocence Project. Former Assistant United States Attorney, Southern District of New York, 1996-2001. E-mail: This essay is the written version of my talk at the Chapman Law Review symposium entitled Miranda at 40: Applications in a Post-Enron, Post-9/11 World. I would like to thank the Chapman University School of Law School and the Chapman Law Review for hosting this outstanding symposium, and all the symposium participants who helped provide feedback that shaped and improved this essay. I would also like to thank UC law student Bobbi Madonna for her outstanding research assistance.
Samuel R. Gross et al., Exonerations in the United States: 1989 Through 2003, 95 J. Crim. L. & Criminology 523, 524 (2005).
Arnold Loewy, Systemic Changes That Could Reduce the Conviction of the Innocent 4 (UNC Legal Studies Research Paper No. 927223, 2006), available at; Gross, supra note 1, at 542. See also Innocence Project–Eyewitness Misidentification,
Loewy, supra note 2, at 7. See also Innocence Project–Unreliable and Limited Science,
Richard A. Leo et al., Bringing Reliability Back In: False Confessions and Legal Safeguards in the Twenty-First Century, 2006 Wis. L. Rev. 479, 484-85.
479 U.S. 157 (1986).
Mark A. Godsey, Rethinking the Involuntary Confession Rule: Toward a Workable Test for Identifying Compelled Self-Incrimination, 93 Cal. L. Rev. 465, 481-85 (2005).
168 Eng. Rep. 234 (1783).
Id. at 234-35.
Godsey, supra note 6, at 482-88.
4 Del. (4 Harr.) 563 (1845).
Id. at 565.
168 U.S. 532 (1897).
Godsey, supra note 6, at 474-88.
Bram, 168 U.S. at 562.
Id. at 539.
Id. at 564-65.
Id. at 543.
Godsey, supra note 6, at 488-99.
297 U.S. 278 (1936).
479 U.S. 157 (1986).
Brown, 297 U.S. at 281-82, 286.
Id. at 286.
See Leo, supra note 4, at 493-96; Godsey, supra note 6, at 488-91.
See Leo, supra note 4, at 493-94. But see id. at 495-96 (“The due process voluntariness test continued to evolve in the 1950s and 1960s as the Supreme Court began to recognize that an entirely false or unreliable confession could, logically, be considered voluntary and thus admissible into evidence against a criminal defendant.”).
479 U.S. 157 (1986).
Id. at 160-61.
Id. at 167.
Id. (internal citation omitted).
Leo, supra note 4, at 499 (“With Connelly, the Court abandoned the reliability rationale upon which that doctrine originally had been premised.”). In holding that there was no “state action” in the case, because the officer did nothing to induce Connelly’s confession, the Court ignored the admission into evidence of the confession as a possible basis for state action. Connelly, 479 U.S. at 165-67. The admission into evidence of a false confession had been the predicate “state action” in Brown v. Mississippi. See 297 U.S. at 286-87.
See generally Barry Scheck, Peter Neufeld & Jim Dwyer, Actual Innocence: Five Days to Execution and Other Dispatches from the Wrongly Convicted (2000).
Gross, supra note 1, at 524.
Leo, supra note 4, at 484.
Melba Newsome, True Crimes, False Confessions, O, The Oprah Mag., Apr. 2006, at 235.
For a detailed discussion of the Central Park Jogger Case, see generally Leo, supra note 4.
John Grisham, The Innocent Man: Murder and Justice in a Small Town 85-100 (2006).
Id. at 91, 93.
Id. at 93.
For citations to the empirical studies on false confessions, see Leo, supra note 4, at 512-20.
See Leo, supra note 4, at 530. See also generally Christopher Slobogin, Toward Taping, 1 Ohio St. J. Crim. L. 309 (2003) (discussing the need for taping police interrogations and outlining three constitutional grounds which would mandate taping as a routine practice).
Leo, supra note 4, at 520-39.
Godsey, supra note 6, at 474-99.