THE CASE FOR THE ABOLITION OF CRIMINAL CONFESSIONS
Guha Krishnamurthi*

Confessions are often considered the “gold standard” of evidence in criminal
proceedings. But in truth, confession evidence imposes significant harms on our
criminal justice system, through false convictions and other violations of
defendant’s due process and moral rights. Moreover, our current doctrine is unable
to eliminate or even curb these harms.
This Article makes the case for the abolition of confession evidence in criminal
proceedings. Though it may seem radical, abolition is both sensible and best
furthers our penological goals. As a theoretical matter, confession evidence has
low probative value, but it is prejudicially overvalued by juries and judges.
Consequently, this overvaluation means both that innocent defendants are
systemically pressured into proclaiming their guilt and that juries are so swayed
by it — even in light of countervailing evidence — that they render wrongful
convictions. Indeed, as practice and empirical evidence demonstrate, this is not
merely a theoretical possibility: false confessions and resulting miscarriages of
injustice occur with disturbing frequency. Moreover, confession evidence, and the
methods to obtain it, impose significant harms on defendants in terms of their due
process and moral rights, due to the pressures of interrogation, investigation, and
jeopardy. And our current constitutional and evidentiary doctrine is incapable of
addressing these harms, for it fails to recognize that false confessions are often
caused by overwhelming pressures endemic to our criminal justice system.
Consequently, solving these problems requires a comprehensive, properly focused
solution that goes far beyond our current doctrinal hodgepodge.
The abolition of confession evidence meets that demand. Compared to other
solutions that have been proposed, such as further limiting law enforcement and
prosecutorial conduct or introducing expert testimony and evidence, the abolition
of such evidence best apprehends and mitigates the epistemic and moral concerns
arising from confession evidence and interrogation. In addition, it coheres with and
flows from the Constitution’s due process requirement of voluntariness in
confessions and the evidentiary requirements of reliability. Finally, it preserves and
improves key features of our criminal justice system, namely interrogation, plea
bargaining, and the assessment of evidence.

*
I would like to thank Susannah Barton Tobin, Warigia M. Bowman, Joshua Braver, Martin Brenner, Jenny
Carroll, Russell Christopher, Ryan William Copus, Colin Doyle, Sheldon Evans, Thomas Ward Frampton, Stephen
Galoob, Charanya Krishnaswami, the Honorable Goodwin H. Liu, Alexander Platt, Betsy Rosenblatt, Blaine Saito,
Peter Salib, Seth Stoughton, and Carly Zubrzycki. I am grateful and indebted to Alexandre Diniz de Oliveira for
excellent research assistance. All errors and omissions are mine.

Electronic copy available at: https://ssrn.com/abstract=3730499

I.

Introduction

NEMO TENETUR SEIPSUM PRODERE.1 It is the rule that no person shall be compelled to
incriminate themselves.2 It finds place in the Constitution’s Fifth Amendment privilege against
self-incrimination,3 and it is one that most everyone is familiar with, due to the famous phrase
from Miranda: “the right to remain silent.”4 The nemo tenetur rule found footing in England,
through the supplantation of ecclesiastical courts and practice by common law courts and
practice.5 One of the principal reasons for the genesis of the nemo tenetur rule was to avoid
imposing upon the accused the cruel choice that criminal inquiry had forced upon them. The
ecclesiastical courts would force the accused to adopt an oath to truthfully answer all questions
posed, on pain of damnation, in order to learn of the formal charges that were made against them.
The purpose of the oath was to extract a confession, and refusal to accept the oath was itself
nearly dispositive evidence of the accused’s guilt. This posed a cruel choice for the accused to
either “cut[] one’s throat with one’s tongue” or “suffer[] eternal damnation.” 6 The imposition by
the state that a defendant testify and therefore choose between worldly punishment and eternal
damnation was considered below the dignity of the state. This is also referred to as the “cruel
trilemma”: the Hobson’s choice between perjury, contempt, or self-incrimination. 7 In American

1

John H. Wigmore, Nemo Tenetur Seipsum Prodere, 5 HARV. L. REV. 71 (1891).

2

Id.; Andrew J. M. Bentz, The Original Public Meaning of the Fifth Amendment and Pre-Miranda Silence, 98
VA. L. REV. 897, 899–900 (2012).
3

U.S. CONST., amend V (“No person . . . shall be compelled in any criminal case to be a witness against
himself . . . .”).
4

Miranda v. Arizona, 384 U.S. 436 (1966).

5

LEONARD W. LEVY, ORIGINS OF THE FIFTH AMENDMENT 43–47 (1968).

6

Laurence A. Benner, Requiem for Miranda: The Rehnquist Court’s Voluntariness Doctrine in Historical
Perspective, 67 WASH. U. L.Q. 59, 75 (1988) (quoting L. LEVY, THE ORIGINS OF THE FIFTH AMENDMENT 330
(1968)).
7

Peter Westen, Answer Self-Incriminating Questions or Be Fired, 37 AM. J. CRIM. L. 97, 112 (2010) (“The
‘cruel trilemma’ is a Hobson’s choice between (1) responding truthfully and being penalized on the basis of one’s

2

Electronic copy available at: https://ssrn.com/abstract=3730499

history, the most stark example is the Salem Witch Trials of 1692, where a number of people,
mostly women, were tried for witchcraft on the basis of confessions extracted through torture
and threats of damnation.8
These considerations evolved into the enshrinement of a defendant’s right against selfincrimination in the Fifth Amendment and the requirement that a defendant’s incriminating
statements be made voluntarily.9 Yet, despite these general advances in protecting against selfincrimination, a similarly odious choice arises for defendants in our criminal justice system.
Consider an innocent defendant investigated for a serious crime, with substantial punishment at
stake. The defendant is brought in for interrogation and through the course of law-enforcement’s
admonitions, the defendant becomes aware of certain facts: that they are a suspect in the crime
and that conviction for the crime can result in substantial punishment. Two further points are
impressed upon the defendant: There is evidence sufficient for an actual jury to convict the
defendant. And if the defendant confesses, the defendant will receive a substantially lighter
punishment than they faced otherwise. Here, the innocent defendant has a terrible choice: Risk a
prosecution, which may result in the harsh penalty, or falsely confess and gain a significant
chance of escaping the harsher penalty.
In theory, this is where procedural protections enter to protect the defendant from this
choice. Requirements of independent evidentiary corroboration, ensuring the voluntariness of the
confession as a matter of due process, the right to remain silent and the right to an attorney,
before and during interrogation, and the right to access any exculpatory information that the

answers; (2) responding falsely and being penalized for it; and (3) remaining silent in the face of incriminating
questions and being penalized for it.”)
8

ELIZABETH REIS, DAMNED WOMEN: SINNERS AND WITCHES IN PURITAN NEW ENGLAND 131 (1999).

9

See infra Part VI.A for a fuller discussion of the history of the development of these legal protections.

3

Electronic copy available at: https://ssrn.com/abstract=3730499

government has all are supposed to minimize the possibility or the gravity of this terrible choice.
Indeed, actual practice shows their implementation has utterly failed to safeguard defendants’
due process rights.
There have been multitudes of verified false confessions where tried and true law
enforcement techniques have succeeded in obtaining false confessions. Consider the Central Park
jogger case, in which10 five Black teenagers, Raymond Santana, Kevin Richardson, Yusef
Salaam, Antron McCray, and Korey Wise, were detained and subjected to intense, protracted
interrogation in connection with the grievous rape and assault of Trisha Meili. 11 As a result, all of
them confessed, were convicted, and were incarcerated. But they weren’t the perpetrators, as
confirmed by DNA evidence. 12 The five teens had falsely confessed. 13 Why? Because the NYPD
had used a battery of techniques, including intimidation, isolation, and threats and inducements
regarding punishment.14 And this is no rarity: It is the design of interrogation and investigation.
Both law enforcement and the prosecution use techniques to capitalize on the pressures of
investigation and adjudicative jeopardy to induce confessions from defendants. The results are
devastating: defendants are pressured into confessing, and that confession evidence, as potent as
it is unreliable, leads to false convictions. In addition, apart from the fact that confession
evidence is perilously unreliable, the process of obtaining confession evidence through

10

Robert J. Norris & Allison D. Redlich, Seeking Justice, Compromising Truth? Criminal Admissions and the
Prisoner’s Dilemma, 77 ALB. L. REV. 1005, 1006 (2014).
11

Id.

12

Id.

13

Id.

14

Id. at 1006–07.

4

Electronic copy available at: https://ssrn.com/abstract=3730499

interrogation and investigation that coerces defendants violates defendants’ due process rights
and subjects them to moral harm.
Consequently, I argue that we should abolish — or categorically exclude — confession
evidence from criminal prosecutions. First, I contend that confession evidence is of low
probative value, yet is prejudicially overvalued by juries and judges. Consequently, any
supposed epistemic benefits are vastly outweighed by the costs, especially given our
commitments to minimizing the punishment of the innocent. Second, as shown by practice and
empirical evidence, false confessions, and resulting false convictions, are prevalent to such a
degree to undermine confidence in the criminal justice system. Third, confession evidence, and
the law-enforcement and adjudicative processes of acquiring it, impose significant harms on
defendants in terms of their due process and moral rights. Fourth, our constitutional and
evidentiary doctrine is critically misfocused, such that it is incapable of addressing the persisting
harms of confession evidence. Indeed, that misfocus is foundational, and thus even extensions of
our current doctrine are doomed to fail.
This Article proceeds in six further Parts. In Part II, I set forth preliminaries and
definitions that shape the contours of the analysis. In Part III, I present the theoretical case that
confession evidence is epistemically weak. This is primarily because there are rational reasons
for any defendant to confess, and thus the fact of confession does not differentiate the innocent
from the guilty. Because confession evidence is popularly viewed as dispositive of a defendant’s
guilt, jurors and judges are likely to overestimate the value of confession evidence, which in turn
creates incentives for the state to deploy coercive tactics to extract confessions from innocent
defendants. In Part IV, I bolster this theoretical analysis with practical and empirical data. I show
that law enforcement practice manuals on interrogation encourage exploiting the defendant to
5

Electronic copy available at: https://ssrn.com/abstract=3730499

induce confessions, that defendants often encounter the pressures of the criminal jury system in
deciding whether to confess, and that studies show a surprising commonality in the occurrence of
false confessions. In Part V, I contend that separate from the perils of false confession and
conviction, the process of obtaining confession evidence, by interrogation and investigation that
coerces defendants, is practically certain to cause moral harms to the defendant and thus violate
the defendant’s rights. For this, I marshal our understanding of torture to elucidate how these due
process and moral harms arise.
In Part VI, I explain how current doctrine is failing to address the continuing harms of
confession evidence. I show that, under current law, Miranda’s prophylactic rules, the due
process voluntariness limitations, and other evidentiary limitations requiring corroboration are
too weak or porous to make a difference. I also contend that most all of our current doctrines to
combat the harms of confession evidence are misfocused by concentrating on wrongful official
conduct or improperly relying on juries to rationally assess evidence that is by its nature
prejudicial.
In Part VII, I set forth the solution to the harms of confession evidence: abolishing
confession evidence from criminal proceedings. I first provide a prima facie justification for the
abolition, explaining how it address the problem of the pressures defendants feel in interrogation
by reducing the stakes of interrogation and removing the terrible choice from the defendant’s
hands. Then I show how abolition is superior to other extant solutions. Finally, I show that
abolition does little to disturb key features of our criminal justice system.
II.

Defining the Abolition of Criminal Confessions

I will make the case that we should “abolish confessions from criminal prosecutions.” For
our purposes, I define a “confession” as a statement made by a defendant claiming that the
6

Electronic copy available at: https://ssrn.com/abstract=3730499

defendant committed a crime, or satisfied particular elements of a crime, with the knowledge that
the statement will be used by the government in a criminal prosecution of the defendant to
establish or help establish an element of a crime, based on prior conduct. 15 Importantly here, the
“confession” excludes any corroborating information by the defendant — it is only the statement
by the defendant that the defendant committed the crime or satisfied an element of the crime.
One consequence of the definition of confession is that statements made by a defendant
to an undercover officer or an informant that the defendant engaged in particular criminal
conduct are not confessions, because the defendant would not know that such statements would
be used by the government. Also, the definition of confession requires that the defendant know
that the statement made will be made to the government in order to establish or help establish an
element of a crime. If a defendant, for example, makes false statements to law enforcement and
prosecuting authorities because the defendant seeks to avoid liability, and law enforcement and
prosecuting authorities determine that they are false, the ban on confession evidence would not
necessarily exclude those false statements by the defendant from evidence.
By “abolish,” I mean that we should implement a rule that confession evidence should be
categorically excluded without exception, when the particular conditional is satisfied.
Importantly, the abolition of confession evidence is ban on such evidence as confession
evidence. It is a separate question whether such evidence can be introduced by the prosecution

15

Black’s Law Dictionary defines confession as follows: “confession n. (14c) 1. A criminal suspect’s oral or
written acknowledgment of guilt, often including details about the crime.” BLACK’S LAW DICTIONARY (11th ed.
2019).
There is an important distinction between false and untrustworthy confessions: “A false confession may be
defined as one in which the facts admitted in the confession appear to be either totally incorrect or materially
inaccurate. An untrustworthy confession, on the other hand, should be defined as one that is obtained under
circumstances that provide significant doubt as to its accuracy.” Welsh S. White, False Confessions and the
Constitution: Safeguards Against Untrustworthy Confessions, 32 HARV. C.R.-C.L. L. REV. 105, 109 (1997).

7

Electronic copy available at: https://ssrn.com/abstract=3730499

for other purposes. For example, if a defendant takes the stand and testifies contrary to a prior
confession, the abolition of confession evidence does not necessarily exclude using the prior
confession to impeach the defendant. 16 For purposes of this Article, this abolition is limited to
criminal prosecutions of the putative confessor. I am not here addressing the question of whether
such evidence should be allowed in civil trials, in criminal prosecutions of other individuals, or
in any other kind of legal proceeding.
Finally, the term “based on prior conduct” is included to criminal offenses based on a
concurrent statement to law enforcement and prosecuting authorities. For example, if an
individual tells a material lie to authorities, such that the individual commits perjury, then the
government will have to use that statement to purse the perjury charge. 17 The abolition of
confession evidence is therefore restricted to statements about prior crimes, and thus does not
address situations where the statements themselves constitute part of the criminal act.
The thesis can be stated as follows: we should exclude all confessions — statements
made by defendants knowingly made so that the government may use the statement to establish
or help establish elements of a crime based on prior conduct — from evidence in criminal
proceedings against the defendant, insofar as they are used for the purpose of directly
establishing elements of a crime.

16

This is similar to the Supreme Court’s holding that a confession obtained in violation of Miranda’s strictures
on interrogations is still admissible to impeach inconsistent testimony by a defendant. Kansas v. Ventris, 556 U.S.
586, 593–94 (2009). That said, I am not taking the position that such evidence should be admissible — it’s just a
separate question from the one I address here. There may be compelling reasons to exclude such evidence.
17

The crime of “perjury” is generally defines as making a statement, under oath, on a material matter that the
utter does not believe to be true. See 18 U.S.C. § 1621. Without the “prior crimes” limitation, we can envision
paradoxical situations where a defendant perjures themselves, and thus in some sense knows that the government
may use those statements to prosecute the defendant. The naïve abolition would restrict the government from using
those statements. But that is not the intention of abolition, hence the restriction to prior crimes.

8

Electronic copy available at: https://ssrn.com/abstract=3730499

III.

The Epistemic Frailty of Confession Evidence

Confession evidence is epistemically weak — so weak that it in fact corrupts the factfinding process. There are two distinct senses to how confession evidence exhibits epistemic
weakness. First, where the potential punishment is substantial, confession evidence is
untrustworthy and therefore cannot meaningfully supply the evidence necessary to prove the
criminal charge beyond a reasonable doubt. This means that confession evidence cannot actually
make a significant difference in the determination as to whether the criminal charge is proven.
Second, despite this, confession evidence is weighed greatly by jurors beyond its rational merit.
Consequently, confession evidence further skews the rational weighing of evidence in a criminal
proceeding.
A.

Confession Evidence Cannot Prove the Charge Beyond a Reasonable
Doubt

I contend that confession evidence cannot make the difference in proving a charge
beyond a reasonable doubt. To understand this, we first need to understand the meaning of the
beyond-a-reasonable-doubt (“BARD”) standard. That is a substantial question in itself. Larry
Laudan has called the BARD standard “obscure, incoherent, and muddled.” 18 James Whitman,
while acknowledging BARD as fundamental and familiar, describes the standard as “vexingly
difficult to interpret and apply.”19 “Not surprisingly, jurors, whose understanding of the standard
is perhaps most important, are not immune to BARD’s complexities either. Oftentimes jurors
have “only the haziest notion” of what BARD truly means and are “understandably baffled”

18

LARRY LAUDAN, TRUTH, ERROR, AND CRIMINAL LAW: AN ESSAY IN LEGAL EPISTEMOLOGY 30 (2006)
[hereinafter LAUDAN, TRUTH, ERROR, AND CRIMINAL LAW].
19
JAMES Q. WHITMAN, THE ORIGINS OF REASONABLE DOUBT: THEOLOGICAL ROOTS OF THE CRIMINAL TRIAL 1
(2008).

9

Electronic copy available at: https://ssrn.com/abstract=3730499

when applying it.20 Compounding the problem, judicial instruction on BARD is typically less
than helpful, and often introduces conceptual and various other errors into the equation. 21
The courts themselves have provided little guidance on the nature and contours of the
BARD standard.22 For example, in a trilogy of cases, the Supreme Court has made the attempt to
further elucidate the BARD standard, but with little success. Prior to Winship, in 1954, the Court
in Holland v. United States allowed an instruction defining reasonable doubt as “the kind of
doubt . . . which you folks in the more serious and important affairs of your own lives might be
willing to act upon.”23 The Court then explained that the better instruction would have been to
cast the doubt as “the kind of doubt that would make a person hesitate to act,” instead of in terms
of willingness.24
Then, in 1990, in Cage v. Louisiana, the Court considered instructions to the jury that a
reasonable doubt was “an actual substantial doubt” that “would give rise to a grave uncertainty,”
and required proof to a “moral certainty.”25 The Court held that this violated the Due Process
Clause, because it required a higher level of doubt for the jury to acquit than under the BARD
standard.26 Finally, in Victor v. Nebraska, the Court considered jury instructions defining
reasonable doubt as “an actual and substantial doubt arising from the evidence” and again

20

LAUDAN, TRUTH, ERROR, AND CRIMINAL LAW, supra note 18, at 31; WHITMAN, supra note 19, at 1.

21

LAUDAN, TRUTH, ERROR, AND CRIMINAL LAW, supra note 18, 51–62.

22

Marc L. Miller & Ronald F. Wright, CRIMINAL PROCEDURES: CASES, STATUTES, AND EXECUTIVE MATERIALS
1319 (3d ed. 2007) (“Most state and federal courts discourage or prohibit trial judges from efforts to quantify the
reasonable doubt standard.”)
23

Holland v. United States, 348 U.S. 121, 140 (1954).

24

Id.

25

Cage v. Louisiana, 498 U.S. 39, 41 (1990).

26

Id.

10

Electronic copy available at: https://ssrn.com/abstract=3730499

relating reasonable doubt with “moral certainty.” 27 This time, the Court found these instructions
permissible, if accompanied with other instructions on how to apply the BARD standard
properly, such as with the Holland definition.28 The majority opinion, authored by Justice
O’Connor, went on to say that “the Constitution neither prohibits trial courts from defining
reasonable doubt nor requires them to do so as a matter of course.” 29 So the best formulation the
Supreme Court has given us is that reasonable doubt is “a doubt that would cause a reasonable
person to hesitate to act.” But, as Justice Ginsburg observed in her concurrence in Victor, this is
still ambiguous and unfamiliar to jurors. 30
There are two prominent scholarly accounts of how to understand the BARD standard as
an epistemic matter. The first, which I call the narrative account, considers whether there is a
plausible explanation of the evidence that is consistent with the defendant not committing the
crime, i.e. innocence.31 The second, which I call the probabilistic account, considers whether the
likelihood that the defendant committed the crime is over a certain probabilistic threshold — say
90% or 95%.32 I contend that under both accounts, confession evidence cannot make a practical

27

Victor v. Nebraska, 511 U.S. 1, 18, 21 (1994).

28

Id. at 20–22.

29

Id. at 5.

30

Id. at 23–25 (Ginsburg, J., concurring). Justice Ginsburg herself suggested an instruction from the Federal
Judicial Center’s Pattern Criminal Jury Instructions: “Proof beyond a reasonable doubt is proof that leaves you
firmly convinced of the defendant’s guilt. There are very few things in this world that we know with absolute
certainty, and in criminal cases the law does not require proof that overcomes every possible doubt. If, based on
your consideration of the evidence, you are firmly convinced that the defendant is guilty of the crime charged, you
must find him guilty. If on the other hand, you think there is a real possibility that he is not guilty, you must give
him the benefit of the doubt and find him not guilty.” Id. at 26–27 (Ginsburg, J., concurring). This instruction may
be better, but the terms “firmly convinced” and “real possibility” may still leave jurors flummoxed.
31
See, e.g., Michael S. Pardo & Ronald J. Allen, Juridical Proof and the Best Explanation, 27 LAW & PHIL.
223, 230 (2008).
32

Jacob Schuman, Probability and Punishment: How to Improve Sentencing by Taking Account of Probability,
18 NEW CRIM. L. REV. 214, 220–21 (2015) (providing a summary of using threshold probabilities to understand the
beyond-a-reasonable-doubt standard).

11

Electronic copy available at: https://ssrn.com/abstract=3730499

difference in proving the case beyond a reasonable doubt: either (1) confession evidence in
addition to the other evidence will not meet the BARD standard; or (2) confession evidence in
addition to the other evidence will meet the BARD standard, but so would the other evidence
without the confession evidence. In either case, confession evidence cannot make a significant
impact in proving the criminal charge to the BARD standard.
1.

The Narrative Account of BARD

As discussed, the narrative account of the BARD standard is formulated in terms of
whether there is a plausible explanation of the evidence consistent with the innocence of the
defendant. To understand this formulation, we need first to investigate its foundations, which are
based in inference to the best explanation (“IBE”).33 IBE, also known as abductive reasoning, is
a method of choosing between competing candidate explanations. 34 IBE states that, when
confronted with different explanations for phenomena, we should examine the explanatory
virtues of each of the respective explanations — such as, consilience, simplicity, coherence, lack
of ad hocery, testability, and internal consistency — and defeasibly accept as true the best
explanations.35 Such an inference pattern is common in the reasoning conducted during daily
life.36
Inference to the best explanation is grounded in the notion that there are criteria for what
makes an explanation a good one, and our assessment of an explanation on those criteria can

33

G. Harman, The Inference to The Best Explanation, 74 PHIL. REV. 88, 89 (1965) (coining the term).

34

David A. Schum, Species of Abductive Reasoning in Fact Investigation in Law, 22 CARDOZO L. REV. 1645,
1659 (2001); Guha Krishnamurthi, Jon Reidy, Michael J. Stephan, Bad Romance: The Uncertain Promise of
Modeling Legal Standards of Proof with the Inference to the Best Explanation, 31 REV. LITIG. 71, 72 (2012)
[hereinafter Bad Romance].
35

Harman, supra note 33, at 90.

36
Leo Groarke, Informal Logic, in STANFORD ENCYCLOPEDIA PHIL. (2008), available at http://
plato.stanford.edu/archives/fall2008/entries/logic-informal/.

12

Electronic copy available at: https://ssrn.com/abstract=3730499

warrant our inference that such an explanation is true. 37 It is an inference — so it allows us to
move from certain premises to others in a way that generally preserves truth. When there are
multiple explanations for a phenomenon, we may infer that the best explanation for the
phenomenon, in light of the relevant criteria, is true.38 Schematically, IBE has the following
general structure:
P1. f1, f2, f3, . . ., fn are facts that require explanation.
P2. h1, h2, h3, . . ., hn are each distinct explanations of the set of facts {f1, f2,
f3, . . ., fn}.
P3. The set of explanations {h1, h2, h3, . . ., hn} is the product of an earnest, good
faith search and contains all discovered explanations.
P4. h1 is the best explanation among {h1, h2, h3, . . ., hn} for {f1, f2, f3, . . ., fn}.
C. Ergo, h1 is probably true.39
At this juncture, there is very little substance to IBE, because the question remains: What
are the criteria that make an explanation good? Unsurprisingly, there are myriad accounts of how
to assess the explanatory virtue of an explanation. Some contend that the “goodness” of an
explanation is dependent on its simplicity, plausibility, and the absence of ad hocery.40 Others
have argued that goodness is based on coherence with background beliefs, consilience,

37

Pardo & Allen, supra note 31, at 223–24.

38

Id.

39
Bad Romance, supra note 34, at 72 (citing Larry Laudan, Strange Bedfellows: Inference to the Best
Explanation and the Criminal Standard of Proof 1 (Univ. of Tex. Sch. of Law Pub. Research, Working Paper No.
143, 2010), available at http:// papers.ssrn.com/sol3/papers.cfm?abstract_id=1153062). This formulation of IBE is
inherently comparative, but there is a formulation of IBE for a scenario with only one explanation, wherein we must
have that the candidate explanation is a “good” explanation in terms of the explanatory virtues.
40

Pardo & Allen, supra note 31, at 230.

13

Electronic copy available at: https://ssrn.com/abstract=3730499

testability, and simplicity.41 And others look to predictive power and internal consistency. 42
What the particular explanatory virtues are is of course controversial, but we needn’t fix that
here. All that is important is that there is some plausible set of explanatory virtues on which we
can rely.
There are two phases of IBE, which already generally mirror trial structure. In the first
phase, we generate candidate explanations.43 In the second phase, we evaluate those
explanations.44 This maps onto two stages of a trial: (1) the presentation of evidence by involved
parties and (2) the evaluation of the evidentiary presentations by judge or jury. 45 The
identification of facts to be explained, the search for explanations, and the generation of a set of
candidate explanations are all accomplished when the involved parties present their respective
cases.46 The identification of the candidate explanation which provides the best explanation and
selecting that explanation as the winning explanation is accomplished during the judge or jury’s
evaluation of the competing cases presented. 47
Thus, we can formulate the narrative account of BARD as follows: Jurors are searching
for a “sufficiently plausible” explanation of the facts that is consistent with a defendant’s
innocence. If such an explanation is found, then the defendant is acquitted, and if no such

41
WILLIAM LYCAN, JUDGMENT AND JUSTIFICATION 129–30 (1988); Paul Thagard, Why Wasn’t O.J. Convicted?
Emotional Coherence in Legal Inference, 17 COGNITION & EMOTION 361, 362–63 (2003).
42

John R. Josephson, On the Proof Dynamics of Inference to the Best Explanation, 22 CARDOZO L. REV. 1621,
1626 (2001).
43

Pardo & Allen, supra note 31, at 234–35; Schum, supra note 34, at 1659.

44

Pardo & Allen, supra note 31, at 234–35.

45

Id.

46

Id.

47

Id.

14

Electronic copy available at: https://ssrn.com/abstract=3730499

explanation is found, a defendant is determined to be guilty. 48 Jurors, unlike in civil trials, are not
searching for the best explanation of the events, all things considered.
Josephson further elucidates this when he writes, “Guilt has been established beyond a
reasonable doubt when there is no plausible alternative explanation for the data that does not
imply the guilt of the defendant. An explanation is plausible if it is internally consistent,
consistent with the known facts, not highly implausible, and it must represent a ‘real possibility’
rather than a mere logical possibility. A real possibility does not suppose the violation of any
known law of nature, nor does it suppose any behavior that is completely unique and
unprecedented, nor any extremely improbable chain of coincidences.” 49
So, now consider how a trial proceeds. Evidence is presented along with candidate
explanations of the evidence. Imagine a tree of possible explanations of the evidence, and as
further evidence is added, some of the potential explanations are pruned away.
Consider an example. In a murder case, police officers testify that, while on patrol, they
heard three gunshots. They approach an alley and come upon a dead body in the street. They exit
the vehicle and see three additional men alive, uninjured, and standing around the body. (Call
them Alex, Eric, and Reid.) There is a gun on the ground, hot to the touch, with three bullet
casings, matching the number of wounds on the dead body.
At this point, jurors may have a number of plausible explanations of the evidence thus
far: Alex was the shooter, Eric was the shooter, and Reid was the shooter. There also may be
other plausible explanations: They each took a turn and shot the victim. There also might be
48

Of course, there must be some plausible explanation of guilt as well. But generally if there is no plausible
explanation of guilt, we can generate a plausible explanation of innocence too.
49
John R. Josephson, On the Proof Dynamics of Inference to the Best Explanation, 22 CARDOZO L. REV. 1621,
1642 (2001).

15

Electronic copy available at: https://ssrn.com/abstract=3730499

other explanations: The true shooter ran away. Or the victim shot themselves thrice, and these
three were rushing to offer assistance.
Then suppose a forensics expert testifies that DNA from skin remnants on the gun’s
meshed handle match Alex’s DNA and that there is evidence that Alex purchased the gun
recently. Moreover, no skin remnants were found on the gun’s meshed handle that did not match
Alex’s DNA. Furthermore, there is testimony by nearby shop owners that locate Eric and Reid at
a pizza place together very shortly before the shots. After this evidence, the putative explanations
that either Eric or Reid were the shooter may be pruned away — those do not seem plausible.
Then suppose that there is video evidence showing Alex shooting the gun. After that, all
of the other putative explanations besides Alex being the shooter are again pruned away, leaving
that one. That doesn’t mean Alex was guilty of the crime, because this is just about the actus
reus. But we can replicate this kind of analysis for each of the different elements of the offense.
Suppose however the evidence showed that the meshed gun had skin flakes from both
Alex and the victim and that there was no security footage. If that were all the evidence, at least
one plausible explanation that remains is that the victim, known to Alex, shot themselves. On
that complete record then, because there is an explanation that is consistent with innocence —
Alex didn’t commit the actus reus — the narrative understanding of BARD would tell us that
Alex must be found not guilty.
We can now apply this to the potentiality of confession evidence. Let’s revisit the
prototypical confession scenario: A defendant is suspected of committing a crime, conviction of
which carries a significant punishment. As part of the investigation, the defendant is interrogated.
Through the course of the interrogation, the defendant learns certain facts: that they are a suspect
in the crime, conviction of the crime carries a significant sentence, there is a significant chance
16

Electronic copy available at: https://ssrn.com/abstract=3730499

that the defendant will be convicted of the crime and receive a significant sentence, and that
confessing to the crime may lower the chance of receiving such a significant punishment or that
it may lower the punishment, even if it still is significant. And we will assume that the defendant
confesses to the crime.
Now, the confession alone is not enough to sustain a conviction. The Supreme Court has
required sufficient corroboration for the confession — that is evidence other than the confession
to support the claims of the confession. In Opper v. United States, the Court explained the reason
for the corroboration rule: “In our country the doubt persists that the zeal of the agencies of
prosecution to protect the peace, the self-interest of the accomplice, the maliciousness of an
enemy or the aberration or weakness of the accused under the strain of suspicion may tinge or
warp the facts of the confession.”50 And the corroborating evidence can take many forms: It can
be independent evidence that the defendant committed the crime — like physical evidence or
other witness testimony; evidence about the reliability of the confession itself; 51 and evidence
that “demonstrates the individual has specific knowledge about the crime.” 52

50

Opper v. United States, 348 U.S. 84, 89–90 (1954). Opper concerned a conviction for paying a federal
employee compensation for rendering a governmental service. Id. at 85, 94. Opper admitted to paying money to a
federal employee, but denied that the payments were for a government service. Id. at 88. The government, however,
had proved that the employee provided governmental services to Opper. Id. at 93–94. Indeed, the quantum of proof
from the government likely would have satisfied the corroboration test and the corpus delicti rule. Nevertheless, the
Court thought it fit to dispense with the corpus delicti rule in federal prosecutions.
51

Richard A. Leo, Steven A. Drizin, Peter J. Neufeld, Bradley R. Hall, and Amy Vatner, Bringing Reliability
Back in: False Confessions and Legal Safeguards in the Twenty-First Century, 2006 WIS. L. REV. 479, 509 (2006).
With respect to the reliability of the statement, some factors to consider are “evidence as to the spontaneity of the
statement; the absence of deception, trick, threats, or promises to obtain the statement; the defendant’s positive
physical and mental condition, including age, education, and experience; and the presence of an attorney when the
statement is given.” Id. (quoting State v. Mauchley, 67 P.3d 477, 488 (Utah 2003)).
52

Id. Showing specific knowledge can be accomplished by providing information that leads to the discovery of
unknown evidence, providing information about nonpublic unusual elements of the crime, or providing information
about nonpublic mundane details of the crime. Id. (citing State v. Mauchley, 67 P.3d 477, 488 (Utah 2003) (internal
citations omitted)).

17

Electronic copy available at: https://ssrn.com/abstract=3730499

For simplicity, let’s group the evidence in three buckets: (1) the confession itself; (2) the
other evidence that corroborates the confession; and (3) any other evidence in favor of guilt.
Suppose that all of the evidence taken together is sufficient to establish guilt beyond a
reasonable doubt. It will either be the case that (A) the evidence other than the confession —
buckets (2) and (3) the alternative evidence — will be insufficient by themselves to establish
guilty beyond a reasonable doubt such that (1) the confession is required to establish guilt or (B)
the evidence other than the confession is sufficient to establish guilt beyond a reasonable doubt. 53
We consider each of the cases.
(A) The Nonconfession Evidence Is Insufficient.
Here the corroborating and alternative evidence is not sufficient to establish that the
defendant is guilty beyond a reasonable doubt. Based on the narrative account of the BARD
standard, that means the corroborating and alternative evidence do not prune the set of possible
explanations of the evidence to only explanations in which the defendant committed the crime.
So there is at least one plausible explanation — call it E1 — consistent with the corroborating
and alternative evidence in which the defendant did not commit the crime. So, purportedly, the
addition of the confession evidence establishes guilt beyond a reasonable doubt.
However, it is difficult to see how this could be the case, from the vantage point of a
neutral observer that is the juror. The explanation E1 explains all of the corroborating and
alternative evidence in a way that is consistent with innocence of the defendant. Now, with
respect to the confession, consider the following explanation EC1: The defendant did not commit
the crime, but rather understood that there was a significant chance that the defendant would be

53
This point just comes from the basic logical theorem that either a fact is true or false — A ˅ ¬A. It is also
known as the Law of Excluded Middle.

18

Electronic copy available at: https://ssrn.com/abstract=3730499

convicted and receive substantial punishment and the defendant believed that by confessing they
would have a significant chance of avoiding or lowering that punishment. Here’s another
explanation EC2: The defendant did not commit the crime, but in the moment felt so
overwhelmed and coerced by the nature of the interrogation and believed that by falsely
confessing the interrogation would end and thus did so. These are both plausible explanations of
the confession evidence, consistent with innocence. If EC1 or EC2 are consistent with E1, then we
have an explanation of all of the evidence, call it E1+, that is plausible and consistent with
innocence. This then means that the defendant is not guilty under the beyond-a-reasonable-doubt
standard.
This might seem too slick. One might worry whether it is always the case that E1 and EC1
or E1 and EC2 are always consistent with each other. For example, suppose in the investigation of
a murder, the defendant confesses to committing the murder and states with precision the
location of a firearm that could not be located before. Thereafter, the firearm is found and
confirmed to be the murder weapon. Statements in the confession then seemingly provide us
with more information — that the defendant knew the location of the murder weapon — which
may allow us to prune the universe of plausible explanations. It might be that the only candidate
explanations of innocence E1, E2, . . . , En are inconsistent with the defendant knowing the
precise location of an otherwise unfound murder weapon. Here, it helps to revisit the definition
of the confession: A “confession” is a statement made by a defendant claiming that the defendant
committed a crime or satisfied particular elements of a crime, with the knowledge that the
statement will be used by the government in a criminal prosecution of the defendant to establish
or help establish an element of a crime, based on prior conduct. The defendant’s statements
about the whereabouts of the murder weapon and the defendant’s knowledge of the murder

19

Electronic copy available at: https://ssrn.com/abstract=3730499

weapon are not themselves the confession. The statement that the defendant committed the
murder is the confession. The whereabouts and knowledge points are evidence that corroborate
the confession. In such a case, the corroborating evidence of the firearm, its relation to the crime,
and the defendant’s knowledge will do all the work of pruning the tree of possible explanations
(which also means, against our assumption, that the nonconfession evidence is sufficient to
establish guilt). The defendant’s statement that they committed the murder — i.e., the confession
— does no additional work.
Another objection is that there might be situations in which a confession on the record is
significantly costly, such that the resulting calculus would favor the credibility of the confession.
Consider a heinous crime — for example, a sexual offense against a child. Confessing to such a
crime — even if it resulted in a significantly lower offense — carries a huge cost including social
stigmatization. So, the fact that the benefit of confessing is not as clear cut might make us think
that the confession is more likely true. But this kind of argument is unconvincing. If there is a
benefit, even if marginal, then the defendant would be rational to confess (and perhaps deny it
later). That would favor an explanation like EC1 and at least render it plausible. If there is no
benefit, then the defendant acted irrationally, but then that would favor an explanation like EC2
or at the least render it plausible.
Finally, one might also object that neither EC1 nor EC2 are plausible and thus E1+ too is
implausible. But if the antecedent conditions of EC1 or EC2 — namely that the defendant faces
significant punishment and would have reason to believe that confessing would mitigate the
punishment or the defendant was facing an overwhelming interrogation and would have reason
to believe confessing would stop that kind of interrogation — then these are both

20

Electronic copy available at: https://ssrn.com/abstract=3730499

straightforwardly plausible. There may be a question of whether such antecedent conditions were
manifest, and we will revisit those questions shortly.54
(B) The Nonconfession Evidence Is Sufficient
So then suppose that buckets (2) and (3) — the corroborating and alternative evidence —
are sufficient to establish that the defendant is guilty beyond a reasonable doubt. Thus, the
corroborating and alternative evidence is enough to prune the universe of possible explanations
to only ones in which the defendant committed the crime. Indeed, as we have seen above, it must
be the case, because the confession itself cannot prune away plausible explanations of the
conduct that are consistent with innocence. But then, by hypothesis, the confession is not
required to establish that only the plausible explanations are consistent with guilt. Thus, the
confession does not do any additional work necessary to establish guilt.
Here, one might object that the confession evidence could still play an important role.
Though all of the explanations of the evidence are consistent with guilt, there may be a number
of them. And it may be the case that if actual jurors cannot pick between a number of different
theories, though all of them are theories in which the defendant is guilty, the jurors may
irrationally feel doubt and thus vote to acquit. The defendant’s confession may prune some of the
potential theories and thereby narrow the set of possibilities — which in turn would better ensure
that the juror does not act irrationally. As an initial matter, I think this objection relies upon a
blurring between confession evidence and corroborating or alternative evidence — for example,
a defendant’s statement about where they were is not necessarily confession evidence as defined.
Moreover, I think that this objection is impractical and nevertheless unconvincing. Generally,
there might be lots of explanations of the evidence that are immaterially different. For example,

54

See infra Part IV.

21

Electronic copy available at: https://ssrn.com/abstract=3730499

it may not matter to a case whether the defendant walked to the spot of the incident or took
public transport. It is unlikely that the jury would generate irrational doubt based on such
immaterial differences. However, if there are material differences between two explanations —
perhaps different murder weapons, different locations, different time periods, etc. — then it is
most likely that there will also be some plausible explanation of the evidence consistent with
innocence. Thus, the jurors’ doubt would not be irrational, contra the impractical hypothesis.
Moreover, even if juries were to generate irrational doubt because of having to choose between
different explanation of the evidence in which the defendant is guilty, it is not clear that
confession evidence would help. Jurors could rationally still disbelieve the veracity of the
confession evidence under an explanation, like EC2, that the defendant rambled nonsense
because of being mentally overwhelmed by the criminal process or nature of interrogation.
Indeed, juror irrationality of this form can be tackled more directly by effective advocacy that
explains the government’s standard — it does not require the admission of confession
evidence.55
*

*

*

Thus, under the narrative account of BARD, we have shown that confession evidence
cannot have a significant impact in proving the charge. If the other evidence does not meet the
BARD standard, then the confession evidence will not help attain the BARD standard, because
there is a plausible explanation consistent with innocence. If the other evidence does meet the

55

There is potentially another way in which confession evidence could epistemically aid the jury’s
determination. That a defendant confessed may instruct the jury that the defendant at one point believed the case
against them to be strong. Jurors may have doubts about their own ability to assess the evidence, so a confession
may operate as a kind of confirmation that the case is indeed strong. But admission of the confession for this
purpose is improper because it is a variety of vouching for the strength of the prosecution’s case. See Mary Nicol
Bowman, Mitigating Foul Blows, 49 GA. L. REV. 309, 321 (2015) (observing that prosecution’s vouching for the
strength of the case or credibility of witnesses is improper).

22

Electronic copy available at: https://ssrn.com/abstract=3730499

BARD standard, then the confession evidence is unnecessary. Either way, the confession
evidence is epistemically inert.
2.

The Probabilistic Account of BARD

In the probabilistic account, we can understand the BARD standard as a probabilistic
threshold. In assessing the evidence, if the probabilistic threshold is met, then the juror should
render a verdict of guilty; but short of that threshold, the juror should render a verdict of not
guilty.56 The threshold is usually quantified in the range of 85%–95% likelihood that the
defendant committed the crime.57 Thus, if the odds that the defendant is guilty exceed the
threshold of probability, then they will be convicted. If not, then they will be “‘categorically
acquitted.’”58
The probabilistic account is not without problems. One particularly pressing problem for
operationalizing BARD is the conjunction paradox. 59 The problem arises because of a difference

56

Schuman, supra note 32, at 220–21.

57

See, e.g., Richard O. Lempert, Samuel R. Gross & James S. Liebman, A MODERN APPROACH TO EVIDENCE:
TEXT, PROBLEMS, TRANSCRIPTS AND CASES 1244 n.13 (3d ed. 2000) (discussing studies suggesting that most jurors
place the requirement in the 85% to 90% range); Hal R. Arkes & Barbara A. Mellers, Do Juries Meet Our
Expectations?, 26 Law & Hum. Behav. 625, 630–31 (2002) (citing surveys of judges, where mean and median
numerical responses were at or near 90%, with 90%, 95%, and 100% being the most frequent specific responses).
We can mathematically deduce the proper probability threshold, based on how tolerant we are of wrong
convictions. Consider the Blackstone maxim that “the law holds it better that ten guilty persons escape, than that one
innocent suffer.” 4 WILLIAM BLACKSTONE, COMMENTARIES, at *358. Under this ratio of 10:1 (but not 11:1), we can
calculate that the appropriate probability threshold for BARD is around 91%. At 20:1, it would be about 95%. And
at 100:1, it would be about 99%. More generally, you can calculate the appropriate threshold t, with the formula:
t = 1/((1/r) + 1), where r is the number of guilty individuals we’d rather escape than have an innocent suffer.
Daniel Pi, Francesco Parisi & Barbara Luppi, Quantifying Reasonable Doubt, 72 RUTGERS L. REV.
(forthcoming 2020), draft available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3226479 (explaining
how the derivation of a probabilistic evidentiary threshold from the Blackstonian ratio (on page 30)).
58
Schuman, supra note 32, at 220–21 (citing Talia Fisher, Conviction Without Conviction, 96 MINN. L. REV.
833, 834–35 (2012)).
59

There are also other concerns, including in whether reasonable doubt should be interpreted by the juror as a
subjective standard or as an objective one and whether and to what extent the questions should be individualized or
generalized and statistical. For more on these questions, see Michael S. Pardo, The Paradoxes of Legal Proof: A
Critical Guide, 99 B.U. L. REV. 233 (2019).

23

Electronic copy available at: https://ssrn.com/abstract=3730499

in how the probability threshold is applied: both to the claim as a whole and to the elements of
the claim specifically. So, as a matter of practice, legal doctrine and jury instructions apply
standards of proof “to the individual elements of a claim, crime, or affirmative defense.” 60 Thus,
the prosecution must prove each element of a crime beyond a reasonable doubt. 61 “[P]arties with
the burden of proof will win if they surpass the threshold for each element.” 62
However, as Michael Pardo explains, “[w]hen there are two or more propositions, the
probability of their combination (conjunction) will not only depend on the probabilities of the
individual propositions, it will also depend on the relationship between the propositions. If two
propositions, A and B, are independent of each other (i.e. the probability of one does not affect
the probability of the other), then the probability of the conjunction (A & B) will be A multiplied
by B. For example, if the probability of a coin landing heads up is 0.5, then the probability of
getting two heads in a row is 0.25 (0.5 x 0.5). . . . When the propositions are dependent, then the
probability of both propositions being true (the conjunction) will depend on the conditional
relationship between the propositions. For example, it might be impossible for two propositions
to both be true. Therefore, even if each has a probability of 0.5, the probability of their
conjunction (A & B) may be 0. On the other hand, one proposition may entail the other and thus
the probability of the conjunction (A & B) will be the same as probability of A. The theorem for
dependent propositions states that the probability of the conjunction (A & B) is equal to A
multiplied by B given A.”63

60
Ronald J. Allen & Sarah A. Jehl, Burdens of Persuasion in Civil Cases: Algorithms v. Explanations, 2003
MICH. ST. L. REV. 893, 898–902 (surveying jury instructions).
61

In re Winship, 397 U.S 358, 361 (1970); Pardo, supra note 59, at 267.

62

Pardo, supra note 59, at 267.

63

Id. at 267–69.

24

Electronic copy available at: https://ssrn.com/abstract=3730499

Assume that a crime has four elements, and assume for simplicity’s sake that the
elements are independent of each other. If we apply the BARD standard to each element, then we
will require that each element be shown by the exacting probabilistic threshold — say 90%. But
because of independence and the way probabilities are calculated, this means that there need
only be 66% likelihood of the result (that is, (90%) 4) to convict, and that seems substantively too
low. On the flip side, if we were to require 90% of the overall crime, then each element would
have to be shown to ~97.4% likelihood (that is, (90%)1/4) to convict, which seems substantively
too high.64 Indeed, this is also impacted by the number of elements: a five-element crime could
range from overall 59% likelihood, under the former calculation, to 97.9% likelihood under the
latter calculation.
There are a number of proposed solutions to the conjunction paradox, none of which are
particularly satisfying.65 What’s critical for our purposes is to recognize that whatever solution
must maintain a significantly high threshold for the overall crime — something over 85% (i.e.
the lower part of the designated range). Thus, applying that element-wise, that will require a
significantly high likelihood of each element manifesting, at least 85%, but likely much higher.
With that in mind, we can approach how confession evidence fares under a probabilistic
understanding of BARD. Again, consider the disjunction: either the corroborating and alternative
evidence will not allow a jury to find the defendant guilty beyond a reasonable doubt, or it will.

64

This is the conjunction theorem in probability. E.T. JAYNES, PROBABILITY THEORY: THE LOGIC OF SCIENCE
(2003).
65

Pardo, supra note 59, at 269–75.

25

Electronic copy available at: https://ssrn.com/abstract=3730499

(A) The Nonconfession Evidence Is Insufficient
Again, here, the corroborating and alternative evidence is not sufficient to establish that
the defendant is guilty beyond a reasonable doubt. On the probabilistic account of the BARD
standard, that means the corroborating and alternative evidence allow some nontrivial chance for
the defendant to have not committed the crime. And again, purportedly, the addition of the
confession evidence establishes guilt beyond a reasonable doubt, beyond the probability
threshold.
The problem with this is that the fact of the confession has little independent probative
value and thus it cannot push the prosecution’s case over the threshold. Suppose that the
prosecution’s case is nearly at the threshold without the confession evidence, but not beyond it.
On the state of that information alone, and assuming perfect visibility for the defendant, no
defendant — innocent or guilty — would rationally choose to confess, assuming standardly that
their main objective was to receive as little punishment as possible. 66 On that state of the
evidence, because it is short of the probabilistic threshold, the defendant would be acquitted.
But defendant’s — and prosecutors — often do not have perfect information. And as set
forth before, it may be rational for the defendant to confess. A person who is actually guilty of

66
It might be that the defendant has other reasons to confess, such as an obligatory commitment to being
truthful, a desire to be personally redeemed by society, a hope of avoiding harm and receiving sanctuary by being
imprisoned, etc. These reasons could be completely rational too.

For example, we might have a society in which cultural mores impose on individuals a great internalized sense
of guilt for criminal conduct, and allows for the release of that sense of guilt by confession. As a result, it might be
rational for a defendant to confess. And with such knowledge, one might surmise that there is some higher
likelihood that guilty defendants are more likely to confess than innocent defendants.
I am skeptical that any such norms predominate so that we could find a significantly higher likelihood of
confession among guilty defendants. Indeed, there are other considerations that might lead to confession specific
only to innocent defendants, for example, confession to ensure a loved one escapes investigation. How these varied
considerations relate generally is unknown. Moreover, in the absence of reliable particularized knowledge about the
confessing individual, it will be hard to assess the actual impact of these reasons, for theorists and potential jurors. In
contrast, we do know that defendants seek to avoid criminal investigation, conviction, and punishment.

26

Electronic copy available at: https://ssrn.com/abstract=3730499

the crime may confess, with the awareness that they may be able to avail themselves of a lower
sentence. Or they may confess because of the oppressive nature of the interrogation or law
enforcement and prosecution inquiries. And of course they also might forego that potential
benefit and instead not confess, to see if the state can meet that burden. But very little about this
calculus is actually changed by the fact that the defendant in fact committed the crime. An
innocent defendant also lacks perfect information and may be risk averse or risk taking in the
same way. Consequently, whether a defendant confesses to a crime — by itself — is a poor
probabilistic indicator of whether the defendant in fact committed the crime.
Consider an example. Suppose Alex is a defendant facing a prosecution for an aiding and
abetting in an armed burglary, which carries a 20-year sentence. The circumstances are murky:
Alex was in a vehicle with his friends, and the vehicle stopped at a house, and his friends went
into the house and robbed it. Alex remained in the car.
The BARD threshold is at, say, 90%, and the state of the evidence would support that
Alex was 80% likely to have committed the crime. Suppose Alex is offered a deal: if he
confesses to the crime, he will get a 3-year sentence instead of the full 20 years. Alex could
rationally take this deal, because he may not be able to appreciate the 10% deficit or because he
thinks that jurors are fallible and may not apprehend the 10% deficit — and so the resulting risk
calculus favors confessing. But that calculus is not obviously changed based on whether Alex
committed the crime.67
Now it is possible that Alex may have something against confessing, if he did not do the
crime, which would as a result make him less likely to confess if he were innocent. But, as a
theoretical matter, I see no reason to assume this is generally the case, and I think such intuitions

67

As Red said, “Everyone in here is innocent.” THE SHAWSHANK REDEMPTION (Castle Rock 1994).

27

Electronic copy available at: https://ssrn.com/abstract=3730499

are built largely on a disagreement with a premise of the hypothetical that an innocent defendant
and a guilty defendant would ever be so similarly situated. Moreover, based on Bayes’s
Theorem,68 the difference between the likelihood that a defendant would confess when actually
guilty and that a defendant would confess when actually innocent would have to be significant to
increase the prosecution’s case over the threshold. Filling in some plausible numbers, if the
prosecution’s evidence shows that Alex is 80% likely to have committed the crime, and they
need to attain 90% likelihood to convict, then a guilty defendant must be 2.3 times as likely to
confess than an innocent defendant. All other things being equal, if an innocent defendant is, say
30% likely to confess, then a guilty defendant must be 69% likely to confess, for the confession
to make an evidentiary difference.69 Even if there was a naturally occurring gap, that innocent
defendants would be unwilling to admit to something that they did not do while guilty
defendants would be so willing, if that bore out favorably in prosecutions, guilty defendants
would be quick to mimic that behavior and avail themselves of the same benefits — and vice
versa. Thus, we would predict any such gap to collapse or narrow significantly.
Notably, this argument and the supporting calculations do not assume that innocent
defendants and guilty defendants facing the same risk calculus appear in equal or similar
numbers. There will certainly be many more guilty defendants in this situation than innocent
ones — assuming that our ways of collecting and assessing evidence are truth adaptive. 70 By
hypothesis, Alex is 80% likely to have committed the crime on the available evidence, so we

68

James Joyce, Bayes’ Theorem, STANFORD ENCYCLOPEDIA PHIL. (2003), available at
https://plato.stanford.edu/entries/bayes-theorem/.
69

This can be calculated using Bayes’s theorem. E.T. JAYNES, PROBABILITY THEORY: THE LOGIC OF SCIENCE
(2003).
70
Welsh S. White, Miranda’s Failure to Restrain Pernicious Interrogation Practices, 99 MICH. L. REV. 1211,
1230 (2001) (making the same point).

28

Electronic copy available at: https://ssrn.com/abstract=3730499

should assume that there will be 4 times as many guilty defendants in this position than innocent
ones. But the point remains that the confession evidence itself is not doing any significant work
in separating guilty from innocent defendants.
Consider an analogy from medicine: Suppose two patients come in showing very similar
symptoms — cough, running nose, and fever — and we want to determine whether they have
influenza. Most people who show these symptoms indicate influenza, but we add one more test:
We hand them a box of tissues, and if they take one, then we deem that they have influenza.
That’s a ridiculous test. We wouldn’t expect them to act any differently — both patients
show the same symptoms. Of course, most people who take the tissue will have influenza, but
that’s solely because they came in with symptoms for influenza. The tissue-box test did nothing.
That’s confession evidence: Its epistemic value is tissue thin.71
(B) The Nonconfession Evidence Is Sufficient
Assume on the other hand that the corroborating evidence and the alternative evidence is
sufficient to show that the defendant is guilty beyond a reasonable doubt. On the probabilistic
model, that means that on the available evidence, the defendant’s likelihood of committing the
crime surpasses the probabilistic threshold. If in fact, the defendant’s likelihood of committing
the crime is a high probability event, then we do not require the confession evidence to establish
the guilt of the defendant. The confession evidence is not necessary. 72
One might object that the confession evidence could still be important in bolstering the
other evidence, to show the increased likelihood that the defendant committed the crime. But just

71

One distinction is that the patient may not know whether they have influenza while the defendant actually
does know whether they committed the crime. But that distinction, I think, doesn’t lead to a disanalogy, because the
defendant still may not know whether the evidence is sufficient for conviction — and that is what drives whether
they will confess. I thank Betsy Rosenblatt for raising this point.
72

There may of course be differences in the way that various jurors interpret the information. For present
purposes, I assume a prototypical rational juror, but the framework can be adopted for different jurors: If there is any

29

Electronic copy available at: https://ssrn.com/abstract=3730499

as before, the confession evidence cannot do that because of its epistemic weakness. Innocent
defendants have all the same rational reasons to behave as guilty ones do, and if there is any gap
in their probable behaviors, it’s as a practical matter negligible, unpredictable, and unreliable.
And just as we observed with the narrative understanding of BARD, the better way to tackle the
problems of jurors not apprehending the strength of the evidence is to persuade them to properly
understand the strength of the evidence.
*

*

*

Thus, under the probabilistic understanding of BARD, the same conclusions hold true:
The addition of confession evidence cannot take a quantum of other evidence that would not
sustain a determination of guilt below the BARD threshold to above the BARD threshold. And if
the quantum of other evidence is such that it would sustain a determination of guilty above the
BARD threshold, then the confession evidence is superfluous.
B.

Confession Evidence Is Trusted Beyond Its Probative Value

Thus far, I have shown that confession evidence cannot make a significant practical
difference in a rational jury’s assessment of whether the defendant is guilty beyond a reasonable
doubt. The reason for this is that there are scenarios where a defendant has a rational reason to
confess, if the confession may provide them with an opportunity to get a significantly lower
punishment than they might otherwise face. That rational reason exists whether the defendant is
guilty or innocent. Accordingly, the fact that a defendant confesses provides little reason to think
that the defendant is guilty rather than innocent. As a factual matter, it may be the case that guilty
defendants are more likely to confess than innocent ones, because for example people have a

reasonable juror for whom the evidence is insufficient, then we are in the (A) variety of case. I thank Thomas
Frampton for this question and insight.

30

Electronic copy available at: https://ssrn.com/abstract=3730499

fidelity to the truth that overwhelms their desire to obtain less punishment, because they have
other goals beyond obtaining less punishment, or because innocent defendants have a stronger
belief in the propriety of the justice system. Nevertheless, I contended that such conditions are of
little probabilistic moment. But the fact that evidence is of little value does not necessitate it be
excluded. One might ask then, what is the harm in allowing low-value evidence, along with all
the other evidence?
The problem is that juries do not understand confession evidence as low-value evidence.
Far from it. Indeed, as the Supreme Court said, “A confession is like no other evidence. Indeed,
‘the defendant’s own confession is probably the most probative and damaging evidence that can
be admitted against him . . . . [T]he admissions of a defendant come from the actor himself, the
most knowledgeable and unimpeachable source of information about his past conduct. Certainly,
confessions have profound impact on the jury, so much so that we may justifiably doubt its
ability to put them out of mind even if told to do so.’” 73 As the California Supreme Court has
observed, “the confession operates as a kind of evidentiary bombshell which shatters the
defense.”74 Both Supreme Courts are surely right that the information is the most damaging
evidence.75 Juries take confession evidence at face value, assuming that the defendant would not
confess if they did not in fact commit the crime. 76 Confessions were found in one study to be

73

Arizona v. Fulminante, 499 U.S. 279, 296 (1991) (quoting Bruton v. United States, 391 U.S. 123, 139–140
(1968) (White, J., dissenting)).
74

People v. Schader, 401 P.2d 665, 674 (Cal. 1965). The rule in Schader was that erroneous admission of a
confession was reversible per se under state law. The Supreme Court of California later overturned that rule, in
People v. Cahill, 853 P.2d 1037, 1059–60 (1993).
75

See also Richard J. Ofshe & Richard A. Leo, The Decision to Confess Falsely: Rational Choice and
Irrational Action, 74 DENV. U. L. REV. 979, 983–84 (1997).
76

Eugene R. Milhizer, Confessions After Connelly: An Evidentiary Solution for Excluding Unreliable
Confessions, 81 TEMP. L. REV. 1, 66 (2008); Richard A. Leo & Richard J. Ofshe, The Consequences of False
Confessions: Deprivations of Liberty and Miscarriages of Justice in the Age of Psychological Interrogation, 88 J.
CRIM. L. & CRIMINOLOGY 429, 476 (1988); Welsh S. White, False Confessions and the Constitution: Safeguards

31

Electronic copy available at: https://ssrn.com/abstract=3730499

more prejudicial than other forms of evidence, more than eyewitness testimony and character
evidence.77 “Confession evidence (regardless of how it was obtained) is so biasing that juries
will convict on the basis of confession alone, even when no significant or credible evidence
confirms the disputed confession and considerable significant and credible evidence disconfirms
it.”78 This has led to the famous maxim “confessio est regina probationum, confession is the
queen of proof.”79
The fact that the jury would overvalue confession evidence is not entirely mysterious.
One explanation is simply that jurors do not understand or credit the fact that the defendant may
have other reasons to confess — such as the opportunity to lessen punishment or to escape the
oppressive nature of law enforcement inquiry and interrogation. 80 Additionally, it is practically
difficult for the defendant to challenge confession evidence. To deny the confession or to
question its reliability requires that the defendant question their own statements — which in turn
impacts the credibility of the defendant, as both a litigant and a moral agent. 81 And in a similar
vein, jurors may be more likely to believe that the statement against the defendant’s interest is

Against Untrustworthy Confessions, 32 HARV. C.R.-C.L. L. REV. 105, 138-40 (1997) (contending that juries believe
that someone would only confess if they actually committed the crime).
77
Saul M. Kassin & Katherine Neumann, On the Power of Confession Evidence: An Experimental Test of the
Fundamental Difference Hypothesis, 21 LAW & HUM. BEHAV. 469, 476–481 (1997).
78

Steven A. Drizin & Richard A. Leo, The Problem of False Confessions in the Post-DNA World, 82 N.C. L.
REV. 891, 923 (2004) [hereinafter Drizin & Leo, The Problem of False Confessions].
79

John H. Langbein, Torture and Plea Bargaining, 46 U. CHI. L. REV. 3, 14 (1978).

80
Eugene R. Milhizer, Confessions After Connelly: An Evidentiary Solution for Excluding Unreliable
Confessions, 81 TEMP. L. REV. 1, 66 (2008).
81

See Sharon L. Davies, The Reality of False Confessions—Lessons of the Central Park Jogger Case, 30
N.Y.U. REV. L. & SOC. CHANGE 209, 209–11 (2006) (“[The] instinct to assert one’s innocence when one is innocent
thus leaves many people skeptical of false confession claims . . . .”).

32

Electronic copy available at: https://ssrn.com/abstract=3730499

true.82 Also, jurors are more likely to privilege claimed personal knowledge over other kinds of
evidence, thus further bolstering the evidentiary impact of the confession. 83
Hence, when the Court in Fulminante recited that “the defendant’s own confession is
probably the most probative [] evidence that can be admitted against him,” the Court was correct,
insofar as it meant that the jury understands confession evidence as the most probative. But
therein lies the disconnect, because as we have seen there are good reasons to think that the
defendant’s confession is not significantly probative.
IV.

The Practical and Empirical Tragedy of Confession Evidence

I have made the case that confession evidence is epistemically weak as a matter of theory.
Specifically, I have shown that confession evidence is that, under standard assumptions, has little
probative value and thus is low-value evidence. Moreover, even as theoretically low-value
evidence, we have seen that juries credit confession evidence with enormous weight, far
outweighing any rational probative value that it could have. As a result, confession evidence is
potentially dangerous in causing incorrect and irrational verdicts and results.
Here I contend that this is not merely theoretical or potential — the practical and
empirical evidence makes clear that confession evidence actually causes incorrect and irrational
results. First, I argue that the evidence of practice shows us that the theoretical case is one that is
truly manifest. The theoretical account was conditional on the assumptions that defendants
generally have the aim of obtaining as little punishment as possible for their crime; that
defendants who confess may be doing so for the opportunity to lower their probable punishment;

82

Saul M. Kassin & Katherine Neumann, On the Power of Confession Evidence: An Experimental Test of the
Fundamental Difference Hypothesis, 21 LAW & HUM. BEHAV. 469, 482 (1997).
83

Id.

33

Electronic copy available at: https://ssrn.com/abstract=3730499

and that defendants who confess may be doing so to escape the oppressive, overwhelming law
enforcement interrogation and inquiry. As shown below, practical evidence shows us that these
assumptions are sound, and thus that the theoretical argument against confession evidence
presents serious concerns. Additionally, I contend that the direct empirical evidence shows us
that false confessions are extraordinarily common. Combining that with the robust strength of a
confession and its relationship to a probable conviction, confession evidence commonly results
in false convictions. Together, this practical and empirical evidence shows that confession
evidence a real and present danger to the just operation of the criminal justice system.
A.

The Terrible Choice Is Real

The actual practices of law enforcement and prosecutors’ offices in obtaining confessions
from defendants commonly imposes upon defendants the choice to confess and potentially obtain
lower punishment, or continue the prosecution and potentially face enormous punishment.
To understand this, we first begin with an understanding of what happens when a
defendant is under suspicion for having committed a crime. Once under suspicion, at some point,
the defendant is made aware of this fact. At that juncture, the defendant has reason to be
concerned and anxious, even if completely innocent of the investigated crime. Federal and state
criminal codes are voluminous,84 and so even if the defendant will easily be cleared of the
suspected crime, the defendant may be potentially liable for other conduct that becomes

84

Shon Hopwood, Clarity in Criminal Law, 54 AM. CRIM. L. REV. 695, 699 (2017) (discussing the problem of
overcriminalization); Gary Fields & John R. Emshwiller, Many Failed Efforts to Count Nation’s Federal Criminal
Laws, Wall St. J. (June 23, 2011),
https://www.wsj.com/articles/SB10001424052702304319804576389601079728920 (detailing the difficulty in
enumerating the potential federal crimes).

34

Electronic copy available at: https://ssrn.com/abstract=3730499

discovered through the course of investigation. This may create an incentive for the defendant to
cooperate, in order to avoid further examination by law enforcement.
Now suppose that the defendant is questioned by law enforcement. If the case against the
evidence reaches some threshold of plausibility,85 law enforcement will seek to interrogate the
defendant. As explained by scholars Steven A. Drizin and Richard A. Leo, “the goal of
interrogation is to elicit incriminating statements, admissions [or] confessions through the use of
psychological methods that are explicitly confrontational, manipulative, and suggestive. The
purpose of interrogation is not to determine whether a suspect is guilty; rather, police are trained
to interrogate only those suspects whose guilt they presume or believe they have already
established. The purpose of interrogation, therefore, is not to investigate or evaluate a suspect’s
alibi or denials. Nor is the purpose of interrogation necessarily to elicit or determine the truth.
Rather, the singular purpose of American police interrogation is to elicit incriminating statements
and admissions — ideally a full confession — in order to assist the State in its prosecution of the
defendant.”86 The canonical model of law enforcement interrogation, based on empirical
research in social psychology and microeconomics, understands the defendant’s decision making
is shaped by: “(1) how the social influence techniques of interrogation cause [them] to perceive
his available courses of action, (2) the [defendant’s] subjective perception of the probability of

85

At an early juncture, where there is no particular suspicion of the defendant, law enforcement may simply
interview the defendant — using non-accusatorial, open-ended questions to gather information. During this time,
law enforcement is often attempting to determine whether the defendant is being truthful. For this, they may use the
Reid technique, which counsels law enforcement to make a determination based on demeanor and behavior. FRED
INBAU, JOHN E. REID, JOSEPH P. BUCKLEY, BRIAN C. JAYNE, CRIMINAL INTERROGATION AND CONFESSION 209–347
(4th ed. 2001). However, such judgments are notoriously unreliable and weak. Richard A. Leo, False Confessions:
Causes, Consequences, and Implications, 37 J. AM. ACAD. PSYCHIATRY LAW 332, 333 (2009); see also Margaux
Joselow, Promise-Induced False Confessions: Lessons from Promises in Another Context, 60 B.C. L. REV. 1641,
1647 (2019).
When law enforcement has determined there is substantial suspicion, law enforcement will escalate to
interrogation tactics. Drizin & Leo, The Problem of False Confessions, supra note 78, at 911–12.
86

Id.

35

Electronic copy available at: https://ssrn.com/abstract=3730499

each course of action actually occurring, and (3) the utility values or benefits (as well as
corresponding harms) associated with each course of action.” 87 Law enforcement’s main course
of action is to utilize the dichotomy between negative and positive incentives. Interrogators
attempt to convince the defendant should confess because they will be inevitably be convicted
due to the strength of the case against them, while at the same time intimating that the magnitude
of punishment will be lessened if the defendant confesses. 88 Saul Kassin and Karlyn McNall
describes this as involving “maximization” — exaggerating the costs of not confessing — versus
“minimization” — downplaying the costs of, as well as suggesting the benefits of, confessing. 89
Indeed, the interrogation handbooks and manuals offer details of how law enforcement
accomplishes this. The key steps in the interrogation are shifting the defendant from “confident
to hopeless” and then eliciting the confession from the hopeless defendant. To do this, law
enforcement uses various psychological techniques.90 They will isolate the defendant for some
time.91 They will use continued, strong accusations that cut off and overwhelm the defendant’s
denials.92 They will attack the defendant’s alibi claims.93 They will also make general claims of

87

Id. at 913; see also GISLI GUDJONSSON, THE PSYCHOLOGY OF INTERROGATIONS AND CONFESSIONS 120–22
(2003). Gudjonsson reviews five different models of confession, but they all generally utilize this framework, albeit
in different terms.
88

Id. at 911–12.

89

Saul Kassin & Karlyn McNall, Police Interrogations and Confessions: Communicating Promises and Threats
by Pragmatic Implication, 15 LAW & HUM. BEHAV. 233, 234-35 (1991).
90

Drizin & Leo, The Problem of False Confessions, supra note 78, at 912 (stating that law enforcement uses a
“laundry list” of techniques to interrogate defendants).
91

NATHAN GORDON & WILLIAM FLEISHER, EFFECTIVE INTERVIEWING & INTERROGATION TECHNIQUES 27–30
(2002); FRED INBAU, JOHN E. REID, JOSEPH P. BUCKLEY, BRIAN C. JAYNE, CRIMINAL INTERROGATION AND
CONFESSION 209–347 (4th ed. 2001).
92
Richard J. Ofshe & Richard A. Leo, The Decision to Confess Falsely: Rational Choice and Irrational Action,
74 DEN. L. REV. 979, 1004–08 (1997).
93

NATHAN GORDON & WILLIAM FLEISHER, EFFECTIVE INTERVIEWING & INTERROGATION TECHNIQUES 27–30
(2002); FRED INBAU, JOHN E. REID, JOSEPH P. BUCKLEY, BRIAN C. JAYNE, CRIMINAL INTERROGATION AND
CONFESSION 209–347 (4th ed. 2001).

36

Electronic copy available at: https://ssrn.com/abstract=3730499

strong evidence, which may be true or false. 94 Law enforcement might appeal to the defendant’s
demeanor, to suggest that the defendant is guilty and that they will be viewed that way by other
judges and jurors.95 They may also simply appeal to fabricated evidence during the interrogation,
such as “nonexistent eyewitnesses, false fingerprints, make-believe videotapes, fake polygraph
results, and so on.”96 And, for any evidence that law enforcement have, they will exaggerate the
amount and strength of the evidence.97
Then they will also make explicit or implicit inducements to the defendant. 98 With
respect to intangible inducements, law enforcement may appeal to moral concerns or social and
community standing, that the defendant can improve or benefit by confessing. 99 These may
proceed by appeals to community catharsis, societal forgiveness, forgiveness from any victims,
and being truthful to family and friends. Then there are systemic inducements, in which law
enforcement suggests that the defendant’s case will fare better in the criminal justice system by
confessing — because law enforcement, prosecutors, and judges look more favorably on one
who tells the truth and seeks remorse. 100 Finally, there are transactional inducements, where law
enforcement will more directly communicate that the defendant will receive less punishment —
through avoiding charge by the prosecution, the ability to avoid a conviction through some

94

Richard J. Ofshe & Richard A. Leo, The Decision to Confess Falsely: Rational Choice and Irrational Action,
74 DEN. L. REV. 979, 1008–14 (1997); Drizin & Leo, The Problem of False Confessions, supra note 78, at 915
(“The most effective technique used to persuade a suspect that his situation is hopeless is to confront him with
seemingly objective and incontrovertible evidence of his guilt, whether or not any actually exists.”)
95

Richard J. Ofshe & Richard A. Leo, The Decision to Confess Falsely: Rational Choice and Irrational Action,
74 DEN. L. REV. 979, 1014–18 (1997)
96

Drizin & Leo, The Problem of False Confessions, supra note 78, at 915.

97

Id.

98

Richard A. Leo, Inside the Interrogation Room, 86 J. CRIM. L. & CRIMINOLOGY 266, 300 (1996).

99

Richard J. Ofshe & Richard A. Leo, The Decision to Confess Falsely: Rational Choice and Irrational Action,
74 DEN. L. REV. 979, 1056–60 (1997).
100

Id. at 1060–66.

37

Electronic copy available at: https://ssrn.com/abstract=3730499

theory of defense, or leniency in sentencing. This may also come in the converse formulation,
that not confessing will increase punishment. 101 There are of course guardrails that purportedly
limit law enforcement’s use of these techniques, but as discussed below, 102 in practice these
limitations are largely ineffective to protect defendants from various pressures that impact the
confession calculus.
Law enforcement uses techniques that effectively create the rational calculus in the
defendant where falsely confessing to criminal conduct for the opportunity for less punishment is
a rational choice. Indeed, creating that calculus is law enforcement’s goal in interrogation.
Moreover, even if law enforcement attentively and vigilantly ensures not to make explicit or
implicit inducements to the defendant, a defendant may themselves come to the belief that
falsely confessing may bring them favor to avoid great punishment. For any falsely confessing
defendant, given the practical knowledge of how law enforcement operates, it is highly likely
that the defendant confronted the choice of whether to confess for purposes of obtaining lesser
punishment.103

101

Id. at 1077–85.

102

See infra Part VI.

103

There have been innovations in the style of interrogation that look to increase the accuracy of information
obtained. One prominent example is the PEACE technique, developed in the United Kingdom. “The PEACE
technique is a nonconfrontational approach to interrogation developed by police officers, lawyers, and
psychologists. The technique consists of five stages: ‘preparation and planning,’ ‘engage and explain,’ ‘account,’
‘closure,’ and ‘evaluate.’ The PEACE method instructs interrogators to use a nonaccusatory approach, in which new
information is compared to the suspects' previous statements and other available evidence. PEACE does not permit
police to lie to suspects and prioritizes obtaining accurate information over eliciting confessions. This approach
would prevent interrogators from using coercive techniques on juveniles and help reduce the risk of false
confessions. The goal of PEACE is not to decrease the rate of confession, but merely to improve the accuracy of the
confessions obtained.” Hannah Brudney, Confessions of A Teenage Defendant: Why A New Legal Rule Is Necessary
to Guide the Evaluation of Juvenile Confessions, 92 S. CAL. L. REV. 1235, 1267 (2019) (internal citations omitted).
The use of the PEACE approach however has not been adopted broadly in the United States. See, e.g., Note, Dylan
J. French, The Cutting Edge of Confession Evidence: Redefining Coercion and Reforming Police Interrogation
Techniques in the American Criminal Justice System, 97 TEX. L. REV. 1031, 1047 (2019) (observing that the
movement for non-adversarial techniques like the PEACE model have shown some success in limited trials in the

38

Electronic copy available at: https://ssrn.com/abstract=3730499

B.

The Overwhelming Pressure Is Real

Separate and apart from the choice of whether to confess for lesser punishment,
defendants face another force that may influence their actions. The atmosphere of the
interrogation and the investigation may itself be so overwhelming that a defendant feels
compelled to confess in order to escape that condition. Indeed, it is clear that law enforcement
uses techniques to create an overwhelming atmosphere of discomfort. The law enforcement
techniques of isolation, continued rejection of the defendant’s explanations and positions,
confrontation with claims of strong evidence against the defendant are all used to make the
defendant feel the weight of the case against them — to make the defendant feel abject, alone,
and without support.104 “Over and over again, the investigator conveys the message that the
suspect has no meaningful choice but to admit to some version of the crime because continued
resistance — in light of the extensive and irrefutable evidence against him — is simply futile.” 105
The implicit or explicit promise, then, is that the confession will allow that pressure to dissipate
and the defendant will be relieved. And this is not hypothetical — this is the very design of the
techniques law enforcement employs.106
The prospect of escaping this overwhelming atmosphere through confessing falsely is
easily understandable. In particular, the atmosphere may be so acutely overwhelming — akin to
the experience of being tortured — that the act of (false) confession may be reflexive and

U.S., like with the formation of the High-Value Detainee Interrogation Group, but have not been implemented
broadly).
104

Drizin & Leo, The Problem of False Confessions, supra note 78, at 911; Richard J. Ofshe & Richard A. Leo,
The Decision to Confess Falsely: Rational Choice and Irrational Action, 74 DENV. U. L. REV. 979, 1004–08 (1997).
105

Drizin & Leo, The Problem of False Confessions, supra note 78, at 915.

106

Id. at 911 (“Because it is designed to break the anticipated resistance of an individual who is presumed
guilty, police interrogation is stress-inducing by design; it is intentionally structured to promote isolation, anxiety,
fear, powerlessness, and hopelessness.”).

39

Electronic copy available at: https://ssrn.com/abstract=3730499

involuntary. In that case, there is no question about whether it is rational or not, because the
defendant has little to no control over that act of confession. But beyond that, it still may be
rational to falsely confess, as a matter of voluntary choice, because the experience is so
overwhelmingly terrible and terrifying. An innocent defendant may believe that falsely
confessing is rational because it will relieve the pressure, yet at the same time be ineffectual,
because they did not in fact commit the crime and that should be borne out by the other
evidence.107 Ultimately, the practical evidence reveals clearly that it is very likely that defendants
face an atmosphere of overwhelming pressure and discomfort that in turn incentivizes the
defendant to falsely confess.
On this picture, it is no surprise then that many of the defendants that capitulate to these
pressures are young or mentally disabled. As the Gross, Jacoby, Matheson & Montgomery study
of exonerations reveals, “False confessions are heavily concentrated among the most vulnerable
groups of innocent defendants.” Of the exonerees in their comprehensive study, 42% of minors
falsely confessed; 69% of 12-to-15-year-olds falsely confessed; and 69% of those with severe
mental disabilities falsely confessed.108 But the fact is that many, including high-functioning
adults, could fall prey to these tactics and pressures.109

107

Id. at 978–79 (discussing the false confession of Frank Kuecken, made on this reasoning).

108

Samuel R. Gross, Kristen Jacoby, Daniel J. Matheson & Nicholas Montgomery, Exonerations in the United
States 1989 through 2003, 95 J. CRIM. L. & CRIMINOLOGY 523, 545 (2005).
109

Scholar Gisli H. Gudjonsson has argued that certain people with vulnerabilities are particularly susceptible:
those with mental disorders, like mental illness or personality disorders; those with abnormal mental states,
including anxiety, drug-related issues, or depression; low intellectual functioning; and certain personality traits such
as suggestibility and high compliance and acquiescence. Gisli H. Gudjonsson, Psychological Vulnerabilities During
Police Interviews. Why Are They Important?, 15 LEGAL & CRIMINOLOGICAL PSYCHOL. 161, 166–68 (2010).

40

Electronic copy available at: https://ssrn.com/abstract=3730499

C.

The Disastrous Results are Real

The prior sections reveal that the theoretical reasons for an innocent defendant to falsely
confess are made manifest by law enforcement techniques and by the nature of the criminal
process. Given that, and assuming rational actors, we would predict that this would result in a
significant risk of false confessions by innocent defendants, which would then also translate into
a substantial number of false confessions and consequent wrongful convictions. The empirical
data tells us that these conclusions are in fact strongly supported. From a number of studies of
exonerated convictions, we know the following: (1) that an innocent defendant would falsely
confess is plausible and there’s a significant chance that it would occur; (2) that such a false
confession may be done for reasons of rational calculus and overwhelming pressure; and (3) that
the evidentiary value of a confession is very impactful to, and greatly overestimated by, jurors
and judges.
That said, there are serious limitations to what we can learn from empirical data. For one,
we have no good sense of how many false confessions there are. One obvious, antecedent reason
is that we do not have a good sense of how many false convictions there are. 110 A huge number,
and percentage, of cases are resolved by plea agreement — which often involve some kind of

110

Some commentators, such as Laurie Magid, argue that there is not enough reason to believe that false
confessions are a problem to rectify. Laurie Magid, Deceptive Police Interrogation Practices: How Far is Too Far?,
99 MICH. L. REV. 1168, 1187, 1194–95 (2001). But such a standard fails to take into account how difficult
discovering false convictions are. The direct empirical evidence shows a significant number of miscarriages of
justice, and the indirect empirical evidence demonstrates that the conditions that would give rise to false confessions
are commonly present.

41

Electronic copy available at: https://ssrn.com/abstract=3730499

confession evidence,111 or functionally similar actions by the defendant.112 If a significant
percentage of these cases are in fact false convictions based on false confessions, then we may
have a large number of false confessions of which we may never be aware. Moreover, as
observed by Leo, most of our exonerations come from DNA evidence, but “the documented
cases appear to represent the proverbial tip of the iceberg.” 113 There are many cases where there
is no DNA to test. Indeed, there is good reason to think that, for example, many robbery
convictions — where DNA evidence is less common — are false convictions. 114 “They also do
not include false confessions that were dismissed or disproved before trial, . . . those given for
crimes that were not subject to postconviction review (especially less serious crimes), and those
given in cases that contain confidentiality provisions (e.g., juvenile proceedings).” 115 Selection
biases predominate here: Our exonerations tend to also focus on crimes with serious penalties —
like those where the defendant has been sentenced to death. 116 That raises the concern that we

111
See, e.g., Lucian E. Dervan, Bargained Justice: Plea-Bargaining’s Innocence Problem and the Brady SafetyValve, 2012 UTAH L. REV. 51, 84 (2012) (stating, based on federal prosecution statistics of how non-dismissed
defendants’ cases are resolved, that “over 95 percent of defendants in the criminal justice system plead guilty and, in
most cases, such confessions are prompted by offers of leniency or other benefits from the prosecution”); Anna
Roberts, Arrests As Guilt, 70 ALA. L. REV. 987, 1011 (2019) (explaining that a high proportion of non-dismissed
defendants’ cases are resolved by guilty or nolo contendere pleas).
112

An Alford plea and a nolo contendere plea, which allow the defendant to plead guilty without admitting or
denying guilt, are possibilities that do not require the defendant to confess. But these pleas are functionally very
similar to guilty pleas, with the main difference being that they allow for some saving face for the defendant. Mark
Gurevich, Justice Department’s Policy of Opposing Nolo Contendere Pleas: A Justification, 6 CAL. CRIM. L. REV. 2,
10 (2004); Note, Curtis J. Shipley, The Alford Plea: A Necessary but Unpredictable Tool for the Criminal
Defendant, 72 IOWA L. REV. 1063, 1063–65 (1987).
113

Richard A. Leo, False Confessions: Causes, Consequences, and Implications, 37 J. AM. ACAD. PSYCHIATRY
LAW 332, 332 (2009).
114

Samuel R. Gross, Kristen Jacoby, Daniel J. Matheson & Nicholas Montgomery, Exonerations in the United
States 1989 through 2003, 95 J. CRIM. L. & CRIMINOLOGY 523, 530–31 (2005) (comparing robbery to rape and
explaining that many of the risk factors for false rape convictions manifest, and are bigger risks, in robberies, but
that robberies generally lack the potential for DNA-evidence exonerations).
115

Richard A. Leo, False Confessions: Causes, Consequences, and Implications, 37 J. AM. ACAD. PSYCHIATRY
LAW 332, 332 (2009).
116
Samuel R. Gross, Kristen Jacoby, Daniel J. Matheson & Nicholas Montgomery, Exonerations in the United
States 1989 through 2003, 95 J. CRIM. L. & CRIMINOLOGY 523, 531–33 (2005).

42

Electronic copy available at: https://ssrn.com/abstract=3730499

may be missing scores of false convictions for lesser crimes — where false confessions or
pleading guilty is much more prevalent. Similarly, resources towards exoneration focus on the
most promising cases, and so cases involving confessions or guilty pleas face an obstacle in
demanding attracting the attention of those working for exoneration, because those cases seem
prima facie hard to win. But with that caveat, the empirical evidence is robust enough to give us
strong reason to support the above conclusions and, consequently, to be concerned about the
epistemic value and practical dangers of confession evidence.
First, the empirical evidence shows that it is entirely plausible that an innocent defendant
would falsely confess, with a substantial chance of it occurring. Consider Brandon Garrett’s
monumental study of the first 200 cases involving exoneration based on DNA evidence. 117 Of
these 200 cases, the vast majority involved serious crimes: 141 involved rape, 44 involved rapemurder, and 12 involved murder.118 Among other reasons, these crimes also are the most likely
to involve DNA evidence.119 Of the 200 cases, 15.5% involved false confessions, and a further
6.5% involved self-inculpatory remarks short of a confession, for a total of 22% of these
convictions tainted with the defendant’s false confession or self-inculpatory statements. 120
Focusing on the cases involving murder and rape-murder, a staggering 40% of cases involved
false confessions.121 This conclusion is reinforced by the data collected on the National Registry

117

Brandon L. Garrett, Judging Innocence, 108 COLUM. L. REV. 55, 59 (2008).

118

Id. at 74.

119

Id. at 73.

120

Id. at 76. See also Edward Connors, Thomas Lundregan, Neal Miller & Tom McEwen, U.S. Department of
Justice, Convicted by Juries, Exonerated by Science (1996) (detailing that approximately 21%, or 6 of 28,
exonerations involved false confessions or self-inculpatory statements).
121
See also Welsh S. White, Confessions in Capital Cases, 2003 U. ILL. L. REV. 979, 984 (2003) (stating that,
according to the Cardozo Innocent Project, out of 35 intentional homicide case exonerations, 23 (over 65%)
involved false confessions); Samuel R. Gross, Kristen Jacoby, Daniel J. Matheson & Nicholas Montgomery,
Exonerations in the United States 1989 through 2003, 95 J. CRIM. L. & CRIMINOLOGY 523, 530–31 (2005) (finding
over 20% of murder exonerations involved false confessions); Hugo Adam Bedau & Michael L. Radelet,

43

Electronic copy available at: https://ssrn.com/abstract=3730499

of Exonerations, where over 12% of exonerations involve a false confession, and over 22% of
murder convictions involve false confessions. 122 And as discussed, there is good reason to think
that exonerations are under-representative with respect to false confessions. 123 Thus, the
empirical data strongly points to the conclusion that it is entirely plausible that an innocent
defendant will actually falsely confess.
What’s more, there is good reason to think that the chances that an innocent defendant
will falsely confess are nearly the same as the chances that a guilty defendant will truly confess.
Garrett included a “matched comparison group” as part of his study, which paired other
convictions with exonerations from the innocence group that had written decisions. Garrett’s
method was to randomly select convictions from the set of reported decisions that “matched” the
exoneration cases, in terms of the same criminal charges, in the same state, and in the same time
period as the paired exoneration case.124 These matched cases could not themselves be verified
as innocent or guilty. Notwithstanding, in this group, 19% of the cases involved confessions. 125

Miscarriages of Justice in Potentially Capital Cases, 40 STAN. L. REV. 21, 57 (1987) (stating that, out of 336
intentional homicide exonerations with sufficient records for analysis, in 49 cases (14.3%), the false confession was
the primary or contributing cause of the false conviction).
The discrepancy in confession rates between rape, on one hand, and murder and rape-murder, on the other, is
perhaps surprising. One potential explanation is that in serious cases, law enforcement and the prosecution have
prima facie less interest in incentivizing the defendant to confess through the promise of less punishment. This
default remains in cases of rape, where there is more likely to be victim identification evidence, that obviates the
need for a confession. In contrast, in murder or rape-murder cases, a confession may be more necessary, and so law
enforcement and the prosecution may be more likely to incentivize or pressure the defendant to obtain a confession.
122

National Registry of Exonerations, http://www.law.umich.edu/special/exoneration/Pages/about.aspx.

123

See supra notes 111–116 and accompanying text.

124
The “match[ing]” here does not seem to look at the actual evidence in the case, like whether there is a
confession or not. If Garrett’s matching does look at the evidence, then we cannot make strong conclusions about
the likelihood of confession in a random (true) conviction, but we can observe that the conditions under which some
innocent people and some guilty people confess are materially the same.
125

Brandon L. Garrett, Judging Innocence, 108 COLUM. L. REV. 55, 78 (2008).

44

Electronic copy available at: https://ssrn.com/abstract=3730499

Assuming that all of these cases involve true guilty convictions, the confession rate of the guilty
defendants is strikingly similar to the proven innocent defendants. 126
Second, the empirical data tells us that false confessions do occur for the reasons of
rational calculus and overwhelming pressure. Unfortunately, there is not clear visibility on the
nature of the interrogations in the various studies of false confessions and exonerations — so it
can be hard to discern the confessors motivations or reasons. But from what is available, it is
clear that the combination of rational calculus and overwhelming pressure are often at play. In a
study by Richard A. Leo and Richard J. Ofshe of 29 false confessions that resulted in conviction,
7, or 24%, pleaded guilty in order to avoid a harsher penalty — typically the death penalty. 127
That is the prototypical example of rational calculus — because falsely confessing to avoid the
death penalty is obviously rational. Welsh S. White has noted that threats and promises regarding
whether a capital defendant will be executed and misrepresentations of forensic evidence — both
tactics that appeal to the defendant’s rational calculus — appear to be particularly problematic in
leading to unreliable and false confessions.128 In Drizin and Leo’s study of false confessions in
which the length of the interrogation could be determined, over 84% of interrogations that led to
false confessions lasted over 6 hours long, with average and median lengths of 16 hours and 12
hours, respectively.129 These numbers are shocking. And, given what we know about the nature
of law enforcement interrogation methods, it is pellucid that defendants experience an
overwhelming atmosphere of pressure to confess. In the Gross, Jacoby, Matheson &

126

Id. at 76.

127

Richard A. Leo & Richard J. Ofshe, Consequences of False Confessions, 88 J. CRIM. L. & CRIMINOLOGY
429, 478–79 (1998).
128

White, supra note 121, at 1008–17.

129

Drizin & Leo, The Problem of False Confessions, supra note 78, at 948.

45

Electronic copy available at: https://ssrn.com/abstract=3730499

Montgomery study of exonerations, of 33 cases involving false confessions with sufficient
records, 28 (over 84%) revealed law enforcement coercion. 130 The empirical studies confirm
what we would predict in light of the theoretical motivations of rational defendants and the
practices of law enforcement — defendants are regularly incentivized and pressured, both in
terms of rational decision making on punishment and in terms of the atmosphere of anxiety,
isolation, and hopelessness, to confess.
Third, and finally, the evidentiary value of a confession is greatly overestimated by
judges and jurors. One particularly telling fact is that, from Garrett’s study, when there was a
false confession, the government will frequently rely on very little else to convict. Out of 31
convictions involving a false confession, in 7 the confession was the central evidence of guilt,
and in 9 cases, “the confession was accompanied by only one other type of evidence (jailhouse
snitch, eyewitness, or blood or hair evidence).” 131 This tells us that both law enforcement and
prosecution officials often view the confession as so powerful that further investigation and
evidentiary support is unnecessary to determine the true perpetrator or to convict the defendant.
Now, as Drizin and Leo’s study of 125 “proven false” confessions shows, false confessions do
not always lead to convictions.132 But nevertheless, the data reveal the potency of confession
evidence: Out of the 125 proven false confessions, 44 (or 35%) still led to convictions. 133 Given
that these confessions were provably false and given that the high BARD standard of review, we
would expect these false confessions to have led to false convictions far fewer than 35% of the

130
Samuel R. Gross, Kristen Jacoby, Daniel J. Matheson & Nicholas Montgomery, Exonerations in the United
States 1989 through 2003, 95 J. CRIM. L. & CRIMINOLOGY 523, 544 n.47 (2005) (noting that there were 51 false
confessions in total: 16 with insufficient records, 5 voluntary confessions, and 28 that involved coercion).
131

Brandon L. Garrett, Judging Innocence, 108 COLUM. L. REV. 55, 89 (2008).

132

Drizin & Leo, The Problem of False Confessions, supra note 78, at 950–52.

133

Id.

46

Electronic copy available at: https://ssrn.com/abstract=3730499

time. The disconnect occurs because all of the relevant actors misapprehend the probative value
of false confessions. With respect to law enforcement, that misapprehension is in large part
because of the documented misconception by officials that they can detect truth from falsity in
defendants’ statements.134
The problem is compounded by the public’s misperception of confession evidence as
strong, because of the “myth” that false confession would only occur to the imposition of torture
on the defendant or the defendant’s mental illness. 135 In Jacqueline McMurtrie’s words, [t]he
idea that an individual would [falsely] confess to a crime, particularly a horrific crime such as
murder or rape, without being subject to physical torture, runs counter to the intuition of most
people.”136 An experimental study by Saul M. Kassin and Katherine Neumann showed that
confession evidence was “uniquely potent,” compared to eyewitness testimony and character
evidence.137 That study provided 62 undergraduate psychology students summaries of four
criminal trials (murder, rape, aggravated assault, and automobile theft). 138 Each trial contained
weak circumstantial evidence plus a confession, an eyewitness identification, a character witness,
or nothing further (as control). In every case the confession was far more likely to garner a
conviction than the character witness; and in all but the automobile theft case, the confession was

134

Richard A. Leo, False Confessions: Causes, Consequences, and Implications, 37 J. AM. ACAD. PSYCHIATRY
LAW 332, 342 (2009).
135

Drizin & Leo, The Problem of False Confessions, supra note 78, at 910.

136

Jacqueline McMurtrie, The Role of the Social Sciences in Preventing Wrongful Convictions, 42 AM. CRIM. L.
REV. 1271, 1280 (2005).
137

Saul M. Kassin & Katherine Neumann, On the Power of Confession Evidence: An Experimental Test of the
Fundamental Difference Hypothesis, 21 LAW & HUM. BEHAV. 469, 475–476 (1997).
138

Id.

47

Electronic copy available at: https://ssrn.com/abstract=3730499

stronger than the eyewitness identification, where they were similar in strength. 139 Another
study, by Sara C. Appleby, Lisa E. Hasel, and Saul M. Kassin, further confirmed this potency of
confession evidence.140 There 141 college students served as jurors in a mock trial and were
presented a summary of an experimental rape-murder case. They were partitioned in nine groups,
receiving either no confession or one of eight written versions of a signed confession that
differed along three dimensions (presence or absence of details; presence or absence of motive
explanation; and presence or absence of apology). In the no-confession group, 30% of the jurors
voted to convict, but in the other eight confession groups, 95% voted to convict, with little
difference among the confession groups — even bare admissions of guilt sufficient to render a
guilty verdict.141
Furthermore, “[w]hen courts fail to dismiss these false confession cases at the pretrial
stage, the overwhelming majority of defendants will be wrongfully convicted. In a 1998 study of
sixty false confessions, 73 percent of the false confessors whose cases went to trial were wrongly
convicted.”142 Indeed, in Drizin and Leo’s study of 125 false confessions, of the 35 cases that
went to jury trial, the jury convicted 28 (or 80%) of them. 143 And the specific empirical results
arise from cases where it has become clear that there was a wrongful conviction, but there may

139

Id. The researchers also suggested that the eyewitness testimony may have been stronger in the automobile
theft case, because the control summary of the circumstantial evidence in the automobile theft case included an
eyewitness, unlike the others, so the additional eyewitness served to corroborate the control summary. Id. at 476.
140

Sara C. Appleby, Lisa E. Hasel & Saul M. Kassin, Police-Induced Confessions: An Empirical Analysis of
Their Content and Impact, 19 PSYCH., CRIME & L. 111, 111 (2013).
141

Id. at 119–21.

142
Richard A. Leo, Steven A. Drizin, Peter J. Neufeld, Bradley R. Hall, and Amy Vatner, Bringing Reliability
Back in: False Confessions and Legal Safeguards in the Twenty-First Century, 2006 WIS. L. REV. 479, 484–85
(2006).
143
Drizin & Leo, The Problem of False Confessions, supra note 78, at 910 (stating that of the 37 false
confessions that went to trial, 28 were convicted by jury, 2 by judge, and 7 acquitted).

48

Electronic copy available at: https://ssrn.com/abstract=3730499

be numerous cases which escape such realization, which then further increase the true likelihood
of conviction in light of a false confession.
The conclusion is clear: whether it be law enforcement, prosecutors, or jurors, the impact
of a confession is dramatic and often dispositive of the defendant’s case, beyond the rational and
probative weight that the confession actually confers.
V.

The Moral Hazards of Confession Evidence

Apart from the potential for engendering false confessions and consequent false
convictions, confession evidence also inflicts moral harms on defendants. To understand the
nature of that harm, consider again the choice that innocent defendants in our criminal justice
system may face: Risk a prosecution, which may result in the harsh penalty, or falsely confess
and gain a significant chance of escaping the harsher penalty. That is an odious choice for the
defendant and that is why we need robust procedural protections, to ensure that innocent
defendants are not subjected to wrongful punishment. But it is more than the consequences of the
choice that cause harm. The presentation of the choice to the defendant is itself harmful.
Consider the definition of torture under the international treaty, The 1984 Convention
Against Torture and Other Forms of Cruel, Inhuman or Degrading Treatment and Punishment
(“CAT”):
For the purposes of this Convention, torture means any act by which severe pain
or suffering, whether physical or mental, is intentionally inflicted on a person for
such purposes as obtaining from him or a third person information or a
confession, punishing him for an act he or a third person has committed or is
suspected of having committed, or intimidating or coercing him or a third person,
or for any reason based on discrimination of any kind, when such pain or
suffering is inflicted by or at the instigation of or with the consent or acquiescence
49

Electronic copy available at: https://ssrn.com/abstract=3730499

of a public official or other person acting in an official capacity. It does not
include pain or suffering arising only from, inherent in or incidental to lawful
sanctions.144
Other philosophically minded definitions of torture include an assault on the defenseless 145; the
creation of the experience of pain, fear, and uncertainty by use of the target’s own body and
emotions — in an act of self-betrayal146; and the “intentional saturation of [the target’s]
consciousness with panic.”147 There are a number of theories on the locus of the immorality of
torture, including in violating human dignity, violating human autonomy, and treating people as
means rather than ends.
As we have seen from practice and empirical evidence, interrogation imposes significant
mental harms — in the forms of pain, suffering, fear, uncertainty, anxiety, and panic — and it
demands of defendants that they confess to potentially alleviate themselves of these harms.
Furthermore, as we have seen, the canonical interrogation techniques used by law enforcement
are indeed designed to isolate defendants and create feelings of hopelessness and anxiety in
them.148 Moreover, even without explicit use of these techniques, in the context of our criminal
justice system, interrogation in the course of an investigation itself would succeed in inflicting
the same impact on defendants.149 Indeed, by creating outlets for the defendant to potentially

144

Convention against Torture and Other Cruel, Inhuman or Degrading Treatment or Punishment art. 16, § 1,
Dec. 10, 1984, 108 Stat. 382, 1465 U.N.T.S. 85.
145

Henry Shue, Torture, 7 PHIL. & PUB. AFF. 124, 130 (1978).

146

David Sussman, What’s Wrong with Torture?, 33 PHIL. & PUB. AFF. 1, 4 (2005).

147

Jacob Bronsther, Torture and Respect, 109 J. CRIM. L. & CRIMINOLOGY 423, 447 (2019).

148

See supra Parts IV.A & IV.B.

149

Indeed, the Court in Miranda recognized that even without law enforcement utilizing particular stratagems to
induce confessions, custodial interrogation itself imposes pressures on defendants and trades on their weaknesses.
Miranda v. Arizona, 384 U.S. 436, 455 (1966). My point extrapolates to say that even when a defendant is not in
custody, interrogation in the course of an investigation can impose many of those pressures.

50

Electronic copy available at: https://ssrn.com/abstract=3730499

escape the harms, namely by confession, it capitalizes on the defendant’s self-betrayal as well.
And, as discussed below,150 the inefficacy of procedural protections to insulate defendants from
the overwhelming pressures of the criminal justice system leave defendants who are subject to
interrogation by and large defenseless. They must either bear the brunt of the mental pain caused
by those pressures or capitulate. Consequently, interrogation in our criminal justice system —
when it exploits and manipulates defendants by use of mental pain — inflicts moral harms on
defendants.151
Importantly, interrogation in our criminal justice system does not merely pose the risk of
such moral harms — they are a practical certainty. The moral harms persist even if the defendant
does not capitulate and confess, because they would still be subjected to the experience of mental
pain. Indeed, the moral harms persist even if law enforcement does not use tactics that overtly
seek to exploit and manipulate the defendant through mental pain, because endemic to the
criminal justice system are pressures, such as the rational calculus of punishment and the
overwhelming nature of investigation, that interrogation will inevitably capitalize on. Indeed, the
moral harms persist even if the defendant is not innocent of the crime. Prior to conviction, the
defendant enjoys due process rights and a presumption of innocence that should shield the
defendant from such tactics that impose harm on the defendant. 152
Now, one might object that some of these pressures are unavoidable, for they are, as I
have contended, embedded in the foundations of the criminal justice system. This is true,

150

See infra Part VI.B.

151
Such moral illegitimacy is presumptive. There may be particular situations where torture is morally justified.
That is a controversial question. Seumas Miller, Torture, in STANFORD ENCYCLOPEDIA PHIL. (2008), available at
https://plato.stanford.edu/entries/torture/. But what should be uncontroversial is that it is morally wrongful for the
state generally to torture criminal defendants through interrogation as a matter of course.
152

See, e.g., Shima Baradaran, Restoring the Presumption of Innocence, 72 OHIO ST. L.J. 723, 728 (2011).

51

Electronic copy available at: https://ssrn.com/abstract=3730499

criminal investigation does impose mental and physical harms on defendants. And we should
endeavor to reduce those harms as much as possible. This is precisely my contention with respect
to confession evidence. Given its severe epistemic inadequacies and the moral harms it causes,
we should exclude confession evidence. As I discuss below, excluding confession evidence is no
panacea to the problem of moral harm — some of these moral harms will arise from
interrogation and investigation itself. But in excluding the potent confession evidence, we may
reduce the incentives of law enforcement and thus mitigate the moral harms that defendants face.
VI.

The Doctrinal Inefficacy in Ensuring Reliable Confessions (and Protecting
Defendants)

I have shown that there is a distinct epistemic weakness to confession evidence and that
interrogation inevitably imposes moral harms on defendants, whether they are innocent or guilty.
Confession evidence is inert as a theoretical matter, but its probative value is vastly
overestimated by the relevant actors in our criminal justice system — law enforcement,
prosecutors, and jurors. What’s more, this is not merely theoretical: The practical and empirical
evidence tells us that conditions that give rise to false confessions — such as the difficult rational
calculus and the overwhelming pressure of investigation — are created and exploited by
government actors. As a consequence, many defendants are so pressured and false confessions
likely abound. And the use of overwhelming pressure, that results in mental pain in the form of
fear, uncertainty, anxiety, and panic, to exploit and manipulate defendants to confess is morally
wrongful and imposes moral harms on defendants.
Of course, the dangers of confession evidence are not unknown. And there are myriad
doctrines that attempt to limit these problems of confessions — in terms of the impact of
interrogation on defendants, the occurrence of false confessions, and the devastating results. But
52

Electronic copy available at: https://ssrn.com/abstract=3730499

these doctrines have been substantially ineffective in solving these problems. Specifically, there
are three key doctrinal buckets aimed at addressing these problems: (1) due process limitations
on confession evidence to ensure that the evidence was produced from the defendant’s voluntary
action; (2) prophylactic rules on interrogation, primarily those from Miranda, to ensure the
defendant’s voluntariness and intelligence in producing the confession evidence; and (3)
evidentiary limitations to ensure the reliability of the confession evidence. I contend that these
doctrines fail to protect against the harms of confession evidence. First, the doctrines have
stringent preconditions, are applied with capacious discretion, and have evolved to incorporate
various exceptions such that they do not practically stop the kind of law enforcement and
prosecutorial behavior that engenders the problems of confession. Second, insofar as these
doctrines focus narrowly on the behavior of government officials, these doctrines fail to address
the most serious concerns of the rational calculus and oppressive atmosphere, which arise from
the foundations of the criminal justice system.
A.

The History and Landscape of the Doctrine on Confession Evidence
1.

Constitutional Doctrines

The starting point, of course, is the privilege against self-incrimination. That comes from
the nemo tenetur maxim and finds itself in the text of the Fifth Amendment. 153 The text of the
Fifth Amendment states in relevant part, “No person . . . shall be compelled in any criminal case
to be a witness against himself.” 154 The Supreme Court made clear that the self-incrimination
privilege extends beyond the courtroom in Bram v. United States.155 Decided in 1897, the case

153

See supra notes 1–3 and accompanying text.

154

U.S. CONST. amend. V.

155

168 U.S. 532 (1897).

53

Electronic copy available at: https://ssrn.com/abstract=3730499

concerned a murder on the high seas. The defendant Bram, who was part of the crew of a ship,
allegedly murdered the captain and two others. 156 He was apprehended by the Canadian police
when the ship docked in Halifax, Nova Scotia, Canada. 157 There he was interrogated, with the
confrontation of the supposed testimony of an eye witness and the suggestion that Bram should
say if he had an accomplice, to “not have the blame of this horrible crime on your own
shoulders.”158 Bram denied the claims, but in so doing, he made incriminating statements that
were treated as a confession.159 At a jury trial, Bram was convicted and then appealed his
conviction to the Supreme Court on the basis that the interrogation that led to the incriminating
remarks violated the Fifth Amendment’s self-incrimination clause. 160 The Supreme Court agreed
and reversed his conviction.161 Most relevantly, in so doing, the Court introduced a voluntariness
standard in determining whether the incriminating remarks were admissible. 162 The Court quoted
an influential treatise to state, “‘But a confession, in order to be admissible, must be free and
voluntary; that is, must not be extracted by any sort of threats or violence, nor obtained by any
direct or implied promises, however slight, nor by the exertion of any improper influence. . . . A
confession can never be received in evidence where the prisoner has been influenced by any
threat or promise; for 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

156

Id. at 537.

157

Id. at 538.

158

Id. at 539.

159
Id. at 537. The statements made by Bram likely are beyond what I consider “confession” evidence. They did
not consist of “I did it” statements, but rather they may have unknowingly implied he committed the crime.
160

Id. at 556–59.

161

Id. at 569.

162
Id. at 542–44. Bram was also the first case from the Supreme Court that held that the Fifth Amendment
applied beyond the courtroom to other kinds of interrogations, including by law enforcement.

54

Electronic copy available at: https://ssrn.com/abstract=3730499

influence has been exerted.’”163 In light of that, the Court stated that the language of “not
hav[ing] the blame of this horrible crime on your own shoulders” implicitly suggested a benefit
of mitigated punishment in exchange for confession, thus rendering the incriminating statements
involuntary.164
Notably, Bram’s theory of involuntariness is considerably robust and broad, because even
an oblique suggestion at the possibility of the benefits of confession was enough to render the
confession involuntary and therefore inadmissible. Bram unearthed this principle from the
English and early American common law — indeed, citing the nemo tenetur maxim — and
imported it into the Fifth Amendment.165
Following Bram’s limitation on confessions arising from the Fifth Amendment, the
Supreme Court decided cases arising from state prosecutions, such as Brown v. Mississippi166
and Chambers v. Florida, 167 finding limits in the Fourteenth Amendment’s Due Process Clause.
Brown, decided in 1936, dealt with murder charges against Black defendants, who confessed
after being whipped, beaten, and tortured by a police officer and local mob. 168 Thereafter, they
were convicted, based only on the confessions. 169 On appeal, the Supreme Court reversed the
sham convictions, reasoning that they violated the Due Process clause of the Fourteenth
Amendment.170 Similarly, in Chambers, four Black defendants were arrested on a murder

163

Id. (quoting 3 RUSSELL, CRIMES 478 (6th ed.)).

164

Id. at 564–66.

165

Id. at 545–62.

166

297 U.S. 278 (1936).

167

309 U.S. 227 (1940).

168

297 U.S. at 282–84.

169

Id.

170

Id. at 288.

55

Electronic copy available at: https://ssrn.com/abstract=3730499

charge.171 They were then jailed without any formal charges and questioned for five days. During
the five days, they were subjected to numerous interrogations, on the backdrop of the threat of
mob violence; in the Court’s words, the police interrogators showed “relentless tenacity which
‘broke’ petitioners’ will and rendered them helpless to resist their accusers further.” 172 Thus, the
Court reversed the convictions under the Due Process Clause of the Fourteenth Amendment,
stating that “[t]o permit human lives to be forfeited upon confessions thus obtained would make
of the constitutional requirement of due process of law a meaningless symbol.” 173
Both of these cases arose under the Fourteenth Amendment in part due to a historically
contingent circumstance: At the time they were decided, the Fifth Amendment was only
applicable to the federal government (as in Bram) and had not been incorporated against the
states.174 Thus, the Court looked to due process, in the Fourteenth Amendment, to develop a
requirement of voluntariness and hold these confessions constitutionally inadmissible. 175 The
doctrine on voluntariness continued to develop, through cases like Spano v. New York,176

171

309 U.S. 227, 227 (1940).

172

Id. at 229–35, 240.

173

Id. at 240.

174

Mark A. Godsey, Rethinking the Involuntary Confession Rule: Toward A Workable Test for Identifying
Compelled Self-Incrimination, 93 CAL. L. REV. 465, 488 (2005).
175

Id. at 488–89.

176

360 U.S. 315, 320–21 (1959) (holding that an eight-hour long confession, persisting through defendant’s
repeated requests of counsel, violated due process, not only because of untrustworthiness of the confession but
because “police must obey the law while enforcing the law” and “that in the end life and liberty can be as much
endangered from illegal methods used to convict”).

56

Electronic copy available at: https://ssrn.com/abstract=3730499

Colorado v. Connelly,177 Arizona v. Fulminante,178 and others.179 Ultimately, the due process
voluntariness limitations require a showing, in the totality of the circumstances, that law
enforcement conduct caused the will of the defendant to be overborne so that the defendant’s
statement were not of free and voluntary acts. 180
Then came the watershed case of Miranda v. Arizona.181 That case, decided in 1966,
involved rape and kidnapping charges against Ernesto Miranda, a Mexican immigrant living in
Phoenix, Arizona.182 After a two-hour long interrogation where he was unrepresented by
counsel, Miranda signed a written confession and identified the victim in person. 183 Based
primarily on the confession, Miranda was convicted. He appealed to the Supreme Court, which
reversed his conviction (and three others in similar cases). In so doing, the Court set forth the
requirement of its now famous Miranda warnings:

177

479 U.S. 157, 167 (1986) (rejecting defendant’s claim that his confession was involuntary as it was made
when mentally ill, because in order for due process–voluntariness to be violated there must be wrongful police
conduct that caused the confession).
178
499 U.S. 279 (1991) (holding that defendant’s incriminating statements violated due process because they
were made to a co-inmate, a paid informant, on the promise of protection from other inmates, which was enough to
raise a credible threat of violence that in turn made the confession a product of coercion).
179
Paul Marcus, It’s Not Just About Miranda: Determining the Voluntariness of Confessions in Criminal
Prosecutions, 40 Val. U. L. Rev. 601, 642 (2006).
180

See, e.g., id. at 605–41; Stephen J. Schulhofer, Confessions and the Court, 79 MICH. L. REV. 865, 865–69
(1981) (explaining the development of the due process–voluntariness test).
181

384 U.S. 436 (1966). The majority opinion in Miranda expressly relies on the Fifth Amendment, but there
has been some confusion about whether it also implicates the Sixth Amendment right to counsel as well. Indeed,
Justice Harlan’s dissent raises this concern, arguing that Miranda is actually based on the Sixth Amendment which
should be inapplicable to police interrogation. Id. at 510 (Harlan, J., dissenting). See also Eve Brensike Primus,
Disentangling Miranda and Massiah: How to Revive the Sixth Amendment Right to Counsel as A Tool for
Regulating Confession Law, 97 B.U. L. REV. 1085, 1086 (2017).
With respect to my exposition, I use the facts from Ernesto Miranda’s case as illustrative, but in its opinion the
Court decided three other similar cases.
182

Id. at 491–92.

183

Id. at 492.

57

Electronic copy available at: https://ssrn.com/abstract=3730499

“[T]he prosecution may not use statements, whether exculpatory or inculpatory,
stemming from custodial interrogation of the defendant unless it demonstrates the
use of procedural safeguards effective to secure the privilege against selfincrimination. . . . Prior to any questioning, the person must be warned that he has
a right to remain silent, that any statement he does make may be used as evidence
against him, and that he has a right to the presence of an attorney, either retained
or appointed. The defendant may waive effectuation of these rights, provided the
waiver is made voluntarily, knowingly and intelligently. If, however, he indicates
in any manner and at any stage of the process that he wishes to consult with an
attorney before speaking there can be no questioning. Likewise, if the individual
is alone and indicates in any manner that he does not wish to be interrogated, the
police may not question him.”184
A striking feature of Miranda was that the Court’s Miranda warnings were prophylactic in
nature, to safeguard potential victims from constitutional violations before they would
prototypically happen. Thus, in the wake of Miranda, many questioned whether the Court had
the power to prescribe such a requirement and whether it could be overruled by statute. 185 Then
in Dickerson v. United States, the Court put these questions to rest, reaffirming that Miranda was
a constitutional decision that could not be overruled. 186
The strictures of Miranda and the limitations on law enforcement conduct continued to
evolve. For example, in Michigan v. Mosley, the Court held that law enforcement is not required

184

Id. at 444.

185

See, e.g., Paul G. Cassell, The Statute that Time Forgot: 18 U.S.C. § 3501 and the Overhauling of Miranda,
85 IOWA L. REV. 175, 227 (1999) (arguing that Miranda is prophylactic and thus could and should be overruled
through legislation); Joseph D. Grano, Prophylactic Rules in Criminal Procedure: A Question of Article III
Legitimacy, 80 NW. U. L. REV. 100, 145 (1985) (arguing that setting prophylactic rules are outside the Article III
power).
186

Dickerson v. United States, 530 U.S. 428, 431–33 (2000). Interestingly, Dickerson mainly affirmed Miranda
on the basis of stare decisis, offering no other affirmative grounds for why Miranda was normatively a good
decision. Id. at 429 (“Whether or not this Court would agree with Miranda’s reasoning and its rule in the first
instance, stare decisis weighs heavily against overruling it now.”).

58

Electronic copy available at: https://ssrn.com/abstract=3730499

to cease interrogation upon the defendant’s invocation of the right to remain silent; they can
reapproach some time after.187 And then in Edwards v. Arizona, the Court held that questioning
must cease immediately after a defendant’s exercise of the right to counsel, and can only
recommence when the defendant themselves reinitiates.188 And the story of that development
continues on.189
2.

The Evidentiary Doctrine

Apart from the constitutional limitations, there are also evidentiary doctrines to ensure
the reliability of confession evidence. The most important of these is the requirement of
corroboration of the confession. The corroboration rule has its foundations in the corpus delicti
rule that requires that the prosecution “introduce some evidence independent of the confession to
establish that the crime described in the confession actually occurred.” 190 The rule originated in
the late 1600s in England, “to prevent the conviction of those who confessed to non-existent
crimes as a result of coercion or mental illness. By the end of the nineteenth century, the corpus
delicti rule had been adopted in some form by almost all American jurisdictions.” 191

187

423 U.S. 96, 96 (1975).

188

451 U.S. 477, 477 (1981).

189
See also Godsey, supra note 174, at 474–510 (drawing a similar story of the development of constitutional
limitations on confession evidence); Catherine Hancock, Due Process before Miranda, 70 TUL. L. REV. 2195 (1995)
(explaining the development of the Due Process Clause cases); Eve B. Primus, The Future of Confession Law:
Toward Rules for the Voluntariness Test, 114 MICH. L. REV. 1, 10–15 (2015) (explaining the development of the
Miranda line of cases); Stephen J. Schulhofer, Confessions and the Court, 79 MICH. L. REV. 865, 865–69 (1981)
(explaining the development of the due process–voluntariness test).

There is also the Sixth Amendment protection that a defendant’s confession may be excluded if questioned after
being indicted outside the presence of defendant’s counsel. Massiah v. United States, 377 U.S. 201, 201 (1964). But
this is in practice irrelevant, because most interrogation happened before the defendant’s indictment. Richard A.
Leo, Miranda and the Problem of False Confessions, in THE MIRANDA DEBATE: LAW, JUSTICE, AND POLICING 271,
275 (Richard A. Leo & George C. Thomas III eds., 1998).
190

David A. Moran, In Defense of the Corpus Delicti Rule, 64 OHIO ST. L.J. 817, 817 (2003). “Corpus delicti”
literally means “body of the crime.” Id. (citing BLACK’S LAW DICTIONARY (7th ed. 1999)).
191

Id. at 818.

59

Electronic copy available at: https://ssrn.com/abstract=3730499

Thereafter, the corpus delicti rule seemed to have lost some of its footing, with concerns
that it was overbroad and too stringent. In Opper v. United States, the Supreme Court rejected
the corpus delicti rule in favor of a more lenient corroboration rule. 192 The Court explained,
“[W]e think the better rule to be that the corroborative evidence need not be sufficient,
independent of the statements, to establish the corpus delicti. It is necessary, therefore, to require
the Government to introduce substantial independent evidence which would tend to establish the
trustworthiness of the statement. Thus, the independent evidence serves a dual function. It tends
to make the admission reliable, thus corroborating it while also establishing independently the
other necessary elements of the offense. . . . It is sufficient if the corroboration supports the
essential facts admitted sufficiently to justify a jury inference of their truth. 193
Jurisdictions across the United States are split on adoption of the corroboration rule or the
corpus delicti rule.194 But even the jurisdictions that have continued with the corpus delicti rule
have relaxed the threshold of evidence required. 195 And part of this is because there is a sense
that the constitutional doctrines to safeguard the confessing defendant have made the concerns
animating the corpus delicti rule less salient and concerning. 196
Apart from the corpus delicti/corroboration requirements, the operation of the other
evidentiary rules and the adversarial system may limit the dangers of confession evidence. For

192

348 U.S. 84, 84 (1954).

193

Id. at 93.

194

Moran, supra note 190, at 832 & ns. 103, 106.

195
Thomas A. Mullen, Rule Without Reason: Requiring Independent Proof of the Corpus Delicti as a Condition
of Admitting an Extrajudicial Confession, 27 U.S.F. L. REV. 385, 416–18 (1993).
196

1 MCCORMICK ON EVIDENCE § 145, at 563–64 (4th ed. 1992) (stating “[g]iven the development of other
confession law doctrines, especially Fifth Amendment protections as promulgated in Miranda and the voluntariness
requirement, concerns regarding law enforcement interrogation practices do not provide significant support for the
corroboration requirement”).

60

Electronic copy available at: https://ssrn.com/abstract=3730499

example, the test of relevance, prototypically in Federal Rules of Evidence Rule 401, requires
that the evidence “has any tendency to make a fact more or less probable than it would be
without the evidence.”197 Rule 402 provides the default is to admit all relevant evidence, but
Rule 403 allows for exclusion if the evidence would be unduly prejudicial. 198 Thus, if confession
evidence shows obvious failures of reliability, and is not captured by a constitutional filter, Rules
401, 402, and 403 could provide an additional protection. Moreover, presenting expert testimony
on the nature of false confessions, presented under the prototypical Rule 702, 199 may be useful to
defendants seeking to undermine the jury’s credence in any confession evidence offered against
the defendant. That said, courts have been split on whether such testimony passes muster for
admission.200
B.

The Practical Problems with the Actual Doctrines

Each of these three types of doctrines has serious and substantial problems in
safeguarding defendants from the harms of confession evidence. Of course, as a preliminary
point, the practical and empirical evidence already presented already robustly establishes the
inefficacy of the current doctrine. That is, the practical and empirical evidence demonstrates that
false confessions occur at a significantly high rate; that the conditions of interrogation,
investigation, and prosecution are fertile ground for false confessions to occur; and that when
false confessions occur they often lead to false conviction. 201 But I contend that this is not merely

197

FEDERAL RULES OF EVIDENCE RULE 401.

198

FEDERAL RULES OF EVIDENCE RULES 402, 403.

199

FEDERAL RULES OF EVIDENCE RULE 702.

200

For a comprehensive analysis of the state of the case law on admission of expert testimony on confessions,
see Nadia Soree, When the Innocent Speak: False Confessions, Constitutional Safeguards, and the Role of Expert
Testimony, 32 AM. J. CRIM. L. 191, 238–55 (2005).
201

See supra Part IV.

61

Electronic copy available at: https://ssrn.com/abstract=3730499

an abnormal or historically contingent result that we should expect to change with proper
application of the doctrine. Instead, the problems of the extant doctrines are directly related to
the continuing problems of confession evidence.
To this end, I contend that the problems of the doctrines are that they have stringent
preconditions, are applied with capacious discretion, and have substantial exceptions.
Specifically, the Miranda prophylactic rules can be waived by the defendant and require clear
exercise by the defendant. The due process–voluntariness protections are only triggered by
government action, generally require extreme conduct, and provide unclear guidance about what
is allowed. Finally, the evidentiary limitations are capacious, without an understanding of the
jury’s capabilities.
1.

Miranda Prophylactic Rules Are Absent

The structure of Miranda’s prophylactic rules require that the defendant who faces
interrogation and investigation exercise the prophylactic protections afforded. As a consequence,
the primary way in which Miranda is supposed to protect the defendant is through the
defendant’s behavior. That is, if law enforcement delivered Miranda warnings, but did nothing
further to change their interrogation style, and no defendants exercised their rights under
Miranda, then we would expect no changes in outcomes.202 It is the defendant’s use of the
Miranda rights that serves as the primary pathway for the proffered prophylaxis.

202

One might think that the provision of the Miranda warnings would substantially change law enforcement
behavior, to then reduce violative behavior by law enforcement and thus reduce the potential for the harms of
confession evidence. But there is good reason to think this is not the case. Of course, Miranda warnings likely do
reduce the incidence of outright coercive conduct by law enforcement. But Miranda has not seemed to significantly
reduce the use of tactics by law enforcement that bring about false confessions. Indeed, manuals on interrogation
instruct law enforcement on methods to dull or circumvent the force of the Miranda instructions to the defendant —
and continue with interrogation practices that we saw engender false confessions.

62

Electronic copy available at: https://ssrn.com/abstract=3730499

Consequently, there are at least two ways in which these Miranda rules do not take
important effect: waiver and the requirement of clear and continued exercise. First, Miranda
rights can be waived by defendants, which then blocks the primary prophylactic effect. And the
data tells us that defendants very frequently waive their Miranda rights. In a study by Leo, over
78% of defendants waived their Miranda rights and proceeded with interrogation.203 In an
empirical study by Paul G. Cassell and Bret S. Hayman, over 83% of defendants waived their
Miranda rights and proceeded with interrogation.204 Thus, for a great number of cases,
Miranda’s primary pathway of protecting defendants does not materialize. What’s more, the
suspects least likely to give a false confession — those who have been prior acquainted with the
criminal justice system and interrogation — are also those most likely to exercise their Miranda
rights to terminate questioning and seek counsel.205 Indeed, after the Miranda warnings are
given, Miranda “provides virtually no restrictions on interrogation practices designed to induce
Miranda waivers.”206 Thus, for the people who are most vulnerable to the perils of interrogation,
Miranda has little effect, because it rarely comes into play.
Second, the post-Miranda case law has required of defendants a high degree of clarity
and resolve in the exercise of their Miranda rights. Indeed, the Supreme Court has allowed law
enforcement tools to continue pressure on defendants, even after they have attempted to invoke
their rights. For example, in Michigan v. Mosley, the Court held that law enforcement can
reapproach a defendant after the defendant exercises their right to remain silent. And in Davis v.

203

Richard A. Leo, Inside the Interrogation Room, 86 J. CRIM. L. & CRIMINOLOGY 266, 286 (1996).

204

Paul G. Cassell & Bret S. Hayman, Police Interrogation in the 1990s: An Empirical Study of the Effects of
Miranda, 43 UCLA L. REV. 839, 859 (1996).
205

Richard A. Leo, Inside the Interrogation Room, 86 J. CRIM. L. & CRIMINOLOGY 266, 286 (1996).

206

White, supra note 70, at 1217.

63

Electronic copy available at: https://ssrn.com/abstract=3730499

United States, the Court required that a defendant articulate the desire for counsel’s presence
sufficiently clearly that a reasonable officer would understand it to be a request for an attorney.
In the abstract, these seem like reasonable enough requirements. But the impact of these
decisions in practice is far-reaching. Law enforcement interrogators wield a great deal of power
in the actual interrogation.207 Miranda itself recognized this: “The circumstances surrounding incustody interrogation can operate very quickly to overbear the will of one merely made aware of
his privilege by his interrogators.”208
As White describes, “the practices employed by seasoned interrogators often have the
effect of undermining a suspect’s ability to assert rights.” 209 Through controlling the pace and
topics of discussion and using the common tactics of interrogation, interrogators can steer
defendants away from invocation of their rights, and this can happen even in the most critical
points of the interrogation.210 Moreover, the most vulnerable targets are at significant
disadvantages to exercise their rights. As Justice Souter observed in his Davis concurrence, “A
substantial percentage of them lack anything like a confident command of the English language,
many are ‘woefully ignorant,’; and many more will be sufficiently intimidated by the
interrogation process or overwhelmed by the uncertainty of their predicament that the ability to
speak assertively will abandon them.” 211
Indeed, the manuals on interrogation instruct investigators how to deflect and circumvent
potential invocations of Miranda by defendants. For an illustrative example, suppose a defendant

207

Id. at 1215.

208

Miranda v. Arizona, 384 U.S. 436, 469 (1966).

209

White, supra note 70, at 1215.

210

Id.

211

Davis v. United States, 512 U.S. 452, 469–70 (1994) (Souter, J., concurring).

64

Electronic copy available at: https://ssrn.com/abstract=3730499

says, “Maybe I need an attorney.” Post-Miranda case law would allow interrogators to ignore the
comment and proceed, or respond to convince the defendant otherwise. The influential Inbau
Interrogation Manual suggests inter alia that the interrogator could respond, “I’m only looking
for the truth, and if you’re telling the truth, that’s it. You can handle this by yourself.” 212 Here
again, the defendants most vulnerable to falsely confessing will often not be able to exercise the
kind of resolve over time and clarity of desire to invoke Miranda rights, especially against law
enforcement tactics that are specifically devised to counter the defendants’ ability to do so. As a
result, we have an even further diminished ability of defendants to exercise their Miranda rights.
Thus, we are left with a situation where the Miranda prophylactic rules simply do not
take effect in the vast majority of cases. This poses a serious problem, then, because it is the
exercise of these Miranda rights that we expect to lower the risk of false confessions.
2.

Due Process Voluntariness Limitations Are Failing

The due process–voluntariness limitations serve as a backstop to the Miranda
prophylactic rules, guarding against violations when the prophylactic rules are ineffectual. And,
as we have seen, because Miranda rights are often waived or not exercised, the due process–
voluntariness limitations must operate in order to protect defendants from the harms of
confession evidence. However, because of the requirements of government action and extreme
conduct, and the lack of clear guidance on these requirements, the due process–voluntariness
limitation has proven ineffectual.

212

FRED E. INBAU & JOHN E. REID, CRIMINAL INTERROGATION AND CONFESSIONS 112 (1962).

These requirements of clarity in exercising Miranda rights also impact waivers, because most Miranda waivers
are oral. See, e.g., Cassell & Hayman, supra note 204, at 859 (observing in the study that virtually all of the waivers
were verbal and not written). As such, law enforcement can interpret a failure to clearly exercise Miranda rights as
operationally a waiver.

65

Electronic copy available at: https://ssrn.com/abstract=3730499

As observed, the Court in Connelly made a prerequisite for the due process–voluntariness
test that there be some offending conduct by law enforcement for purposes of the constitutional
protection.213 George Dix noted that this was a striking change, because the focus no longer was
the defendant’s mental state, but rather the nature of the official conduct in coercing the
defendant.214 However, given the potentially oppressive nature of interrogation and the broader
investigation, a focus on official conduct rather than the defendant defangs the due process–
voluntariness test. The practical facts reveal that law enforcement can dress their conduct in a
way that meets the strictures of any requirements imposed upon them and yet still impose a great
deal of pressure on the defendant. This has the obvious impact that the due process–voluntariness
test, by searching for wrongful official conduct, will be blind to myriad occasions where
voluntariness is lacking.
This is made worse by the nature of the examination of official conduct. Under the due
process–voluntariness test, what is clearly excluded is extreme conduct by law enforcement, such
as confessions induced by force, threats of force, or inducements of protection from force and
confessions conducted under odious conditions. 215 This kind of conduct — likely due to the
operation of the due process–voluntariness test itself — is considerably rare. And though it is
laudable that the due process–voluntariness test has curbed the occurrence of this kind of
extreme conduct, we know that such extreme conduct is not necessary for the harms of
confession evidence to manifest.

213

George Dix, Federal Constitutional Confession Law: The 1986 and 1987 Supreme Court Terms, 67 TEX. L.
REV. 251, 272 (1988).
214

Id.

215

White, supra note 70, at 1218.

66

Electronic copy available at: https://ssrn.com/abstract=3730499

Absent such extreme conduct, however, the due process–voluntariness test operates as a
totality-of-circumstances test, that considers both the nature of the putative offending official
conduct and the characteristics and circumstances of the defendant. 216 But such a totality-ofcircumstances test has offered little guidance to law enforcement and defendants. This seems not
to have constrained law enforcement, who have largely decided to employ tactics that have not
been explicitly ruled impermissible.217 And a review of the case law shows that law enforcement
is still allowed to use tactics that can pressure and overwhelm defendants. 218
This is further bolstered by the history — where it was the ineffectiveness of the due
process–voluntariness test that led to Miranda.219 But, after Miranda, the due process–
voluntariness test has not become any more restrictive on law enforcement than its pre-Miranda
incarnation. Indeed it is arguably less restrictive because of Miranda, as the Miranda warnings
and rights were supposed to serve to protect the defendant. 220 As such, there has seemingly been
some weakening of the due process–voluntariness test, after Miranda.221

216

Id.

217

Id. at 1218–19; Lawrence Herman, The Supreme Court, the Attorney General, and the Good Old Days of
Police Interrogation, 48 OHIO ST. L.J. 733, 752 (1987).
218
See, e.g., United States v. Villalpando, 588 F.3d 1124, 1129 (7th Cir. 2009) (holding that an officer’s
promise to defendant that the officer would aid defendant in avoiding revocation of probation did not render the
confession involuntary); United States v. Turner, 674 F.3d 420, 433 (5th Cir. 2012) (holding that an officer’s
promise that if the defendant could “get it straight,” he could see his four-year-old daughter’s first day of school did
not render defendant’s incriminating statements involuntary); United States v. Binford, 818 F.3d 261, 266, 272 (6th
Cir. 2016) (holding that an officer’s statement of “you help me, I help you” to defendant did not render
incriminating statements involuntary); United States v. Swan, 842 F.3d 28, 31, 34 (1st Cir. 2016) (holding that an
officer’s promise to defendant that the officer would inform the prosecutor of defendant’s cooperation did not render
defendant’s incriminating statements involuntary).
219

Welsh S. White, supra note 70, at 1219; Geoffrey R. Stone, The Miranda Doctrine in the Burger Court, 1977
SUP. CT. REV. 99, 100 (stating that the limitations of the voluntariness test led the Court to decide Miranda).
220

White, supra note 70, at 1219.

221
Id.; Louis Michael Seidman, Brown and Miranda, 80 CAL. L. REV. 673, 745–466 (1992) (observing that the
ritual of Miranda was traded for a weaker voluntariness test).

67

Electronic copy available at: https://ssrn.com/abstract=3730499

What emerges from this is a due process–voluntariness test that is blind to a plethora of
potential cases of harm arising from confession evidence. That is because the due process–
voluntariness test only applies to cases with putative offending official conduct, and only clearly
prohibits extreme conduct. Consequently, there are a number of cases — where the pressure
arises from the broad nature of the interrogation and investigation or where the law enforcement
conduct is borderline — that the due process test does not certainly reach and thus does not
protect the defendant.
3.

The Evidentiary Limitations Are Feeble

With the ineffectiveness of the constitutional protections, the evidentiary limitations are
the last line of defense. But these too have failed to do the required work. The main problem with
these evidentiary limitations is that they leave too much to the jury, where the jury lacks
precisely the rational capabilities to competently assess the evidence.
The corpus delicti rule and the more limited corroboration rule are useful insofar as they
exclude confessions in obviously wrong cases, but they do not significantly protect against the
current problems of confession evidence. That is because satisfying the corpus delicti and
corroboration rules is relatively easy. So long as there is independent evidence of a crime’s
commission, that is sufficient to pass through the tests. 222 That serves an important purpose. It
ensures baseless false confessions, which are the ones most likely caused by egregious conduct,
do not result in false convictions. However, these cases are the ones also most likely to be culled
by operation of the constitutional doctrines (and so the tests may still provide a useful backup).
But the corpus delicti and corroboration rules do not help much in the troublesome cases that

222

See supra notes 194–196 and accompanying text.

68

Electronic copy available at: https://ssrn.com/abstract=3730499

have persisted. That is because in prototypical cases of interrogation, there is sufficient suspicion
of the defendant to subject them to interrogation223 — and that will be enough to pass through
the corpus delicti and corroboration tests.
Similarly, the test for relevance of evidence is not practically any significant bar to
confession evidence.224 Though I have shown that confession evidence is of low probative value,
with high potential for undue prejudice, courts have rarely barred confession evidence on the
basis of relevance or prejudice.225 Insofar as these rules do restrict confession evidence, that
would likely be duplicative of the constitutional rules. That is, for example, a confession
produced by means of physical threat may be also restricted as a matter of reliability, in addition
to the constitutional violations.
From an evidentiary perspective, the prevailing attitude and justification for the leniency
of these rules is that the adversarial system is the best check on the epistemic value of the
evidence and that the jury should be allowed to make the determination. However, this is
precisely the wrong view, given the competencies of the jury. As shown above, juries irrationally
overestimate the probative value of evidence to prove the necessary facts, and are the infected
with prejudicial bias.226 In particular, jurors do not properly understand the rates of confessions
and false confessions, the nature of interrogation, and the potential reasons for confession.
What’s more, the most potent evidence for the defendant on these issues — namely, the

223

See supra note 85.

224

See FEDERAL RULES OF EVIDENCE RULES 401, 402, 403.

225

I have not been able to find any case, state or federal, in WestLaw’s database where a confession was barred
for failure of relevance on the basis of its unreliability.
226

See supra Part III.B and Part IV.C.

69

Electronic copy available at: https://ssrn.com/abstract=3730499

defendant’s own testimony — is severely compromised, because the defendant must discredit
their own words (in the confession) and thus themselves.
The other evidence that may be useful on these topics is expert testimony. But courts
have been mixed in their reception of expert evidence on confessions. 227 They either do not think
that the testimony is reliable under scientific standards, prototypically under Rule 702 and the
Daubert or Frye standards, or courts believe that such evidence would be potentially confusing
and prejudicial to the jury.228 In cases where the expert evidence is barred, there does seem to be
an overarching sensibility in judges that juries are competent to assess confessions subjected to
the adversarial process, without the need for expert intervention. 229 That is because anything that
experts may add could be brought to the jury’s attention through case-specific facts, about the
nature of the interrogation and the psyche of the confessing defendant.
The most promising pathway is the use of expert evidence. It has the best chance of
educating jurors about the pressures and decision making of the confessing defendant. But it is
not yet clear how such expert evidence impacts the jurors’ understanding. 230 One cause of

227
Nadia Soree, When the Innocent Speak: False Confessions, Constitutional Safeguards, and the Role of
Expert Testimony, 32 AM. J. CRIM. L. 191, 238–55 (2005).
228

Id.

229

Id.; Brian Cutler, Keith A. Findley & Danielle Loney, Expert Testimony on Interrogation and False
Confession, 82 UMKC L. REV. 589, 600 (2014) (“Frequently, however, the testimony is excluded on the basis that
the evidence will not ‘help’ the jury because it is within the common experience and knowledge of ordinary people
and therefore the testimony invades the province of the jury to decide credibility questions for itself.”); Danielle E.
Chojnacki, Michael D. Cicchini, & Lawrence T. White, An Empirical Basis for the Admission of Expert Testimony
on False Confessions, 40 ARIZ. ST. L. J. 1, 22 (2008); Christopher Slobogin, The Structure of Expertise in Criminal
Cases, 34 SETON HALL L. REV. 105, 113–15 (2003) (examining the admissibility of expert testimony on false
confessions). See also United States v. Adams, 271 F.3d 1236 (10th Cir. 2001); People v. Son, 93 Cal. Rptr.2d 871,
883 (Ct. App. 2000); People v. Gilliam, 670 N.E.2d 606, 619 (Ill. 1996); People v. Polk, 942 N.E.2d 44 (Ill. App.
Ct. 2010); State v. Davis, 32 S.W.3d 603, 608–09 (Mo. Ct. App. 2000); State v. Free, 798 A.2d 83, 84 (N.J. Super.
Ct. App. Div. 2002).
230

See, e.g., Neil Vidmar & Shari Seidman Diamond, Juries and Expert Evidence, 66 BROOK. L. REV. 1121,
1158 (2001) (surveying studies on expert testimony on eyewitness testimony, which show some increased sensitivity
to relevant considerations, but not an overall increased skepticism of eyewitness testimony); Kelsey S. Henderson &
Lora M. Levett, Can Expert Testimony Sensitize Jurors to Variations in Confession Evidence?, 40 LAW & HUM.

70

Electronic copy available at: https://ssrn.com/abstract=3730499

concern it that, because both sides will submit experts, expert testimony will simply result in the
dreaded “battle of the experts.”231 In such a battle, both sides may be able to inundate the jury
with information to an equal duel, and the expert testimony does little work in educating the
jury.232
Ultimately, in practice, the standard evidentiary limitations — the corpus
delicti/corroboration rule and the requirements of relevance/prejudice — do little to protect
defendants beyond the constitutional protections. They do not tackle the main problem — the
gap between the real probative value of confession evidence and potential for prejudice and the
jury’s estimation of these facts. Expert testimony offers potential in this regard, but the realities
of the adversarial system may also undercut the ability to practically educate the jury and
improve its decision making.
C.

The Misfocus of Doctrinal Limitations

I have argued that the doctrinal limitations — both the constitutional and evidentiary
doctrines — are insufficient to protect defendants from the harms of confession evidence. In
assessing the failures of these doctrines to protect against the present harms of confession
evidence, there is one key problem: The doctrines fail to understand the key point that innocent
criminal defendants may have strong reasons to falsely confess, even without any intervening

BEHAV. 638, 648 (2016) (showing some promise of expert testimony in sensitizing jurors to features of
confessions); Stephen J. Ceci & Richard D. Friedman, The Suggestibility of Children: Scientific Research and Legal
Implications, 86 CORNELL L. REV. 33, 103 (2000) (arguing for expert testimony regarding the suggestibility of
children but recognizing the problem of the battle of the experts).
231
Neil Vidmar & Shari Seidman Diamond, Juries and Expert Evidence, 66 BROOK. L. REV. 1121, 1164 (2001)
(reviewing empirical studies of how juries assessed adversarial expert evidence).
232

See, e.g., Mark S. Brodin, Behavioral Science Evidence in the Age of Daubert: Reflections of A Skeptic, 73
U. CIN. L. REV. 867, 928 (2005) (discussing the concerns of the “battle of experts” with respect to eyewitness
testimony). See infra Part VII.B (discussing the problems with the solution of expert evidence at greater length).

71

Electronic copy available at: https://ssrn.com/abstract=3730499

offensive conduct by law enforcement and the prosecution, but rather due to pressures inherent
in the criminal justice system.
As seen, the constitutional doctrines focus on law enforcement conduct, seeking to
eliminate a narrow set of practices. Outside of those practices, the constitutional doctrine is inert.
But this fails to reach the problem that the entirety of the criminal justice system itself — among
other things, the nature of interrogation and investigation, the overcriminalization of various
types of conduct, the exceedingly lengthy sentences for crimes and the nature of plea bargaining,
and the horrific conditions of incarceration — imposes sufficient pressures on the innocent
defendant to falsely confess.233 The evidentiary rules fail to recognize that jurors are by-andlarge unfamiliar with the fact that innocent criminal defendants have strong reasons to falsely
confess, instead assuming that the adversarial system is capable of bringing this to light and that
juries are competent to consider this in assessing confession evidence. But, unfortunately, these
assumptions have proven to be false, because the system often does not admit evidence or

233
See, e.g., Darryl K. Brown, Democracy and Decriminalization, 86 TEX. L. REV. 223, 223–24 (2007)
(discussing the dangers of overcriminalization); William J. Stuntz, The Pathological Politics of Criminal Law, 100
MICH. L. REV. 505, 519 (2001) (discussing the positive relationship between expansive laws and the rate of plea
bargaining); Stephanos Bibas, Plea Bargaining Outside the Shadow of Trial, 117 HARV. L. REV. 2463, 2486 (2004)
(explaining how high maximums impact plea bargaining); James E. Robertson, A Clean Heart and an Empty Head:
The Supreme Court and Sexual Terrorism in Prison, 81 N.C. L. REV. 433 (2003) (detailing the prevalence of sexual
violence in U.S. prisons); Charles Fried, Reflections on Crime and Punishment, 30 SUFFOLK U. L. REV. 681, 683–92
(1997) (describing the deplorable conditions of prisons); Matt Ford, The Everyday Brutality of America’s Prisons,
NEW REPUBLIC (Apr. 5, 2019), available at https://newrepublic.com/article/153473/everyday-brutality-americasprisons; HUMAN RIGHTS WATCH, PRISON CONDITIONS IN THE UNITED STATES (1991), available at
https://www.hrw.org/sites/default/files/reports/US91N.pdf; Thea Johnson, Fictional Pleas, 94 IND. L.J. 855, 868
(2019) (explaining the practices of “creative plea bargaining” and “fictional pleas,” where defendants plea to
criminal acts with sanctions more serious than they might generally obtain or where defendants plea to criminal acts
they did not commit in order to avoid collateral consequences, including deportation and sex offender registration).

That is, defendants are confronted with overwhelming atmospheric pressures — including isolation,
hopelessness, and depression — through the nature and design of investigation and interrogation; they are made to
feel that conviction and punishment is inevitable through investigation, interrogation, and the fact of
overcriminalization; they then confront the horrors of punishment and incarceration; and thus they may be persuaded
that falsely confessing is rational and appropriate, to avoid the strong likelihood of a huge increase in punishment.
To be sure, not all defendants will succumb to these pressures, but it is rational and predictable that some defendants
will.

72

Electronic copy available at: https://ssrn.com/abstract=3730499

instructions that explain the nature of the systemic pressures and jurors nevertheless prejudicially
overestimate the probative value of confession evidence. Without rectifying these
misapprehensions, confession evidence continues, and will continue, to pose serious risks of
harm to defendants.
VII.

The Solution of Abolishing Confession Evidence

I have made the case that confession evidence poses a significant potential for harm to
defendants. As a matter of theory, confession evidence has low probative value in showing a
defendant’s guilt, because the pressures of the criminal justice system have the potency to force
false confessions. As a matter of practice, we know that law enforcement and prosecutors exploit
the conditions of the criminal justice system to push defendants to confess. All the while, the
probative value of confession evidence is vastly overestimated by judges and juries. All of this is
indeed confirmed by the weight of the available empirical evidence. Moreover, confession
evidence poses the risks of imposing substantial moral harms on defendants, apart from the risk
of false conviction. And the doctrine is ineffectual in handling these problems of confession
evidence.
In light of these dire perils and consequences, I propose a simple solution: We must
abolish confession evidence against the defendant in criminal trials. Recall that by “confession,”
I mean a statement made by a defendant claiming that the defendant committed a crime, or
satisfied particular elements of a crime, with the knowledge that the statement will be used by
the government in a criminal prosecution of the defendant to establish or help establish an
element of a crime, based on prior conduct. Thus, I suggest the following text for the rule of the
abolition of confession evidence:

73

Electronic copy available at: https://ssrn.com/abstract=3730499

In any criminal proceeding, confession evidence is not admissible. Confession evidence
is defined as:
(1) A statement by a defendant;
(2) claiming that the defendant committed a crime, or satisfied particular elements of a
crime;
(3) where the defendant knows that the statement will be used by the government in a
criminal prosecution of the defendant to establish or help establish an element of a
crime based on prior conduct.234
As stated above, this rule excludes incriminating statements made by a defendant to an
undercover agent or informant, for example, because in such a situation, the defendant does not
have knowledge that the statement would be used to establish the crime. Also, importantly the
proposed rule does not necessarily exclude all incriminating statements. Suppose in the course of
questioning, a guilty defendant reveals knowledge of key facts of the crime, without explicitly
stating that they committed the crime. Perhaps, the defendant was attempting to provide a false
alibi, but mistakenly revealed these key facts. That is not confession evidence covered by the
rule, because the defendant did not provide that information knowing that it would be used by
the government to establish the crime. That is not to say that there are no dangers arising from
such interrogation — but they are not the focus here.
As foreshadowed above, the confession abolition rule could and should apply to plea
bargains as well. The language of the rule says “any criminal proceeding,” which includes plea
allocutions as well — not just trials. Allowing confessions, or their functional equivalent, in plea
bargains but excluding them from trials may not change the detrimental results from confession

234

The rule is itself inspired by and modeled on the Federal Rules of Evidence Rule 412, entitled “Sex-Offense
Cases: The Victim” and governing sexual proclivity and predisposition evidence against a victim of a sex offense.
FEDERAL RULES OF EVIDENCE RULE 412.

74

Electronic copy available at: https://ssrn.com/abstract=3730499

evidence. That is, it could be that the pressures imposed by law enforcement and felt by
defendants at the investigatory stage can be levied by prosecutors at the adjudicatory stage. This
is especially true of the rational calculus of punishment, but it may also be true of the pressures
of being subjected to the jeopardy of trial.
We could envision a rule that only applies to trial proceedings and not plea proceedings.
And such a rule may have useful impact. Though few criminal cases go to trial and the vast
majority are handled by plea bargain, the rule’s impact on what is available at trial would then
also impact the negotiation of pleas. If confession evidence is not allowed, because of the
potency of confession evidence, the chances of conviction are diminished, even if they remain
high. Defendants can use that to negotiate from a position of greater strength than they would
have had before. But such a proposed rule — that only applies to evidence at trial proceedings
and not plea proceedings — would be little more than a further limitation on law enforcement
conduct to elicit confessions. The idea is that, in order to be sure that a defendant is voluntarily
confessing, we insist that they do it in a plea proceeding (and perhaps live testimony at trial),
with the benefit to the right of counsel and court oversight. As we have seen, however, this again
commits the same fallacious misfocus as the current doctrine: It fails to recognized that the
pressures of the criminal justice system — in terms of the sentences that a defendant faces and
the overwhelming nature of investigation and jeopardy — can lead to false confession. Further
restricting law enforcement conduct in obtaining confessions does not substantially mitigate
those pressures, and thus does not adequately solve the problems from confession evidence.
A.

The Affirmative Case for the Abolition of Confession Evidence

The rule may at first glance seem radical. In one sense, it is: It would be a great departure
from practice as usual. Confessions, and pleas that rely on confessions or their functional
75

Electronic copy available at: https://ssrn.com/abstract=3730499

equivalents, occur in the vast majority of criminal cases. Thus, the proposal would impact most
all criminal cases. But in the more important sense — what rationally comports with our legal
commitments to defendants — abolishing confession evidence is not radical. It is the most
sensible solution.
As we have shown above, confession evidence is, theoretically and empirically, weak in
terms of probative value, but prejudicially overestimated by juries and judges. This is precisely
when evidence should be inadmissible. Consider Federal Rules of Evidence Rule 403: “The
court may exclude relevant evidence if its probative value is substantially outweighed by a
danger of one or more of the following: unfair prejudice, confusing the issues, misleading the
jury, undue delay, wasting time, or needlessly presenting cumulative evidence.” 235 The term
“misleading the jury” refers “to the possibility of the jury incorrectly evaluating the probative
value of a particular item of evidence, usually by overvaluing.” 236 Now, exclusion of evidence
under Rule 403 is a considerably rare remedy, 237 but the language of the rule is clear and, given
what we know about confession evidence, demands exclusion.
Of course, Rule 403 operates on individualized determinations of pieces of evidence. But
the Federal Rules of Evidence also include Rule 412, which operates to exclude most evidence
regarding the sexual proclivity and sexual predisposition of victims of sexual offenses in cases
involving alleged sexual misconduct.238 The animating idea behind Rule 412 was, in Congress’s
view, to protect victims from “invasion[s] of privacy, potential embarrassment[,] and sexual

235

FEDERAL RULES OF EVIDENCE RULE 403.

236

MICHAEL GRAHAM, EVIDENCE 22–23 (2002).

237

Id. at 21 (“Exclusion of relevant evidence under Rule 403 is employed sparingly as it is an extraordinary
remedy.”)
238

FEDERAL RULES OF EVIDENCE RULE 412.

76

Electronic copy available at: https://ssrn.com/abstract=3730499

stereotyping.”239 This would protect victims, and thus encourage reporting, and it would
safeguard the proceedings from unnecessary information that could corrupt the fact finding
mission of the jury.240
Rule 412 also provides an analogy in that it has a significant moral component. One of
the motivating reasons for its promulgation was that victims of sexual assault would not suffer
re-traumatization through the process of investigation and adjudication of the crimes they
suffered and to counteract the wrongful association of sexual behavior and untrustworthiness or
desert of criminal violation.241 That is, allowing evidence of victims’ unrelated sexual behavior
tended to cause these undeserved harms to victims and that it was morally wrongful to allow that
to continue. This too is similar with confession evidence — not only does it have detrimental
epistemic results, but its continued use causes significant moral harms.
Consider also Rule 404(b)’s prohibition that “[e]vidence of a crime, wrong, or other act is
not admissible to prove a person’s character in order to show that on a particular occasion the
person acted in accordance with the character.” 242 Though Rule 404 is subject to potentially
capacious exceptions in criminal cases, the justification for excluding the prior-act-propensity
evidence in the first instance is on point: “Character evidence is of slight probative value and

239

MICHAEL GRAHAM, EVIDENCE 384 (2002).

The history of Rule 412 is fascinating. Unlike most evidence rules, Rule 412 was not promulgated by the
Supreme Court or through an Advisory Commission. Rather, the rule was passed directly by Congress. The original
rape shield statute was part of the Privacy Protection for Rape Victims Act of 1978, which was designed “protect
rape victims from the admission of certain types of prejudicial evidence.”
240

Harriett R. Galvin, Shielding Rape Victims in the State and Federal Courts: A Proposal for the Second
Decade, 70 MINN. L. REV. 763, 773–77 (1986) (explaining that the rationale behind the law was to avoid
embarrassment of the victim, which was harmful in itself, but also that such “character assassination” resulted in
unfair prejudice against the victim that skewed the results of trials).
241

Id. at 795, 799 (explaining that the rationale behind the law was to avoid embarrassment and
retraumatization of the victim).
242

FEDERAL RULES OF EVIDENCE RULE 404(b)(1).

77

Electronic copy available at: https://ssrn.com/abstract=3730499

may be very prejudicial. It tends to distract the trier of fact from the main question of what
actually happened on the particular occasion. It subtly permits the trier of fact to reward the good
man [and] to punish the bad man because of their respective characters despite what the evidence
in the case shows actually happened.” 243 That is what we have here: evidence with low probative
value, but highly and prejudicially overvalued by the jury.
The proposed rule can operate as an evidentiary rule. In that form, the proposed rule
could be passed by the appropriate legislature. 244 But I think that the rule could be properly
imposed as a constitutional matter as well. As a constitutional matter, a defendant has the right to
due process of law. The Supreme Court has stated “the Due Process Clauses of the Fifth and
Fourteenth Amendments may constitute a further bar to admission of, for example, unreliable
evidence.”245 That is precisely what we have with confession evidence: unreliable evidence. 246
The truth is that the interventions to protect defendants from the harms of confession
evidence have always been radical, or at least perceived that way. For example, scholars have
observed that the decision in United States v. Bram was substantially untethered to the prior case

243

Advisory Committee Note to Fed. R. Evid. Rule 404.

244
For state criminal proceedings, the state legislature would need to pass the legislation. For federal criminal
proceedings, Congress would have to ultimately pass the legislation. Under 28 U.S.C. §§ 2071–74, the Supreme
Court may pass rules, subject to the disapproval of Congress, unless the rule “create[es], abolish[es], or modif[ies]
an evidentiary privilege” in which Congress’s express approval is required. At first glance, this would appear to
create a privilege, which would then need Congress’s approval.
245

See, e.g., Michigan v. Bryant, 562 U.S. 344, 371 n.13 (2011).

246

In an insightful article, David Crump explains the affirmative reasons why we do allow confession evidence
in, homing in on their usefulness as evidence in prosecution, the fact that when not compelled they are “fair”
evidence, and the civic duty of citizens to account for their behavior by admitting their criminality. ” David Crump,
Why Do We Admit Criminal Confessions into Evidence?, 43 SEATTLE U. L. REV. 71, 72 (2019). My arguments here
undercut the first two reasons for the admission of confession evidence. As far as the civic duty rationale, I am
skeptical that citizens have such a duty, but even they did, I contend that confession evidence would not be more
likely to lead to upholding of that duty, rather than capitulation in the face of the pressures of the criminal justice
system. Thus, the third rationale cannot alone support the admission of confession evidence, in the face of its
unreliability and potential for harm. Crump also raises the concern that widespread exclusion of confession evidence
would lead to legitimacy concerns. Id. at 97. But I think that is countermanded by the legitimacy concerns of false
confessions.

78

Electronic copy available at: https://ssrn.com/abstract=3730499

law and understanding of the Fifth Amendment. 247 Similarly, Miranda is often understood as a
“watershed” moment in the history of the Supreme Court. 248 Indeed, the nemo tenetur, and thus
the Fifth Amendment privilege against self-incrimination, were dramatic grants of rights against
the state.249 Tackling the harms of confession evidence is a behemoth, which has historically
required exceptional action.
Notwithstanding, a rule abolishing confession evidence would not be unmoored from the
law. Recall the sweeping language of Bram: “‘But a confession, in order to be admissible, must
be free and voluntary; that is, must not be extracted by any sort of threats or violence, nor
obtained by any direct or implied promises, however slight, nor by the exertion of any improper
influence. . . . A confession can never be received in evidence where the prisoner has been
influenced by any threat or promise; for 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.’”250 Abolition embodies this principle, for it
recognizes that overwhelming pressures on defendants to falsely confess pervade the criminal
justice system. Thus, the conditions set forth by Bram are nearly always present and thus require
the exclusion of confession evidence.
“Nearly” is not always, and so one might question why we need abolition, as opposed to
individualized determinations, based on new criteria for determining voluntariness that take into
account the rational calculus or the pressure of investigation and jeopardy. Suppose, for example,

247

Godsey, supra note 174, at 477.

248

See, e.g., Mark R. Brown, Weathering Constitutional Change, 2000 U. ILL. L. REV. 1091, 1103 (2000)
(using Miranda and Brown v. Board of Education, 347 U.S. 483, 495 (1954), as examples of “watershed” cases
from the Court).
249

Godsey, supra note 174, at 479.

250

168 U.S. 532, 542–44 (1897) (quoting 3 RUSSELL, CRIMES 478 (6th ed.)).

79

Electronic copy available at: https://ssrn.com/abstract=3730499

there is strong evidence that a criminal defendant was not impacted by these pressures of the
criminal justice system and confessed voluntarily. Why should we eliminate confession evidence
in those cases?
This kind of case should not concern us, for several reasons. First, most all cases will
involve defendants who are impacted by the rational calculus or the pressures of jeopardy. And if
that does impact the defendants, the problem — as enunciated by Bram — is that we cannot
easily quantify that impact and thus must exclude. The cases where the confessing defendant is
not affected by the pressures of the criminal justice system would be exceedingly rare. And then
we must ask what is motivating the defendant to confess. Among other reasons, it could be that
the defendant seeks atonement, that the defendant seeks to protect some other party, that the
defendant wants notoriety or political goals, or it could be that the defendant has no rational
reason at all.251 The first such reason for confession may coordinate with truth, but we have no
obligation to provide a criminal defendant atonement through the criminal justice system. And,
given that such a reason being the sole reason is rare, we shouldn’t contort the rule to save this
possibility. The other reasons do not actually coordinate with truth: a defendant confessing to
protect another could be sacrificing themselves, a defendant seeking notoriety or some political
goal could also be falsely confessing, and a defendant confessing with no rational reason is
generally unpredictable such that their confession is unreliable.
Second, such a confessing defendant could still otherwise provide useful information to
investigators and prosecutors that would allow for their proper conviction. Recall that the
definition of “confession” includes only statements by a defendant knowingly stating the

251
See, e.g., Gail Johnson, False Confessions and Fundamental Fairness: The Need for Electronic Recording of
Custodial Interrogations, 6 B.U. PUB. INT. L.J. 719, 726 (1997)

80

Electronic copy available at: https://ssrn.com/abstract=3730499

defendant’s guilt for the purpose of helping the prosecution establish guilt. That would
prototypically be a defendant’s statement of the form “I did it.” Other statements by the
defendants — for example, evincing uncommon knowledge of the crime — is corroborating and
supporting evidence, which prima facie can still be admitted under the abolition rule.252
Third, in a similar vein, any harm in excluding the evidence is de minimis anyway. The
confession evidence is, as shown above, so low in actual probative value, that excluding it is no
matter. Certainly, the abolition will impact conviction rates, because juries do value confession
evidence so highly, but that is a feature of the proposal — not a bug — because confession
evidence is irrationally overvalued by juries. The corroborating and supporting evidence is the
probative evidence — and that is still prima facie allowed in as it is beyond the scope of
confession evidence.
My arguments here, however, may generate a concern from the other side: If confession
evidence is so limited in definition, will this abolition rule do anything? I maintain that it will.
The evidence of the bare confession is extremely powerful evidence, without the actual probative
value to back that up. But what can make a confession much more reliable is corroborating
evidence — such as when confessing defendants know particular details of the crime that
increase the chances, that increase the chance that the defendant actually committed the crime. 253
If the defendant knows, say, where the bodies of victims are buried or what and where the
murder weapon was, that increases the chance that the defendant was the culprit. Such
corroborating evidence is epistemically useful. But the fact of confession itself — the “I did it”
— adds very little in terms of probative value. Omitting that confession evidence from the

252

See supra note 52 and accompanying text.

253

See infra note 263 and accompanying text.

81

Electronic copy available at: https://ssrn.com/abstract=3730499

evidentiary record, but keeping the corroborating evidence, gets us substantially closer to an
epistemically fair assessment of the evidence by the jury. Lose the bad, keep the good — it’s a
good strategy.
That said, there are potential gray areas. Courts may have to use their discretion to
exclude or alter the presentation of statements that have some probative value, but may strongly
indicate that the defendant did in fact confess. Consider an example: Suppose in the opening
statement, the prosecution says it will show that the “defendant knew the victim, hated the
victim, met the victim that day, and murdered the victim.” Then the prosecutor introduces
statements from an interrogation where the defendant stated that they “knew the victim”; “hated
the victim”; and “met the victim that day.” That may indicate to the jury that the defendant also
confessed to murdering the victim. Or it may not. But if it does convey to the jury, with a wink
and a nod, that there was a confession, the court could and should, in the vigilant application of
the abolition, either omit this evidence — because perhaps it is cumulative — or ask the
prosecution to alter the opening or presentation of evidence. These can be tough, judgment calls,
but trial courts are equipped to handle them. What this makes clear is that the abolition of
confession evidence — though sensible and beneficial — is not self-executing, but it is
nevertheless workable.
B.

The Superiority of Abolition Over Other Solutions

I have thus far explained why the abolition of confession evidence is justified. And part
of that justification was observing that the current existing doctrine is not a solution to the harms
of confession evidence. Thus, another solution is necessary. But, nevertheless, abolition may
seem like an axe when a scalpel would do. To see why that is not the case, it’s useful to see why
the other commonly proposed solutions are lacking.
82

Electronic copy available at: https://ssrn.com/abstract=3730499

Further limiting or scrutinizing law enforcement behavior. One schema of solution by
scholars is to call for further limiting what law enforcement can do in interrogation. White has
suggested that interrogators be strictly limited in being able to make threats or offers of leniency,
make threats of adverse consequences to loved ones, or misrepresent the quality of evidence to
the defendant.254 Others have suggested limitations on law enforcement’s use of lies and
deception.255
And many scholars have suggested electronically recording interrogation proceedings. 256
Though I would welcome further limitations, I do not think that type of solution is sufficient. For
one, law enforcement is sufficiently capable to cope with and circumvent these limitations on
behavior. Ultimately, as the manuals on interrogation show, law enforcement will be able to
communicate to the defendant the necessary information to implant the understanding of the
rational calculus of punishment or the pressures of jeopardy. There will also be vague lines, like
with whether law enforcement has properly represented the quality of evidence, and that will

254

White, supra note 70, at 1232–46. See also White, supra note 121, at 1007 (arguing that in capital cases,
interrogators should not be allowed to threaten or offer removing the death penalty from punishment, deceive
regarding the strength of evidence, or interrogate over a protracted time).
Eve Primus has argued for a clarification in the two components of the voluntariness test and application of
those components, involving a recognition of the problems of deontologically offensive law enforcement conduct
and consequentialist concerns about reliability. But the focus of both is on law enforcement behavior. Primus, supra
note 189, at 34–55.
255

Note, Amelia Courtney Hritz, “Voluntariness with A Vengeance”: The Coerciveness of Police Lies in
Interrogations, 102 CORNELL L. REV. 487, 502 (2017).
In a similar vein, some have advocated for implementation of other models of interrogation techniques, like the
PEACE model. See supra note 103. Here too, I welcome law enforcement innovation that moves to better
techniques, but I maintain that the various pressures of the system — on law enforcement and the prosecution to
obtain convictions and on defendants to capitulate and confess — still create perils that cannot be solved by
voluntary adoption of improved techniques by law enforcement. Defendants require further protections. Moreover,
rules mandating the use of such improved techniques like the PEACE model will be subject to potential
circumvention as discussed above.
256

Id. at 1026–28; Johnson, supra note 251, at 750; Richard A. Leo & Richard J. Ofshe, The Consequences of
False Confessions: Deprivations of Liberty and Miscarriages of Justice in the Age of Psychological Interrogation,
88 J. CRIM. L. & CRIMINOLOGY 429, 494 (1988).

83

Electronic copy available at: https://ssrn.com/abstract=3730499

result in very little practical oversight. Moreover, with respect to recording interrogations, I also
worry that much will happen beyond the scope of the recording — as happens from time to time
with body cameras, for example.257 And more fundamentally, these solutions focusing narrowly
on law enforcement conduct miss the forest for the trees: the criminal justice as a whole imposes
these pressures on defendants and that is sufficient to cause the harms of confession evidence.
Presumption or per se exclusion in potential death penalty cases. Another solution
embraces the points about the rational calculus to suggest exclusion of confession evidence when
the potential for the death penalty is on the table. 258 The animating idea is that the defendant’s
actions, in the face of the possibility of death, cannot be a reliable indicator of truth. And with
that, I wholeheartedly agree. But I see no reason why that is limited to the death penalty. If a
defendant faces a statutory maximum of 50 years and confesses with the hope of getting 5 or 10
years, it is the same calculus of punishment that is motivating the decision to confess. The death
penalty makes the rationality of that calculus stark and obvious, but it is no less true of other
defendants facing (what they perceive to be) stiff sentences. This holds similarly with the
pressures of jeopardy.259

257

E.g., Connie Fossi & Willard Shepard, Body Cameras Turned Off During Miami PD Arrest, NBC MIAMI
(Feb. 20, 2020), https://www.nbcmiami.com/investigations/body-cameras-turned-off-during-miami-pdarrest/2193518/; Megan Cassidy, San Francisco Police Turned Off Body Cameras Before Illegal Raid On
Journalist, Memo Says, S.F. CHRONICLE (June 18, 2020), https://www.sfchronicle.com/crime/article/San-Franciscopolice-turned-off-body-cameras-15349795.php; Eddy Rodriguez, Chicago Mayor Calls Out Officers Who Turn Off
Body Cams, Says ‘We Will Strip You of Your Police Powers’, NEWSWEEK (June 6, 2020),
https://www.newsweek.com/chicago-mayor-calls-out-officers-who-turn-off-body-cams-says-we-will-strip-youyour-police-1509167; Claudia Vargas, Body Cameras Are Ineffective Because Some Philly Cops Misuse Them,
Advocates Say, NBC PHILADELPHIA (July 6, 2020), https://www.nbcphiladelphia.com/investigators/body-cameraspolice-departments-philadelphia-septa-protests/2458223/.
258

Note, Lauren Morehouse, Confess or Die: Why Threatening A Suspect with the Death Penalty Should
Render Confessions Involuntary, 56 AM. CRIM. L. REV. 531, 532 (2019).
259
One might wonder what about short sentences or situations with obdurate statutory minimums for
punishment — where the rational calculus is skewed such that confession is not rational. First, we should inquire
about what the defendant thought — because the defendant may not have understood those niceties and instead
persisted with the belief that confession was rational. Additionally, we should ask whether the pressures of jeopardy
affected the defendant. But in the absence of those reasons, we should still endeavor to discover the motivating

84

Electronic copy available at: https://ssrn.com/abstract=3730499

Expert testimony. A now familiar solution is to introduce expert testimony regarding the
nature of confessions, the tactics used by law enforcement in interrogation, and the pressures
confessing defendants face.260 This is a promising solution, because it can elucidate for the jury
the heart of the problem about how the criminal justice system as a whole can instigate false
confessions from defendants. But as explained above, the problem is with the efficacy of expert
evidence in an adversarial system. Both sides get to present expert evidence, which results in the
infamous “battle of the experts.” That can bury the critical information for juries, such that
misconceptions about confessions persist. Indeed, because the fact of confession is so viscerally
powerful, expert evidence may not be able pierce through to the jury. 261 At the baseline, we
know that juries are not able to rationally process the true probative value of confession
evidence. Among other reasons, that occurs because a defendant challenging their own
confession is self-defeating in the sense that it impacts their own credibility and moral standing.
Adding dueling experts is unlikely to significantly change this baseline, because many of these
jury intuitions are about ground facts of credibility that are fixed, or at least resistant to
reassessment.
What’s more, given the powerful potential of confessions in bringing about a conviction,
in comparison to its low probative value, the continuing risk of harm, even with expert evidence,

reason. As I explained before, other motivating reasons seem relatively rare and often not truth adaptive. Thus, I
contend that the categorical abolition rule is still the best solution.
260

See, e.g., Brian Cutler, Keith A. Findley & Danielle Loney, Expert Testimony on Interrogation and False
Confession, 82 UMKC L. REV. 589, 590–91 (2014).
261

There are studies that suggest that expert testimony is able to make jurors more sensitive to issues with
criminal evidence — mainly confession evidence and witness identifications. A substantial methodological question
with these studies is that they do not seem to provide dueling adversarial expert testimony, as would occur in actual
criminal trials. Without that, the real-world impact of expert testimony is uncertain at best. Indeed, given the relative
resource advantages between the government and criminal defendants, it may be the case that expert testimony
further disadvantages criminal defendants.

85

Electronic copy available at: https://ssrn.com/abstract=3730499

is not worth the inclusion of confession evidence. The comparisons with Rule 412 and Rule 404
are apt. In both cases, we might think that theoretically, it’s enough to allow the adversarial
process to inform juries about the (ir)relevance of evidence of sexual proclivity and sexual
predisposition of victims of sexual offenses or past act evidence as illustrative of criminal
defendants’ character. However, our experience tells us that attorney argument or even expert
testimony in the adversarial system are not enough to overcome the potential prejudice — and
that’s why such evidence must be excluded from the jury’s consideration. 262 The same holds of
confession evidence.
Requiring more indicia of reliability from confessions. A final solution to consider is the
requirement that confessions exhibit further indicia of reliability. Leo and Ofshe suggest indicia
such as whether the confession leads law enforcement to new evidence, includes details about
non-public unusual elements of the crime, or includes details about non-public mundane
elements of the crime.263 The idea here is that, as you cannot squeeze blood from a stone, you
cannot obtain such details from an innocent defendant. One concern about this potential solution
is, with respect to known information, that law enforcement may, in the course of interrogation,
suggest to the defendant information about the crime, which in turn the defendant parrots

262

It may be the case that abolition need only be a temporary solution. As the public becomes more educated,
such that juries would come to understand the ways in which confession evidence is epistemically weak, then the
concern about overvaluing confession evidence becomes less concerning. In theory, the same is true of Rules 412
and 404. I think this is totally agreeable, but as the empirical evidence makes clear, we are not there yet.
263

Richard A. Leo & Richard J. Ofshe, The Consequences of False Confessions: Deprivations of Liberty and
Miscarriages of Justice in the Age of Psychological Interrogation, 88 J. Crim. L. & Criminology 429, 438–39
(1988).
Eugene Milhizer suggested an evidentiary rule that would look to individualized confessions and exclude them
if they were unreliable, on the preponderance of the evidence standard. Eugene R. Milhizer, Confessions After
Connelly: An Evidentiary Solution for Excluding Unreliable Confessions, 81 TEMP. L. REV. 1, 47 (2008). I do not
think this is much of a change from the present evidentiary standards. In principle, such a rule may work, but given
the considerations I have noted, the test would regularly be satisfied and thus confessions regularly excluded. For the
stated reasons, I prefer the categorical approach.

86

Electronic copy available at: https://ssrn.com/abstract=3730499

back.264 This would undercut the truth adaptiveness of the putative indicia of reliability. But
perhaps that too can be rectified with recording all the proceedings of interrogation, with a
special focus on whether the indicia of reliability were indeed genuine. But there is another point
here: This proposed solution is not far from the abolition of confession evidence. The principal
difference is the exclusion of the confession, the “I did it.” The indicia of reliability — that is,
the corroboration evidence — is doing the epistemic work. Abolition preserves the admissibility
of that evidence. And is now familiar, the confession evidence itself is low probative value,
while highly and irrationally overestimated by the jury. If it is the indicia of reliability that is
useful, keep it and use it. But that provides no warrant for the use of the epistemically flawed
confession evidence.265
*

*

*

Finally, none of these alternative solutions deal with the moral harm of confession
evidence. By allowing confession evidence, all of these other solutions still allow for and indeed
encourage the process and pathologies of interrogation. That puts the defendant in harm’s way.
Abolition does best in substantially mitigating these harms. Because no confession is admissible,
the innocent defendant will not likely face the same kind of choice of confessing for lowering

264
An examination of exonerations where there were false confessions reveals that the defendants often
provided detailed and accurate information of the crimes. It is most likely that the defendants were informed of these
details through the interrogation process. Brandon L. Garrett, The Substance of False Confessions, 62 STAN. L. REV.
1051, 1053–54 (2010) (examining false confessions in exonerations and concluding that suspects must have been
informed of the facts of the crime).
265
Another potential solution would be for the prosecution to vigilantly pursue perjury charges against falsely
confessing defendants. Drizin & Leo, The Problem of False Confessions, supra note 78, at 993–95 (discussing cases
of David Saraceno and Teresa Sornberger, where false confessors were prosecuted for obstructing justice after
falsity came to light). That might change the calculus of innocent defendants who are considering whether to confess
to obtain a lighter punishment, for example. But this is a problematic solution because it requires the acceptance by
prosecutors, who institutionally benefit from defendant’s being pressured by the punishment calculus to confess.
Moreover, it does not obviously address the defendants who confess because of the pressure of interrogation and
investigation. Indeed, it seems particularly unjust to subject defendants to the overwhelming pressures of
interrogation, and then prosecute them for succumbing to those pressures.

87

Electronic copy available at: https://ssrn.com/abstract=3730499

punishment. Moreover, the stakes of interrogation are reduced, leading to less overt pressure
from law enforcement and the prosecution, which in turn leads to less pressures on the
defendant.
C.

The Consequence of the Abolition of Confession Evidence

Abolition is by design a sweeping change, to fix the pervasive pathologies created by
confession evidence. But an overarching, far-reaching solution such as the abolition of
confession evidence may have drastic peripheral effects on the criminal justice system. I contend
that these consequences are appropriately limited. Most pressingly, abolition remains consistent
with and leaves room for interrogation and plea bargaining — two key features of the practices
of our criminal justice system most likely impacted by the proposed rule.
Interrogation. The abolition of confession evidence does not directly impact the practices
of law enforcement interrogation. It merely states that particular evidence produced from
interrogation — namely confession evidence of the form “I did it” — is not admissible in
criminal proceedings. That may mean interrogation is less useful in the production of potent
evidence against a defendant. And, thus, it may change the stakes of interrogation and the way
law enforcement approaches interrogation. Interrogation would be much more about the
production of actual and true information about the crime. That is all welcome change.
But it does not mean interrogation is useless. Interrogation can still be extraordinarily
helpful in genuine investigation — in learning the facts of the crime. By operation of the rule, the
prosecution may not be allowed to use the defendant’s confessing statements to affirmatively
prove the case, but the prosecution can use the information gleaned from an interrogation to
build the theory of the case. Moreover, the proposed rule could allow a defendant’s statements to

88

Electronic copy available at: https://ssrn.com/abstract=3730499

be used for other purposes, such as impeachment. 266 That is still extraordinarily useful, as it helps
to ensure that the defendant does not lie. Ultimately, thus the most substantial change abolition
will have on interrogation is to make it more focused on truth-adaptive information.
Plea bargaining. One of the more striking consequences of the proposed rule is that
confession evidence would not be allowed in any criminal proceeding, including in plea
allocutions. However, given that most cases are handled by guilty plea, where the plea is
supported by the confession and slight other corroborating evidence (which may be little more
than the stated crime occurred), what would this mean for plea bargaining?
There would be a room, and indeed an important place, for plea bargaining, but it would
be different and improved. Under current law, a guilty plea must still be supported by
evidence,267 and even if that evidence is a confession, it must be corroborated. 268 If there was no
confession, then the plea would still need to be supported with other evidence. Thus, under the
abolition rule, the prosecution must build a case that convinces the court that the defendant is
guilty of the charged crime beyond a reasonable doubt.
This is not an altogether foreign concept for our criminal justice system. We already have
a type of plea that approximates this — the Alford plea. Under the Alford plea, the defendant
pleads guilty, but does not admit to committing the criminal act or professes innocence. 269 As a
consequence, the prosecution must present a case to the court that would sustain a conviction

266

As stated above, I have not taken a firm position as to whether confession evidence should be allowed for
impeachment purposes, but I am open to that possibility. See supra note 16 and accompanying text.
267

McCarthy v. United States, 394 U.S. 459, 467 (1969) (requiring, under Rule 11, that the judge “personally
inquire” into whether the defendant understands the charges and must “satisfy [themselves] that there is a factual
basis for the plea”).
268

See supra note 50 and accompanying text.

269

North Carolina v. Alford, 400 U.S. 25, 31, 37 (1970).

89

Electronic copy available at: https://ssrn.com/abstract=3730499

beyond a reasonable doubt.270 Now, in some Alford pleas, even though the defendant did not
plead guilty, the sustaining evidence may include a confession — say, obtained from an
interrogation. That would not be permissible here, under the abolition rule. Importantly, for the
abolition rule to operate meaningfully, it does require that courts vigilantly ensure that the case
built against the defendant, without any confession evidence and without taking into account the
defendant’s own resignation, actually meets the BARD standard. Otherwise, we have simply
have the functional equivalent of a confession. 271
Thus, the abolition rule allows for meaningful plea bargaining. It requires law
enforcement and prosecution to build a case against the defendant, without use of confession
evidence. That may take further resources, which in turn may mean that less perpetrators are
charged and convicted. But that is primarily because law enforcement and prosecution would be
required to meet their evidentiary burden without the use of extraneous, illegitimate background
pressures that arise from the criminal justice system. Insofar as that is the case, it is a design
feature of the abolition rule — that gives effect to our societal beliefs about criminal justice. 272

270

Id. See also Note, Curtis J. Shipley, The Alford Plea: A Necessary but Unpredictable Tool for the Criminal
Defendant, 72 IOWA L. REV. 1063, 1063–65 (1987).
271
If the court were not to vigilantly require a sufficient factual basis, and instead took the plea itself to mean
that the defendant was guilty, then the defendant is essentially confessing by pleading guilty.

Another type of plea, the nolo contendere, allows defendants to not contest guilt, without admitting guilt. The
important distinction between nolo contendere pleas and Alford pleas is that nolo contendere pleas do not require an
inquiry into the factual basis for the plea. F. R. CRIM. P. 11(f) and Advisory Committee Note to Rule 11. Thus, to
effectuate the abolition rule, nolo contendere pleas would have to be reformed to require a factual basis inquiry. But
there are good reasons to independently require this anyway. John L. Barkai, Accuracy Inquiries for All Felony and
Misdemeanor Pleas: Voluntary Pleas but Innocent Defendants?, 126 U. PA. L. REV. 88, 133 (1977).
272
In particular, requiring firm and reliable evidence not afflicted with prejudice is how we ensure the BARD
standard is meaningfully applied, and consequently that the Blackstonian ratio is upheld.

90

Electronic copy available at: https://ssrn.com/abstract=3730499

D.

Is the Comprehensive Argument a Reductio Ad Absurdum?

Finally, there is the concern of how this form of reasoning might apply to other evidence.
Is this applicable to other forms, or most other forms, of evidence too? Would that mean much
commonly used evidence would be abolished and consequently that criminally prosecutions
severely encumbered?
To these questions, the answer is mostly no. As seen through these many foregoing
pages, the argument is that confession evidence is theoretically and empirically weak in
epistemic value, but overvalued by juries and judges; is as a matter of practice exploited by law
enforcement and prosecution; is morally harmful; and is unsolved by current doctrine, by their
logical extensions, or by other less extreme solutions. Converting this into criteria for the
exclusion of other evidence, we arrive at an exacting test.
So let us consider another type of evidence that has come under scrutiny in terms of its
apparent epistemic problems: eyewitness identifications.273 Eyewitness identification has an
arguable theoretical weakness in that it relies on the potentially flawed perceptions of other
humans — who may suffer from all sorts of pathologies. It is empirically confirmed that
eyewitness identifications — especially cross-racial identifications — are suspect. Moreover,
eyewitness identification evidence has a strong impact on juries, and arguably beyond their
suspect weight. But eyewitness testimony does not present any obvious moral harms beyond its
epistemic problems. Furthermore, current evidentiary doctrine has adequate safeguards to
counter eyewitness testimony that may be faulty. Advocates can cross-examine eyewitnesses and

273

Sheri Lynn Johnson, Cross-Racial Identification Errors in Criminal Cases, 69 CORNELL L. REV. 934 (1984);
Taki V. Flevaris & Ellie F. Chapman, Cross-Racial Misidentification: A Call to Action in Washington State and
Beyond, 38 SEATTLE U. L. REV. 861, 866 (2015) (noting the overwhelming evidence of high error rates in
eyewitness identifications);

91

Electronic copy available at: https://ssrn.com/abstract=3730499

doing so does not have the double-edged repercussion as with casting doubt on a confession.
Furthermore, expert testimony can elucidate the problems with eyewitness identification for
juries and understanding the reasons for mistaken identifications is not beyond the jury’s ken.
Thus, for these reasons among others, it does not seem that eyewitness identification — one of
the weaker classes of evidence — would require abolition.
Indeed, we could apply this to other kinds of potentially weak evidence, such as
inculpating codefendant testimony and fingerprint evidence, to get a similar result. Confession
evidence is uniquely problematic and meets this exacting, multifarious test of harms. And thus it
demands abolition.
But perhaps I am wrong and there are other forms of evidence — and perhaps the ones I
have listed — that do meet the exacting test and should be abolished. That is not a conclusion I
feel we must resist. Regardless of whether much of our commonly used evidence meets the
aforementioned test for abolition, it is clear that the test — with its multifarious criteria — is an
exacting one. Suppose it is the case that some other class of evidence is epistemically weak and
overvalued in theory and in empirical observation, that it is exploited by the authorities, that it is
morally harmful, and that it cannot be rectified by another doctrinal solution, then that evidence
likely should be abolished. If we take due the beyond-a-reasonable-doubt standard, due process,
and a criminal defendant’s moral standing seriously, then that evidence should be abolished.
That is what a criminal justice system demands.
VIII.

Conclusion

Confession evidence is epistemically weak: it has low probative value but is nevertheless
overvalued by juries and judges. This is true as a theoretical matter and it is also borne out and
verified through the practices of law enforcement and prosecutors in our criminal justice system
92

Electronic copy available at: https://ssrn.com/abstract=3730499

and empirical data on false confessions. Furthermore, confession evidence engenders substantial
moral harms on defendants, because the process of investigation and interrogation in our
criminal justice system imposes mental pain on defendants in order to induce confessions. This
perilous state of affairs calls for action. The doctrinal solutions, both constitutional and
evidentiary, have made significant strides in reducing the most flagrant types of wrongful law
enforcement conduct and consequent unreliable confession evidence. But the further potential of
these doctrines in solving the persisting harms of confession evidence is limited. To best address
the myriad harms introduced by the existence of confession evidence in criminal proceedings, we
must abolish it. Though the solution is drastic in comparison to our current practice, it is also the
most sensible in terms of preserving and furthering the foundational goals of our criminal justice
system.
As dramatic and ambitious as the solution of abolishing confession evidence may seem, it
is in fact merely a patch on our criminal justice system. Indeed, it leaves much to be desired:
From an evidentiary perspective, it is suboptimal in that it isolates information from the jury,
where we would rather have transparency and elucidation. From a moral harms perspective, even
excluding confession evidence leaves remaining the practices of interrogation and investigation,
along with the pressures of jeopardy those impose on defendants. Excluding confession evidence
is a substantially effective solution, given the current state of our criminal justice system, but it
should be a stopgap measure. To rectify still-persistent epistemic and moral harms, the criminal
justice system must be radically transformed. It must be one where judges, juries, and the public
at large understand fully and engage with reflectively the mechanisms by which we attempt to
obtain justice. And it also must be a system that employs mechanisms genuinely interested in and
responsive to the welfare of the public—including the welfare of defendants and their
93

Electronic copy available at: https://ssrn.com/abstract=3730499

comprehensive rehabilitation. In that world, confession may be trustworthy and reliable, as well
as reconciliatory and progressive.

94

Electronic copy available at: https://ssrn.com/abstract=3730499