Identifying the Culprit: Assessing Eyewitness Identification

Visit the National Academies Press online and register for...

Instant access to free PDF downloads of titles from the

Distribution, posting, or copying of this PDF is strictly prohibited without written permission of the National Academies Press.

Unless otherwise indicated, all materials in this PDF are copyrighted by the National Academy of Sciences.

Request reprint permission for this book

10% off print titles

Custom notication of new releases in your eld of interest

Special offers and discounts

NATIONAL ACADEMY OF SCIENCES

NATIONAL ACADEMY OF ENGINEERING

INSTITUTE OF MEDICINE

NATIONAL RESEARCH COUNCIL

This PDF is available from The National Academies Press at http://www.nap.edu/catalog.php?record_id=18891

ISBN

978-0-309-31059-8

160 pages

6 x 9

PAPERBACK (2014)

Identifying the Culprit: Assessing Eyewitness Identification

Committee on Scientific Approaches to Understanding and Maximizing the

Validity and Reliability of Eyewitness Identification in Law Enforcement and

the Courts; Committee on Science, Technology, and Law; Policy and

Global Affairs; Committee on Law and Justice; Division of Behavioral and

Social Sciences and Education; National Research Council

Identifying the Culprit: Assessing Eyewitness Identification

Committee on Scientiﬁc Approaches to Understanding and Maximizing

the Validity and Reliability of Eyewitness Identiﬁcation

in Law Enforcement and the Courts

Committee on Science, Technology, and Law

Policy and Global Affairs

Committee on Law and Justice

Division of Behavioral and Social Sciences and Education

Identifying the Culprit: Assessing Eyewitness Identification

THE NATIONAL ACADEMIES PRESS 500 Fifth Street, NW Washington, DC 20001

NOTICE: The project that is the subject of this report was approved by the Govern-

ing Board of the National Research Council, whose members are drawn from the

councils of the National Academy of Sciences, the National Academy of Engineer-

ing, and the Institute of Medicine. The members of the committee responsible for

the report were chosen for their special competences and with regard for appropriate

balance.

This study was funded by a grant between the National Academy of Sciences and

the Laura and John Arnold Foundation. Any opinions, ﬁndings, conclusions, or rec-

ommendations expressed in this publication are those of the author and do not nec-

essarily reﬂect the views of the organization that provided support for the project.

International Standard Book Number13: 978-0-309-31059-8

International Standard Book Number10: 0-309-31059-8

Library of Congress Control Number: 2014955458

Additional copies of this report are available from the National Academies Press,

500 Fifth Street, NW, Room 360, Washington, DC 20001; (800) 624-6242 or (202)

334-3313; http://www.nap.edu.

in the United States of America

Identifying the Culprit: Assessing Eyewitness Identification

The National Academy of Sciences is a private, nonproﬁt, self-perpetuating society

of distinguished scholars engaged in scientiﬁc and engineering research, dedicated to

the furtherance of science and technology and to their use for the general welfare.

Upon the authority of the charter granted to it by the Congress in 1863, the Acad-

emy has a mandate that requires it to advise the federal government on scientiﬁc

and technical matters. Dr. Ralph J. Cicerone is president of the National Academy

of Sciences.

The National Academy of Engineering was established in 1964, under the charter

of the National Academy of Sciences, as a parallel organization of outstanding en-

gineers. It is autonomous in its administration and in the selection of its members,

sharing with the National Academy of Sciences the responsibility for advising the

federal government. The National Academy of Engineering also sponsors engineer-

ing programs aimed at meeting national needs, encourages education and research,

and recognizes the superior achievements of engineers. Dr. C. D. Mote, Jr., is presi-

dent of the National Academy of Engineering.

The Institute of Medicine was established in 1970 by the National Academy of

Sciences to secure the services of eminent members of appropriate professions in

the examination of policy matters pertaining to the health of the public. The Insti-

tute acts under the responsibility given to the National Academy of Sciences by its

congressional charter to be an adviser to the federal government and, upon its own

initiative, to identify issues of medical care, research, and education. Dr. Victor J.

Dzau is president of the Institute of Medicine.

The National Research Council was organized by the National Academy of Sciences

in 1916 to associate the broad community of science and technology with the

Academy’s purposes of furthering knowledge and advising the federal government.

Functioning in accordance with general policies determined by the Academy, the

Council has become the principal operating agency of both the National Academy

of Sciences and the National Academy of Engineering in providing services to

the government, the public, and the scientiﬁc and engineering communities. The

Council is administered jointly by both Academies and the Institute of Medicine.

Dr. Ralph J. Cicerone and Dr. C. D. Mote, Jr., are chair and vice chair, respectively,

of the National Research Council.

www.national-academies.org

Identifying the Culprit: Assessing Eyewitness Identification

COMMITTEE ON SCIENTIFIC APPROACHES TO

UNDERSTANDING AND MAXIMIZING THE VALIDITY

AND RELIABILITY OF EYEWITNESS IDENTIFICATION

IN LAW ENFORCEMENT AND THE COURTS

Co-Chairs

THOMAS D. ALBRIGHT (NAS), Professor and Director, Vision Center

Laboratory and Conrad T. Prebys Chair in Vision Research, Salk

Institute for Biological Studies

JED S. RAKOFF, Senior Judge, United States District Court for the

Southern District of New York

Members

WILLIAM G. BROOKS III, Chief of Police, Norwood (MA) Police

Department

JOE S. CECIL, Project Director, Division of Research, Federal Judicial

Center

WINRICH FREIWALD, Assistant Professor, Laboratory of Neural

Systems, The Rockefeller University

BRANDON L. GARRETT, Roy L. and Rosamond Woodruff Morgan

Professor of Law, University of Virginia Law School

KAREN KAFADAR, Commonwealth Professor and Chair of Statistics,

University of Virginia

A.J. KRAMER, Federal Public Defender for the District of Columbia

SCOTT McNAMARA, Oneida County (NY) District Attorney

CHARLES ALEXANDER MORGAN III, Associate Clinical Professor of

Psychiatry, Yale University School of Medicine

ELIZABETH A. PHELPS, Silver Professor of Psychology and Neural

Science, New York University

DANIEL J. SIMONS, Professor, Department of Psychology, University of

Illinois

ANTHONY D. WAGNER, Professor of Psychology and Neuroscience

and Co-Director, Center for Cognitive and Neurobiological Imaging,

Stanford University; Director, Stanford Memory Laboratory

JOANNE YAFFE, Professor of Social Work and Adjunct Professor of

Psychiatry, University of Utah

Identifying the Culprit: Assessing Eyewitness Identification

Staff

ANNE-MARIE MAZZA, Study Director and Director, Committee on

Science, Technology, and Law

ARLENE F. LEE, Director, Committee on Law and Justice

STEVEN KENDALL, Program Ofﬁcer, Committee on Science,

Technology, and Law

KAROLINA KONARZEWSKA, Program Coordinator, Committee on

Science, Technology, and Law

ANJALI SHASTRI, Christine Mirzayan Science and Technology Policy

Graduate Fellow

SARAH WYNN, Christine Mirzayan Science and Technology Policy

Graduate Fellow

Identifying the Culprit: Assessing Eyewitness Identification

vii

COMMITTEE ON SCIENCE, TECHNOLOGY, AND LAW

Co-Chairs

DAVID BALTIMORE (NAS/IOM), President Emeritus and Robert

Andrews Millikan Professor of Biology, California Institute of

Technology

DAVID S. TATEL, Judge, U.S. Court of Appeals for the District of

Columbia Circuit

Members

THOMAS D. ALBRIGHT (NAS), Professor and Director, Vision Center

Laboratory and Conrad T. Prebys Chair in Vision Research, Salk

Institute for Biological Studies

ANN ARVIN (IOM), Lucile Packard Professor of Pediatrics and

Microbiology and Immunology; Vice Provost and Dean of Research,

Stanford University

BARBARA E. BIERER, Professor of Medicine, Harvard Medical School

CLAUDE CANIZARES (NAS), Vice President and the Bruno Rossi

Professor of Physics, Massachusetts Institute of Technology

ARTURO CASADEVALL (IOM), Leo and Julia Forchheimer Professor

of Microbiology and Immunology; Chair, Department of Biology and

Immunology; and Professor of Medicine, Albert Einstein College of

Medicine

JOE S. CECIL, Project Director, Program on Scientiﬁc and Technical

Evidence, Division of Research, Federal Judicial Center

R. ALTA CHARO (IOM), Warren P. Knowles Professor of Law and

Bioethics, University of Wisconsin at Madison

HARRY T. EDWARDS, Judge, U.S. Court of Appeals for the District of

Columbia Circuit

DREW ENDY, Associate Professor, Bioengineering, Stanford University

and President, The BioBricks Foundation

MARCUS FELDMAN (NAS), Burnet C. and Mildred Wohlford Professor

of Biological Sciences, Stanford University

JEREMY FOGEL, Director, Federal Judicial Center

HENRY T. GREELY, Deane F. and Kate Edelman Johnson Professor of

Law and Professor, by courtesy, of Genetics, Stanford University

MICHAEL GREENBERGER, Law School Professor and Director, Center

for Health and Homeland Security, University of Maryland

BENJAMIN W. HEINEMAN, JR., Senior Fellow, Harvard Law School

and Harvard Kennedy School of Government

Identifying the Culprit: Assessing Eyewitness Identification

viii

MICHAEL IMPERIALE, Arthur F. Thurnau Professor of Microbiology

and Immunology, University of Michigan

GREG KISOR, Chief Technologist, Intellectual Ventures

GOODWIN LIU, Associate Justice, California Supreme Court

JENNIFER MNOOKIN, David G. Price and Dallas P. Price Professor of

Law, University of California, Los Angeles School of Law

R. GREGORY MORGAN, Vice President and General Counsel,

Massachusetts Institute of Technology

ALAN B. MORRISON, Lerner Family Associate Dean for Public Interest

and Public Service Law, George Washington University Law School

CHERRY MURRAY (NAS/NAE), Dean, School of Engineering and

Applied Sciences, Harvard University

ROBERTA NESS (IOM), Dean and M. David Low Chair in Public

Health, University of Texas School of Public Health

HARRIET RABB, Vice President and General Counsel, The Rockefeller

University

DAVID RELMAN (IOM), Thomas C. and Joan M. Merigan Professor,

Departments of Medicine, and of Microbiology and Immunology,

Stanford University and Chief, Infectious Disease Section, VA Palo

Alto Health Care System

RICHARD REVESZ, Lawrence King Professor of Law; Dean Emeritus;

and Director, Institute for Policy Integrity, New York University

School of Law

MARTINE A. ROTHBLATT, Chairman and Chief Executive Ofﬁcer,

United Therapeutics

DAVID VLADECK, Professor and Co-Director, Institute for Public

Representation, Georgetown Law School

Staff

ANNE-MARIE MAZZA, Director

STEVEN KENDALL, Program Ofﬁcer

KAROLINA KONARZEWSKA, Program Coordinator

Identifying the Culprit: Assessing Eyewitness Identification

COMMITTEE ON LAW AND JUSTICE

Chairs

JEREMY TRAVIS (Chair), President, John Jay College of Criminal

Justice, The City University of New York

RUTH D. PETERSON (Vice-Chair), Professor of Sociology and Director,

Criminal Justice Research Center, Ohio State University

Members

CARL C. BELL, Staff Psychiatrist, St. Bernard’s Hospital; Staff

Psychiatrist, Jackson Park Hospital’s Outpatient Family Practice

Clinic; and Professor of Psychiatry and Public Health, University of

Illinois at Chicago

JOHN J. DONOHUE III, C. Wendell and Edith M. Carlsmith Professor

of Law, Stanford University Law School

MINDY FULLILOVE, Professor of Clinical Psychiatry and Professor of

Clinical Sociomedical Sciences and Co-Director, Community Research

Group, New York State Psychiatric Institute and Mailman School of

Public Health, Columbia University

MARK KLEIMAN, Professor of Public Policy, University of California,

Los Angeles

GARY LAFREE, Director, National Consortium for the Study of

Terrorism and Responses to Terrorism (START) and Professor,

Criminology and Criminal Justice, University of Maryland

JANET L. LAURITSEN, Professor, Department of Criminology and

Criminal Justice, University of Missouri

GLENN C. LOURY, Merton P. Stoltz Professor of the Social Sciences,

Department of Economics, Brown University

JAMES P. LYNCH, Professor and Chair, Department of Criminology and

Criminal Justice, University of Maryland

CHARLES F. MANSKI (NAS), Board of Trustees Professor in Economics,

Department of Economics, Northwestern University

DANIEL S. NAGIN, Teresa and H. John Heinz III University Professor of

Public Policy and Statistics, Carnegie Mellon University

ANNE MORRISON PIEHL, Associate Professor, Department of

Economics and Program in Criminal Justice, Rutgers University

DANIEL B. PRIETO, Director, Cybersecurity and Technology and

Director, Defense Industrial Base Cyber Security/Information

Assurance, Ofﬁce of the Secretary of Defense Chief Information

Ofﬁcer

SUSAN B. SORENSON, Professor, University of Pennsylvania

Identifying the Culprit: Assessing Eyewitness Identification

DAVID WEISBURD, Distinguished Professor, Department of

Criminology, Law and Society and Director, Center for Evidence-

Based Crime Policy, George Mason University; Walter E. Meyer

Professor of Law and Criminal Justice, The Hebrew University

Faculty of Law

CATHY SPATZ WIDOM, Distinguished Professor, Psychology

Department, John Jay College of Criminal Justice, The City

University of New York

PAUL K. WORMELI, Executive Director, Integrated Justice Information

Systems

Staff

ARLENE F. LEE, Director

EMILY BACKES, Research Associate

MALAY MAJMUNDAR, Senior Program Ofﬁcer

STEVE REDBURN, Scholar

JULIE SCHUCK, Senior Program Associate

DANIEL TALMAGE, Program Ofﬁcer

TINA M. LATIMER, Program Coordinator

Identifying the Culprit: Assessing Eyewitness Identification

Acknowledgments

ACKNOWLEDGMENT OF PRESENTERS

The committee gratefully acknowledges the contributions of the fol-

lowing individuals:

Karen L. Amendola, Police Foundation; Steven E. Clark, University

of California, Riverside; Rob Davis, Police Executive Research Forum;

Kenneth Deffenbacher, University of Nebraska at Omaha; Paul DeMuniz,

Oregon Supreme Court; Shari Seidman Diamond, Northwestern University

and American Bar Foundation; John Firman, International Association of

Chiefs of Police; Ronald Fisher, Florida International University; Geoffrey

Gaulkin, Special Master, State v. Henderson (NJ); Kristine Hamann, Na-

tional District Attorney’s Association; Barbara Hervey, Texas Court of

Criminal Appeals; Robert J. Kane, Supreme Judicial Study Group on Eye-

witness Identiﬁcation (MA); Saul Kassin, John Jay College of Criminal

Justice; Peter Kilmartin, State of Rhode Island; David LaBahn, Association

of Prosecuting Attorneys; Elizabeth F. Loftus, University of California,

Irvine; Roy S. Malpass, University of Texas at El Paso; Sheri Mecklenburg,

U.S. Department of Justice; Christian A. Meissner, Iowa State University;

John Monahan, University of Virginia; Steven D. Penrod, John Jay College

of Criminal Justice; P. Jonathon Phillips, National Institute of Standards

and Technology; Joseph Salemme, Chicago Police Department; Daniel L.

Schacter, Harvard University; Barry Scheck, The Innocence Project; Jessica

Snowden, Federal Judicial Center; Nancy K. Steblay, Augsburg College;

Gary L. Wells, Iowa State University; John T. Wixted, University of Cali-

fornia, San Diego; David V. Yokum, University of Arizona.

Identifying the Culprit: Assessing Eyewitness Identification

xii ACKNOWLEDGMENTS

ACKNOWLEDGMENT OF REVIEWERS

This report has been reviewed in draft form by individuals chosen for

their diverse perspectives and technical expertise, in accordance with pro-

cedures approved by the National Academies’ Report Review Committee.

The purpose of this independent review is to provide candid and critical

comments that will assist the institution in making its published report as

sound as possible and to ensure that the report meets institutional standards

for objectivity, evidence, and responsiveness to the study charge. The review

comments and draft manuscript remain conﬁdential to protect the integrity

of the process.

We wish to thank the following individuals for their review of this

report: Art Acevedo, Austin, Texas Police Department; Aaron Benjamin,

University of Illinois at Urbana-Champaign; Vicki Bruce, Newcastle Uni-

versity; Jules Epstein, Widener University; Jeremy Fogel, Federal Judicial

Center; Constantine Gatsonis, Brown University; Henry T. Greely, Stanford

University; Peter Imrey, Cleveland Clinic; Robert Kane, Massachusetts Su-

preme Court; Timothy Koller; Ofﬁce of the Richmond County District At-

torney; Elizabeth Loftus, University of California, Irvine; Robert Masters,

Ofﬁce of the Queens County District Attorney; Geoffrey Mearns, Northern

Kentucky University; and Hal Stern, University of California, Irvine.

Although the reviewers listed above have provided many constructive

comments and suggestions, they were not asked to endorse the conclu-

sions or recommendations, nor did they see the ﬁnal draft of the report

before its release. The review of this report was overseen by David Korn,

Harvard Medical School and Massachusetts General Hospital and Stephen

E. Fienberg, Carnegie Mellon University. Appointed by the National Acad-

emies, they were responsible for making certain that an independent ex-

amination of this report was carried out in accordance with institutional

procedures and that all review comments were carefully considered. Re-

sponsibility for the ﬁnal content of this report rests entirely with the author-

ing committee and the institution.

Identifying the Culprit: Assessing Eyewitness Identification

xiii

Preface

Eyewitness identiﬁcations play an important role in the investigation

and prosecution of crimes, but they have also led to erroneous convictions.

In the fall of 2013, the Laura and John Arnold Foundation called upon

the National Academy of Sciences (NAS) to assess the state of research on

eyewitness identiﬁcation and, when appropriate, make recommendations.

In response to this request, the NAS appointed an ad hoc study committee

that we have been privileged to co-chair.

The committee’s review analyzed relevant published and unpublished

research, external submissions, and presentations made by various experts

and interested parties. The research examined fell into two general cat-

egories: (1) basic research on vision and memory and (2) applied research

directed at the speciﬁc problem of eyewitness identiﬁcation.

Basic research has progressed for many decades, is of high quality,

and is largely deﬁnitive. Research of this category identiﬁes principled and

insurmountable limits of vision and memory that inevitably affect eyewit-

ness accounts, bear on conclusions regarding accuracy, and provide a broad

foundation for the committee’s recommendations.

Through its review, the committee came to recognize that applied

eyewitness identiﬁcation research has identiﬁed key variables affecting the

accuracy of eyewitness identiﬁcations. This research has been instrumental

in informing law enforcement, the bar, and the judiciary of the frailties of

eyewitness identiﬁcation testimony. Such past research has appropriately

identiﬁed the variables that may affect an individual’s ability to make an

accurate identiﬁcation. However, given the complex nature of eyewitness

identiﬁcation, the practical difﬁculties it poses for experimental research,

Identifying the Culprit: Assessing Eyewitness Identification

xiv PREFACE

and the still ongoing evolution of statistical procedures in the ﬁeld of

eyewitness identiﬁcation research, there remains at the time of this review

substantial uncertainty about the effect and the interplay of these variables

on eyewitness identiﬁcation. Nonetheless, a range of practices has been

validated by scientiﬁc methods and research and represents a starting place

for efforts to improve eyewitness identiﬁcation procedures.

In this report, the committee offers recommendations on how law

enforcement and the courts may increase the accuracy and utility of eyewit-

ness identiﬁcations. In addition, the committee identiﬁes areas for future

research and for collaboration between the scientiﬁc and law enforcement

communities.

We are indebted to those who addressed the committee and to those

who submitted materials to the committee, and we are particularly indebted

to the members of the committee. These individuals devoted untold hours

to the review of materials, meetings, conference calls, analyses, and report

writing. This report is very much the result of the enormous contributions

of an engaged community of scholars and practitioners who reached their

ﬁndings and recommendations after many vigorous and thoughtful discus-

sions. We also would like to thank the project staff, Karolina Konarzewska,

Steven Kendall, Arlene Lee, and Anne-Marie Mazza, and editor Susanna

Carey for their dedication to the project and to the work of the committee.

Thomas D. Albright and Jed S. Rakoff

Committee Co-chairs

Identifying the Culprit: Assessing Eyewitness Identification

Contents

SUMMARY 1

1 INTRODUCTION 9

2 EYEWITNESS IDENTIFICATION PROCEDURES 21

3 THE LEGAL FRAMEWORK FOR ASSESSMENT OF

EYEWITNESS IDENTIFICATION EVIDENCE 31

4 BASIC RESEARCH ON VISION AND MEMORY 45

5 APPLIED EYEWITNESS IDENTIFICATION RESEARCH 71

6 FINDINGS AND RECOMMENDATIONS 103

APPENDIXES

A BIOGRAPHICAL INFORMATION OF COMMITTEE

AND STAFF 123

B COMMITTEE MEETING AGENDAS 133

C CONSIDERATION OF UNCERTAINTY IN DATA ON THE

CONFIDENCE–ACCURACY RELATIONSHIP AND THE

RECEIVER OPERATING CHARACTERISTIC (ROC) CURVE 139

Identifying the Culprit: Assessing Eyewitness Identification

xvi CONTENTS

BOXES, FIGURES, AND TABLES

Boxes

1-1 The Ronald Cotton Case, 10

1-2 Charge to the Committee, 12

2-1 Blinding, 26

5-1 The Inﬂuences of Discriminability and Response Bias on Human

Binary Classiﬁcation Decisions, 81

5-2 Analysis of Receiver Operating Characteristics (ROCs), 84

Figures

1-1 Memory accuracy and time, 17

5-1 Contingency table for possible eyewitness identiﬁcation

outcomes, 78

C-1 Data inferred from Juslin, Olsson and Winman, 143

C-2 Data from Brewer and Wells, 147

C-3 Data from Experiment 1A in Mickes, Flowe, and Wixted, 148

C-4 Data from Experiment 2 in Mickes, Flowe, and Wixted, 149

Tables

C-1 Conditions and Logarithms of Reported pAUC Values, 152

C-2 Analysis of Variance Table for log(pAUC), 153

Identifying the Culprit: Assessing Eyewitness Identification

Summary

yewitnesses play an important role in criminal cases when they can

identify culprits.

Yet it is well known that eyewitnesses make mis-

takes and that their memories can be affected by various factors

including the very law enforcement procedures designed to test their memo-

ries. For several decades, scientists have conducted research on the factors

that affect the accuracy of eyewitness identiﬁcation procedures. Basic re-

search on the processes that underlie human visual perception and memory

have given us an increasingly clear picture of how eyewitness identiﬁcations

are made and, more important, an improved understanding of the prin-

cipled limits on vision and memory that may lead to failures of identiﬁca-

tion. Basic research has been complemented by a growing body of applied

research on eyewitness identiﬁcation, which has examined those variables

that particularly affect eyewitnesses to crimes: system variables (conditions

such as the procedures followed to obtain identiﬁcations that can be con-

trolled by law enforcement) and estimator variables (conditions associated

with the actual crime, such as viewing conditions, or factors speciﬁc to the

eyewitness, such as the race of the victim relative to that of the perpetrator,

that cannot be controlled by law enforcement).

Through such scientiﬁc research, we have learned that many factors in-

ﬂuence the visual perceptual experience: dim illumination and brief viewing

times, large viewing distances, duress, elevated emotions, and the presence

of a visually distracting element such as a gun or a knife. Gaps in sensory

Throughout this report, the term identiﬁcation denotes person recognition. Eyewitness

identiﬁcation refers to recognition by a witness to a crime of a culprit unknown to the witness.

Identifying the Culprit: Assessing Eyewitness Identification

2 IDENTIFYING THE CULPRIT

input are ﬁlled by expectations that are based on prior experiences with the

world. Prior experiences are capable of biasing the visual perceptual experi-

ence and reinforcing an individual’s conception of what was seen. We also

have learned that these qualiﬁed perceptual experiences are stored by a sys-

tem of memory that is highly malleable and continuously evolving, neither

retaining nor divulging content in an informational vacuum. The ﬁdelity of

our memories to actual events may be compromised by many factors at all

stages of processing, from encoding to storage to retrieval. Unknown to the

individual, memories are forgotten, reconstructed, updated, and distorted.

Therefore, caution must be exercised when utilizing eyewitness procedures

and when relying on eyewitness identiﬁcations in a judicial context.

In 2013, the Laura and John Arnold Foundation called on the National

Academy of Sciences (NAS) to appoint an ad hoc study committee to:

1. critically assess the existing body of scientiﬁc research as it relates

to eyewitness identiﬁcation;

2. identify any gaps in the existing body of literature and suggest

appropriate research questions to pursue that will further our un-

derstanding of eyewitness identiﬁcation and that might offer ad-

ditional insight into law enforcement and courtroom practice;

3. provide an assessment of what can be learned from research ﬁelds

outside of eyewitness identiﬁcation;

4. offer recommendations for best practices in the handling of eyewit-

ness identiﬁcations by law enforcement;

5. offer recommendations for developing jury instructions;

6. offer advice regarding the scope of a Phase II consideration of neu-

roscience research as well as any other areas of research that might

have a bearing on eyewitness identiﬁcation; and

7. write a consensus report with appropriate ﬁndings and

recommendations.

The committee heard from numerous experts, practitioners, and stake-

holders and reviewed relevant published and unpublished literature as well

as submissions provided to the committee. In this report, the committee

offers its ﬁndings and recommendations for:

• identifying and facilitating best practices in eyewitness procedures

for the law enforcement community;

• strengthening the value of eyewitness identiﬁcation evidence in

court; and

• improving the scientiﬁc foundation underpinning eyewitness

identiﬁcation.

Identifying the Culprit: Assessing Eyewitness Identification

SUMMARY 3

OVERARCHING FINDINGS

The committee is conﬁdent that the law enforcement community, while

operating under considerable pressure and resource constraints, is working

to improve the accuracy of eyewitness identiﬁcations. These efforts, how-

ever, have not been uniform and often fall short as a result of insufﬁcient

training, the absence of standard operating procedures, and the continuing

presence of actions and statements at the crime scene and elsewhere that

may intentionally or unintentionally inﬂuence eyewitness’ identiﬁcations.

Basic scientiﬁc research on human visual perception and memory has

provided an increasingly sophisticated understanding of how these systems

work and how they place principled limits on the accuracy of eyewitness

identiﬁcation.

Basic research alone is insufﬁcient for understanding condi-

tions in the ﬁeld, and thus has been augmented by studies applied to the

speciﬁc practical problem of eyewitness identiﬁcation. Applied research has

identiﬁed key variables that affect the accuracy and reliability of eyewitness

identiﬁcations and has been instrumental in informing law enforcement, the

bar, and the judiciary of the frailties of eyewitness identiﬁcation testimony.

A range of best practices has been validated by scientiﬁc methods and

research and represents a starting place for efforts to improve eyewitness

identiﬁcation procedures. A number of law enforcement agencies have, in

fact, adopted research-based best practices. This report makes actionable

recommendations on, for example, the importance of adopting “blinded”

eyewitness identiﬁcation procedures. It further recommends that standard-

ized and easily understood instructions be provided to eyewitnesses and

calls for the careful documentation of eyewitness’ conﬁdence statements.

Such improvements may be broadly implemented by law enforcement now.

It is important to recognize, however, that, in certain cases, the state of sci-

entiﬁc research on eyewitness identiﬁcation is unsettled. For example, the

relative superiority of competing identiﬁcation procedures (i.e., simultane-

ous versus sequential lineups) is unresolved.

The ﬁeld would beneﬁt from collaborative research among scientists

and law enforcement personnel in the identiﬁcation and validation of new

best practices that can improve eyewitness identiﬁcation procedures. Such

a foundation can be solidiﬁed through the use of more effective research

designs (e.g., those that consider more than one variable at a time, and in

Basic research on vision and memory seeks a comprehensive understanding of how these

systems are organized and how they operate generally. The understanding derived from

basic research includes principles that enable one to predict how a system (such as vision or

memory) might behave under speciﬁc conditions (such as those associated with witnessing a

crime) and to identify the conditions under which it will operate most effectively and those

under which it will fail. Applied research, by contrast, empirically evaluates speciﬁc hypotheses

about how a system will behave under a particular set of real-world conditions.

Identifying the Culprit: Assessing Eyewitness Identification

4 IDENTIFYING THE CULPRIT

different study populations to ensure reproducibility and generalizability),

more informative statistical measures and analyses (i.e., methods from

statistical machine learning and signal detection theory to evaluate the per-

formance of binary classiﬁcation tasks), more probing analyses of research

ﬁndings (such as analyses of consequences of data uncertainties), and more

sophisticated systematic reviews and meta-analyses (that take account of

current guidelines, including transparency and reproducibility of methods).

In view of the complexity of the effects of both system and estimator

variables and their interactions on eyewitness identiﬁcation accuracy, bet-

ter experimental designs that incorporate selected combinations of these

variables (e.g., presence or absence of a weapon, lighting conditions, etc.)

will elucidate those variables with meaningful inﬂuence on eyewitness

performance, which can, in turn, inform law enforcement practice of eye-

witness identiﬁcation procedures. To date, the eyewitness literature has

evaluated procedures mostly in terms of a single diagnosticity ratio or

an ROC (Receiver Operating Characteristic) curve; even if uncertainty is

incorporated into the analysis, many other powerful tools for evaluating

a “binary classiﬁer” are available and worthy of consideration.

Finally,

syntheses of eyewitness research has been limited to meta-analyses that have

not been conducted in the context of systematic reviews. Systematic reviews

of stronger research studies need to conform to current standards and be

translated into terms that are useful for decision makers.

The committee here offers a summary of its key recommendations to

strengthen the effectiveness of policies and procedures used to obtain ac-

curate eyewitness identiﬁcations.

RECOMMENDATIONS TO ESTABLISH BEST PRACTICES

FOR THE LAW ENFORCEMENT COMMUNITY

The committee’s review of law enforcement practices and procedures,

coupled with its consideration of the scientiﬁc literature, has identiﬁed

a number of areas where eyewitness identiﬁcation procedures could be

strengthened. The practices and procedures considered here involve acquisi-

tion of data that reﬂect a witness’ identiﬁcation and the contextual factors

that bear on that identiﬁcation. A recurrent theme underlying the commit-

tee’s recommendations is development of and adherence to guidelines that

are consistent with scientiﬁc standards for data collection and reporting.

T. Hastie, R. Tibshirani, and J. H. Friedman, The Elements of Statistical Learning: Data

Mining, Inference, and Prediction (New York: Springer, 2009).

Identifying the Culprit: Assessing Eyewitness Identification

SUMMARY 5

Recommendation #1: Train All Law Enforcement Ofﬁcers in Eyewitness

Identiﬁcation

The committee recommends that all law enforcement agencies provide

their ofﬁcers and agents with training on vision and memory and the vari-

ables that affect them, on practices for minimizing contamination, and on

effective eyewitness identiﬁcation protocols.

Recommendation #2: Implement Double-Blind Lineup and Photo Array

Procedures

The committee recommends blind (double-blind or blinded) admin-

istration of both photo arrays and live lineups and the adoption of clear,

written policies and training on photo array and live lineup administration.

Recommendation #3: Develop and Use Standardized Witness

Instructions

The committee recommends the development of a standard set of easily

understood instructions to use when engaging a witness in an identiﬁcation

procedure.

Recommendation #4: Document Witness Conﬁdence Judgments

The committee recommends that law enforcement document the wit-

ness’ level of conﬁdence verbatim at the time when she or he ﬁrst identiﬁes

a suspect.

Recommendation #5: Videotape the Witness Identiﬁcation Process

The committee recommends that the video recording of eyewitness

identiﬁcation procedures become standard practice.

RECOMMENDATIONS TO STRENGTHEN THE VALUE OF

EYEWITNESS IDENTIFICATION EVIDENCE IN COURT

The best guidance for legal regulation of eyewitness identiﬁcation evi-

dence comes not from constitutional rulings, but from the careful use and

understanding of scientiﬁc evidence to guide fact-ﬁnders and decision-

makers. The Manson v. Brathwaite test under the Due Process Clause of the

U.S. Constitution for assessing eyewitness identiﬁcation evidence was estab-

lished in 1977, before much applied research on eyewitness identiﬁcation

had been conducted. This test evaluates the “reliability” of eyewitness iden-

Identifying the Culprit: Assessing Eyewitness Identification

6 IDENTIFYING THE CULPRIT

tiﬁcations using factors derived from prior rulings and not from empirically

validated sources. As critics have pointed out, the Manson v. Brathwaite

test includes factors that are not diagnostic of reliability. Moreover, the test

treats factors such as the conﬁdence of a witness as independent markers

of reliability when, in fact, it is now well established that conﬁdence judg-

ments may vary over time and can be powerfully swayed by many factors.

While some states have made minor changes to the due process framework,

wholesale reconsideration of this framework is only a recent development.

Recommendation #6: Conduct Pretrial Judicial Inquiry

The committee recommends that, as appropriate, a judge make basic

inquiries when eyewitness identiﬁcation evidence is offered.

Recommendation #7: Make Juries Aware of Prior Identiﬁcations

The committee recommends that judges take all necessary steps to

make juries aware of prior identiﬁcations, the manner and time frame in

which they were conducted, and the conﬁdence level expressed by the eye-

witness at the time.

Recommendation #8: Use Scientiﬁc Framework Expert Testimony

The committee recommends that judges have the discretion to al-

low expert testimony on relevant precepts of eyewitness memory and

identiﬁcations.

Recommendation #9: Use Jury Instructions as an Alternative Means to

Convey Information

The committee recommends the use of clear and concise jury instruc-

tions as an alternative means of conveying information regarding the fac-

tors that the jury should consider.

RECOMMENDATIONS TO IMPROVE THE

SCIENTIFIC FOUNDATION UNDERPINNING

EYEWITNESS IDENTIFICATION RESEARCH

Basic scientiﬁc research on visual perception and memory provides

important insight into the factors that can limit the ﬁdelity of eyewitness

identiﬁcation. Research targeting the speciﬁc problem of eyewitness iden-

tiﬁcation complements basic scientiﬁc research. However, this strong sci-

entiﬁc foundation remains insufﬁcient for understanding the strengths and

Identifying the Culprit: Assessing Eyewitness Identification

SUMMARY 7

limitations of eyewitness identiﬁcation procedures in the ﬁeld. Many of the

applied studies on key factors that directly affect eyewitness performance

in the laboratory are not readily applicable to actual practice and policy.

Applied research falls short because of a lack of reliable or standardized

data from the ﬁeld, a failure to include a range of practitioners in the estab-

lishment of research agendas, the use of disparate research methodologies,

failure to use transparent and reproducible research procedures, and inad-

equate reporting of research data. The task of guiding eyewitness identiﬁ-

cation research toward the goal of evidence-based policy and practice will

require collaboration in the setting of research agendas and agreement on

methods for acquiring, handling, and sharing data.

Recommendation #10: Establish a National Research Initiative on

Eyewitness Identiﬁcation

The committee recommends the establishment of a National Research

Initiative on Eyewitness Identiﬁcation.

Recommendation #11: Conduct Additional Research on System and

Estimator Variables

The committee recommends broad use of statistical tools that can

render a discriminability measure to evaluate eyewitness performance and

a rigorous exploration of methods that can lead to more conservative

responding. The committee further recommends that caution and care be

used when considering changes to any existing lineup procedure, until such

time as there is clear evidence for the advantages of doing so.

CONCLUSION

Eyewitness identiﬁcation can be a powerful tool. As this report indi-

cates, however, the malleable nature of human visual perception, memory,

and conﬁdence; the imperfect ability to recognize individuals; and policies

governing law enforcement procedures can result in mistaken identiﬁcations

with signiﬁcant consequences. New law enforcement training protocols,

standardized procedures for administering lineups, improvements in the

handling of eyewitness identiﬁcation in court, and better data collection and

research on eyewitness identiﬁcation can improve the accuracy of eyewit-

ness identiﬁcations.

Identifying the Culprit: Assessing Eyewitness Identification

Introduction

ccurate eyewitness identiﬁcations

may aid in the apprehension

and prosecution of the perpetrators of crimes. However, inaccurate

identiﬁcations may lead to the prosecution of innocent persons

while the guilty party goes free. It is therefore crucial to develop eyewitness

identiﬁcation procedures that achieve maximum accuracy and reliability.

Eyewitness evidence is not infallible. In 1932, Yale University law pro-

fessor Edwin M. Borchard documented nearly seventy cases of miscarriage

of justice caused by eyewitness errors in his book, Convicting the Innocent.

Years later, in 1967, the U.S. Supreme Court highlighted the danger of er-

roneous eyewitness identiﬁcation in United States v. Wade, stating, “The

vagaries of eyewitness identiﬁcation are well-known; the annals of criminal

law are rife with instances of mistaken identiﬁcation.”

The Federal Bureau of Investigation (FBI) estimates that U.S. law en-

forcement made 12,196,959 arrests in 2012. The FBI estimates that 521,196

of these arrests were for violent crimes.

Accurate data on the number of

crimes observed by eyewitnesses are not available. If only a fraction of the

violent crimes in the United States involve an eyewitness, the number must

Throughout this report, the term identiﬁcation denotes person recognition. Eyewitness

identiﬁcation refers to recognition by a witness to a crime of a culprit unknown to the witness.

Edwin M. Borchard, Convicting the Innocent: Sixty-Five Actual Errors of Criminal Justice

(New York: Garden City Publishing Company, Inc., 1932).

United States v. Wade, 388 U.S. 230, 288 (1967).

Federal Bureau of Investigation, “Crime in the United States 2012: Persons Arrested,”

available at: http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2012/crime-in-the-u.s.-2012/

persons-arrested/persons-arrested.

Identifying the Culprit: Assessing Eyewitness Identification

10 IDENTIFYING THE CULPRIT

BOX 1-1

The Ronald Cotton Case

In 1984, a college student named Jennifer Thompson was raped in her apart-

ment in Burlington, North Carolina. The police asked her to help create a com-

posite sketch of the rapist. The police then received a tip that a local man named

Ronald Cotton resembled the composite, and shortly after the crime, Thompson

was shown a photo array containing six photos. With some difﬁculty, she chose

two pictures, one of which was of Cotton. Finally, she said, “I think this is the guy,”

pointing to Cotton. “You’re sure,” the lead detective asked, and she responded,

“Positive.” Thompson asked, “Did I do OK?” The detectives responded, “You did

great.” She has described how those encouraging remarks had the effect of mak-

ing her more conﬁdent in her identiﬁcation.

The police then showed Thompson a live lineup. Cotton was the only person

repeated from the prior photo array. This would make Cotton more familiar and

might suggest that he was the prime suspect. Nevertheless, Thompson remained

hesitant and was having trouble deciding between two people. After several

minutes, she told the police that Cotton “looks the most like him.” The lead detec-

tive asked “if she was certain,” and she said, “Yes.” Again, the detectives further

reinforced her decision. The lead detective told Thompson, “It’s the same person

you picked from the photos.” She later described feeling a “huge amount of relief”

when told that she had again picked the right person.

At Ronald Cotton’s criminal trial, Thompson agreed she was “absolutely

sure” that he was the rapist. Cotton was sentenced to life in prison plus 54 years.

He served 10.5 years before DNA tests exonerated him and implicated another

man, Bobby Poole. Not only did the identiﬁcation procedures increase Thompson's

conﬁdence in the mistaken memory event, but they also resulted in her rejection

of the actual culprit. Poole had been presented to Thompson at a post-trial hear-

ing, and she could not recognize him. “I have never seen him in my life,” she said

at the time.

In response to this error, the lead detective in the case, Mike Gauldin, later

as police chief, was the ﬁrst in the state to institute a series of new practices, in-

cluding double-blind lineup procedures. In the years that followed, North Carolina

adopted such practices statewide. Ronald Cotton and Jennifer Thompson have

since written a book, Picking Cotton, that describes their case and experiences.

See, generally, http://www.cbsnews.com/news/eyewitness-how-accurate-is-visual-memory/

and http://www.slate.com/articles/news_and_politics/jurisprudence/features/2011/getting_it_

wrong_convicting_the_innocent/how_eyewitnesses_can_send_innocents_to_jail.html.

Identifying the Culprit: Assessing Eyewitness Identification

INTRODUCTION 11

be sizable. One estimate based on a 1989 survey of prosecutors suggests

that at least 80,000 eyewitnesses make identiﬁcations of suspects in crimi-

nal investigations each year.

Recently, post-conviction DNA exonerations of innocent persons have

dramatically highlighted the problems with eyewitness identiﬁcations.

6,7

In the United States, more than 300 exonerations have resulted from post-

conviction DNA testing since 1989.

According to the Innocence Project,

at least one mistaken eyewitness identiﬁcation was present in almost three-

quarters of DNA exonerations.

In many of these cases, eyewitness identi-

ﬁcation played a signiﬁcant evidentiary role, and almost without exception,

the eyewitnesses who testiﬁed expressed complete conﬁdence that they had

chosen the perpetrator. Many eyewitnesses testiﬁed with high conﬁdence

despite earlier expressions of uncertainty.

For example, in the well-known

case of Ronald Cotton (see Box 1-1), Jennifer Thompson (the victim) has

described how she was initially quite unsure of her eyewitness identiﬁcation

of Cotton, a man later exonerated by DNA testing. She became certain it

was Cotton only after the police made conﬁrmatory remarks and had her

participate in two identiﬁcation procedures where Cotton was the only

person shown both times.

Erroneous eyewitness identiﬁcations can occur across the range of

criminal convictions in which eyewitness evidence is presented, but most

of these cases lack the biological material that can be tested for DNA and

used for exoneration purposes. While eyewitness misidentiﬁcations may

have been a dominant factor in some erroneous convictions, it is important

to note that other factors, including errors at various stages of the legal

and judicial processes, may have contributed to the erroneous convictions.

CHARGE TO THE COMMITTEE

In 2013, the Laura and John Arnold Foundation called on the Na-

tional Research Council (NRC) to assess the state of scientiﬁc research on

A. G. Goldstein, J. E. Chance, and G. R. Schneller, “Frequency of Eyewitness Identiﬁcation

in Criminal Cases: A Survey of Prosecutors,” Bulletin of the Psychonomic Society 27(1): 71,

73 (January 1989).

CNN, “Exonerated: Cases by the Numbers,” December 4, 2013, available at: http://www.

cnn.com/2013/12/04/justice/prisoner-exonerations-facts-innocence-project/.

Taryn Simon, “Freedom Row,” New York Times Magazine, January 26, 2003.

The Innocence Project, “DNA Exoneree Case Proﬁles,” available at: http://www.innocence

project.org/know/.

The Innocence Project, “Eyewitness Identiﬁcation,” available at: http://www.innocence

project.org/ﬁx/Eyewitness-Identiﬁcation.php.

Brandon L. Garrett, Convicting the Innocent: Where Criminal Prosecutions Go Wrong

63–68 (Cambridge, MA: Harvard University Press, 2011).

Identifying the Culprit: Assessing Eyewitness Identification

12 IDENTIFYING THE CULPRIT

eyewitness identiﬁcation and to recommend best practices

for handling

eyewitness identiﬁcations by law enforcement and the courts. The goal of

this effort was to evaluate the scientiﬁc basis for eyewitness identiﬁcation,

to help establish the scientiﬁc foundation for effective real-world practices,

and to facilitate the development of policies to improve eyewitness identiﬁ-

cation validity in the context of the American justice system.

In response to this charge, the NRC appointed an ad hoc committee,

the Committee on Scientiﬁc Approaches to Understanding and Maximiz-

ing the Validity and Reliability of Eyewitness Identiﬁcation in Law En-

forcement and the Courts (hereinafter, the committee), to undertake this

study (see Box 1-2 for the committee’s charge). The committee met three

times, held numerous conference calls, heard from various stakeholders (see

Appendix B), and reviewed extensive research on eyewitness identiﬁcation

before reaching its ﬁndings and recommendations.

For the purposes of this report, the committee characterizes best practice as the adoption

of standardized procedures based on scientiﬁc principles. The committee does not make any

endorsement of practices designated as best practices by other bodies.

BOX 1-2

Charge to the Committee

The charge to the NRC was to:

1. critically assess the existing body of scientiﬁc research as it relates to eyewit-

ness identiﬁcation;

2. identify any gaps in the existing body of literature and suggest, as appropriate,

research questions to pursue that will further our understanding of eyewitness

identiﬁcation and that might offer additional insight into law enforcement and

courtroom practice;

3. provide an assessment of what can be learned from research ﬁelds outside of

eyewitness identiﬁcation;

4. offer recommendations for best practices in the handling of eyewitness identi-

ﬁcations by law enforcement;

5. offer recommendations for developing jury instructions;

6. offer advice regarding the scope of a Phase II consideration of neuroscience

research as well as any other areas of research that might have a bearing on

eyewitness identiﬁcation; and

7. write a consensus report with appropriate ﬁndings and recommendations.

Identifying the Culprit: Assessing Eyewitness Identification

INTRODUCTION 13

SCIENCE AND LAW

Law enforcement ofﬁcers investigating crimes rely on eyewitness iden-

tiﬁcation procedures to verify that a suspect is the individual seen by an

eyewitness.

Such procedures can take place under conditions that may

have signiﬁcant effects on the accuracy and reliability of an eyewitness’

identiﬁcation. Unlike ofﬁcers in the ﬁeld, laboratory researchers have, in

theory, greater control over inﬂuences that might contaminate the visual

perceptual experience and memory of an eyewitness.

Science is a self-correcting enterprise. Researchers formulate and test

hypotheses using observations and experiments, which are then subject to

independent review. In science, evidence and data are analyzed and experi-

ments are repeated to ensure that biases or other factors do not lead to in-

correct conclusions. Scientiﬁc progress results from the review and revision

of earlier results and conclusions.

The culture of scientiﬁc research is markedly different from a legal cul-

ture that must seek deﬁnitive results in individual cases. In 1993, in Daubert

v. Merrell Dow Pharmaceuticals, Inc., the U.S. Supreme Court ruled that,

under Rule 702 of the Federal Rules of Evidence (which covers both civil

and criminal trials in the federal courts), a “trial judge must ensure that

any and all scientiﬁc testimony or evidence admitted is not only relevant,

but reliable.”

Criminal justice and legal personnel have come to rely on eyewitness

evidence. Law enforcement ofﬁcials have ﬁrst-hand experience with eye-

witnesses in criminal investigations and trials, and over the years, some

juridictions have implemented and strengthened practices and procedures

in an attempt to improve acccuracy. Consequently, the law enforcement and

legal communities have made important contributions to our understand-

ing of eyewitness identiﬁcations and the improvements of practices in the

ﬁeld. Researchers have become increasingly involved in assessing eyewit-

ness identiﬁcation procedures as law enforcement, lawyers, and judges

have themselves sought more accurate procedures and approaches. In the

2009 National Research Council report, Strengthening Forensic Science in

the United States: A Path Forward, the committee noted, “in addition to

protecting innocent persons from being convicted of crimes that they did

not commit, we are also seeking to protect society from persons who have

For ease of reading, throughout the report the committee will use the term ofﬁcer to mean

law enforcement ofﬁcials and professionals.

Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993). The Court also noted

that “there are important differences between the quest for truth in the courtroom and the

quest for truth in the laboratory. Scientiﬁc conclusions are subject to perpetual revision. Law,

on the other hand, must resolve disputes ﬁnally and quickly.”

Identifying the Culprit: Assessing Eyewitness Identification

14 IDENTIFYING THE CULPRIT

committed criminal acts.”

This shared common goal of protecting in-

nocent persons and society makes collaboration between the scientiﬁc, law

enforcement, and legal communities critically important.

IDENTIFYING THE CULPRIT

Ofﬁcers typically use three procedures to identify a perpetrator whose

identity is unknown: (1) showups; (2) presentations of photo arrays; and

(3) live lineups. A showup is a procedure in which ofﬁcers present a single

criminal suspect to a witness. This procedure usually occurs near the crime

location and immediately or shortly after the crime has occurred. Ofﬁcers

also use photo arrays and live lineups, in which they ask the witness to view

numerous individuals, one of whom may be the perpetrator. The suspect is

presented along with ﬁllers (known non-suspects). Currently, photo arrays

are used more often than live lineups.

15,16

If the eyewitness makes a positive identiﬁcation during a showup, a

photo array, or a lineup, the identiﬁcation may constitute evidence about a

suspect’s involvement in a crime. The eyewitness identiﬁcation may, when

considered with other available evidence, establish probable cause to sup-

port an arrest. Such evidence may play a pivotal role in enabling the pros-

ecution to meet its burden of proof in a subsequent trial.

In recent years, more law enforcement agencies have created written

eyewitness identiﬁcation policies and have adopted formalized training.

However, there are many agencies that do not have standard written poli-

cies or formalized training for the administration of identiﬁcation proce-

dures or for ongoing interactions with witnesses.

VISION AND MEMORY

At its core, eyewitness identiﬁcation relies on brain systems for visual

perception and memory: The witness perceives the face and other aspects

of the perpetrator’s physical appearance and bearing, stores that informa-

National Research Council, Strengthening Forensic Science in the United States: A Path

Forward (Washington, DC: The National Academies Press, 2009), p. 12.

Police Executive Research Forum, “A National Survey of Eyewitness Identiﬁcation Pro-

cedures in Law Enforcement Agencies,” March 2013, p. 48. The survey indicates that 94.1

percent of responding law enforcement agencies reported that they use photo arrays, while

only 21.4 percent reported using live lineups. Sixty-one point eight percent of agencies re-

ported that they use showups. See also J. S. Neuschatz et al., “Comprehensive Evaluation of

Showups,” in Advances in Psychology and Law, ed. M. Miller and B. Bornstein (New York:

Springer, in press).

Throughout the report, unless otherwise speciﬁed, references to lineups refer to both photo

arrays and live lineups.

Police Executive Research Forum, p. 65.

Identifying the Culprit: Assessing Eyewitness Identification

INTRODUCTION 15

tion in memory, and later retrieves the information for comparison with the

visual percept of an individual in a lineup. Recent years have seen great ad-

vances in our scientiﬁc understanding of the basic mechanisms, operational

strategies, and limitations of human vision and memory. These advances

inform our understanding of the accuracy of eyewitness identiﬁcation.

Human vision does not capture a perfect, error-free “trace” of a wit-

nessed event. What an individual actually perceives can be heavily inﬂu-

enced by bias

and expectations derived from cultural factors, behavioral

goals, emotions, and prior experiences with the world. For eyewitness iden-

tiﬁcation to take place, perceived information must be encoded in memory,

stored, and subsequently retrieved. As time passes, memories become less

stable. In addition, suggestion and the exposure to new information may

inﬂuence and distort what the individual believes she or he has seen.

Several factors are known to affect the ﬁdelity of visual perception and

the integrity of memory. In particular, vision and memory are constrained

by processing bottlenecks and various sources of noise.

Noise comes

from a variety of sources, some associated with the structure of the visual

environment, some inherent in the optical and neuronal processes involved,

some reﬂecting sensory content not relevant to the observer’s goals, and

some originating with incorrect expectations derived from memory. The

concept of noise has profound signiﬁcance for understanding eyewitness

identiﬁcation, as the accuracy of information about the environment gained

through vision and stored in memory is necessarily, and often sharply, lim-

ited by noise.

The recognition of one person by another—a seemingly commonplace

and unremarkable everyday occurrence—involves complex processes that

are limited by noise and subject to many extraneous inﬂuences. Eyewitness

identiﬁcation research confronts methodological challenges that some other

basic experimental sciences do not encounter, as well as practical challenges

Bias is deﬁned as any tendency that prevents unprejudiced consideration of a question

(see Dictionary.com; http://dictionary.reference.com/browse/bias). Response bias is a general

term for a wide range of inﬂuences that moderate the responses of participants away from

an accurate or truthful response. Response bias can be induced or caused by a number of

factors, all relating to the idea that humans do not respond passively to stimuli, but rather

actively integrate multiple sources of information to generate a response in a given situation

[(see M. Orne,“On the Social Psychology of the Psychological Experiment: With Particular

Reference to Demand Characteristics and Their Implications,” American Psychologist 17:

776–783, (1962)]. In research, bias is seen in sampling or testing when circumstances select

or encourage one outcome or answer over another (see Merriam-Webster.com; http://www.

merriam-webster.com/dictionary/bias).

Noise refers here to factors that cause uncertainty on the part of an individual about

whether a particular signal (e.g. a speciﬁc visual stimulus) is present. This use of the term fol-

lows the deﬁnition used in electronic signal transmission, in which noise refers to random or

irrelevant elements that interfere with detection of coherent and informative signals.

Identifying the Culprit: Assessing Eyewitness Identification

16 IDENTIFYING THE CULPRIT

in establishing adequate experimental controls over the numerous variables

that affect visual perception and memory.

APPLIED RESEARCH ON EYEWITNESS IDENTIFICATION:

SYSTEM AND ESTIMATOR VARIABLES

Our understanding of the underlying processes and limits of eyewit-

ness identiﬁcation, derived from basic research on vision and memory, is

complemented by research directed speciﬁcally at the problem of eyewit-

ness identiﬁcation. The modern era of eyewitness identiﬁcation research

began in the 1970s. Today, eyewitness identiﬁcation is generally viewed as

a behavioral output. The accuracy and reliability of eyewitness identiﬁca-

tion are critically modulated by variables that include a witness’ extant

cognition and memory and related psychological and situational factors at

the time of the event, over the ensuing intervals, and at all stages of recall

(see Figure 1-1). Because a crime is an unexpected event, one can draw a

natural distinction between variables that reﬂect the witness’ unplanned

situational or cognitive state at the time of the crime and the variables that

reﬂect controllable conditions and internal states following the witnessed

events. Researchers categorize these factors, respectively, as estimator vari-

ables and system variables.

System variables describe the characteristics of speciﬁc procedures and

practices (e.g., the content and nature of instructions given to witnesses

who are asked if they are able to make an identiﬁcation). The criminal

justice system can exert some control over system variables by follow-

ing standardized procedures that are based on scientiﬁc knowledge and

strengthened through education and training.

One important category of system variables concerns the conditions

and protocols for lineup identiﬁcation. Under current law enforcement

practice, eyewitness identiﬁcation procedures involve having a witness view

individuals or images of individuals. Research indicates that accuracy and

reliability of eyewitness identiﬁcations may be inﬂuenced by the type of

presentation (e.g., lineup) used, the likeness of non-suspect lineup partici-

pants (ﬁllers) to the suspect, the number of ﬁllers, and the suspect’s physical

location in the presentation.

21,22

Eyewitness performance may be affected

by how the lineup images are presented—simultaneously (as a group) or

G. L. Wells, “Applied Eyewitness-Testimony Research: System Variables and Estimator

Variables,” Journal of Personality and Social Psychology 36(12):1546–1557 (1978).

N. K. Steblay et al., “Eyewitness Accuracy Rates in Police Showup and Lineup Presenta-

tions: A Meta-Analytic Comparison,” Law and Human Behavior 27(5): 523–540 (October

2003).

R. J. Fitzgerald et al., “The Effect of Suspect-Filler Similarity on Eyewitness Identiﬁcation

Decisions: A Meta-analysis,” Psychology, Public Policy, and Law 19(2): 151–164 (May 2013).

Identifying the Culprit: Assessing Eyewitness Identification

INTRODUCTION 17

sequentially (one at a time). System variables, such as the nature of the

instructions and feedback provided before and after the identiﬁcation pro-

cedure, may also affect the eyewitness’ identiﬁcation.

Estimator variables affect the accuracy of eyewitness identiﬁcation,

but they are beyond the control of the criminal justice system. Estimator

variables tend to be associated with characteristics of the witness or factors

that are operating either at the time of the criminal event (perhaps relating

to memory encoding) or the retention interval (the time between witness-

ing an event and the identiﬁcation process). Speciﬁc examples include the

eyewitness’ level of stress or trauma at the time of the incident, the light

level and nature of the visual conditions that affect visibility and the clarity

of a perpetrator’s features, and the physical distance between the witness

and the perpetrator. Both system and estimator variables will be discussed

in detail in subsequent chapters.

EFFORTS AT IMPROVEMENT

In response to insights gained from research on erroneous convictions,

there have been attempts to provide recommendations for improving the

reliability and validity of eyewitness identiﬁcations. An effort of particular

note is the National Institute of Justice’s (NIJ) Technical Working Group for

Eyewitness Evidence (TWGEYEE). Called together by then-U.S. Attorney

General Janet Reno in 1998, members of the working group were asked

to develop and publish guidance for improving eyewitness identiﬁcation

FIGURE 1-1 Memory accuracy and time.

SOURCE: Courtesy of Thomas D. Albright.

Identifying the Culprit: Assessing Eyewitness Identification

18 IDENTIFYING THE CULPRIT

procedures.

The working group recognized the role that memory plays in

the mistaken interpretation and remembrance of events and offered guid-

ance based on the practical experiences of the law enforcement community

and insights gained from behavioral and psychological research. The NIJ

provided detailed instructions for each step of the eyewitness identiﬁcation

procedure to the approximately 18,000 state and local law enforcement

agencies across the nation. After the report was issued, only a few states

conducted evaluations and engaged in improvement efforts, including the

implementation of new laws and the issuance of corrective guidelines and

policies. Consequently, eyewitness identiﬁcation policies remain fragmented

by jurisdiction, except in a minority of states that have adopted state-wide

policies. At present, the United States does not have a uniform national set

of protocols.

JUDICIAL CONSIDERATION OF EYEWITNESS

IDENTIFICATION EVIDENCE

The U.S. Supreme Court’s 1977 ruling in Manson v. Brathwaite pro-

vides the current framework for judicial review of eyewitness identiﬁcation

under the Due Process Clause of the U.S. Constitution.

The Manson v.

Brathwaite test asks judges to evaluate the “reliability” of eyewitness iden-

tiﬁcations using factors derived from prior rulings and not from empirically

validated sources. The Manson v. Brathwaite ruling was not based on much

of the research conducted by scientists on visual perception, memory, and

eyewitness identiﬁcation, and it fails to include important advances that

have strengthened standards for judicial review of eyewitness identiﬁcation

evidence at the state level.

In 2011, the Justices of the Massachusetts Supreme Judicial Court con-

vened the Study Group on Eyewitness Identiﬁcation to “offer guidance as

to how our courts can most effectively deter unnecessarily suggestive iden-

tiﬁcation procedures and minimize the risk of a wrongful conviction.” The

report made ﬁve recommendations to minimize inaccurate identiﬁcations:

(1) acknowledge variables affecting identiﬁcation accuracy; (2) develop a

model policy and implement best practices for police departments; (3) ex-

pand use of pretrial hearings; (4) expand use of improved jury instructions;

and (5) offer continuing education.

U.S. Department of Justice, Ofﬁce of Justice Programs, Eyewitness Evidence: A Guide for

Law Enforcement (Washington, DC, 1999).

Police Executive Research Forum, p. 65.

Manson v. Brathwaite, 432 U.S. 98, 114 (1977).

Massachusetts Supreme Judicial Court Study Group on Eyewitness Identiﬁcation, Report

and Recommendations to the Justices, July 24, 2013, available at: http://www.mass.gov/courts/

docs/sjc/docs/eyewitness-evidence-report-2013.pdf.

Identifying the Culprit: Assessing Eyewitness Identification

INTRODUCTION 19

In 2011, the New Jersey Supreme Court issued a unanimous decision

in State v. Larry R. Henderson. The opinion revised the legal framework

for evaluating and admitting eyewitness identiﬁcation evidence and directed

that improved jury instructions be prepared to help jurors evaluate such

evidence. Henderson drew on an extensive review of scientiﬁc evidence

regarding human vision, memory, and the various factors that can affect

the reliability of eyewitness identiﬁcations. In July 2012, the court released

expanded jury instructions and revised court rules relating to eyewitness

identiﬁcations in criminal cases.

In fall 2012, the Oregon Supreme Court also established a new pro-

cedure for evaluating whether eyewitness identiﬁcations could be used in

court. In State v. Lawson, the Court reviewed eyewitness identiﬁcation

research conducted over the past 30 years, determined that the Manson

v. Brathwaite test “does not accomplish its goal of ensuring that only suf-

ﬁciently reliable identiﬁcations are admitted into evidence,” and offered

a revised procedure that requires the court to make a determination of

whether investigators used “suggestive” tactics to get an identiﬁcation and

the extent to which other information supports the identiﬁcation.

Despite these improvements and judicial decisions, policies and prac-

tices across the country remain inconsistent.

ORGANIZATION OF THE REPORT

This report begins with a description of law enforcement protocols for

eyewitness identiﬁcation (Chapter 2). Chapter 3 presents the legal frame-

work for eyewitness identiﬁcation evidence. A discussion of the current

scientiﬁc understanding of visual perception and memory follows in Chap-

ter 4. In Chapter 5, the committee provides an assessment of eyewitness

identiﬁcation research. The report concludes with the committee’s ﬁndings

and recommendations (Chapter 6).

New Jersey Judiciary, “Supreme Court Releases Eyewitness Identiﬁcation Criteria for

Criminal Cases,” July 19, 2012, available at: http://www.judiciary.state.nj.us/pressrel/2012/

pr120719a.htm.

State v. Lawson, 352 Or. 724 (Or. 2012).

Identifying the Culprit: Assessing Eyewitness Identification

Eyewitness Identiﬁcation Procedures

olice in the United States investigate millions of crimes each year.

Only a small percentage of the police-investigated crimes involve

the use of police-arranged identiﬁcation procedures. Identiﬁcation

procedures are unnecessary when, for example, the perpetrator is caught

during the commission of the criminal act, as in the crime of driving while

intoxicated, or when the victim knows the perpetrator, as in crimes of do-

mestic violence.

Police use identiﬁcation procedures for numerous reasons. In some

circumstances, the police identify a suspect during an investigation and use

the identiﬁcation procedure to test a witness’ ability to identify the suspect

as the perpetrator. In other instances, the identiﬁcation procedure is used

as an investigative tool to further an investigation. A positive identiﬁca-

tion might form probable cause for a search warrant or the apprehension

and subsequent questioning of a suspect, or both. Most signiﬁcant for the

purposes of this report are the circumstances in which a witness positively

identiﬁes the police suspect as the perpetrator, and the identiﬁcation serves

as compelling evidence in the prosecution of a case.

Data on the number of eyewitness identiﬁcation procedures are not

systematically or uniformly collected. While the exact number of eyewitness

Federal Bureau of Investigation, “Crime in the United States 2012: Persons Arrested,”

available at: http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2012/crime-in-the-u.s.-2012/

persons-arrested/persons-arrested.

Throughout Chapter 2, the terms law enforcement and police are used interchangeably and

refer to all law enforcement agencies at the local, state, and federal levels.

Identifying the Culprit: Assessing Eyewitness Identification

22 IDENTIFYING THE CULPRIT

identiﬁcation procedures related to crimes involving strangers is unknown,

mistaken identiﬁcations have disastrous effects for those wrongly accused

of crimes and for society should a guilty person go free. Mistaken identiﬁ-

cations may also erode public conﬁdence in the criminal justice system as

a whole.

Recently, some police departments and prosecutors have imple-

mented stringent eyewitness identiﬁcation procedures in an effort to reduce

erroneous convictions.

Police-arranged eyewitness identiﬁcation procedures vary greatly de-

pending on the nature of the case. In some cases, a police-arranged identiﬁ-

cation is conducted at the very early stages of an investigation. For instance,

consider the circumstance in which police respond to a bank robbery in

progress. The perpetrator is described as a white male, approximately

6 feet, 2 inches in height wearing an orange shirt. As the police arrive at

the crime scene, an ofﬁcer observes and apprehends a man ﬂeeing the bank

wearing an orange shirt and exhibiting similar physical characteristics. In

this situation, a police-arranged identiﬁcation procedure may be conducted

on the scene and prior to any signiﬁcant investigation. At the other extreme

are, for example, lengthy homicide or rape cases that include extensive

investigations, forensic testing, and eyewitness interviews conducted over a

protracted period of time. Such efforts may culminate in the identiﬁcation

of a suspect and the suspect’s inclusion in a photo array identiﬁcation pro-

cedure. In such a circumstance, an eyewitness may not be asked to identify

a perpetrator until months after the commission of the crime—and often

after repeated probes of her or his memory by, for example, police, family

members, and others.

Identiﬁcation procedures may be used in different ways for different

purposes. They are not always used to identify an unknown perpetrator of

a crime. The police may, for example, use photo arrays and conﬁrmatory

single photographs to clarify the legal identity (birth name/government

name) of an individual who is well known to a witness, but only by a street

name. In such examples, a witness may know (and may have known) the

perpetrator for years but may only be able to identify him by a common

See, generally, The International Association of Chiefs of Police, “National Summit on

Wrongful Convictions: Building a Systemic Approach to Prevent Wrongful Convictions,”

August 2013.

See The Innocence Project, Eyewitness Identiﬁcation, available at: http://www.innocence

project.org/ﬁx/Eyewitness-Identiﬁcation.php; U.S. Department of Justice, Ofﬁce of Justice

Programs, Eyewitness Evidence: A Guide for Law Enforcement (Washington, DC, 1999); Met-

ropolitan Police—District of Columbia, General Order—Procedures for Obtaining Pretrial

Eyewitness Identiﬁcation, April 18, 2013; New York State District Attorneys Association Best

Practice Committee, New York State Photo Identiﬁcation Guidelines, October 2010; Rhode

Island Police Chiefs Association, Lineup and Showup Procedures (Eyewitness Identiﬁcation),

November 2011; and Innocence Project of Texas, Eyewitness Identiﬁcation Reform, available

at: http://www.ipoftexas.org/eyewitness-id.

Identifying the Culprit: Assessing Eyewitness Identification

EYEWITNESS IDENTIFICATION PROCEDURES 23

street name, such as “Prince.” The police typically will use an identiﬁcation

procedure to identify the “Prince” to which the witness is referring before

they make an arrest or take other investigative measures such as the execu-

tion of a search warrant.

This chapter reviews the eyewitness identiﬁcation procedures com-

monly used by the police and concludes with a brief discussion of situa-

tions in which citizens engage in identifying perpetrators without police

assistance.

PHOTOGRAPHIC ARRAY

The photo array is the most common police-arranged identiﬁcation

procedure used in the United States.

A photo array consists of six to nine

photographs displayed to a witness. An ofﬁcer might create an array by

selecting photographs of persons deemed to resemble the perpetrator.

Ofﬁcers might then display the photographs one at a time to the witness

and ask whether she or he recognizes each one. This method is known as

a sequential procedure. Ofﬁcers might also create photo arrays by cutting

six square holes in a folder and taping the photographs to the back of the

folder so that the faces of the ﬁllers (non-suspects) and suspect are displayed

together. When such photographs are presented simultaneously as a two

by three matrix, this type of array is referred to as a “six pack.” When, as

in this instance, photographs are displayed together, this is referred to as a

simultaneous procedure.

In 1999, Attorney General Janet Reno released the U.S. Department

of Justice, Eyewitness Evidence: A Guide for Law Enforcement,

one of

the earliest efforts to establish standardized procedures for police-arranged

eyewitness identiﬁcation. The guide set forth rigorous criteria and basic

procedures to promote accuracy in eyewitness evidence.

However, after

the guide was released, most police departments in the United States did

not adopt these procedures.

Today, many police departments use computer systems to access image

databases and assemble photo arrays. Ofﬁcers enter physical characteristics

(e.g., race, gender, hair color) speciﬁc to the suspect into a computer, and

the system retrieves ﬁller photographs with the desired attributes. If an of-

ﬁcer determines that a photograph in the array is suggestive or otherwise in-

appropriate, she or he can reject one or more ﬁllers and instruct the system

Police Executive Research Forum, “A National Survey of Eyewitness Identiﬁcation Proce-

dures in Law Enforcement Agencies,” March 2013, p. 48.

Historically, the photographs were mug shots in the possession of a police department.

U.S. Department of Justice, Ofﬁce of Justice Programs, Eyewitness Evidence: A Guide for

Law Enforcement (Washington, DC, 1999).

Ibid, pp. 11–38.

Identifying the Culprit: Assessing Eyewitness Identification

24 IDENTIFYING THE CULPRIT

to provide alternate photographs. Departments may conduct the procedure

without revealing to the witness how many photographs she or he will view.

In recent decades, many police agencies and prosecutors have adopted

sequential presentation of photographs, based on the belief that this ap-

proach improves the performance of an eyewitness. Currently, however,

there is no consensus among law enforcement professionals as to whether

the sequential presentation procedure is superior to the simultaneous pro-

cedure (see Chapter 5). The District of Columbia Metropolitan Police De-

partment, for example, does not endorse either simultaneous or sequential

procedures in its Procedures for Obtaining Pretrial Eyewitness Identiﬁca-

tion.

The District Attorneys Association of the State of New York in 2010

adopted recommended policies for New York State and endorsed the simul-

taneous method.

On the other hand, in North Carolina, legislation was

passed that requires that lineup photographs be presented sequentially,

and in Massachusetts, the Supreme Judicial Court Study Group on Eyewit-

ness Identiﬁcation recommended sequential procedures as best practice for

Massachusetts Police Departments.

The committee was presented with information regarding improvement

efforts from states including New Jersey, Oregon, Rhode Island, Texas,

New York, and Massachusetts. However, the committee is unable to deter-

mine the percentage of police departments that have adopted policies for

eyewitness identiﬁcation procedures and instituted training in these proce-

dures.

Some police departments require that photo arrays be presented to

the witness during a procedure that is either “double blind” or “blinded.”

(See Box 2-1 for a discussion of blinding as used in scientiﬁc practice and

blinding as used in eyewitness identiﬁcation procedures.) Blinding is used

to prevent conscious and unconscious cues from being given to the witness.

In a double-blind procedure, an individual who does not know the identity

of the suspect or the suspect’s position in the photo array shows a photo

array to the eyewitness. In cases where such a double-blind procedure is

See Metropolitan Police—District of Columbia, General Order—Procedures for Obtaining

Pretrial Eyewitness Identiﬁcation, April 18, 2013.

See New York State District Attorneys Association Best Practice Committee, New York

State Photo Identiﬁcation Guidelines, October 2010.

N.C. Gen. Stat. § 15A-284.52 (West 2007).

See Massachusetts Supreme Judicial Court Study Group on Eyewitness Identiﬁcation,

Report and Recommendations to the Justices (2013).

The Police Executive Research Forum’s 2013 survey of eyewitness identiﬁcation proce-

dures in law enforcement agencies [Police Executive Research Forum, A National Survey of

Eyewitness Identiﬁcation Procedures in Law Enforcement Agencies, (2013)], notes that most

agencies that completed the survey have no written policy for eyewitness identiﬁcation proce-

dures and that more agencies provide training to their employees than have written policies.

See pp. 79–80.

Police Executive Research Forum, p. 64.

Identifying the Culprit: Assessing Eyewitness Identification

EYEWITNESS IDENTIFICATION PROCEDURES 25

not feasible, a “blinded” procedure will approximate the condition of

double-blinding. For example, the photo array may be administered by an

individual who knows who the suspect is, but is unable to tell when the

witness is looking at the suspect’s photo and so is unable to provide even

subconscious feedback to the witness. In one common “blinded” proce-

dure, the ofﬁcer places each photo in a separate envelope or folder and

then shufﬂes the envelopes/folders so that only the witness sees the images

therein. Additional recommendations to minimize the possibility of biasing

feedback to the witness include requiring that the ofﬁcer read instructions

to the witness from a pre-printed form.

If the witness identiﬁes someone from the photo array, some depart-

ments ask the witness for a conﬁdence statement. Based upon information

presented to the committee, it appears that police departments do not

always document identiﬁcation procedures in instances when an identiﬁca-

tion is not made. Further, if a witness does make an identiﬁcation, practices

differ as to how such information is documented and preserved. Some

agencies, for example, require ofﬁcers to document this information in a

written report. Others make audio or video recordings of the identiﬁcation

procedure.

LIVE LINEUP

A live lineup is a police-arranged identiﬁcation procedure in which

the physical suspect and ﬁllers stand or sit in front of the witness (either

individually, i.e., sequentially or en masse, i.e., simultaneously). The police

generally use at least ﬁve ﬁllers. Fillers are selected for their physical simi-

larities to the suspect (gender, race, hair length and color, facial hair, height,

skin tone, and other distinguishing features). The ﬁllers are presumed to be

unknown to the witness. Traditionally, the suspect and ﬁllers are seated or

stood in a row, and the witness views the lineup from behind a two-way

mirror. Police use both simultaneous and sequential procedures for live

lineups.

Live lineups are used in some jurisdictions, but they are not the pre-

dominant method used by law enforcement.

The use of these police

identiﬁcation procedures is limited for a variety of reasons. First, in certain

circumstances, legal counsel may be required at a lineup, thereby making

it less attractive to police and prosecutors. Second, in smaller jurisdictions,

it may be difﬁcult to obtain suitable ﬁllers (e.g., those with appropriate

As discussed in Chapter 3, the courts have been sensitive to the potential for misiden-

tiﬁcation resulting from “suggestive” identiﬁcation procedures and have set standards for

admissibility of evidence.

Police Executive Research Forum, p. 48.

Identifying the Culprit: Assessing Eyewitness Identification

26 IDENTIFYING THE CULPRIT

physical similarities to the suspect). Third, conducting a lineup requires

a signiﬁcant amount of time and labor,

thereby making photo arrays a

more attractive alternative that may be undertaken promptly and with less

demand on department resources.

Live lineup construction may be further constrained by the inability to hold a suspect in

custody without probable cause. See Chapter 3.

BOX 2-1

Blinding

Empirical evidence

has shown that the beliefs, desires, and expectations

of researchers can inﬂuence, often subconsciously, how they observe and in-

terpret the phenomena they study and thus the outcomes of experiments. This

evidence has inﬂuenced how scientists carry out their experiments, resulting in

the use of blind or double-blind procedures to control for this form of bias. Blind

assessment

has been used since the late 18th century; an early medical trial in

1835 used double-blind assessment, and psychologists started using blinding in

the 20th century.

By the 1950s, blind assessment in randomized controlled trials

was considered standard procedure in both psychological and medical research.

Currently, virtually all of science uses some form of blinding.

In single-blind experiments, participants do not know which treatment they

are receiving; this form of blinding is used widely across scientiﬁc ﬁelds. In experi-

ments involving humans, as in medical or psychological research, double-blind

procedures are used to guard against “expectancy effects” for both participants

and researchers. In a classic double-blind clinical trial, some patients receive ac-

tive medication and others are given an alternative (either a “standard treatment”

or a similar-looking placebo without active ingredients), but neither researchers

nor participants know who is receiving which treatment.

In an eyewitness identiﬁcation setting, double-blinding can be used to prevent

a lineup administrator from either intentionally or unintentionally inﬂuencing a wit-

ness. In these cases, neither the eyewitness nor the administrator knows which

persons in a photo array or live lineup are the suspected culprits and which are

the ﬁllers.

d,e

In eyewitness identiﬁcation procedures, as in science, the purpose of

double-blinding is to prevent the conscious or subconscious expectations of the

administrator from inﬂuencing the witness or research outcomes.

In a double-blind photo array, the ofﬁcer or detective conducting the inves-

tigation reads a set of standard instructions to the witness. The instructions may

include an advisory that the ofﬁcer about to show the photos does not know

whether any of the photos are of the person who committed the crime. The ofﬁcer

then leaves the room and a second ofﬁcer—perhaps a patrol ofﬁcer—displays the

Identifying the Culprit: Assessing Eyewitness Identification

EYEWITNESS IDENTIFICATION PROCEDURES 27

SHOWUP

A showup is a police-arranged identiﬁcation procedure in which the

police show one person to a witness and ask if she or he recognizes that per-

son. This procedure typically is used when the police locate a suspect shortly

after the commission of a crime and within close proximity to the scene.

Case law limits the time and distance from a crime during which such a

procedure will pass legal standards. In response to such case law, police

typically restrict showups to a two-hour time period after the commis-

photos. It is the duty of this second ofﬁcer (the “blind administrator”) to show the

photos and, if an identiﬁcation is made, document what the witness said and ask

the witness how certain she or he is of their identiﬁcation. Once all photos have

been shown, the ofﬁcer reports the result of the procedure to the investigating

ofﬁcer (preferably out of earshot from the witness).

As an alternative to a double-blind array, some departments use “blinded”

procedures. A blinded procedure prevents an ofﬁcer from knowing when the wit-

ness is viewing a photo of the suspect, but can be conducted by the investigating

ofﬁcer. A common approach is the so-called “folder shufﬂe.” With a six-photo array,

an ofﬁcer uses eight manila folders. A photograph of a ﬁller is placed in the top

folder, and a photograph of the suspect and four additional ﬁllers are placed in

the next ﬁve folders. The six folders are then shufﬂed so that the ofﬁcer does not

know which folder contains the image of the suspect. Two folders with blank paper

are placed on the bottom of the stack so that the witness is led to believe that

there are more than six images in the array (this is referred to as back-loading,

and it prevents the witness from knowing when she or he is about to view the last

photograph). After reading instructions to the witness, the administering ofﬁcer sits

to the witness’ left and hands him or her one folder at a time and instructs him/her

to open each folder and look at the enclosed photo. The cover of the folder blocks

the ofﬁcer from viewing the photo that the witness is viewing. When an identiﬁca-

tion occurs, the ofﬁcer notes the witness’ words and reaction and asks about the

witness’ conﬁdence in his or her identiﬁcation.

R. Rosenthal, Experimenter Effects in Behavioral Research (New York: John Wiley, 1976).

M. Stolberg, “Inventing the Randomized Double-Blind Trial: The Nürnberg Salt Test of

1835,” James Lind Library Bulletin (2006), available at: http://www.jameslindlibrary.org/

illustrating/articles/inventing-the-randomized-double-blind-trial-the-nurnberg-salt.

T. J. Kaptchuk," Intentional ignorance: A History of Blind Assessment and Placebo Controls

in Medicine,\” Bulletin of the History of Medicine 72(3): 389–433 (1998).

P. Kilmartin, Presentation to the committee, February 6, 2014.

K. Hamann, Presentation to the committee, December 2, 2013.

Identifying the Culprit: Assessing Eyewitness Identification

28 IDENTIFYING THE CULPRIT

sion of a crime. Ideally, ofﬁcials take the witness to the location where the

suspect has been detained and do not display the suspect in a suggestive

manner (e.g., not in a police car, not handcuffed, without drawn weapons).

However, as chases, ﬁghts, or disarmaments frequently precede showups,

the apprehension of a suspect can raise safety issues that make it difﬁcult to

adhere to recommended procedures. Further, the nature of a showup does

not lend itself to the use of a blinded procedure. A showup is designed to

promptly clear innocent suspects, thereby sparing them from a prolonged

period of detention as the investigation continues. Delaying the showup to

locate an uninvolved ofﬁcer may defeat that purpose. While some law en-

forcement agencies use a standard procedure with written instructions when

conducting a showup, there is no indication that such procedures are used

uniformly. Courts consider showups highly suggestive, and prosecutors urge

the police to exercise caution when conducting them.

CONFIRMATORY PHOTOGRAPH

Police will, on occasion, display a single photograph to a witness in

an effort to conﬁrm the identity of a perpetrator. Police typically limit this

method to situations in which the perpetrator is previously known to or

acquainted with the witness.

FIELD VIEW

Police also use ﬁeld views in attempts to identify perpetrators. The

method, which involves inviting a witness to view many people in a context

where the perpetrator is thought likely to appear, is used when the police

do not have a suspect but believe that the offender frequents a particular

location. For example, police investigating a purse snatching may obtain

information that the perpetrator frequents a particular recreation site dur-

ing the lunch hour. A plainclothes ofﬁcer or investigator might take the

eyewitness to the site and walk around with him or her during the lunch

hour without directing his or her attention to any speciﬁc individual.

OTHER PROCEDURES—MUG BOOKS AND YEARBOOKS

At times, police use other means to identify perpetrators. In the past,

police sometimes had witnesses review mug shot books. Mug books have

since been largely replaced by digitized images displayed on computer

screens. Nonetheless, there are situations in which the police will have a

witness review a large collection of photographs in an effort to identify a

perpetrator. Witnesses who identify a perpetrator as being a student at a

speciﬁc school might be asked to review a yearbook for that school in an

Identifying the Culprit: Assessing Eyewitness Identification

EYEWITNESS IDENTIFICATION PROCEDURES 29

effort to identify the perpetrator. When using this method, police typically

attempt to mask the names of the students. Similarly, if the offender is be-

lieved to be an individual from a certain profession, then the police might

have the witness review photographs from the suspect’s professional society.

Social media sites also serve as the catalyst for police-arranged identiﬁcation

procedures. If a witness knows that the perpetrator is a “friend” of Jane

Doe through social media, then the police might have the witness review all

friends of Jane Doe to see if she or he recognizes the individual.

All of these additional procedures (i.e., conﬁrmatory photo, ﬁeld view,

mug books, yearbooks) have the potential to introduce biases of the sort

that blind lineup procedures are designed to avoid.

NON-POLICE IDENTIFICATION PROCEDURES

In some cases, the victims or witnesses, or both, identify suspects

without involving the police. A private citizen, organization, or corpora-

tion may conduct an investigation before, during, or even after reporting

a crime to the police. The identiﬁcation of suspects by entities other than

law enforcement has become increasingly common as more businesses

and private citizens use security cameras to identify criminal actors. High-

resolution cameras coupled with high-capacity hard drives allow for real-

time streaming of video with superior clarity. Such systems are relatively

inexpensive and within ﬁnancial reach of many home and business owners.

Additionally, the proliferation of smart phones has put the ability to cre-

ate a spontaneous, high-quality video record of an event into the hands of

more and more people.

The rise of social media has resulted in the rise of private investigations

and identiﬁcations using this resource. In one recent case, a stabbing vic-

tim drew a picture of her assailant and showed it to her husband.

Upon

viewing the picture, the husband believed that the assailant looked familiar

and might be his ex-girlfriend. He obtained several photographs of the ex-

girlfriend from her personal website and showed them to the victim who,

after looking at those and other online images, identiﬁed the suspect at a

lineup and at trial.

CONCLUSION

Many local, state, and federal law enforcement agencies have adopted

policies and practices to address the issue of misidentiﬁcation. However,

efforts are not uniform or systemic.

Many agencies are unfamiliar with

New Jersey v. Chen, 27 A.3d 930 (N.J. 2011).

See Massachusetts Supreme Judicial Court Study Group on Eyewitness Evidence, p. 2.

Identifying the Culprit: Assessing Eyewitness Identification

30 IDENTIFYING THE CULPRIT

the science that has emerged during the past few decades of research on

eyewitness identiﬁcations. Questions remain about the optimal design of

photo array procedures, including the size of the array, the contents of

the photographs, and their relationship to the context of the crime scene.

Similar questions apply to the design of live lineups.

Eyewitness identiﬁca-

tion is further complicated by the increasing number of situations in which

victims and witnesses seek to identify the perpetrator of a crime without

the aid of law enforcement. Such identiﬁcations raise new concerns about

reliability and accuracy of the identiﬁcation of individuals. Inconsistent

and nonstandard practices might easily add noise to the eyewitness iden-

tiﬁcation process, contaminate the witness, and bias the outcome of an

identiﬁcation procedure.

The design of a live lineup is subject to more practical constraints than a photo array.

Identifying the Culprit: Assessing Eyewitness Identification

The Legal Framework for Assessment

of Eyewitness Identiﬁcation Evidence

he admissibility of eyewitness testimony at a criminal trial may be

challenged on the basis of procedures used by law enforcement of-

ﬁcials in obtaining the eyewitness identiﬁcation. The U.S. Supreme

Court, in its 1977 ruling in Manson v. Brathwaite, set out the modern test

under the Due Process Clause of the U.S. Constitution that regulates the

fairness and the reliability of eyewitness identiﬁcation evidence.

The Court

also speciﬁed ﬁve reliability factors, discussed below, that a judge must con-

sider when deciding whether to exclude the identiﬁcation evidence at trial.

Although the constitutional standards for assessing eyewitness tes-

timony have remained unchanged in the decades since the Manson v.

Brathwaite decision, a body of research has shed light on the extent to

which each of the ﬁve reliability factors supports a reliable eyewitness

identiﬁcation. Research has cast doubt, for instance, on the belief that the

apparent certainty displayed in the courtroom by an eyewitness is an indi-

cator of an accurate identiﬁcation, and has found that a number of factors

may enhance the certainty of the eyewitness.

Recently, state courts and lower federal courts have taken the lead in

developing standards relating to the admissibility of expert evidence, jury

instructions, and judicial notice of scientiﬁc evidence. Some states have

adopted more stringent standards for regulating eyewitness identiﬁcation

evidence than the U.S. Constitution requires, either by legislative statutes or

by state court decisions, and have modiﬁed or entirely supplanted the Man-

Manson v. Brathwaite, 432 U.S. 98, 113–114 (1977).

Manson v. Brathwaite at 114.

Identifying the Culprit: Assessing Eyewitness Identification

32 IDENTIFYING THE CULPRIT

son v. Brathwaite test to take account of advances in the growing body of

scientiﬁc research. This chapter describes the changes in the legal standards

for eyewitness identiﬁcation and explores the relationship between the state

of the scientiﬁc research and the law regulating procedures and evidence.

EYEWITNESS EVIDENCE AND DUE PROCESS

UNDER THE U.S. CONSTITUTION

Beginning with rulings in 1967, the U.S. Supreme Court set out a

standard under the Due Process Clause of the Fourteenth Amendment for

reviewing eyewitness identiﬁcation evidence.

In Manson v. Brathwaite,

the Court emphasized that “reliability is the linchpin in determining the

admissibility of identiﬁcation testimony.”

First, the Court instructed judges

to examine whether the identiﬁcation procedures were unnecessarily sug-

gestive. Second, to assess whether an identiﬁcation is reliable, judges were

instructed to examine the following ﬁve factors: (1) the opportunity of the

witness to view the criminal at the time of the crime; (2) the witness’ degree

of attention; (3) the accuracy of the witness’ prior description of the crimi-

nal; (4) the level of certainty demonstrated at the confrontation; and (5) the

time between the crime and the identiﬁcation procedure.

The ﬁve factors

were drawn from earlier judicial rulings and not from scientiﬁc research.

Eyewitness identiﬁcation evidence continues to be litigated primarily

under the ﬂexible two-part Manson v. Brathwaite Due Process test.

It is

In Stovall v. Denno, 388 U.S. 293, 302 (1967), the U.S. Supreme Court ﬁrst set out a due

process rule asking whether identiﬁcation procedures used were “so unnecessarily suggestive

and conducive to irreparable mistaken identiﬁcation.” The Court elaborated that rule in deci-

sions such as Simmons v. U.S., 390 U.S. 377, 384 (1968) and Foster v. California, 394 U.S.

440, 442 (1969), and then adopted an approach setting out “reliability” considerations in

Neil v. Biggers, 409 U.S. 188 (1972). For a description of the development of this doctrine,

see, e.g., B. L. Garrett, “Eyewitnesses and Exclusion,” Vanderbilt Law Review 65(2): 451,

463–467 (2012).

Brathwaite, 423 U.S. at 114.

Id. at 114.

Id. at 114. Justice Thurgood Marshall dissented, noting studies indicated that unnecessarily

suggestive eyewitness identiﬁcations had resulted in “repeated miscarriages of justice result-

ing from juries’ willingness to credit inaccurate eyewitness testimony.” 432 U.S. at 125–27

(Marshall, J., dissenting).

Due process is the most important constitutional right that arises in challenges to eyewit-

ness identiﬁcation, but rights under the Fourth and Sixth Amendments also may be implicated.

The Fourth Amendment protects individuals “against unreasonable searches and seizures,”

and the probable cause typically required to seize and arrest a suspect may arise from an eye-

witness identiﬁcation. U.S. Const. Amend. IV. The few lower courts to address the question

are divided on whether probable cause is needed to place individuals in a live lineup proce-

dure. Biehunik v. Felicetta, 441 F.2d 228, 230 (2d Cir. 1971); but see, e.g., Wise v. Murphy,

275 A.2d 205, 212–15 (D.C. 1971); State v. Hall, 461 A.2d 1155 (N.J. 1983). In contrast,

Identifying the Culprit: Assessing Eyewitness Identification

THE LEGAL FRAMEWORK 33

important to note, however, that the vast majority of criminal cases are

settled through plea bargaining. The role that evidence type and strength

play in plea bargaining is complex and necessarily difﬁcult to study. Because

eyewitness identiﬁcation evidence may never be tested at trial, it is doubly

important for lawyers and judges to understand the credibility of the prof-

fered evidence.

In the most recent U.S. Supreme Court ruling addressing a challenge

to an eyewitness identiﬁcation (Perry v. New Hampshire),

the Court ruled

that a due process analysis was not triggered. In that case, while the police

were obtaining a description of the suspect, the eyewitness looked out of the

apartment window and recognized the suspect standing outside. The police

had not intended to conduct an identiﬁcation procedure. In those circum-

stances, the Court ruled that the Due Process Clause does not require a pre-

liminary judicial review of the reliability of an eyewitness identiﬁcation.

probable cause is not required to place a person’s photograph in an array, since doing so does

not involve a seizure. However, courts may also rule that an illegal stop or seizure renders

a subsequent identiﬁcation inadmissible, absent an “independent” source for the courtroom

identiﬁcation. U.S. v. Crews, 445 U.S. 463, 473 (1980).

In addition, the Sixth Amendment provides that, in all criminal prosecutions, the accused

has the right “to have the assistance of counsel for his defense.” In United States v. Wade,

the Supreme Court held that, once indicted, a person has a right to have a lawyer present

at a lineup, reasoning that the right to counsel applies at all “critical” stages of the criminal

process. 388 U.S. 218, 235–37 (1967). However, the Court subsequently held that a photo

array procedure, of the type now most commonly used by police agencies, does not implicate

the Wade right to counsel. U.S. v. Ash, 413 U.S. 300, 321 (1973).

As the current report demonstrates, a comparative consideration of evidence value is

particularly important in the case of eyewitness identiﬁcation evidence. Similar consideration

should be given when other adjudication mechanisms are used (e.g., bench trials).

Perry v. New Hampshire, 132 S. Ct. 716, 718 (2012). In that case, the eyewitness happened

to look out her window and see the suspect standing at the crime scene where the police had

told him to wait. The Court held that the Due Process Clause did not regulate such a situation,

since the police did not intend to conduct an identiﬁcation procedure. Id. at 729. The Court

indicated that the reliability of the evidence could be addressed by federal and state evidentiary

standards, and added: “In appropriate cases, some States also permit defendants to present

expert testimony on the hazards of eyewitness identiﬁcation evidence.” Id.

Justice Sotomayor dissented, arguing, “Our due process concern . . . arises not from the

act of suggestion, but rather from the corrosive effects of suggestion on the reliability of the

resulting identiﬁcation,” and the manner in which “[a]t trial, an eyewitness’ artiﬁcially inﬂated

conﬁdence in an identiﬁcation’s accuracy complicates the jury’s task of assessing witness

credibility and reliability.” Perry, 132 S. Ct. at 731–32 (Sotomayor, J., dissenting). Justice

Sotomayor also emphasized: “A vast body of scientiﬁc literature has reinforced every concern

our precedents articulated nearly a half-century ago.” Id. at 738.

Identifying the Culprit: Assessing Eyewitness Identification

34 IDENTIFYING THE CULPRIT

STATE LAW REGULATION OF EYEWITNESS EVIDENCE

State Supreme Court Standards

Several state supreme courts have altered or supplemented the federal

Manson v. Brathwaite due process rule to focus more on the effects of sug-

gestion, to emphasize certain factors in speciﬁc circumstances,

or to focus

on showup identiﬁcations in particular.

New Jersey and Oregon have

now supplemented the Manson v. Brathwaite test with separate state law

standards regulating eyewitness identiﬁcation evidence.

In 2011, the New Jersey Supreme Court issued a unanimous decision in

State v. Larry R. Henderson that revised the legal framework for admitting

eyewitness identiﬁcation evidence and directed that revised jury instructions

be prepared to help jurors evaluate such evidence.

The new framework

was based on the record of hearings before a Special Master that considered

an extensive review of scientiﬁc research regarding eyewitness identiﬁca-

tions.

The legal framework established by the Henderson opinion relies

on pretrial hearings to review eyewitness evidence and more comprehensive

jury instructions at trial.

To obtain a pretrial hearing, a defendant must

show some evidence of suggestiveness related to either estimator or system

See State v. Ramirez, 817 P.2d 774, 780–81 (Utah 1991) (altering three of the reliability

factors to focus on effects of suggestion); State v. Marquez, 967 A.2d 56, 69–71 (Conn.

2009) (adopting criteria for assessing suggestion); Brodes v. State, 614 S.E.2d 766, 771 & n.8

(Ga. 2005) (rejecting eyewitness certainty jury instruction); State v. Hunt, 69 P.3d 571, 576

(Kan. 2003) (adopting Utah’s ﬁve factor “reﬁnement” of the Biggers factors); State v. Crom-

edy, 727 A.2d 457, 467 (N.J. 1999) (requiring, when applicable, instruction on cross-racial

misidentiﬁcations).

See, e.g., State v. Dubose, 285 Wis.2d 143, 166 (Wis. 2005); Commonwealth v. Johnson,

650 N.E.2d 1257, 1261 (Mass. 1995); People v. Adams, 423 N.E.2d 379, 383–84 (N.Y. 1981).

State v. Henderson, 27 A.3d 872 (N.J. 2011). The Henderson opinion described criticisms

of the Manson v. Brathwaite test, including that suggestion may itself affect the seeming “reli-

ability” of the identiﬁcation. Id. at 877–78. For examples of scholarly criticism of the Manson

v. Brathwaite test in light of scientiﬁc research, see, e.g., G. L. Wells and D. S. Quinlivan,

“Suggestive Eyewitness Identiﬁcation Procedures and the Supreme Court’s Reliability Test in

Light of Eyewitness Science: 30 Years Later,” Law and Human Behavior 33(1): 1, 16 (Febru-

ary 2009); T. P. O’Toole and G. Shay, “Manson v. Brathwaite Revisited: Towards a New Rule

of Decision for Due Process Challenges to Eyewitness Identiﬁcation Procedures,” Valparaiso

University Law Review 41(1): 109 (2006).

See Report of the Special Master at 16–17, State v. Henderson, No. A-8-08 (N.J. June 18,

2011, available at: http://www.judiciary.state.nj.us/pressrel/HENDERSON%20FINAL%20

BRIEF%20.PDF%20(00621142.pdf.

In the companion case, State v. Chen, 27 A.3d 930, 932 (N.J. 2011), the New Jersey

Supreme Court took an approach that departed from that of the U.S. Supreme Court in

Perry, ruling that the defendant may be entitled to a hearing in a case in which the eyewitness

identiﬁed the defendant using social media, not a police-orchestrated identiﬁcation procedure.

Identifying the Culprit: Assessing Eyewitness Identification

THE LEGAL FRAMEWORK 35

variables that could lead to mistaken identiﬁcation.

At the pretrial hear-

ing, the State must offer proof that the eyewitness identiﬁcation is reliable.

However, the ultimate burden of proving a “very substantial likelihood of

irreparable misidentiﬁcation” is on the defendant.

In July 2012, the New Jersey Supreme Court released an expanded

set of jury instructions and related rules that govern the use of suggestive

identiﬁcations.

The jury instructions state that “[r]esearch has shown that

there are risks of making mistaken identiﬁcations” and noted that eyewit-

ness evidence “must be scrutinized carefully.”

Human memory involves

three stages—encoding, storage, and retrieval. At “each of these stages,

memory can be affected by a variety of factors.”

The Court identiﬁed a

set of factors that jurors should consider when deciding whether eyewit-

ness identiﬁcation evidence is reliable, including estimator variables (e.g.,

stress, exposure duration, weapon focus, distance, lighting, intoxication,

disguises or changed appearance of the perpetrator, time since the incident,

and cross-racial effects) and system variables (e.g., lineup composition,

ﬁllers, use of multiple viewings, presence of feedback, use of double-blind

procedures, and use of showup identiﬁcations). The instructions also noted

the possible inﬂuence of outside opinions, descriptions or identiﬁcations by

other witnesses, and photographs or media accounts.

In 2012, in Oregon v. Lawson, the Oregon Supreme Court established

a new procedure for evaluating the admissibility of eyewitness identiﬁca-

tions. In a unanimous decision, the Court found “serious questions” about

the reliability of eyewitness identiﬁcation, citing research conducted over

the past 30 years.

The Court determined that the Manson v. Brathwaite

two-step process for weighing eyewitness identiﬁcation “does not accom-

plish its goal of ensuring that only sufﬁciently reliable identiﬁcations are

admitted into evidence,” because it relies on an eyewitness’ self-reports to

determine whether the threshold level of suggestiveness is reached, ren-

dering the identiﬁcation unreliable.

The Court set forth a process that

requires the trial court to examine whether investigators used “suggestive”

Henderson, 27 A.3d. at 878.

Id.

New Jersey Criminal Model Jury Instructions, Identiﬁcation (July 19, 2012), available at:

http://www.judiciary.state.nj.us/pressrel/2012/jury_instruction.pdf; New Jersey Court Rule

3:11, Record of an Out-of-Court Identiﬁcation Procedure (July 19, 2012), available at: http://

www.judiciary.state.nj.us/pressrel/2012/new_rule.pdf; New Jersey Court Rule 3:13-3, Discov-

ery and Inspection (July 19, 2012), available at: http://www.judiciary.state.nj.us/pressrel/2012/

rev_rule.pdf.

See New Jersey Criminal Model Jury Instructions, Identiﬁcation, supra at 2.

Id.

Id. at 9.

State v. Lawson, 352 Ore. 724 (Or. 2012).

Id. at 746–748.

Identifying the Culprit: Assessing Eyewitness Identification

36 IDENTIFYING THE CULPRIT

identiﬁcation procedures and whether other factors, such as estimator vari-

ables, may have affected the reliability of the identiﬁcation.

The Court

ruled that “intermediate remedies,” including the use of expert testimony,

should be available even if the trial judge concludes that the identiﬁcation is

admissible. The Court also brieﬂy noted that judges might use “case-speciﬁc

jury instructions.”

Other states continue to explore possible changes to the judicial re-

view of eyewitness identiﬁcation evidence. In 2013, the Massachusetts

Supreme Judicial Court Study Group on Eyewitness Identiﬁcation offered

guidance on the adjudication of eyewitness identiﬁcation evidence.

The

report adopted Lawson’s approach of taking judicial notice of “certain

scientiﬁcally-established facts about eyewitness identiﬁcation.”

The re-

port recommended that trial judges conduct pretrial hearings to determine

whether suggestive identiﬁcation procedures were used, and if so, whether

these procedures impaired the reliability of identiﬁcation evidence. Pretrial

hearings would consider the effects of both estimator variables (relating to

viewing at the crime scene) and system variables (relating to the lineup or

showup procedures) on the identiﬁcation. The report also recommended

that the state adopt a set of recommended practices for conducting identi-

ﬁcation procedures, create new model jury instructions on eyewitness iden-

tiﬁcations, and set limitations on the admissibility of certainty statements

and in-court identiﬁcations.

State Statutes Regulating Identiﬁcation Procedures

Judicial rulings regulating admissibility of eyewitness evidence in the

courtroom do not specify the identiﬁcation procedures to be used by law

enforcement ofﬁcials. However, 14 states have adopted legislation regard-

ing eyewitness identiﬁcation procedures. Of the 14, 11 states (Connecticut,

Illinois, Maryland, North Carolina, Ohio, Texas, Virginia, West Virginia,

Wisconsin, Utah, and Vermont) have enacted statutes directly requiring that

Id. at 747–748, 755–756.

Id. at 759, 763.

See Massachusetts Supreme Judicial Court Study Group on Eyewitness Evidence, Report

and Recommendations to the Justices (2013).

Id. at 48.

Id. at 28. In the courtroom, the eyewitness can easily see where the defendant is sitting.

Thus, in-court identiﬁcations do not reliably test an eyewitness’ memory. Nevertheless, courts

have shown great tolerance of in-court identiﬁcations, deeming them based on “independent”

memory, and even following suggestive out-of-court procedures. Garrett, Eyewitnesses and

Exclusion, supra. For example, the New York Court of Appeals ruled that “[e]xcluding evi-

dence of a suggestive showup does not deprive the prosecutor of reliable evidence of guilt. The

witness would still be permitted to identify the defendant in court if that identiﬁcation is based

on an independent source.” People v. Adams, 423 N.E.2d 379, 384 (N.Y. 1981).

Identifying the Culprit: Assessing Eyewitness Identification

THE LEGAL FRAMEWORK 37

law enforcement ofﬁcials adopt written procedures for eyewitness identiﬁ-

cations and regulating the particular procedures to be used.

Three more

states (Georgia, Nevada, and Rhode Island) have passed statutes recom-

mending further study, tasking a group with developing best practices, or

requiring some form of written policy.

State statutes typically assert that a trial judge may consider the failure

to follow the prescribed procedures as a factor in assessing admissibility

and informing the jury. The statutes rarely require that a trial judge exclude

such identiﬁcation evidence from consideration by the jury. However, some

of the more detailed statutes, such as those in Ohio, North Carolina, and

West Virginia, require that law enforcement ofﬁcials use particular practices

(e.g., eyewitness instructions, a blind administrator). Other statutes require

adherence to model policies or guidelines. Utah requires that lineup pro-

cedures be recorded. Some jurisdictions and departments also have volun-

tarily adopted guidelines or policies regulating eyewitness identiﬁcations.

Several state courts have issued rulings regulating lineup practices (e.g.,

New Jersey’s Supreme Court has required documentation of identiﬁcation

procedures).

AIDING JURORS IN ASSESSMENT OF EYEWITNESS TESTIMONY

Expert Witness Testimony Regarding Eyewitness Identiﬁcation

The standards for assessing the admissibility of testimony by expert

witnesses have undergone great changes in the past two decades. Before

1993, the Frye test allowed scientiﬁc expert testimony in federal courts

if it met the standard of “general acceptance” in the relevant scientiﬁc

community.

In 1993, the Supreme Court, in Daubert v. Merrell Dow

See Conn. Gen. Stat. § 54-1p (West 2012); 725 Ill. Comp. Stat. § 5/107A-5 (West 2003);

Md. Code Ann., Pub. Safety § 3-506 (West 2007); N.C. Gen. Stat. § 15A-284.52 (West 2007);

Ohio Rev. Code Ann. § 2933.83 (West 2010); Tex. Code Crim. Proc. Ann. art. 38.20 (West

2011); Utah Code Ann. §77-8-4 (West 1980); Va. Code Ann. §19.2-390.02 (West 2005); Va

Code Ann. § 9.1-102.54; 13 V.S.A. § 5581; W. Va. Code Ann. § 62-1E-1 (West 2013); Wis.

Stat. § 175.50 (West 2005).

GA. H.R. 352, 149th Gen. Assem., Reg. Sess. (April 20, 2007); Nev. Rev. Stat. § 171.1237

(West 2011); R.I. Gen. Laws § 12-1-16 (West 2012); 2010 Leg. Reg. Sess. (Vt. 2010).

See, e.g., John J. Farmer, Jr., Attorney General of the State of New Jersey, “Letter to All

County Prosecutors: Attorney General Guidelines for Preparing and Conducting Photo and

Live Lineup Identiﬁcation Procedures” (April 18, 2001), available at: http://www.state.nj.us/

lps/dcj/agguide/photoid.pdf; CALEA Standards for Law Enforcement Agencies: 42.2.11 Line-

ups, available at: http://www.calea.org/content/standards-titles; International Association of

Chiefs of Police, Model Policy: Eyewitness Identiﬁcation (2010).

State v. Delgado, 188 N.J. 48, 63–64, 902 A.2d 888 (2006).

Frye v. United States, 54 App. D.C. 46, 293 F. 1013 (1923).

Identifying the Culprit: Assessing Eyewitness Identification

38 IDENTIFYING THE CULPRIT

Pharmaceuticals, Inc.,

ruled that, under Federal Rule of Evidence 702, a

“trial judge must ensure that any and all scientiﬁc testimony or evidence

admitted is not only relevant, but reliable.”

Judges determine reliability

by assessing the scientiﬁc foundation of the expert’s testimony prior to trial,

so that “evidentiary reliability will be based upon scientiﬁc validity.”

Many states have adopted Daubert, and many of those that have not for-

mally adopted Daubert have revised their Frye test to adopt much of the

Daubert standard. In turn, Federal Rule of Evidence 702 has been revised

to incorporate the holding in Daubert.

Federal and state courts remain

divided on whether expert testimony on eyewitness identiﬁcations is ad-

missible under Daubert or Frye, and on the proper exercise of trial court

discretion when deciding whether to admit such expert testimony. Appellate

rulings emphasize that a trial judge should use discretion when deciding

whether proffered expert evidence satisﬁes the Daubert or Frye standards.

An increasing number of rulings emphasize the value of presenting expert

testimony regarding eyewitness identiﬁcation. Some courts have held that

it can be an abuse of discretion for a trial judge to bar the defense from

admitting such testimony.

Detailed descriptions of the relevant scientiﬁc

research ﬁndings accompany such decisions.

There are also many federal

and state courts that continue to follow the traditional approach, emphasiz-

ing that credibility of eyewitnesses is a matter within the “province of the

jury” and insisting that information regarding valid scientiﬁc research in

this area will not assist the jury in its task.

509 U.S. 579 (1993).

Id. at 589.

Id. at 590 n.9.

Fed. R. Evid. 702. Rule 702 now provides:

A witness who is qualiﬁed as an expert by knowledge, skill, experience, training, or educa-

tion may testify in the form of an opinion or otherwise if: (a) the expert’s scientiﬁc, technical,

or other specialized knowledge will help the trier of fact to understand the evidence or to de-

termine a fact in issue; (b) the testimony is based on sufﬁcient facts or data; (c) the testimony

is the product of reliable principles and methods; and (d) the expert has reliably applied the

principles and methods to the facts of the case.

See, e.g., Tillman v. State, 354 S.W.3d 425, 441 (Tex. Crim. App. 2011); People v. Le-

Grand, 835 N.Y.S.2d 523, 524 (2007); State v. Clopten, 223 P.3d 1103, 1117 (Utah 2009);

U.S. v. Smithers, 212 F.3d 306, 311–14 (6th Cir. 2000).

See, e.g., State v. Copeland, 226 S.W.3d 287, 299–300 (Tenn. 2007); Tillman, 354 S.W.3d

at 441; Clopten, 223 P.3d at 1108.

For scholarly examination of this case law, see, e.g., “The Province of the Jurist: Judicial

Resistance to Expert Testimony on Eyewitnesses as Institutional Rivalry,” Harvard Law

Review 126(8): 2381 (2013); R. Simmons, “Conquering the Province of the Jury: Expert

Testimony and the Professionalization of Fact-Finding,” University of Cincinnati Law Review

74: 1013 (2006); G. Vallas, “A Survey of Federal and State Standards for the Admission of

Expert Testimony on the Reliability of Eyewitnesses,” American Journal of Criminal Law

39(1): 97 (2011).

Identifying the Culprit: Assessing Eyewitness Identification

THE LEGAL FRAMEWORK 39

The trend is toward greater acceptance of expert testimony regarding

the factors that may affect eyewitness identiﬁcation. In a 2012 decision,

the Connecticut Supreme Court disavowed earlier rulings restricting expert

testimony and stated that such rulings are now “out of step with the wide-

spread judicial recognition that eyewitness identiﬁcations are potentially

unreliable in a variety of ways unknown to the average juror.”

Similarly,

the Pennsylvania Supreme Court recently held that expert testimony on

eyewitness identiﬁcations was no longer per se inadmissible, emphasizing

that “courts in 44 states and the District of Columbia have permitted such

testimony at the discretion of the trial judge,” and that “all federal circuits

that have considered the issue, with the possible exception of the 11th

Circuit, have embraced this approach.”

As the Seventh Circuit Court of

Appeals recently explained:

It will not do to reply that jurors know from their daily lives that memory

is fallible. The question that social science can address is how fallible,

and thus how deeply any given identiﬁcation should be discounted. That

jurors have beliefs about this does not make expert evidence irrelevant;

to the contrary, it may make such evidence vital, for if jurors’ beliefs are

mistaken then they may reach incorrect conclusions. Expert evidence can

help jurors evaluate whether their beliefs about the reliability of eyewitness

testimony are correct.

Courts also have allowed expert witnesses to testify about particular is-

sues concerning eyewitness identiﬁcations, such as cross-race effects, stress,

weapons focus, suggestive lineup procedures, and the like.

Rarely have

experts conducted eyewitness identiﬁcation research related to the speciﬁc

case before the court. However, in one such case, in which an experiment

State v. Guilbert, 306 Conn. 218, 234 (Conn. 2012). Prior to that decision, the Connecti-

cut Supreme Court had long ruled that “the reliability of eyewitness identiﬁcation is within

the knowledge of jurors and expert testimony generally would not assist them in determining

the question” (State v. Kemp, supra 199 Conn. at 473, 477), and that factors affecting eyewit-

ness memory are “nothing outside the common experience of mankind” (State v. McClendon,

supra 248 Conn. at 572, 586).

Com. v. Walker, 2014 WL 2208139 *13 (Pa. 2014) (collecting authorities).

U.S. v. Bartless, 567 F.3d 901, 906 (7th Cir. 2009). Other federal courts have found it a

proper exercise of discretion to exclude expert testimony on eyewitness identiﬁcations. See,

e.g., United States v. Lumpkin, 192 F.3d 280, 289 (2d Cir. 1999). Most federal courts treat

the subject as one of considerable trial discretion; see, e.g., United States v. Rodriguez-Berrios,

573 F.3d 55, 71–72 (1st Cir. 2006). For a survey of federal decisions, see Lauren Tallent, Note,

Through the Lens of Federal Evidence Rule 403: An Examination of Eyewitness Identiﬁcation

Expert Testimony Admissibility in the Federal Circuit Courts, Washington & Lee Law Review

68 (2): 765 (2011); see also Walker, 2014 2208139 *13.

See, e.g., Loftus, Doyle & Dysart at § 14-8[a]-[b] p. 408 n. 41–42, 410, n. 53 (5th Edi-

tion, 2013) (collecting cases).

Identifying the Culprit: Assessing Eyewitness Identification

40 IDENTIFYING THE CULPRIT

was conducted with the actual photo array used in the case, the federal

courts found expert testimony admissible where it was directed not only to

general research, but also by the question of whether suggestive procedures

affected the identiﬁcation in that case.

Expert witnesses who explain the complications of eyewitness identiﬁ-

cation can be expensive. Most criminal defendants are indigent and cannot

afford such assistance.

In Ake v. Oklahoma, the Supreme Court held that

an indigent defendant has a constitutional due process right to assistance by

an expert witness only if that expert assistance is so crucial to the defense

(or such a “signiﬁcant factor”) that its denial would deprive the defendant

of a fundamentally fair trial.

In federal courts, funding for expert wit-

nesses is available, and requests by indigent defendants are common.

state courts, such assistance is uncommon, especially in state courts that

rarely ﬁnd denial of expert assistance on eyewitness matters to be a due

process violation.

Expert testimony on eyewitness memory and identiﬁcations has many

advantages over jury instructions as a method to explain relevant scientiﬁc

framework evidence to the jury: (1) Expert witnesses can explain scientiﬁc

research in a more ﬂexible manner, by presenting only the relevant research

to the jury; (2) Expert witnesses are familiar with the research and can de-

scribe it in detail; (3) Expert witnesses can convey the state of the research

at the time of the trial; (4) Expert witnesses can be cross-examined by the

other side; and (5) Expert witnesses can more clearly describe the limita-

tions of the research. The beneﬁts of expert testimony are offset somewhat

by the expense. However, conﬂicting testimony by opposing experts may

lead to confusion among the jurors. Nonetheless, trial judges have discre-

tion to determine whether the potential beneﬁts of expert testimony out-

weigh the cost.

Jury Instructions Regarding Eyewitness Identiﬁcation

Some courts restricting expert testimony have found jury instructions

regarding the fallible nature of eyewitness identiﬁcations to be an accept-

able substitute for expert testimony.

At the conclusion of a criminal trial,

Newsome v. McCabe, 319 F.3d 301 (7th Cir. 2003).

See, e.g., Bureau of Justice Statistics, “Indigent Defense,” available at: http://www.bjs.gov/

index.cfm?ty=pbdetail&iid=995.

470 U.S. 68, 82–83 (1985). Even if an indigent defendant receives funding to retain an

expert, the judge may ultimately decide that the expert testimony is not admissible at trial.

18 U.S.C. § 3006A(e)(1).

See, e.g., U.S. v. Jones, 689 F.3d 12, 20 (1st Cir. 2012) (“The judge was fully entitled to

conclude that this general information could be more reliably and efﬁciently conveyed by

instructions rather than through dueling experts.”).

Identifying the Culprit: Assessing Eyewitness Identification

THE LEGAL FRAMEWORK 41

the trial judge can instruct jurors on the factors that may result in an erro-

neous identiﬁcation while also offering instructions on the legal principles

jurors must apply when assessing the factual record. Such instructions may

be given when the witness testiﬁes. Judges tend to rely on model or pattern

instructions, because any departure from these standard instructions may

be a ground for appellate reversal.

The New Jersey Supreme Court viewed jury instructions as preferable

to expert testimony.

The New Jersey instructions adopted, following the

Henderson decision, are by far the most detailed set of jury instructions

regarding eyewitness identiﬁcation evidence. Traditionally, instructions re-

garding eyewitness identiﬁcations have been brief and remind the jurors to

consider the following: (1) the credibility of an eyewitness is like that of any

other witness and (2) any eyewitness identiﬁcation is part of the prosecu-

tor’s burden of proof in a criminal case.

Many state courts have held that,

although general jury instructions regarding credibility and the burden of

proof are appropriate, more speciﬁc instructions on eyewitness identiﬁca-

tions are considered an inappropriate judicial comment on the evidence.

Following the U.S. Supreme Court’s decision in Manson v. Brathwaite,

some state courts supplemented their jury instructions by including the ﬁve

reliability factors named by the Supreme Court.

In 1972, in U.S. v. Telfaire, the D.C. Circuit Court of Appeals adopted

a set of inﬂuential model jury instructions to be used in appropriate federal

cases involving eyewitness identiﬁcations.

The instructions emphasized

the following:

You must consider the credibility of each identiﬁcation witness in the same

way as any other witness, consider whether he is truthful, and consider

The New Jersey Supreme Court indicated: “Jury charges offer a number of advantages:

they are focused and concise, authoritative (in that juries hear them from the trial judge, not

a witness called by one side), and cost-free; they avoid possible confusion to jurors created by

dueling experts; and they eliminate the risk of an expert invading the jury’s role or opining on

an eyewitness’ credibility.” Henderson, 27 A.3d at 925.

New Jersey courts used such instructions a decade before Henderson. See, e.g., State

v. Robinson, 165 N.J. 32, 46–47 (N.J. 2000). Some states have also approved instructions

informing the jury that there may be an “independent source” for an in-court identiﬁcation.

See, e.g., State v. Cannon, 713 P.2d 273, 281 (Ariz. 1985).

Brodes v. State, 279 Ga. 435, 439 & n.6 (Ga. 2005) (surveying state case law).

State v. Tatum, 219 Conn. 721 (1991).

U.S. v. Telfaire, 469 F.2d 552, 558 (D.C. Cir. 1972). Some federal courts follow that ap-

proach, while others adopt a “ﬂexible approach.” See, e.g., United States v. Luis, 835 F.2d

37, 41 (2d Cir. 1987). Some more recent federal model instructions include added detail,

reﬂecting variables such as stress and cross-race identiﬁcations. See, e.g., Third Circuit Model

Criminal Jury Instructions, 4.15 (Jan. 2014), available at: http://www.ca3.uscourts.gov/sites/

ca3/ﬁles/2013%20Chapter%204%20ﬁnal%20revised.pdf.

Identifying the Culprit: Assessing Eyewitness Identification

42 IDENTIFYING THE CULPRIT

whether he had the capacity and opportunity to make a reliable observa-

tion on the matter covered in his testimony.

The Telfaire instructions departed from the brief traditional instruction

by adding that the jury should consider factors related to the initial sighting,

including “how long or short a time was available, how far or close the

witness was, how good were lighting conditions, [and] whether the witness

had had occasion to see or know the person in the past.” The decision also

noted that an identiﬁcation is more reliable if the witness is able to pick the

defendant out of a group, rather than at a showup, and that the jury should

consider the length of time between the crime and the identiﬁcation.

Some states have adopted cautionary instructions on speciﬁc issues

related to eyewitness identiﬁcation evidence. In State v. Ledbetter, the

Connecticut Supreme Court ordered lower courts to use a special instruc-

tion in cases in which law enforcement failed to instruct the eyewitness

that the perpetrator may or may not be present in a lineup.

The Georgia

Supreme Court concluded in 2005 that one particular use of the Manson v.

Brathwaite factors must no longer be permitted: “we can no longer endorse

an instruction authorizing jurors to consider the witness’ certainty in his/

her identiﬁcation as a factor to be used in deciding the reliability of that

identiﬁcation.”

Other courts have done the same.

In 1999, the New

Jersey Supreme Court ruled in State v. Cromedy that instructions on cross-

racial identiﬁcations are required in certain cases.

Expert testimony on eyewitness memory and identiﬁcations appears to

have many advantages when used as a method to explain relevant scientiﬁc

framework evidence to the jury. However, when expert testimony is not

available to the defense, jury instructions may be a preferable alternative

means to inform the jury of the ﬁndings of scientiﬁc research in this area.

U.S. v. Telfaire, 469 F.2d at 559.

Id. at 558.

State v. Ledbetter, 275 Conn. 534, 579–580 (2005) (The instruction reads, in part, “the

individual conducting the procedure either indicated to the witness that a suspect was present

in the procedure or failed to warn the witness that the perpetrator may or may not be in the

procedure. Psychological studies have shown that indicating to a witness that a suspect is pres-

ent in an identiﬁcation procedure or failing to warn the witness that the perpetrator may or

may not be in the procedure increases the likelihood that the witness will select one of the indi-

viduals in the procedure, even when the perpetrator is not present. Thus, such behavior on the

part of the procedure administrator tends to increase the probability of a misidentiﬁcation.”)

Brodes, 279 Ga. at 442.

See, e.g., supra Commonwealth v. Payne, 426 Mass. 692 (1998); State v. Romero, 191

N.J. 59 (2007).

State v. Cromedy, 158 N.J. 112 (1999); see also Innocence Project, “Know the Cases:

McKinley Cromedy,” available at: http://www.innocenceproject.org/Content/McKinley_

Cromedy.php.

Identifying the Culprit: Assessing Eyewitness Identification

THE LEGAL FRAMEWORK 43

Brief instructions may not, however, provide sufﬁcient guidance to explain

the relevant scientiﬁc evidence to the jury, but lengthy instructions may be

cumbersome and complex.

More research is warranted to better understand how best to com-

municate to jurors the factors that may affect the validity of eyewitness

testimony and support a more sensitive discrimination of the strengths and

weaknesses of eyewitness testimony in individual cases. Indeed, research

ﬁndings on the effectiveness of jury instructions on assessment of eyewitness

identiﬁcation evidence have been mixed. In general, such studies ﬁnd that

jury instructions cause jurors to become more suspicious of all eyewitness

identiﬁcation evidence.

A recent study of the effect of the New Jersey jury

instructions used in Henderson found that the instructions reduced juror re-

liance on both strong and weak eyewitness identiﬁcation evidence.

Among

the few studies ﬁnding that jury instructions succeed in increasing jurors’

sensitivity to the strength of such evidence are those that study the effect of

jury instructions presented before the eyewitness testimony rather than at

the end of the case before deliberation.

Such studies also have examined

instructions that use visual aids rather than rely on a judge’s recitation of

written instructions.

In addition, research studies might explore the use

of videotape as an alternative way to present such information

and the

effects of moving jury instructions to precede the introduction of the testi-

mony by the eyewitness.

For a review of this research, see K. A. Martire and R. I. Kemp, “The Impact of Eyewitness

Expert Evidence and Judicial Instruction on Juror Ability to Evaluate Eyewitness Testimony,”

Law and Human Behavior 33:225–236, 226 (reviewing studies of jury instructions on eyewit-

ness identiﬁcation and concluding that increased skepticism and confusion is a common re-

sult); see also J. L. Devenport, C. D. Kimbrough, and B. L. Cutler, “Effectiveness of traditional

safeguards against erroneous conviction arising from mistaken eyewitness identiﬁcation,” in

Expert testimony on the psychology of eyewitness identiﬁcation, ed. B. L. Cutler (New York:

Oxford University Press, 2009), 51–68 (summarizing research studying the Telfair jury instruc-

tion and concluding that “cautionary jury instructions may be an ineffective safeguard against

erroneous convictions resulting from mistaken eyewitness identiﬁcations.”).

A. P. Papailiou, D. V. Yokum, C. T. Robertson, “The Novel New Jersey Eyewitness In-

struction Induces Skepticism But Not Sensitivity,” August 2014, available at: http://papers.

ssrn.com/sol3/papers.cfm?abstract_id=2475217.

See, e.g., N. B. Pawlenko, M. A. Safer, R. A. Wise, and B. Holfeld, “A Teaching Aid for

Improving Jurors’ Assessments of Eyewitness Accuracy,” Applied Cognitive Psychology 27(2):

190–197. Other studies are reviewed in Martire and Kemp, supra note 105 at 226.

Pawlenko et al., supra note 107.

For an example of videotaped instructions, see Federal Judicial Center, The Patent Process:

An Overview for Jurors, available at: http://www.youtube.com/watch?v=ax7QHQTbKQE.

Identifying the Culprit: Assessing Eyewitness Identification

44 IDENTIFYING THE CULPRIT

CONCLUSION

The Manson v. Brathwaite test under the Due Process Clause of the

U.S. Constitution set out the modern test that regulates the fairness and

the reliability of eyewitness identiﬁcation evidence. The test evaluates the

“reliability” of eyewitness identiﬁcations using factors derived from prior

rulings and not from empirically validated sources. It includes factors that

are not diagnostic of reliability and treats factors such as the conﬁdence of

a witness as independent markers of reliability when, in fact, it is now well

established that conﬁdence judgments may vary over time and can be pow-

erfully swayed by many factors. The best guidance for legal regulation of

eyewitness identiﬁcation evidence comes not, however, from constitutional

rulings, but from the careful use and understanding of scientiﬁc evidence

to guide fact-ﬁnders and decision makers.

Identifying the Culprit: Assessing Eyewitness Identification

Basic Research on Vision and Memory

ccurate eyewitness identiﬁcation requires that a witness to a crime

correctly sense, perceive, and remember objects and events that

occurred and recall them later. The veracity of the witness’ identi-

ﬁcation thus depends on the limits of sensation, perception, and memory.

Recent scientiﬁc studies have yielded great advances in our understanding

of how vision and memory work. This chapter provides a brief overview

of current knowledge, identiﬁes areas in which vision and memory are

imperfect, and describes implications for the accuracy of eyewitness iden-

tiﬁcation. These implications, in turn, have guided much of the applied

research on this topic (see Chapter 5) and provide a general framework for

the recommendations made herein (see Chapter 6).

VISION AND MEMORY IN CONTEXT

This chapter begins by offering a concrete example to place the body

of basic scientiﬁc research on vision and memory in context so as to better

communicate its relevance to eyewitness identiﬁcation. In the sections that

follow the example, the different functional steps of the sequence (high-

lighted in italics) are dissected in some detail, with special reference to its

limitations and the ways in which it may fail to deliver accurate eyewitness

identiﬁcation.

While returning home late, you hear a mufﬂed scream from around the

street corner. Seconds later, you come face-to-face with a man turning the

corner and moving swiftly past you. Instantaneously, properties of the

Identifying the Culprit: Assessing Eyewitness Identification

46 IDENTIFYING THE CULPRIT

scene are conveyed to you through patterns of light cast on the backs of

your eyes and sensed by photoreceptors in your retina. Only a fraction of

the information sensed is selected for further processing; in this case you

focus your attention on certain features of the man’s face. Those features

are integrated and interpreted to yield a coherent percept of the man.

As you round the corner, you perceive, through an identical process, the

victim slumped lifelessly against a wall. You quickly grasp the meaning of

these perceptual experiences, and they immediately elicit both cognitive

and visceral components (e.g., increased heart rate) of fear and anxiety.

Your percepts are initially encoded in short-term working memory, where

content is limited and labile. Your elevated level of arousal may cause

interference and some loss of content, but with time and recognition of

the importance of the experience, your percepts are consolidated into long-

term memory. Long-term memories are maintained in storage but subject

to ongoing updates and modiﬁcations resulting from new experiences and

perhaps distortions caused by sustained levels of stress.

At a later date, you are asked to look at a police lineup that includes a

suspect apprehended near the crime scene. Visual features of the men in the

lineup are sensed, selectively attended, and perceived, using the same visual

processes engaged on the night in question. Some of these features—the

high brow and sharp cheekbones of one man in the lineup—elicit retrieval

of memories of your visual experiences on the night of the crime. The si-

multaneously perceived and retrieved experiences are implicitly compared,

leading to a cycle of greater visual scrutiny of the man in front of you and

retrieval of additional details of the original percept. The context of the

lineup procedure, the sight of the man, and the retrieved memories trigger

latent emotions and anxiety, which may interfere with your comparison

of percept and memory. Eventually, the comparison reaches your internal

criterion for identiﬁcation: You decide, with an implicit level of certainty,

that your current visual percept and the percept from the night of the

crime were caused by the same external source (the man now in front of

you), and you assert that you have identiﬁed the person you witnessed at

the crime scene.

VISION

Functional Processes of Vision

To understand the contributions and limitations of vision to eyewit-

ness identiﬁcation, it is useful to consider the workings of three functional

stages of visual processing—sensation, attention, and perception—bearing

in mind that they comprise highly interdependent elements of a continuous

operation. Sensation is the initial process of detecting light and extracting

basic image features. Sensations themselves are evanescent, and only a small

Identifying the Culprit: Assessing Eyewitness Identification

BASIC RESEARCH ON VISION AND MEMORY 47

fraction of what is sensed is actually perceived. Attention is the process by

which information sensed by the visual system is selected for further pro-

cessing. Perception is the process by which attended visual information is

integrated, linked to environmental cause, made coherent, and categorized

through the assignment of meaning, utility, value, and emotional valence. In

addition, memories and emotions resulting from prior experiences with the

world can inﬂuence all stages of visual processing and thus deﬁne a thread

that weaves throughout the following discussions.

All of the functional processes of vision are beset by noise, which

affects the quality and types of information accessible from the visual en-

vironment, and bears heavily on the validity of eyewitness identiﬁcation.

Before considering the processes of sensation, attention, and perception in

greater detail, consideration is given to the concept of noise in visual pro-

cessing and to ways of interpreting its impact on visual experience.

The Fundamental Role of Noise

Vision is usefully understood as the process of detecting informative

signals about the external world and using those signals to recognize ob-

jects, make decisions, and guide behavior. As with any signal detection,

there are occasionally factors that lead to uncertainty on the part of the

observer about whether a particular signal is present. These factors are

generically termed noise, following the deﬁnition used in electronic signal

transmission, in which noise refers to random or irrelevant elements that

interfere with detection of coherent and informative signals. In vision, noise

comes from a variety of sources, some associated with the structure of the

visual environment (e.g., occluding surfaces, glare, shadows), some inher-

ent to the optical and neuronal processes involved (e.g., scattering of light

in the eye), some reﬂecting sensory content not relevant to the observer’s

goals (e.g., a distracting sign or a loud sound), and some originating with

incorrect expectations derived from memory. Consider, for example, the

seemingly simple problem of detecting a green light while waiting at a

trafﬁc signal. In this case, your ability to “see” the green light may be

compromised by glare or dust on your windshield, by poor visual acuity,

by your eyes having been aimed instead at the driver of the adjacent car, by

the presence of other (irrelevant) colored lights in your ﬁeld of view (e.g., a

trafﬁc signal at a different intersection or the lights of a nearby restaurant),

by a cell phone conversation, or by the news on the car radio. The signiﬁ-

cance of this view for eyewitness identiﬁcation is profound, as it helps us

to realize that the accuracy of information about the environment—the face

Identifying the Culprit: Assessing Eyewitness Identification

48 IDENTIFYING THE CULPRIT

of a criminal, for example—gained through vision is necessarily, and often

sharply, limited by noise.

The fact that vision is noise-limited suggests a familiar statistical frame-

work—signal detection theory—for assessing and understanding the effects

of noise on visual perception and recognition ability.

Signal detection

theory has long been successfully applied to analogous problems in elec-

tronic signal reception.

To illustrate these principles as applied to sensory

processing, consider the problem of detecting a vibrating cell phone in

your pocket. Anyone who has operated a cell phone in vibrate mode will

be familiar with two types of signal detection errors: (1) the occasional

sense that the phone is vibrating in your pocket, only to discover that it is

not, and, conversely, (2) the phone call that is sometimes missed because

you attribute the vibration to some other cause. Signal, in this example, is

a subtle tactile stimulus resulting from an incoming phone call. Noise, in

this example, is all of the other things in your environment that may also

lead to subtle tactile stimulation, such as vibration of your car seat, a shift

of keys in your pocket, or the touch of another person.

Signal detection theory posits that there are three main factors that

determine whether a signal will be detected: (1) the distribution of stimuli

(e.g., the variety of stimulus magnitudes) that reﬂect noise only, (2) the

distribution of stimuli that reﬂect signal, and (3) the observer’s criterion

for “deciding” that a speciﬁc stimulus resulted from noise sources or sig-

nal. An important factor for the ﬁdelity of signal detection is the degree to

which noise and signal distributions overlap with one another. In the case

of the vibrating cell phone, if the distributions of tactile stimuli resulting

from noise and signal overlap, as is often the case, then there will always

be some cases in which you believe the phone is vibrating when it is not

(noise stimuli attributed to signal source), and there will be some cases in

which the phone is vibrating and you miss the call (signal stimuli attributed

to noise source).

The third factor that inﬂuences signal detection in the presence of noise

is the observer’s decision criterion, which is simply the value (e.g., stimulus

amplitude) above which a stimulus is attributed to signal, and below which

a stimulus is attributed to noise. In the same sense that your car radio is

programmed to “decide” (and allow you to hear) when informative patterns

of electromagnetic radiation (signal) are sufﬁciently different from random

ﬂuctuations (noise), an observer adopts a criterion for deciding whether a

W. S. Geisler, “Sequential Ideal-Observer Analysis of Visual Discriminations,” Psychologi-

cal Review 96(2): 267–314 (1989).

D. M. Green and J. A. Swets, Signal Detection Theory and Psychophysics (New York:

Wiley, 1966).

W. W. Peterson, T. G. Birdsall, and W. C. Fox, “The Theory of Signal Detectability,”

Proceedings of the IRE Professional Group on Information Theory 4(4): 171–212 (1954).

Identifying the Culprit: Assessing Eyewitness Identification

BASIC RESEARCH ON VISION AND MEMORY 49

stimulus is caused by a signal or is simply a manifestation of noise. This

criterion reﬂects the level of precision acceptable for the observer’s needs,

given uncertainty about whether a given stimulus reﬂects a real signal.

In practice, the criterion

used is determined by a host of factors unique

to the circumstances, including psychological and social demands and be-

havioral goals. These factors collectively determine the relative “costs” of

incorrect attributions of signal as noise (“misses”) and of noise as signal

(“false alarms”).

If an individual places high value on not missing a phone call, then she

or he will adopt a very liberal criterion, in which all stimuli reﬂecting real

incoming calls (signal) are successfully detected, but many noise stimuli

(e.g., shifting keys in a pocket) are erroneously (and frustratingly) believed

to be incoming calls. By contrast, if an individual places little value on de-

tecting incoming phone calls, she or he will adopt a conservative criterion,

in which many calls are missed and noise stimuli rarely elicit an effort to

answer the phone, which may be of value to the individual who wishes to

avoid distraction.

The example of the signal detection logic used for the vibrating cell

phone applies similarly to all aspects of visual perceptual experience, in-

cluding the conditions of witnessing criminal events. The uncertainty about

visual events caused by manifold sources of noise will inevitably lead to

inaccurate visual perceptual experiences, which result from conditions in

which an observer fails to detect a critically informative stimulus as “real”

(attributing the stimulus instead to a source of noise) or conﬁdently per-

ceives a noise stimulus to have originated from an informative source.

The latter instance is problematic because it increases the likelihood that

observers will unwittingly “construct,” on the basis of expectations derived

from memory and situational context, perceptual experiences to account

for noise erroneously interpreted as signal.

What follows from this consideration of uncertainty and decision cri-

teria for visual perception is that the actual impact of factors that limit the

amount of visual information available to an eyewitness (factors considered

in more detail below) will depend on the criterion adopted. The criterion

may reﬂect the values and prejudices of the eyewitness, his or her motiva-

tional and emotional state, and a variety of behavioral goals. In principle,

the observer’s criterion can be altered by instruction or incentives, but it is

important to note that the criterion held by an observer witnessing a crime

scene cannot be anticipated, nor can it be altered after the fact. It is an

“estimator variable,” which simply needs to be recognized and understood

when evaluating eyewitness reports. By contrast, the decision criterion held

The criterion is sometimes referred to as bias.

Identifying the Culprit: Assessing Eyewitness Identification

50 IDENTIFYING THE CULPRIT

by an observer at the time of identiﬁcation can be controlled, and there may

be valid reasons for doing so (see Chapter 5).

In the following discussions of sensation, attention, and perception, the

various means and conditions under which many different types of noise

introduce uncertainty in visual signal detection (and thus fundamentally

limit the accuracy of eyewitness identiﬁcation) are addressed.

Visual Sensation

When an observer views an object of any sort (such as a person) or

events involving the object (a criminal act), patterns of light reﬂected from

the environment are focused by the lens at the front of the eye and projected

onto the back surface of the eye (the retina) to form the retinal image. Light

in the image is initially “sensed” by the activation of photoreceptors, and

early stages of sensory processing function to detect spatial and temporal

contrast along a number of dimensions, including intensity and wavelength

of light.

These contrast measurements are integrated by subsequent pro-

cessing stages in the brain to yield representations of basic image features,

or primitives, such as oriented image contours.

Several sources of noise, or factors that limit the ratio of signal to

noise, can restrict the visual information accessible to these early sensory

processes. Some factors are inherent to the visual system and largely un-

controllable (e.g., the scattering of light by the ﬂuid and tissues of the eye)

and can be exacerbated by common observer-speciﬁc visual deﬁcits (e.g.,

myopia, poor contrast sensitivity, or color blindness). Others factors are

dependent on viewing conditions (e.g., the effects of viewing time and level

of illumination).

Both of these types of factors predictably inﬂuence the

quantity of information—the visual signal strength—that a viewer gains

from a visual scene, and thus the degree to which the perceptual experi-

L. Mickes, H. D. Flowe, and J. T. Wixted, “Receiver Operating Characteristic Analysis

of Eyewitness Memory: Comparing the Diagnostic Accuracy of Simultaneous and Sequential

Lineups,” Journal of Experimental Psychology: Applied 18(4): 361–376 (2012).

M. Meister and M. Tessier-Lavigne, “Low-level Visual Processing: The Retina,” in Prin-

ciples of Neuroscience, 5th Edition, ed. E. Kandel, J. H. Schwartz, T. M. Jessell, S. A.

Siegelbaum, and A. J. Hudspeth (New York: McGraw-Hill Professional, 2012), 577–601.

C. D. Gilbert, “Intermediate-level Visual Processing and Visual Primitives,” in Principles of

Neuroscience, 5th Edition, ed. E. Kandel, J. H. Schwartz, T. M. Jessell, S. A. Siegelbaum, and

A. J. Hudspeth (New York: McGraw-Hill Professional, 2012), 602–620.

D. G. Pelli, “Uncertainty Explains Many Aspects of Visual Contrast Detection and Dis-

crimination” Journal of the Optical Society of America A2(9): 1508–32 (1985). D. G. Pelli,

“The Quantum Efﬁciency of Vision,” in Vision: Coding and Efﬁciency, ed. C. Blakemore

(Cambridge: Cambridge University Press, 1990), 3–24. G. Sperling, “The Information Avail-

able in Brief Visual Presentations,” Psychological Monographs: General and Applied 74(11,

Whole No. 498): 1–29 (1960).

Identifying the Culprit: Assessing Eyewitness Identification

BASIC RESEARCH ON VISION AND MEMORY 51

ence can accurately reﬂect the properties of the external world.

At the ex-

treme, short viewing times and low levels of illumination simply reduce the

number of correlated photons reaching the retina to the point where they

scarcely exceed photon noise, and uncertainty is very high.

At slightly

longer viewing times and greater illumination levels, signal-to-noise levels

improve, but there may remain marked limits on visual sensitivity. Visual

acuity, for example, which is a measure of the ability to resolve the ﬁne

spatial details of a visual pattern, is known to decline signiﬁcantly with

decreases in illumination.

Signal-to-noise loss can depend on the direction of the observer’s gaze.

Visual acuity is highest at the observer’s center of gaze. The center is the

part of your visual system that is used for ﬁne sensing, such as reading

or scrutinizing faces in a social context. Acuity drops off markedly with

angular distance from this center, such that the quality and quantity of

information sensed a mere 10 degrees from center are far less than what is

available at the center of gaze.

Under unrestricted viewing conditions, the movements of the eyes

largely overcome the effects of gaze direction. However, under the viewing

conditions associated with a typical crime, this source of noise may place

severe limitations on the ability of the observer to sense key pieces of infor-

mation that are not present at the center of gaze. To appreciate the impact

of these limitations, consider that patients with macular degeneration are

effectively blinded in the region of the visual ﬁeld possessing highest acuity,

and must rely instead on the much-reduced quality of visual information

gained from the peripheral visual ﬁeld. To compensate for this clinical loss,

images and text must be greatly magniﬁed to enable comprehension—an

option that is clearly not available to an eyewitness.

Visual Attention

Light falling on all parts of the retina is available to be sensed—and

must be sensed for it to be available for further processing—but only a

G. Sperling, “A Signal-to-Noise Theory of the Effects of Luminance on Picture Memory:

Comment on Loftus,” Journal of Experimental Psychology: General 115(2): 189–192 (1986).

S. Hecht, S. Schlaer, and M. H. Pirenne, “Energy, Quanta, and Vision,” Journal of General

Physiology 25(6): 819–840 (1942).

P. W. Cobb, “The Inﬂuence of Illumination of the Eye on Visual Acuity,” American Journal

of Physiology 29: 76–99 (1911). S. Hecht, “A Quantitative Basis for the Relation Between Vi-

sual Acuity and Illumination,” Proceedings of the National Academy of Sciences 13: 569–574

(1927). S. Shlaer, “The Relation Between Visual Acuity and Illumination,” Journal of General

Physiology 21 (2): 165–188 (1937).

H. Strasburger, I. Rentschler, and M. Jüttner, “Peripheral Vision and Pattern Recognition:

A Review,” Journal of Vision 11(5):13, 1–82 (2011).

Identifying the Culprit: Assessing Eyewitness Identification

52 IDENTIFYING THE CULPRIT

small fraction of the information sensed reaches awareness or is used by the

observer for recognition, action, or storage in memory. This limited access

to visual sensory information is a product of selective attention.

Attention

is an active process that can be directed by external factors—visual attri-

butes with high salience, such as a bright light or an unfamiliar object—or

by internal control.

If you are searching for a coffee cup, for example,

you may explicitly direct your attention to the table where it was last seen.

Attention can be directed to different types of image content, including spe-

ciﬁc locations in space,

speciﬁc image features (such as a speciﬁc color),

or to speciﬁc objects (such as the coffee cup).

Attended image content is transiently enhanced to increase the ﬁdelity

of visual experience.

Attention interacts with sensory processing, for ex-

ample, by selectively enhancing contrast

and potentially overcoming low

signal-to-noise levels resulting from limited viewing time or illumination.

The effects of attention on contrast enhancement can be potentiated further

when attention is commanded by emotionally laden stimuli.

Image con-

W. James, Principles of Psychology (New York: Henry Holt, 1890); H. Pashler, J. John-

ston, and E. Ruthruff, “Attention and Performance,” Annual Review of Psychology 52:

629–651 (2001).

M. I. Posner, “Orienting of Attention,” Quarterly Journal of Experimental Psychology

32: 3–25 (1980).

Ibid.

A. F. Rossi and M. A. Paradiso, “Feature-speciﬁc Effects of Selective Visual Attention,”

Vision Research 35(5): 621–634 (1995).

J. Duncan, “Selective Attention and the Organization of Visual Information,” Journal of

Experimental Psychology: General 113(4): 501–517 (1984).

H. Pashler, J. Johnston, and E. Ruthruff, “Attention and Performance,” Annual Review

of Psychology 52: 629–651 (2001).

M. Carrasco et al.,“Attention Alters Appearance” Nature Neuroscience 7: 308–313

(2004).

M. I. Posner, C. R. Snyder, and B. J. Davidson, “Attention and the Detection of Signals,”

Journal of Experimental Psychology 109(2): 160–174 (1980). M. Carrasco and B. McElree,

“Covert Attention Accelerates the Rate of Visual Information Processing,” Proceedings of

the National Academies of Science 98(9): 5363–5367 (2001). Y. Yeshurun and M. Carrasco,

“Attention Improves or Impairs Visual Performance by Enhancing Spatial Resolution,” Nature

396, 72–75 (1998). M. Carrasco et al., “Covert Attention Increases Spatial Resolution with

or without Masks: Support for Signal Enhancement,” Journal of Vision 2(6): 467–79 (2002).

E. Blaser et al., “Measuring the Ampliﬁcation of Attention,” Proceedings of the National

Academies of Science 96(20): 11681–11686 (1999). K. Anton-Erxleben and M. Carrasco,

“Attentional Enhancement of Spatial Resolution: Linking Behavioural and Neurophysiological

Evidence,” Nature Reviews Neuroscience 14(3):188–200 (2013). J. W. Couperus and G. R.

Mangun, “Signal Enhancement and Suppression During Visual-Spatial Selective Attention,”

Brain Research 1359:155–177 (2010).

E. A. Phelps, S. Ling, and M. Carrasco, “Emotion Facilitates Perception and Potentiates:

The Perceptual Beneﬁts of Attention,” Psychological Science 17(4): 292 (2006).

Identifying the Culprit: Assessing Eyewitness Identification

BASIC RESEARCH ON VISION AND MEMORY 53

tent not falling within the focus of attention is processed with less ﬁdelity.

In some cases, unattended content is effectively invisible: It does not reach

awareness, it is not perceived, and it is not available for use in guiding deci-

sions or actions, or for storage in memory.

Different pieces of visual information compete for selection,

as their

attributes of physical salience, location in space, novelty, and relevance

to the observer’s needs and behavioral goals are always changing.

The

outcome of the competition is highly susceptible to noise (in this instance,

noise is deﬁned as uncontrolled factors that bias the focus of attention and

create uncertainty about the content of a visual scene), because the infor-

mational content of the visual image vastly exceeds what can be attended

at any point in time. The implications of such noise for eyewitness identi-

ﬁcation are profound. An observer must “select” what to attend to, often

within a short window of time, without advance warning, in the presence

of many novel objects and events, and under such confounding inﬂuences

as anxiety and fear.

The signal detection framework is readily adaptable to the problem

of noise in visual attention and provides some insights into the limits of

attentional selection in the presence of noise.

In essence, this signal detec-

tion approach quantiﬁes the extent to which multiple items competing with

one another for attention affect attentional enhancement for any one of the

items.

Reductions in efﬁciency are common under such noise conditions.

Indeed, sensitivity to unattended items can be markedly reduced under

conditions of high “perceptual load,” in which there are many objects si-

Posner, Snyder, and Davidson, “Attention and the Detection of Signals.” Y. Yeshurun

and M Carrasco, “Attention Improves or Impairs Visual Performance by Enhancing Spatial

Resolution,” Nature 396: 72–75 (November 1998).

A. Mack and I. Rock, Inattentional Blindness (Cambridge, MA: MIT Press, 1998).

R. Desimone and J. Duncan, “Neural Mechanism of Selective Visual Attention,” Annual

Review of Neuroscience 18: 193–222 (March 1995).

J. M. Wolfe and T. S. Horowitz, “What Attributes Guide the Deployment of Visual Atten-

tion and How Do They Do It?” Nature Reviews Neuroscience 5: 495–501 (June 2004). H. E.

Egeth and S.Yantis, “Visual Attention: Control, Representation, and Time Course,” Annual

Review of Psychology 48(1): 269–297 (February 1997). M. I. Posner, “Orienting in Atten-

tion,” Quarterly Journal of Experimental Psychology 32(1): 3–25 (1980). A. Treisman and

G. Gelade, “A Feature Integration Theory of Attention,” Cognitive Psychology 12(1):97–136

(January 1980). L. Itti and C. Koch, “A Saliency-based Search Mechanism for Overt and Co-

vert Shifts of Visual Attention,” Vision Research 40(10–12): 1489–1506 (June 2000).

G. Sperling and M. J. Melchner, “The Attention Operating Characteristic: Examples from

Visual Search,” Science 202(4365): 315–318 (October 1978). G. Sperling and B. A. Dosher,

“Strategy and Optimization in Human Information Processing,” in Handbook of Perception

and Human Performance, ed. K. Boff, L. Kaufman, and J. Thomas (New York: Wiley, 1986).

Ibid.

Identifying the Culprit: Assessing Eyewitness Identification

54 IDENTIFYING THE CULPRIT

multaneously competing for attention.

The spacing of items in the visual

ﬁeld also impacts visual sensitivity.

When objects are closely spaced, their

discriminability is reduced. One explanation offered for this “crowding

effect” is that the spacing of visual items is smaller than the resolution

of visual attention.

The visual phenomenon of crowding suggests that

a crime committed in a visually complex scene, such as a sporting event,

could easily place limits on the ability of a witness to accurately perceive

the facial features of a perpetrator.

A related consequence of attentional noise is that competing interests

can readily hijack the attentional focus. The technique of misdirection—

one of the original mainstays of performance magic—directs attention to

uninformative image content and exploits the invisibility of unattended

features.

The well-studied inattentional blindness effect is another ex-

ample of this phenomenon, in which attention that is pre-directed to one

behaviorally signiﬁcant property of a visual scene precludes awareness of

other features that also may be important.

(For a dramatic demonstration

of this effect, produced by Simons and Chabris,

see http://tinyurl.com/

inattentional-blindness.)

Inattentional blindness effects translate well to real-world interactions

between people. An individual can be surprisingly unaware of surreptitious

changes to the physical appearance of another person while engaged in con-

versation.

One demonstration of this phenomenon involved two strang-

ers (experimenter and pedestrian) in a brief face-to-face conversation on a

sidewalk. At some point in the conversation an opaque door was carried

between the two individuals, and another person with different appearance,

clothing, and voice quickly replaced the experimenter. More than half of

N. Lavie, “Perceptual Load as a Necessary Condition for Selective Attention,” Journal of

Experimental Psychology: Human Perception and Performance 21(3): 451–468 (June 1995).

J. W. Couperus, “Perceptual Load Inﬂuences Selective Attention Across Development,” De-

velopmental Psychology 47(5):1431–1439 (September 2011).

D. M. Levi, “Crowding—An Essential Bottleneck for Object Recognition: A Mini-review,”

Vision Research 48: 635–654 (2008).

J. Intriligator and P. Cavanagh, “The Spatial Resolution of Visual Attention,” Cognitive

Psychology 43: 171–216 (2001).

G. Kuhn et al., “Misdirection in Magic: Implications for the Relationship Between Eye

Gaze and Attention,” Visual Cognition 16(2–3): 391–405 (2008). S. L. Macknik, S. Martinez-

Conde, and S. Blakeslee, Sleights of Mind: What the Neuroscience of Magic Reveals About

Our Everyday Deceptions (New York: Henry Holt and Co., 2010).

A. Mack and I. Rock, Inattentional Blindness (Cambridge, MA: MIT Press, 1998). U.

Neisser and R. Becklen, “Selective Looking: Attending to Visually Speciﬁed Events,” Cognitive

Psychology 7(4): 480–494 (October 1975). D. Simons, “Attentional Capture and Inattentional

Blindness,” Trends in Cognitive Sciences 4(4): 147–155 (April 2000).

D. J. Simons and C. F. Chabris, “Gorillas in Our Midst: Sustained Inattentional Blindness

for Dynamic Events,” Perception 28: 1059–1074 (1999).

D. J. Simons and D. T. Levin, “Failure to Detect Changes to People During a Real-World

Interaction,” Psychonomic Bulletin and Review 5(4): 644–649 (1998).

Identifying the Culprit: Assessing Eyewitness Identification

BASIC RESEARCH ON VISION AND MEMORY 55

the participants (pedestrians) failed to notice that their conversation part-

ner had changed. This ﬁnding suggests that naturally occurring events that

brieﬂy divert attention have the potential to markedly impair the accuracy

of eyewitness identiﬁcations.

Attentional hijacking is particularly characteristic of stimuli that elicit

strong emotional responses, such as fear and arousal.

Visual stimuli that

trigger fear responses act as powerful external cues that command atten-

tion.

While this potentiates sensitivity to those stimuli, at the considerable

expense of sensitivity to others, it is often the case that the attended emo-

tional stimuli are not the ones with relevant informational content.

The

so-called weapon focus is a real-world case in point for eyewitness iden-

tiﬁcation, in which attention is compellingly drawn to emotionally laden

stimuli, such as a gun or a knife, at the expense of acquiring greater visual

information about the face of the perpetrator (see also discussion of weapon

focus in Chapter 5).

(One might argue that this is an adaptation that

beneﬁts immediate action or engagement with a threatening stimulus, but

is surely detrimental to one’s efforts to bear witness.)

Visual Perception

Visual perception is the conscious functional result of efforts to identify

the environmental causes of the pattern of light cast onto the back of the

eye.

Perception does not reﬂect the sensory world passively, as camera ﬁlm

detects patterns of light. On the contrary, visual perception is constructive

C. H. Hansen and R. D. Hansen, “Finding the Face in the Crowd: An Anger Superior-

ity Effect,” Journal of Personality and Social Psychology 54: 917–924 (1988). E. Fox et al.,

“Facial Expressions of Emotion: Are Angry Faces Detected More Efﬁciently?” Cognition and

Emotion 14(1): 61–92 (2000). R. Compton, “The Interface Between Emotion and Attention:

A Review of Evidence from Psychology and Neuroscience,” Behavioral and Cognitive Neu-

roscience Reviews 2(2): 115–129 (2003). R. L. Bannerman, E. V. Temminck, and A. Sahraie,

“Emotional Stimuli Capture Spatial Attention But Do Not Modulate Spatial Memory,” Vision

Research 65: 12–20 (15 July 2012).

J. A. Easterbrook, “The Effects of Emotion on Cue Utilization and the Organization of

Behavior,” Psychological Review 66(3): 183–201 (1959).

E. Ferneyhough et al., “Anxiety Modulates the Effects of Emotion and Attention on Early

Vision,” Cognition and Emotion 27(1): 166–176 (2013). G. Pourtois and P. Vuilleumier, “Dy-

namics of Emotional Effects on Spatial Attention in the Human Visual Cortex,” Progress in

Brain Research 156: 67–91 (2006).

T. Kramer, R. Buckhout, and P. Eugenio, “Weapon Focus, Arousal, and Eyewitness

Memory: Attention Must Be Paid,” Law and Human Behavior 14(2): 167–184 (1990). R. S.

Truelove, “Do Weapons Automatically Capture Attention,” Applied Cognitive Psychology

20(7): 871–893 (2006). E. F. Loftus, G. R. Loftus, and J. Messo, “Some Facts About ‘Weapon

Focus’,” Law and Human Behavior 11(1): 55–62 (1987).

W. James, Principles of Psychology (New York: Henry Holt, 1890). S. Harnad, ed., Cat-

egorical Perception: The Groundwork of Cognition (New York: Cambridge University Press,

1987). T. D. Albright, “Perceiving,” Daedalus (in press).

Identifying the Culprit: Assessing Eyewitness Identification

56 IDENTIFYING THE CULPRIT

and entails (1) integrating and segmenting attended attributes of the visual

image into objects, (2) complementing and interpreting the product with

expectations derived from memory of prior experiences with the world,

and (3) assigning meaning and emotional valence by reference to prior

knowledge of function and value.

All of these perceptual processes are

affected by noise. Because the things perceived are the things we place into

memory, perceptual noise can dramatically limit the accuracy of eyewitness

identiﬁcation.

The process of feature integration and interpretation may be dis-

torted by images of an object unique to a speciﬁc angle of view.

The

retinal pattern generated by a face viewed directly from the front differs

considerably—with changes in aspect ratio and relative placement of fa-

cial features—from that generated by a face viewed from an oblique side

angle. Viewing a face from an angle above or below center (as might be

the case if the criminal were standing over you, or below you on the stairs)

also yields retinal distortions of facial features. In this case, the distortions

prominently mimick facial gestures of smiling versus frowning, and perhaps

cause incorrect inferences about the emotional state of the person observed

and his or her intentions and motivations. (This distortion is the basis for

the Japanese Noh Theatre mask effect, in which a rigid mask tilted forward

leads to the appearance of a smile and backward leads to the appearance

of a frown—an effect you can simulate by simply looking into the mirror

and tilting your face up or down.)

Viewing conditions can also affect the perception of face, gender, and

age.

Investigators found that faces that were physically identical—and

particularly those bordering on androgyny—were perceived as unambigu-

ously male or female depending on where they appeared in the observer’s

visual ﬁeld. The spatial patterning of these effects was distinctive and

stable for each observer. Perceptual distortions of this sort are a source of

noise that may have important implications for the accuracy of eyewitness

identiﬁcation.

Perceptual distortions also may be introduced through memory recall.

C. D. Gilbert, “The Constructive Nature of Visual Processing,” in Principles of Neurosci-

ence, 5th Edition, ed. E. Kandel, J. H. Schwartz, T. M. Jessell, S. A. Siegelbaum, and A. J.

Hudspeth (New York: McGraw-Hill Professional, 2012). T. D. Albright, “On the Perception

of Probable Things: Neural Substrates of Associative Memory, Imagery, and Perception,”

Neuron 74 (2): 227–245 (2012).

W. G. Hayward and P. Williams, “Viewpoint Dependence and Object Discriminability,”

Psychological Science 11(1): 7–12 (2000).

M. J. Lyons et al., “The Noh Mask Effect: Vertical Viewpoint Dependence on Facial

Expression Perception,” Proceedings of the Royal Society B: Biological Sciences 267(1459):

2239–2245 (2000).

A. Afraz, M. Vaziri-Pashkam, and P. Cavanagh, “Spatial Heterogeneity in the Perception

of Face and Form Attributes,” Current Biology 20(23): 2112–2116 (2010).

Identifying the Culprit: Assessing Eyewitness Identification

BASIC RESEARCH ON VISION AND MEMORY 57

The way an observer experiences a visual scene—the setting, the people,

and the actions associated with a crime —is commonly inﬂuenced as much

by expectations from prior experience with the world as it is by the precise

patterns of light cast upon the retina. There are good reasons why this is

true. As noted above, the sensory input (the pattern of light received) is

often noisy, incomplete, and ambiguous, and memories of what is likely to

be out there, given the context, are called on to ﬁll in the blanks, recon-

cile ambiguities, and leave clear and coherent percepts.

This perceptual

completion is probabilistic.

It is an hypothesis, and the accuracy naturally

depends on the degree to which the observer’s expectations match the noisy

sensory data.

What is implied is that the same mechanism that grants the certainty of

perceptual experience in the face of noise and ambiguity is also capable of

implicitly fabricating content that does not correspond to external reality

and yet is experienced with no less certainty. Performance magic relies on

this constructive nature of perceptual experience, and that nature is also

the foundation for many visual illusions and forms of visual art.

In a

classic experiment that drives home the point, Bruner and Postman looked

at the ability of observers to recognize ‘‘trick’’ playing cards.

The trick

cards were created by altering the color of a given suit (e.g., a red seven

of spades). Observers were shown a series of cards with brief presenta-

tions. Some cards were trick, and the remainder normal. With astonish-

ing frequency, observers reported that the trick cards were normal. When

questioned, observers defended their reports, even after being allowed to

scrutinize the trick cards, thus demonstrating that learned properties of

the world are capable of sharply altering our experience and, moreover,

reinforcing our convictions about what we have seen, even in the face of

countermanding sensory evidence. In view of this inherent dependence of

perception on prior experiences and context—and, importantly, the fact

that the viewer is commonly none the wiser when perception differs from

Albright, “On the Perception of Probable Things.”

D. C. Knill and W. Richards, Perception as Bayesian Inference, ed. D. C. Knill and W.

Richards (Cambridge: Cambridge University Press, 1996). D. Kersten “High-level Vision as

Statistical Inference,” in The New Cognitive Neurosciences, 2nd Edition, ed. M. S. Gazzaniga

(Cambridge: MIT Press, 1999), 353–363. D. Kersten, P. Mamassian, and A. Yuille, “Object

Perception as Bayesian Inference,” Annual Review of Psychology 55: 271–304 (February

2004).

E. H. Gombrich, Art and Illusion. A Study in the Psychology of Pictorial Representation

(London: Phaidon 1960). T. D. Albright, “The Veiled Christ of Cappella Sansevero: On Art,

Vision and Reality,” Leonardo 46(1): 19–23 (2013). Macknik, Martinez-Conde, and Blakeslee,

Sleights of Mind: What the Neuroscience of Magic Reveals About Our Everyday Deceptions

(New York: Henry Holt and Co., 2010).

J. S. Bruner and L. Postman, “On the Perception of Incongruity: A Paradigm,” Journal of

Personality 18(2): 206–223 (1949).

Identifying the Culprit: Assessing Eyewitness Identification

58 IDENTIFYING THE CULPRIT

the “ground truth” of the external world—it appears that accurate eyewit-

ness identiﬁcation may be difﬁcult to achieve.

Additional noise (in this case deﬁned as uncertainty resulting from loss

of perceptual resolution) may result from the fact that visual perception is

categorical.

Although the objects of our experience vary broadly along

multiple sensory dimensions, we lump them into categories based upon

prior associations, many of which stem from common functions, physical

properties, meanings, or emotional valence. Apples in a basket or the many

typographic fonts for the letter “A” are visually distinct, yet we readily

perceive them as categorically identical. For most behavioral and cognitive

goals, perceptual processing is greatly simpliﬁed by treating all members

of a category as the same, despite their differences. It rarely matters, for

example, whether the apple we choose is dappled on one side or irregular

in shape, nor does the font used bear greatly on our ability to read. One

of the functional corollaries of categorical perception is that observers are

far better at discriminating between objects from different categories than

objects from the same category.

Evidence indicates that the structure of

object memory is also categorical, suggesting that perceived objects are

encoded in memory as a category type, often without speciﬁc detail.

Perceptual categorization naturally applies to faces.

We readily cat-

egorize faces by distinctions along the obvious dimensions of gender, age,

W. James, Principles of Psychology (New York: Henry Holt, 1980). S. Harnad, ed.,

Categorical Perception: The Groundwork of Cognition (New York: Cambridge University

Press, 1987).

R. Goldstone, “Inﬂuences of Categorization on Perceptual Discrimination,” Journal of

Experimental Psychology General 123(2): 178–200 (1994). R. Goldstone, Y. Lippa, and

R. M. Shiffrin, “Altering Object Representations Through Category Learning,” Cognition

78(1): 27–43 (2001).

E. Tulving, “Episodic and Semantic Memory,” in Organization of Memory, ed. E. Tulving

and W. Donaldson (New York: Academic Press, 1972), 381–403. L. K. Tyler et al., “Processing

Objects at Different Levels of Speciﬁcity,” Journal of Cognitive Neuroscience 16(3): 351–362

(2004). M. J. Farah and J. L. McClelland, “A Computational Model of Semantic Memory

Impairment: Modality Speciﬁcity and Emergent Category Speciﬁcity,” Journal of Experimen-

tal Psychology: General 120 (4): 339–357 (1991). C. Gerlach et al., “Categorization and

Category Effects in Normal Object Recognition: A PET Study,” Neuropsychologia 38(13):

1693–1703 (2000). G. W. Humphreys and E. M. Forde, “Hierarchies, Similarity, and Inter-

activity in Object Recognition: ‘Category-Speciﬁc’ Neuropsychological Deﬁcits,” Behavioral

and Brain Sciences 24(3): 453–476 (2001).

J. M. Beale and F. C. Keil, “Categorical Effects in the Perception of Faces,” Cognition

57(3): 217–239 (1995). D. T. Levin, “Classifying Faces by Race: The Structure of Face Cat-

egories,” Journal of Experimental Psychology: Learning, Memory, and Cognition 22(6):1364–

1382 (1996). D. T. Levin and J. Beale, “Categorical Perception Occurs in Newly Learned

Faces, Cross-Race Faces, and Inverted Faces,” Perception and Psychophysics 62: 386–401

(2000). M. A. Webster et al., “Adaptation to Natural Facial Categories,” Nature 428(6982):

557–561 (2004). Y. Lee et al., “Broadly Tuned Face Representation in Older Adults Assessed

by Categorical Perception,”Journal of Experimental Psychology: Human Perception and

Performance 40(3): 1060–1071 (2014).

Identifying the Culprit: Assessing Eyewitness Identification

BASIC RESEARCH ON VISION AND MEMORY 59

and race, but we also draw distinctions along dimensions such as skin tone,

hair color and style, presence and type of facial hair, such subtler factors

as shape of cheeks and jaw, and subjective qualities such as attractiveness.

The practical consequence of this for eyewitness identiﬁcation is that the

precision of a perceptual experience may be reduced within any of these cat-

egories, particularly because we typically witness criminal events for such

a brief period of time. The ensuing memory of the experience will likely

reﬂect that reduced precision, and the memory retrieved may regress to a

category prototype or to other exemplars of the perceived category.

The

witness may categorically perceive a square jawed man with a moustache,

but the ﬁne details needed for individuation of a suspect are neither per-

ceived nor encoded in memory. For example, although you may have seen

the iconic Marlboro Man countless times on billboards and in magazines,

it is unlikely that you could distinguish him in a lineup from other square

jawed mustachioed men.

MEMORY

Functional Processes of Memory

Conscious visual perceptual experiences, rendered by the processes

described in the previous section on vision, are commonly stored as declara-

tive memories, meaning that they can be consciously accessed and expressed

as knowledge about the world (as distinct from procedural memories, such

as motor skills).

Declarative memories are of two types, semantic and

episodic, reﬂecting a distinction between memories of meanings, facts,

and concepts versus memories of events (such as those witnessed during a

crime).

Declarative memories are conceptualized as involving three core

processes—encoding, storage, and retrieval—which refer to the placement

of items in memory, their maintenance therein, and subsequent access to

the stored information.

Like vision, memory is also beset by noise. Encoding, storage, and re-

membering are not passive, static processes that record, retain, and divulge

J. Huttenlocher, L. V. Hedges, and J. L. Vevea, “Why Do Categories Affect Stimulus Judge-

ment?” Journal of Experimental Psychology: General 129(2): 220–241 (2000). R. Goldstone,

Y. Lippa, and R. M. Shiffrin, “Altering Object Representations Through Category Learning,”

Cognition 78(1): 27–43 (2001).

W. James, Principles of Psychology (New York: Henry Holt, 1890). B. Milner, Physiologie

de l’hippocampe, ed. P. Passouant (Paris: Centre National de la Recherche Scientiﬁque, 1962),

257–272. L. R. Squire and J. Wixted, “The Cognitive Neuroscience of Human Memory since

H.M.,” Annual Review of Neuroscience 34: 259–288 (2011).

Tulving, “Episodic and Semantic Memory.”

E. Tulving, “Organization of Memory: Quo vadis?” in The Cognitive Neurosciences, ed.

M. S. Gazzaniga (Cambridge, MA: MIT Press, 1995), 839–847.

Identifying the Culprit: Assessing Eyewitness Identification

60 IDENTIFYING THE CULPRIT

their contents in an informational vacuum, unaffected by outside inﬂuences.

The contents cannot be treated as a veridical permanent record, like pho-

tographs stored in a safe. On the contrary, the ﬁdelity of our memories for

real events may be compromised by many factors at all stages of process-

ing, from encoding through storage, to the ﬁnal stages of retrieval. Without

awareness, we regularly encode events in a biased manner and subsequently

forget, reconstruct, update, and distort the things we believe to be true.

The following sections discuss memory encoding, storage, and retrieval,

with emphasis on the limits of these processes as they pertain to eyewitness

identiﬁcation. Emotions can strongly inﬂuence these processes of memory;

some speciﬁc actions are highlighted. The phenomenon of “recognition

memory” is also discussed. This refers to the speciﬁc type of memory re-

trieval in which a stimulus (e.g., a face) is used to probe memory, and the

rememberer (e.g., an eyewitness) must decide whether the strength of the

elicited memory evidence is sufﬁcient to declare that the stimulus was pre-

viously encountered or is novel. Recognition memory underlies eyewitness

identiﬁcation, as the witness must make a recognition decision.

Memory Encoding

Memory encoding refers to the process whereby perceived objects and

events are initially placed into storage. The encoding process involves two

stages, which are commonly distinguished by the quantity of information

stored, the duration of storage, and the susceptibility to interference.

Short-term or working memory is the conscious content of recent percep-

tual experiences or information recently recalled from long-term storage.

Information that remains at the focus of attention persists in and forms the

contents of short-term memory. This form of memory is of limited duration

J. T. Wixted, “The Psychology and Neuroscience of Forgetting,” Annual Review of

Psychology 55: 235–269 (2004). E. Tulving and D. M. Thomson, “Encoding Speciﬁcity and

Retrieval Processes in Episodic Memory,” Psychological Review 80(5): 352–373 (1973). Y.

Dudai, “Reconsolidation: The Advantage of Being Refocused,” Current Opinion in Neurobi-

ology 16(2): 174–178 (2006). E. F. Loftus, “Planting Misinformation in the Human Mind: A

30-Year Investigation of the Malleability of Memory,” Learning and Memory 12(4): 361–366

(2005). R. A. Bjork, “Interference and Memory,” in Encyclopedia of Learning and Memory,

ed. L. R. Squire (New York: Macmillan, 1992), 283–288. J. A. McGeoch, “Forgetting and

the Law of Disuse,” Psychological Review 39(4): 352–370 (1932). J. G. Jenkins and K. M.

Dallenbach, “Obliviscence during Sleep and Waking,” The American Journal of Psychology

35(4): 605–612 (1924). B. J. Underwood and L. Postman, “Extra-Experimental Sources of

Interference in Forgetting,” Psychological Review 67 (2): 73–95 (1960).

R. C. Atkinson and R. M. Shiffrin, “Human Memory: A Proposed System and its Control

Processes,” in The Psychology of Learning and Motivation (Volume 2), ed. K. W. Spence and

J. T. Spence (New York: Academic Press,1968), 89–195. W. James, Principles of Psychology

(New York: Henry Holt, 1890). A. Baddeley, “Working Memory: Looking Back and Look-

ing Forward,” Nature Reviews Neuroscience 4(10): 829–839 (2003). A. Baddley, Working

Memory (New York: Oxford University Press, 1986).

Identifying the Culprit: Assessing Eyewitness Identification

BASIC RESEARCH ON VISION AND MEMORY 61

and capacity

and labile, decaying quickly with time and easily disrupted

by other perceptual or cognitive processes.

Through cellular and molecu-

lar events that play out over time, the contents of short-term memories

may be encoded and consolidated into long-term memory,

which is more

enduring (albeit evolving with ongoing experience), and of greater capacity.

The structure of an individual’s full library of long-term declarative

memories can be thought of as a collection of associations between items of

speciﬁc semantic (e.g., the fact that that person X is a 34-year-old female) or

episodic content (e.g., the fact that person X was at location Y on the night

of the witnessed crime).

As the individual gains new experiences, long-

term declarative memories may be updated by adding new content to the

existing library or by forming new associations between existing content.

Memories are particularly labile during the encoding process. The con-

tents of short-term memory are limited and highly subject to interference by

subsequent sensory, cognitive, emotional, or behavioral events; the contents

can also be biased by prior knowledge, expectations, or beliefs, resulting in

a distorted representation of experience. Short-term memories of events that

happened early in a witnessed proceeding may simply be forgotten with the

passage of time or badly compromised by attention directed to subsequent

emotional events or cognitive and behavioral demands (e.g., anxiety, fear,

the need to escape). In such cases, the compromised information may never

be consolidated fully into long-term storage or that storage may contain

distorted content.

At the same time, the quality of encoding of stimuli that

are attended is commonly enhanced by highly emotional content.

G. A. Miller, “The Magical Number Seven,” The Psychological Review 63(2): 81–97

(1956).

J. Jonides et al., “The Mind and Brain of Short-Term Memory,” Annual Review of Psy-

chology 59: 193–224 (2008).

E. Kandel and L. Squire, Memory: From Mind to Molecules (New York: Scientiﬁc Ameri-

can Library, 2008).

J. R. Anderson, The Architecture of Cognition (Cambridge: Harvard University Press,

1983). J. R. Anderson and C. Lebiere, The Atomic Components of Thought (Mahwah:

Lawrence Erlbaum Associates, 1998).

M. P. Walker et al., “Dissociable Stages of Human Memory Consolidation and Reconsoli-

dation,” Nature 425: 616 (2003).

J. L. McGaugh, “Memory—a Century of Consolidation,” Science 287(5451): 248–251

(2000). J. L. McGaugh and B. Roozendaal, “Role of Adrenal Stress Hormones in Forming

Lasting Memories in the Brain,” Current Opinion in Neurobiology 12(2): 205–210 (2002).

K. N. Ochsner, “Are Affective Events Richly Recollected or Simply Familiar? The Experi-

ence and Process of Recognizing Feelings Past,” Journal of Experimental Psychology: General

129 (2): 242–261 (2000). D. Talmi, et al., “Immediate Memory Consequences of the Effect of

Emotion on Attention to Pictures,” Learning and Memory 15(2008): 172–182. E. A. Kens-

inger and D. L. Schacter, “Neural Processes Supporting Young and Older Adults’ Emotional

Memories,” Journal of Cognitive Neuroscience 7 (2008): 1–13. E. A. Phelps. “Emotion and

Cognition: Insights from Studies of the Human Amygdala,” Annual Review of Psychology

57: 27–53 (2006).

Identifying the Culprit: Assessing Eyewitness Identification

62 IDENTIFYING THE CULPRIT

Memory Storage

Memory storage refers to the long-term retention of information after

encoding. The stability of stored information is continuously challenged

and subject to modiﬁcation. We forget, qualify, or distort existing memories

as we acquire new perceptual experiences and encode new content and as-

sociations into memory.

Forgetting can be partially mitigated, and memories stabilized, by hab-

its of retrieval (or reactivation) and reconsolidation, which happen when-

ever we tell the story of our experiences.

Reactivation is not perfect. With

each implicit retrieval or explicit telling of a story, we may unconsciously

smooth over inconsistencies or modify content based on our prior beliefs,

the accounts of others, or through the lens of new information. We may

add embellishments that reﬂect opinions, emotions, or prejudices

rather

than observed facts; or we may simply omit disturbing content and pass

over ﬁne details.

A second threat to the stability of long-term memories is, ironically,

our life-long ability to learn new things. Because memory mechanisms are

inherently plastic throughout life, content stored for the long term is sur-

prisingly labile in the face of new information. Our memories are thus an

ever-evolving account of our experiences. A memory that reﬂects witnessing

person X at location Y on a particular evening might be readily and notably

updated by subsequent learning that location Y is the home of a business

associate of person X. Our memories of the witnessed actions of person

J. T. Wixted, “The Psychology and Neuroscience of Forgetting,” Annual Review of Psy-

chology 55: 235–269 (2004). Tulving and Thomson, “Encoding Speciﬁcity and Retrieval Pro-

cesses.” Y. Dudai, “Reconsolidation: The Advantage of Being Refocused,” Current Opinion

in Neurobiology 16(2): 174–178 (2006). E. F. Loftus, “Planting Misinformation in the Human

Mind: A 30-Year Investigation of the Malleability of Memory,” Learning and Memory 12(4):

361–366 (2005). R. A. Bjork, “Interference and Memory,” in Encyclopedia of Learning and

Memory, ed. L. R. Squire (New York: Macmillan, 1992), 283–288. J. A. McGeoch, “Forget-

ting and the Law of Disuse,” Psychological Review 39(4): 352–370 (1932). J. G. Jenkins

and K. M. Dallenbach, “Obliviscence During Sleep and Waking,” The American Journal

of Psychology 35 (1924): 605–612. B. J. Underwood and L. Postman, “Extra-Experimental

Sources of Interference in Forgetting,” Psychological Review 67(2): 73–95 (1960). E. F. Loftus,

“The Malleability of Human Memory,” American Scientist 67(3): 312–320 (1979). D. J. Yi

et al., “When a Thought Equals a Look: Refreshing Enhances Perceptual Memory,” Journal

of Cognitive Neuroscience 20(8): 1371–1380 (2008).

C. M. Alberini, Memory Reconsolidation (Waltham: Academic Press, 2013).

D. L. Schacter, Psychology, Second Edition (New York: Worth Publishers, 2011), 253–

254. E. F. Loftus and H. G. Hoffman, “Misinformation and Memory, the Creation of New

Memories,” Journal of Experimental Psychology 118(1): 100–104 (1989). G. Mazzoni and A.

Memon, “Imagination Can Create False Autobiographical Memories,” Psychological Science

14(2): 186–188 (2003).

F. C. Bartlett, Remembering: A Study in Experimental and Social Psychology (London:

Cambridge University Press, 1932).

Identifying the Culprit: Assessing Eyewitness Identification

BASIC RESEARCH ON VISION AND MEMORY 63

X may be qualiﬁed by new knowledge of his or her life history. Moreover,

because new content can be added and the source of that content forgot-

ten, we may attribute our updated memories to the originally witnessed

events—in some cases substantially changing what we believe we have

seen.

It is thus not surprising that newly incorporated information need

not be true to fact. Research on false memories shows that it is possible

to plant fabricated content in memory, which leads us to recall things we

never experienced.

The emotional content of stored memories is a factor that appears

to promote long-term retention; memories of highly arousing emotional

stimuli, such as those associated with a witnessed crime, tend to be more

enduring than memories of non-arousing stimuli.

Highly salient, un-

expected, or arousing events—such as the Kennedy assassination or the

Space Shuttle disaster—are commonly more strongly stored in memory,

and their later retrieval is often associated with the subjective experience

D. S. Lindsay and M. K. Johnson, “Recognition Memory and Source Monitoring,” Bul-

letin of the Psychonomic Society 29(3): 203–205 (1991). D. L. Schacter and C. S. Dodson,

“Misattribution, False Recognition and the Sins of Memory,” Philosophical Transactions

of the Royal Society: Biological Sciences 356(1413): 1385–1393 (2001). L. A. Henkel, N.

Franklin, and M. K. Johnson, “Cross-Modal Source Monitoring Confusions Between Per-

ceived and Imagined Events,” Journal of Experimental Psychology: Learning, Memory, and

Cognition 26(2): 321–335 (2000). D. L. Schacter, ed., Memory Distortion: How Minds,

Brains, and Societies Reconstruct the Past (Cambridge, MA: Harvard University Press, 1995).

K. J. Mitchell and M. K. Johnson, “Source Monitoring: Attributing Mental Experiences,” in

The Oxford Handbook of Memory, ed. E. Tulving and F. I. M. Craik (New York: Oxford

University Press, 2000), 179–195. H. L. Roediger III and K. B. McDermott, “Creating False

Memories: Remembering Words Not Presented in Lists,” Journal of Experimental Psychology:

Learning, Memory, and Cognition 21(4): 803–814 (1985).

Loftus, “Planting Misinformation in the Human Mind.” E. F. Loftus and J. E. Pickrell,

“The Formation of False Memories,” Psychiatric Annals 25(12): 720–725 (1995). M. K.

Johnson and C. L. Raye, “False Memories and Confabulation,” Trends in Cognitive Sciences

2(4): 137–145 (1998).

L. J. Kleinsmith and S. Kaplan, “Paired-Associate Learning as a Function of Arousal and

Interpolated Interval” Journal of Experimental Psychology 65(2): 190–193 (1963). M. W.

Eysenck, “Arousal, Learning, and Memory,” Psychological Bulletin 83(3): 389–404 (1976).

F. Heuer and D. Reisberg, “Vivid Memories of Emotional Events: The Accuracy of Remem-

bered Minutiae,” Memory and Cognition 18(5): 496–450 (1990). T. Sharot and E. A. Phelps,

“How Arousal Modulates Memory: Disentangling the Effects of Attention and Retention,”

Cognitive, Affective, and Behavioral Neuroscience 4(3): 294–306 (2004). E. A. Kensinger,

R. J. Garoff-Eaton, and D. L. Schacter, “Memory for Speciﬁc Visual Details Can Be Enhanced

by Negative Arousing Content,” Journal of Memory and Language 54(1): 99–112 (2006). E.

Kensinger, “Remembering Emotional Experiences: The Contribution of Valence and Arousal,”

Reviews in the Neurosciences 15(4): 241–251 (2004).

Identifying the Culprit: Assessing Eyewitness Identification

64 IDENTIFYING THE CULPRIT

of high vividness and a sense of reliving

(although not necessarily with

greater accuracy, as detailed below). The stronger encoding and storage of

emotional memories results from the engagement of a specialized system

of stress hormones (glucocorticoids) which is triggered by arousing content

and has potentiating effects on the neuronal processes underlying memory

consolidation and storage.

Despite the vividness and the sense of reliving

that characterizes retrieval of emotional memories, there are many indica-

tions that such memories are just as prone to errors.

This may reﬂect, in

part, memory enhancements, of the sort described above, which accompany

frequent re-consolidation or re-telling of the story of the emotional experi-

ence, and often include details (some true to fact, some not) learned after

the experience.

Although emotional memories are often inaccurate in

detail, one important corollary of their vividness is that they are frequently

G. Wolters and J. J. Goudsmit, “Flashbulb and Event Memory of September 11, 2001:

Consistency, Conﬁdence and Age Effect,” Psychological Report 96: 605–619 (2005). E. A.

Kensinger, A. C. Krendl, and S. Corkin, “Memories of an Emotional and a Nonemotional

Event: Effects of Aging and Delay Interval,” Experimental Aging Research 32: 23–45 (2006).

U. Neisser and N. Harsch, “Phantom Flashbulbs: False Recollections of Hearing the News

about Challenger,” in Affect and Accuracy in Recall: Studies of “Flashbulb” Memories, ed. E.

Winograd and U. Neisser (New York: Cambridge University Press, 1992): 9–31. K. S. LaBar

and E. A. Phelps, “Arousal-Mediated Memory Consolidation: Role of the Medial Temporal

Lobe in Humans,” Psychological Science 9(6): 490–493 (1998).

J. L. McGaugh, “Memory: A Century of Consolidation,” Science 287(5451): 248–251

(2000). J. L. McGaugh and B. Roozendaal, “Role of Adrenal Stress Hormones in Forming

Lasting Memories in the Brain,” Current Opinion in Neurobiology 12(2): 205–210 (2002).

E. A. Kensinger, “Remembering the Details: Effects of Emotion,” Emotion Review 1(2):

99–113 (2009). T. Sharot, M. R. Delgado, and E. A. Phelps, “How Emotion Enhances the

Feeling of Remembering,” Nature Neuroscience 7(12): 1376–1380 (2004). H. Schmolck, E. A.

Buffalo, and L. R. Squire, “Memory Distortions Develop over Time: Recollections of the O. J.

Simpson Trial Verdict after 15 And 32 Months,” Psychological Science 11 (1): 39–45 (2000).

S. R. Schmidt, “Autobiographical Memories for the September 11th Attacks: Reconstructive

Errors and Emotional Impairment of Memory,” Memory and Cognition 32(3): 443–454

(2004). T. W. Buchanan and R. Adolphs, “The Role of the Human Amygdala in Emotional

Modulation of Long-Term Declarative Memory,” in Emotional Cognition: From Brain to Be-

havior, ed. S. Moore and M. Oaksford (Amsterdam: John Benjamins Publishing, 2002), 9–34.

E. Soleti et al., “Does Talking About Emotions Inﬂuence Eyewitness Memory? The Role

of Emotional vs. Factual Retelling on Memory Accuracy,” Europe’s Journal of Psychology

8(4): 632–640 (2012).

Identifying the Culprit: Assessing Eyewitness Identification

BASIC RESEARCH ON VISION AND MEMORY 65

held with high conﬁdence.

This breakdown of the relationship between

accuracy and conﬁdence can obviously undermine eyewitness accounts.

The enduring plasticity of stored memories is a serious concern for the

validity of eyewitness identiﬁcation. A witness’ inevitable interactions with

law enforcement and legal counsel, not to mention communications from

journalists, family, and friends, have the potential to signiﬁcantly modify

the witness’ memory of faces encountered and of other event details at the

scene of the crime.

Thus, the ﬁdelity of retrieved events—and the accuracy

of identiﬁcation—is likely to be greater when retrieval occurs closer to the

time of the witnessed events. The conclusion above has important implica-

tions for law enforcement and the legal process and calls into question the

validity of in-court identiﬁcations and their appropriateness as statements

of fact.

Memory Retrieval

Memory retrieval refers to the process by which stored information is

accessed and brought into consciousness, where it can be used to make deci-

sions and guide actions. Retrieval of long-term declarative memories is of-

ten triggered through association with an external stimulus (i.e., a retrieval

cue).

For example, the slight stubble on a lineup participant’s face may

be enough to elicit retrieval of a suspect’s entire face. These same retrieval

processes can also be engaged internally—a verbally triggered stream of

thought related to a witnessed crime may readily bring to mind visual fea-

tures of the perpetrator. A corollary of this association-based phenomenon

is that memory retrieval is often context dependent; a memory may be more

U. Rimmele et al., “Emotion Enhances the Subjective Feeling of Remembering, Despite

Lower Accuracy for Contextual Details,” Emotion 11(3): 553–562 (2011). Kensinger, “Re-

membering the Detail.” Neisser and Harsh, Affect and Accuracy in Recall. E. A. Phelps and

T. Sharot, “How (and Why) Emotion Enhances the Subjective Sense of Recollection,” Current

Directions in Psychological Science 17(2): 147–152 (2008).

K. A. Houston et al., “The Emotional Eyewitness: The Effects of Emotion on Speciﬁc

Aspects of Eyewitness Recall and Recognition Performance,” Emotion 13(1): 118–128 (2013).

R. B. Edelstein et al., “Emotion and Eyewitness Memory,” in Memory and Emotion, ed. D.

Reisberg and P. Hertel (New York: Oxford University Press, 2004): 308–346. S-A. Christian-

son, “Emotional Stress and Eyewitness Memory: A Critical Review,” Psychological Bulletin

112(2): 284–309 (1992).

M. S. Zaragoza and S. M. Lane, “Sources of Misattribution and Suggestibility of Eyewit-

ness Memory,” Journal of Experimental Psychology: Learning, Memory, and Cognition 20

(4): 934–945 (1994). W. C. Thompson, K. A. Clarke-Stewart, and S. J. Lepore, “What Did

the Janitor Do? Suggestive Interviewing and the Accuracy of Children’s Accounts,” Law and

Human Behaviour 21(4): 405–426 (1997). D. S. Lindsay and M. K. Johnson, “The Eyewitness

Suggestibility Effect and Memory for Source,” Memory and Cognition 17(3): 349–358 (1989).

E. Tulving and Z. Pearlstone, “Availability Versus Accessibility of Information in Memory

for Words,” Journal of Verbal Learning and Verbal Behaviour 5: 381–391 (1966).

Identifying the Culprit: Assessing Eyewitness Identification

66 IDENTIFYING THE CULPRIT

readily retrieved if the observer is in physical surroundings that are the same

as or similar to those in which the original experiences took place (because

the surroundings provide additional cues to trigger memory retrieval).

Memory retrieval is heavily affected by various sources of noise. Simi-

larities of meaning or appearance between retrieval cues and items in

memory can easily lead to retrieval of the wrong item, producing a false

memory.

This is particularly a problem given the categorical nature of

memory.

The rugged mustachioed man in the lineup may lead to retrieval

of the familiar categorical prototype—the Marlboro Man—rather than the

speciﬁc person perceived at the scene of the crime, which in turn could

interfere with or lead to errors in recognition (i.e., identiﬁcation). Another

type of memory retrieval failure is caused by “intrusion errors,” in which

information known to be commonly associated with events of a general

type becomes incorporated into the retrieved content of a speciﬁc memory

(and subsequently incorporated into the reconsolidated memory). For ex-

ample, because guns are often associated with robbery, an observer may

readily and unwittingly incorporate a gun into the retrieved version of his

or her memory of a witnessed robbery.

Intrusion errors are one manifestation of a larger retrieval problem in

which there is loss of information about the source of a memory. In cases

of “source memory failure,” we effectively forget how we know things

(forget when and where we learned the content of our memories). What this

means practically is that we may attribute later acquisition of information

to earlier experiences. An eyewitness might learn from the police or some

other source that a potential suspect has a moustache and then attribute

D. Godden and A. Baddeley, “Context Dependent Memory in Two Natural Environ-

ments,” British Journal of Psychology 66(3): 325–331 (1975). S. M. Smith and E. Vela,

“Environmental Context-Dependent Eyewitness Recognition,” Applied Cognitive Psychology

6: 125–139 (1992). S. M. Smith and E. Vela, “Environmental Context-Dependent Memory:

A Review and Meta-Analysis,” Psychonomic Bulletin Review 8 (2): 203–220 (2001). Tulving

and Thomson, “Encoding Speciﬁcity and Retrieval Processes.”

J. R. Anderson, “A Spreading Activation Theory of Memory,” Journal of Verbal Learning

and Verbal Behavior 22(3): 261–295 (1983). A. M. Collins and E. F. Loftus, “A Spreading-

Activation Theory of Semantic Processing,” Psychological Review 82(6):407–428 (1975).

H. L. Roediger III, D. A. Balota, and J. M. Watson, “Spreading Activation and Arousal of

False Memories,” in The Nature of Remembering: Essays in Honor of Robert G. Crowder,

ed. H. L. Roediger III, J. Nairne, I. Neath, and A. Surprenant (Washington, DC: American

Psychological Association, 2001): 95–115. C. J. Brainerd and V. F. Reyna, The Science of False

Memory (New York: Oxford University Press, 2005).

Tulving, “Episodic and Semantic Memory.” M. J. Farah and J. L. McClelland, “A Compu-

tational Model of Semantic Memory Impairment: Modality Speciﬁcity and Emergent Category

Speciﬁcity,” Journal of Experimental Psychology: General 120(4): 339–357 (1991). G. W.

Humphreys and E. M. Forde, “Hierarchies, Similarity, and Interactivity in Object Recogni-

tion: ‘Category-Speciﬁc’ Neuropsychological Deﬁcits,” Behavioral and Brain Sciences 24(3):

453–476 (2001).

Identifying the Culprit: Assessing Eyewitness Identification

BASIC RESEARCH ON VISION AND MEMORY 67

that knowledge to the witnessed events, which may, in turn, have disastrous

consequences for the ability of the eyewitness to accurately report what she

or he has seen.

As for the processes of memory encoding and storage, the emotional

content of memory also affects memory retrieval. As noted above, memory

retrieval is commonly context dependent. A related and well-documented

phenomenon that bears on emotional memories is state dependent memory,

in which retrieval accuracy is best if the individual’s cognitive state at the

time of retrieval matches cognitive state at the time of encoding.

When

memories have an emotional component, retrieval may be best when the

individual is induced to a corresponding emotional state (mood dependent

memory),

which is accomplished by verbally or physically placing him

or her in the same context, and may offer a valuable investigative tool for

probing eyewitness accounts.

Recognition Memory

Recognition memory is a speciﬁc type of declarative memory retrieval

in which a sensory stimulus (a “cue” stimulus) elicits a memory of the

stimulus stored following a prior encounter and often the sequence of

events involving the stimulus, the spatial context in which the stimulus was

experienced, and the presence of other objects, people, or thoughts that

had appeared with the stimulus during the event.

Recognition memory

decisions are based on the retrieved memory evidence, which can be trig-

gered by the stimulus and can also emerge from an active search of items

D. W. Goodwin et al., “Alcohol and Recall: State-Dependent Effects in Man,” Science

163(3873): 1358–1360 (1969). Tulving and Thomson, “Encoding Speciﬁcity and Retrieval

Processes.” Psychological Review 80(5): 352–373 (1973). E. Girden and E. Culler, “Condi-

tioned Responses in Curarized Striate Muscle in Dogs,” Journal of Comparative Psychology

23(2): 261–274 (1937). D. A. Overton, “State-Dependent or ‘Dissociated’ Learning Produced

with Pentobarbital,” Journal of Comparative and Physiological Psychology 57(1): 3–12

(1964).

P. M. Kenealy, “Mood State-Dependent Retrieval: The Effects of Induced Mood on Mem-

ory Reconsidered,” The Quarterly Journal of Experimental Psychology Section A: Human

Experimental Psychology 50(2): 290–317 (1997). P. A. Lewis and H. D. Critchley, “Mood-

Dependent Memory,” Trends in Cognitive Sciences 7(10): 431–433 (2003). G. H. Bower,

“Mood and Memory,” American Psychologist 36(2): 129–148 (1981). F. I. M. Craik and

R. S. Lockhart, “Levels of Processing: A Framework for Memory Research,” Journal of Verbal

Learning and Verbal Behavior 11(6):671–684 (1972). Kensinger, “Remembering the Detail.”

K. A. Leight and H. C. Ellis “Emotional Mood States, Strategies, and State-Dependency in

Memory,” Journal of Verbal Learning and Verbal Behavior 20(3): 251–266 (1981).

S. M. Smith and E. Vela, “Environmental Context-Dependent Eyewitness Recognition,”

Applied Cognitive Psychology 6: 125–139 (1992).

G. Mandler, “Recognizing: The Judgment of Previous Occurrence,” Psychological Review

87(3): 252–271 (1980).

Identifying the Culprit: Assessing Eyewitness Identification

68 IDENTIFYING THE CULPRIT

in memory. One factor affecting the strength of the evidence retrieved is the

similarity between the cue stimulus and the stimulus or stimuli that was/

were previously encountered during the event. An observer engaged in this

process holds an implicit criterion for the strength of evidence required to

reach a positive decision. In the case of eyewitness identiﬁcation, this pro-

cess is routinely elicited by viewing faces in a lineup. When the evidence

retrieved is insufﬁcient to reach a decision, this can lead to a cycle of ever-

greater scrutiny of the cue stimulus and efforts to recollect additional details

of the original event. Ultimately a decision must be made about whether the

retrieved evidence is sufﬁcient to declare that the stimulus was previously

experienced (or previously experienced in the particular event of interest)

or whether the stimulus is novel (or not from the event of interest). If a

recognition event occurs—that is, if the memory search triggered by one of

the faces in the lineup leads to a strong enough subjective experience that

the face is familiar and/or the recollection of sufﬁcient event details—then

the witness may declare that they recognize the face as having been previ-

ously encountered. Recognition memory decisions can thus be thought of

as the ﬁnal stage in the process of eyewitness identiﬁcation.

Because it is a form of memory retrieval, recognition memory is sus-

ceptible to all of the factors summarized above that are known to interfere

with retrieval. Recognition memory differs from other forms of retrieval

(such as recalling a phone number or a cake recipe), however, in that a

comparison must be made between the retrieved evidence and a decision

threshold. That is, as noted above, recognition judgments require a decision

criterion, an understanding of which presents a unique set of challenges for

eyewitness identiﬁcation (and recognition memory, generally). In particular,

an observer’s report of recognition (or, in a lineup setting, of identiﬁcation)

is inﬂuenced not simply by the strength or quality of the recalled memory

evidence. The report of recognition (identiﬁcation of a lineup member) is

also inﬂuenced by the level of evidence that the observer ﬁnds acceptable

to reach such a decision, i.e., by his or her decision criterion, or bias. An

observer who holds a liberal criterion will likely recognize many true targets

(i.e., the guilty), but will frequently err by reporting recognition of many

false targets (i.e., innocents). Conversely, an observer who holds a conserva-

tive criterion will avoid the problem of erroneous recognition (identiﬁca-

tion), but will fail to identify some true targets. Estimating (or controlling)

the observer’s decision criterion is thus a critical step in efforts to judge the

validity of an identiﬁcation (see also Chapter 5).

Recognition memory for faces differs greatly between familiar and un-

familiar faces.

Because we often identify familiar individuals with ease,

P. J. B. Hancock, V. Bruce, and A. M. Burton, “Recognition of Unfamiliar Faces,” Trends

in Cognitive Sciences 4(9): 330–337 (2000).

Identifying the Culprit: Assessing Eyewitness Identification

BASIC RESEARCH ON VISION AND MEMORY 69

we tend to think we are generally very good at face recognition. However,

we are not as good with unfamiliar faces.

All of the sources of noise that

inﬂuence perception and memory contribute to these difﬁculties, and they

are exacerbated by the attempts by criminals to conceal their identity (even

a change in hairstyle and clothing can have a major effect on recognition).

The ability to recognize unfamiliar faces differs widely across individu-

als. At one extreme are those people, referred to as “super recognizers,”

who rarely forget a face.

At the other end of the spectrum are “face-blind

people (prosopagnosics),” who have great difﬁculty recognizing even highly

familiar faces.

Current estimates of the fraction of the general population

afﬂicted by prosopagnosia are as high as ~2 percent.

The ability of an

eyewitness to identify a suspect may thus differ greatly from individual to

individual simply as a consequence of general variations in face recognition

ability.

CONCLUSION

The shortcomings of eyewitness identiﬁcation present a societal prob-

lem that has profound implications for our systems of law and justice.

Ultimately, a solution to this problem must be informed by a thorough

understanding of human vision and memory. The processes of vision and

memory, which are fundamental to human experience, have been frequent

targets of scientiﬁc investigation since the 19th century. The past few de-

cades have seen an explosion of additional research that has led to impor-

tant insights into how vision and memory work, what we see and remember

best, and what causes these processes to fail. The committee has reviewed

much of this research, as it pertains to eyewitness identiﬁcation, and has

identiﬁed restrictions on what can be seen under speciﬁc environmental

and behavioral conditions (e.g., as poor illumination, limited viewing dura-

tion, viewing angle), factors that impede the ability to attend to critically

informative features of a visual scene (e.g., the deleterious effect of an

attention-grabbing element, such as a weapon, on the ability to correctly

perceive the features of the assailant’s face), distortions of perceptual ex-

perience derived from expectations, and ways in which emotion and stress

enhance or suppress speciﬁc perceptual experiences. Memory is often far

V. Bruce, “Changing Faces: Visual and Non-Visual Coding Processes in Face Recognition,”

British Journal of Psychology 73: 105–116 (1982).

R. Russell, B. Duchaine, and K. Nakayama, “Super-Recognisers: People with Extraordi-

nary Face Recognition Ability,” Psychonomic Bulletin and Review 16(2): 252–272 (2009).

T. Susilo and B. Ducahine, “Advances in Developmental Prosopagnosia Research,” Cur-

rent Opinion in Neurobiology 2(3):423–429 (2013).

I. Kennerknecht et al., “First Report of Prevalence of Nonsyndromic Hereditary Prosop-

agnosia (HPA),” American Journal of Medical Genetics Part A 140(15): 1617–1622 (2006).

Identifying the Culprit: Assessing Eyewitness Identification

70 IDENTIFYING THE CULPRIT

from a faithful record of what was perceived through the sense of sight:

its contents can be forgotten or contaminated at multiple stages, it can be

biased by the very practices designed to elicit recall, and it is heavily swayed

by emotional states associated with witnessed events and their recall. From

this analysis, the committee must conclude that there are insurmountable

limits on vision and memory imposed by our biological nature and the

properties of the world we inhabit. With this knowledge, it is possible to

more fully appreciate the value and risks associated with eyewitness reports

and accordingly advise those who collect, handle, defend, consider, and

adjudicate such reports.

Identifying the Culprit: Assessing Eyewitness Identification

Applied Eyewitness

Identiﬁcation Research

he committee was tasked with (1) critically assessing the existing

body of scientiﬁc research on eyewitness identiﬁcation; (2) identi-

fying gaps in the literature; and (3) suggesting other research that

would further the understanding of eyewitness identiﬁcation and improve

law enforcement and courtroom practice. Eyewitness identiﬁcation research

resides in both the scientiﬁc literature and the law and justice-related schol-

arly literature. Although experiential, anecdotal, and some administrative

records from law enforcement and the judiciary could contribute to a

better understanding of eyewitness identiﬁcation, the committee did not

comprehensively review this more qualitative material. The committee did,

however, examine select examples of law enforcement policies and inﬂuen-

tial judicial rulings.

In late 2013, the committee compiled an extensive and comprehensive

bibliography from the following nine electronic databases, with the search

limited to publications over the past two decades (i.e., since 1993): Aca-

demic Search Premier (EBSCO), Embase (Elsevier), MEDLINE (National

Library of Medicine), NCJRS Abstracts Database (U.S. Department of Jus-

tice), PsycINFO (American Psychological Association), PubMed (National

Institutes of Health), Scopus (Elsevier), Web of Science (Thomson Reuters),

and LexisNexis.

Papers were drawn from such ﬁelds as social science,

cognitive science, behavioral science, neuroscience, criminology, and law

The law review literature was represented by the citations from the LexisNexis search.

While all these materials were not reviewed in detail, several of the documents informed

Chapter 3 of this report (The Legal Framework for Assessment of Eyewitness Identiﬁcation).

Identifying the Culprit: Assessing Eyewitness Identification

72 IDENTIFYING THE CULPRIT

using Boolean-logic-based search strategies designed to identify empirical

research reports, review articles, systematic reviews, meta-analyses, and

articles in law reviews and legal journals.

The committee concentrated its review on the subset of the bibliography

deemed most important to its task, focusing more on the scientiﬁc literature

than on the law review literature. These materials included meta-analyses

and systematic reviews and primary research in neuroscience, statistics,

and eyewitness identiﬁcation. This report also was informed by several

early foundational papers and written comments from, and presentations

to, the committee by representatives from science, law enforcement, state

courts and government, private organizations, and other interested parties.

The comments and presentations revealed additional highly relevant new

ﬁndings, some recently published or in press and others in submission. The

agenda for each committee meeting is available in Appendix B. All materials

submitted to the committee are retained in the Academies’ public access ﬁle

and are available upon request.

COMMITTEE ASSESSMENT

Many factors affect eyewitness accuracy. Some factors are related to

protocols within the law enforcement and legal systems, while others are

related to characteristics associated with the crime scene, perpetrator, and

witness.

System variables are those that the criminal justice system can inﬂuence

through the enforcement of standards and through education and training

of law enforcement personnel in the use of best practices

and procedures

(e.g., by specifying the content and nature of instructions given to witnesses

prior to a lineup identiﬁcation). Estimator variables include factors operat-

ing either at the time of the criminal event (relating to visual experience

or memory encoding) or during the retention interval (the time between

witnessing an event and the identiﬁcation process). Speciﬁc examples in-

clude the eyewitness’ level of stress or trauma at the time of the incident,

the light level and nature of the visual conditions that affect visibility and

clarity of a perpetrator’s features, similarity of age and race of the witness

and perpetrator, presence or absence of a weapon during the incident, and

the physical distance separating the witness from the perpetrator.

A scientiﬁc consensus about the effects of some factors has emerged,

but no such consensus exists for many other factors. One method of assess-

ing scientiﬁc consensus is by surveys of experts. A 2001 survey collected

As noted in Chapter 1, for the purposes of this report, the committee characterizes best

practice as the adoption of standardized procedures based on scientiﬁc principles. The commit-

tee does not make any endorsement of practices designated as best practices by other bodies.

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 73

responses from 64 psychologists about their courtroom experiences and

their opinions on 30 eyewitness-related phenomena to determine the “gen-

eral acceptance” of these phenomena within the eyewitness identiﬁcation

research community.

General acceptance is relevant to whether scientiﬁc

testimony is admissible as evidence in court (see Chapter 3). The survey

revealed substantial agreement about which ﬁndings these experts felt were

sufﬁciently reliable to present in court.

The committee examined the scientiﬁc literature on eyewitness identi-

ﬁcation, focusing ﬁrst on quantitative syntheses, largely systematic reviews

and meta-analyses, which were identiﬁed in a comprehensive search of

electronic databases designed to locate research on both estimator and

system variables. In addition, primary research studies were identiﬁed in

this database search, many of which were also highlighted in the relevant

systematic reviews and meta-analyses. Finally, some researchers forwarded

manuscripts to the committee that have been submitted for peer-review or

are in press. In their examination of this body of literature, the committee

examined the quality of the identiﬁed research and, where possible, worked

to derive summary empirical generalizations related to variables of interest.

Quantitative Syntheses of Eyewitness Identiﬁcation Research

The committee ﬁrst evaluated the consistency of research ﬁndings

across studies for system and estimator variables by studying published

quantitative reviews of empirical research. Systematic reviews, which collect

and appraise available research on speciﬁc hypotheses or research ques-

tions, are efforts to synthesize the effects of variables across studies. Within

systematic reviews, meta-analysis is often, but not always, used to compute

the effects of variables as well as to identify factors that explain differ-

ences across studies. When assumptions about consistency of data collected

across studies are met, meta-analysis provides a quantitative summary of

empirical ﬁndings by statistically averaging effect sizes across individual

studies, thereby increasing the precision of the effect size estimate as well

S. M. Kassin et al., “On the ‘General Acceptance’ of Eyewitness Testimony Research: A

New Survey of the Experts,” American Psychologist 56(5): 405–416 (2001).

Kassin et al. also compared the reliability assessments of the 2001 survey to assessments

from a similar 1989 survey and noted that, for the 17 propositions retested, there was a

remarkable degree of consistency: “most experts saw as sufﬁciently reliable expert testimony

on the wording of questions,” lineup instructions, attitudes and expectations, the accuracy-

conﬁdence correlation, the forgetting curve, exposure time, and unconscious transference.

“There was less, if any, consensus on the effects of color perception in monochromatic light,”

“observer training, high levels of stress, the accuracy of hypnotically refreshed testimony, and

event violence.” The authors observed that two phenomena were seen as signiﬁcantly more

reliable than had been the case when the initial survey was conducted: weapon focus effect

and hypnotic suggestibility effects. See p. 410.

Identifying the Culprit: Assessing Eyewitness Identification

74 IDENTIFYING THE CULPRIT

as the statistical power to detect effects. Done well, systematic reviews with

or without meta-analysis provide evidence for practice and policy for such

ﬁelds as health care,

crime and justice, social welfare, and education.

The

utility of systematic reviews for informing practice and policy is predicated

on the included studies being transparently reported, conducted so as to

minimize risk of bias, and representing as complete a sample as possible of

research conducted on the central question, including both published and

unpublished studies. In turn, systematic reviews should specify inclusion

criteria and data extraction procedures a priori, use independent and dupli-

cate procedures for study selection and data extraction, rigorously evaluate

potential biases in included studies, and interpret results of meta-analyses

in terms that are useful to decision-makers. Further, meta-analyses should

not be conducted outside the context of systematic reviews. In short, both

systematic reviews and the studies they include need to be transparent and

reproducible in order to best inform practice and policy decisions about

eyewitness identiﬁcation.

The committee examined quantitative reviews that covered decades of

research on both estimator variables (exposure duration,

retention interval,

stress,

weapon focus,

own-race bias,

and own-age bias

) and system

variables (identiﬁcation test medium, i.e., live lineup versus photo array,

See the Cochrane Collaboration, available at: http://www.cochrane.org.

See the Campbell Collaboration, available at: http://www.campbellcollaboration.org.

B. H. Bornstein et al., “Effects of Exposure Time and Cognitive Operations on Facial

Identiﬁcation Accuracy: A Meta-Analysis of Two Variables Associated with Initial Memory

Strength,” Psychology, Crime and Law 18(5): 473–490 (2012).

K. A. Deffenbacher et al., “Forgetting the Once-Seen Face: Estimating the Strength of an

Eyewitness’s Memory Representation,” Journal of Experimental Psychology: Applied 14(2):

139–150 (2008).

K. A. Deffenbacher et al., “A Meta-Analytic Review of the Effects of High Stress on Eyewit-

ness Memory,” Law and Human Behavior 28(6): 687–706 (2004).

J. M. Fawcett et al., “Of Guns and Geese: A Meta-Analytic Review of the ‘Weapon Focus’

Literature,” Psychology, Crime and Law 19(1): 35–66 (2013).

C. A. Meissner and J. C. Brigham, “Thirty Years of Investigating the Own-Race Bias

in Memory for Faces—A Meta-Analytic Review,” Psychology, Public Policy, and Law 7(1):

3–35 (2001).

M. G. Rhodes and J. S. Anastasi, “The Own-Age Bias in Face Recognition: A Meta-

Analytic and Theoretical Review,” Psychological Bulletin 138(1): 146–174 (2012).

B. L. Cutler et al., “Conceptual, Practical, and Empirical Issues Associated with Eyewit-

ness Identiﬁcation Test Media,” in Adult Eyewitness Testimony: Current Trends and Devel-

opments, ed. D. F. Ross (New York: Press Syndicate of the University of Cambridge, 1994),

163–181.

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 75

biased and unbiased lineup instructions,

post-identiﬁcation feedback,

si-

multaneous versus sequential lineup presentation,

target absent versus tar-

get present lineups,

foil similarity,

blinding,

showup versus lineup,

prior mug shot exposure,

verbal description and identiﬁcation,

and the

cognitive interview

). Many of these quantitative reviews were published

recently, with more than one-third published since 2010. However, none

of the reviews met all current standards for conducting and reporting sys-

S. E. Clark, “A Re-Examination of the Effects of Biased Lineup Instructions in Eyewit-

ness Identiﬁcation,” Law and Human Behavior 29(4): 395–424 (2005). S. E. Clark, “Costs

and Beneﬁts of Eyewitness Identiﬁcation Reform: Psychological Science and Public Policy,”

Perspectives on Psychological Science 7(3): 238–259 (2012). N. K. Steblay, “Social Inﬂuence in

Eyewitness Recall: A Meta-Analytic Review of Lineup Instruction Effects,” Law and Human

Behavior 21(3): 283–297 (1997). N. K. Steblay, G. L. Wells, and A. B. Douglass, “The Eyewit-

ness Post Identiﬁcation Feedback Effect 15 Years Later: Theoretical and Policy Implications,”

Psychology, Public Policy, and Law 20(1): 1–18 (2014).

S. E. Clark and R. D. Godfrey, “Eyewitness Identiﬁcation Evidence and Innocence Risk,”

Psychonomic Bulletin and Review 16(1): 22–42 (2009). A. B. Douglass and N. K. Steblay,

“Memory Distortion in Eyewitnesses: A Meta-Analysis of the Post-Identiﬁcation Feedback

Effect,” Applied Cognitive Psychology 20(7): 859–869 (2006).

Clark, “Costs and Beneﬁts of Eyewitness Identiﬁcation Reform.” S. E. Clark, R. T. Howell,

and S. L. Davey, “Regularities in Eyewitness Identiﬁcation,” Law and Human Behavior 32(3):

187–218 (2008). N. K. Steblay et al., “Eyewitness Accuracy Rates In Sequential and Simulta-

neous Lineup Presentations: A Meta-Analytic Comparison,” Law and Human Behavior 25(5):

459–473 (2001). N. K. Steblay et al., “Seventy-two Tests of the Sequential Lineup Superiority

Effect: A Meta-Analysis and Policy Discussion,” Psychology, Public Policy, and Law 17(1):

99–139 (2011).

Clark, “A Re-Examination of the Effects of Biased Lineup Instructions in Eyewitness

Identiﬁcation.” Clark, Howell, and Davey, “Regularities in Eyewitness Identiﬁcation.” Clark

and Godfrey, “Eyewitness Identiﬁcation Evidence and Innocence Risk.”

Clark, “Costs and Beneﬁts of Eyewitness Identiﬁcation Reform.” Clark and Godfrey,

“Eyewitness Identiﬁcation Evidence and Innocence Risk.” Clark, Howell, and Davey, “Regu-

larities in Eyewitness Identiﬁcation.” R. J. Fitzgerald et al., “The Effect of Suspect-Filler Simi-

larity on Eyewitness Identiﬁcation Decisions: A Meta-Analysis,” Psychology, Public Policy,

and Law 19(2): 151–164 (2013). S. L. Sporer et al., “Choosing, Conﬁdence, and Accuracy:

A Meta-Analysis of the Conﬁdence-Accuracy Relation in Eyewitness Identiﬁcation Studies,”

Psychological Bulletin 118(3): 315–327 (1995).

Clark, “Costs and Beneﬁts of Eyewitness Identiﬁcation Reform.”

Clark, “Costs and Beneﬁts of Eyewitness Identiﬁcation Reform.” N. K. Steblay et al.,

“Eyewitness Accuracy Rates in Police Showup and Lineup Presentations: A Meta-Analytic

Comparison,” Law and Human Behavior 27(5): 523–540 (2003).

K. A. Deffenbacher et al., “Mugshot Exposure Effects: Retroactive Interference, Mugshot

Commitment, Source Confusion, and Unconscious Transference,” Law and Human Behavior

30(3): 287–307 (2006).

C. A. Meissner, S. L Sporer, and K. J. Susa, “A Theoretical Review and Meta-Analysis

of the Description-Identiﬁcation Relationship in Memory for Faces,” European Journal of

Cognitive Psychology 20(3): 414–455 (2008).

A. Memon et al., “The Cognitive Interview: A Meta-Analytic Review and Study Space

Analysis of the Past 25 Years,” Psychology, Public Policy, and Law 16(4): 340–372 (2010).

Identifying the Culprit: Assessing Eyewitness Identification

76 IDENTIFYING THE CULPRIT

tematic reviews,

and few met even a majority of these standards, making

assessment of the credibility of their ﬁndings problematic.

After examining the reviews, the committee concluded that the ﬁndings

may be subject to unintended biases and that the conclusions are less cred-

ible than was hoped. In many cases, the data from the studies cited were not

readily available or were not clearly presented. Nevertheless, these reviews

were helpful in highlighting some of the issues associated with speciﬁc

research questions and in identifying primary studies that might be both

credible and important.

RESEARCH STUDIES ON SYSTEM VARIABLES

After its assessment of the systematic reviews and meta-analytic studies,

the committee’s review focused on the most-studied system variables. Key

system variables, such as lineup procedures (e.g., simultaneous vs. sequen-

tial lineups, blinded vs. non-blinded lineup administration) and the collec-

tion/use of witness conﬁdence statements, can have a marked inﬂuence over

the validity of eyewitness identiﬁcations. In the following section, one of the

most important practical issues raised by this inﬂuence is addressed: What

is the best way to evaluate the effects of system variables on the diagnostic

accuracy of eyewitness reports, and how might we use the results of such an

evaluation to optimize the states of key system variables and thus maximize

performance of an eyewitness? This question is, in principle, relevant to

all system variables, but we address it ﬁrst in the timely and controversial

context of simultaneous versus sequential lineup presentations and in the

role of eyewitness conﬁdence judgments in evaluation of identiﬁcation per-

formance. This examination of lineup procedures and conﬁdence reports is

followed by a brief discussion of the effects on eyewitness performance of

another important system variable: the extent and content of communica-

tions between the witness and the larger community (law enforcement, legal

defense, the press, family and friends, etc.).

Evaluating Eyewitness Performance

Perhaps the most important empirical question that can be asked about

eyewitness identiﬁcation is: How well do witnesses perform as a function

of different system and estimator variables? For example, do factors such

as the structure of a lineup, stress, or weapon focus affect the ability of

See, e.g., Institute of Medicine, Finding What Works in Health Care: Standards for System-

atic Reviews (Washington, DC: The National Academies Press, 2011) and B. J. Shea et al., Devel-

opment of AMSTAR: A Measurement Tool to Assess the Methodological Quality of Systematic

Reviews, BMC Medical Research Methodology 2007, 7:10 doi:10.1186/1471-2288-7-10.

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 77

a witness to provide reliable information? If so, what practices will yield

the best performance? The issues are multifaceted, and the answers likely

depend upon many factors. Given the complexity of these issues, the experi-

mental literature to date has focused largely on one of the more tractable

problems: How do different lineup identiﬁcation procedures affect witness

identiﬁcations? The committee will use this focus (and its eminent practi-

cal relevance) to illustrate how one might go about evaluating eyewitness

performance generally.

Most lineup identiﬁcation procedures take one of two forms: simul-

taneous or sequential. In a simultaneous procedure, the witness views

all individuals in the lineup at the same time and either identiﬁes one (or

more) as the perpetrator or reports that the person she or he saw at the

crime scene was not in the lineup. In a sequential procedure, the witness

views individuals one at a time and reports whether or not each one is the

person from the crime scene. Rigorous evaluation of eyewitness identiﬁca-

tion performance as a function of these two procedures requires a formal

understanding of the task that the witness confronts, and it requires criteria

for assessing the outcome.

The task of a witness viewing a lineup is an example of what is known

as a binary classiﬁcation problem.

Each eyewitness faces two possible (bi-

nary) states associated with each person in the lineup (guilt or innocence),

and the witness must assign each person to one of two classes (guilty or

innocent). For each decision, the witness can be correct or incorrect, yield-

ing four possible outcomes: a correct classiﬁcation as guilty (“hit”), an

incorrect classiﬁcation as guilty (“false alarm”), a correct classiﬁcation as

innocent (“correct rejection”), and an incorrect classiﬁcation as innocent

(“miss”). These outcomes are commonly presented in a contingency table

(see Figure 5-1), and the frequencies in each part of that table are the raw

data used to evaluate performance on a binary classiﬁcation task, such as

eyewitness identiﬁcation.

There are many different performance measures that can be derived

from data of this sort—indeed, the ﬁelds of statistical classiﬁcation and ma-

chine learning are replete with tools for the evaluation of binary classiﬁers.

The binary classiﬁer in this context is deﬁned as the witness operating under a speciﬁc set

of conditions, such as lineup procedures.

Also termed “confusion matrix.”

The prevalence or “base-rate”—the fraction of individuals in each category (guilty or in-

nocent, in the eyewitness problem) in the population is also a factor that may come into play

when evaluating binary classiﬁcation performance.

See, e.g., T. Hastie, R. Tibshirani, and J. H. Friedman, The Elements of Statistical Learn-

ing: Data Mining, Inference, and Prediction (New York: Springer, 2009) and A. Smola and

S. V. N. Vishwanathan, Introduction to Machine Learning (Cambridge: Cambridge University

Press, 2008).

Identifying the Culprit: Assessing Eyewitness Identification

78 IDENTIFYING THE CULPRIT

The preferred measure will depend to a large degree upon the criteria one

adopts for performance evaluation.

Perhaps the simplest measure of binary classiﬁcation performance is

the ratio of hit rates (HR) to false alarm rates (FAR), i.e., HR/FAR.

The

magnitude of this measure, which is known in the eyewitness identiﬁcation

literature as the “diagnosticity ratio,” is proportional to the likelihood that

a classiﬁcation is correct, i.e., that the person identiﬁed as guilty is actually

guilty.

The diagnosticity ratio is appealing if the most critical criterion is

avoiding erroneous identiﬁcations.

The “rate” associated with each cell of the contingency table is computed as the number

of counts within that cell (e.g., number of people correctly classiﬁed as guilty) divided by

the number of instances that are truly in that class (e.g., total number of guilty people being

classiﬁed). Thus, hit rates (HR) = number of hits / (number of hits+number of misses), and

false alarm rate (FAR) = number of false alarms / (number of false alarms+number of correct

rejections).

The “diagnosticity ratio” is also known in other disciplines by other names; e.g., “posi-

tive likelihood ratio” or “LR+ = Likelihood Ratio of a Positive Call;” see Peter Lee, Bayesian

Statistics: An Introduction (Chichester: Wiley, 2012), Sec 4.1.

FIGURE 5-1 Contingency table for possible eyewitness identiﬁcation outcomes.

SOURCE: Courtesy of Thomas D. Albright.

Witness

Classiﬁcaon of

Lineup Parcipant

True Status

of Lineup

Parcipant

guilty

innocent

“Hit”

(true posive)

“False Alarm”

(false posive)

“Miss”

(false negave)

“Correct Rejecon”

(true negave)

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 79

Not surprisingly, the diagnosticity ratio was adopted in pioneering

efforts to identify lineup conditions that would yield better witness identi-

ﬁcation performance.

Most laboratory-based studies and meta-analyses

of the effects of lineup procedures on eyewitness identiﬁcation performance

show that, with standard lineup instructions informing the witness that the

perpetrator may or may not be present, the sequential procedure produces

a higher diagnosticity ratio.

That is, when considering only those cases in

which a witness actually selects someone from a lineup, the ratio of correct

to false identiﬁcations is commonly higher with the sequential than with

the simultaneous procedure.

A higher diagnosticity ratio could result from a higher hit rate, a lower

false alarm rate, or some combination of the two. Some early reports sug-

gested that sequential procedures (relative to simultaneous) lead to fewer

false alarms without changing the frequency of hits, which would result in

a higher diagnosticity ratio.

More recent laboratory-based studies and

meta-analyses typically show that sequential procedures (relative to simul-

taneous) are associated with a somewhat reduced hit rate accompanied by

a larger reduction in the false alarm rate, thereby resulting in diagnosticity

ratios higher than those yielded by simultaneous procedures.

In other

R. C. L. Lindsay and G. L. Wells, “Improving Eyewitness Identiﬁcations from Lineups:

Simultaneous Versus Sequential Lineup Presentation,” Journal of Applied Psychology 70(3),

556–564 (1985).

Steblay et al. “Eyewitness Accuracy Rates in Sequential and Simultaneous Lineup Presenta-

tions.” Steblay, et al., “Seventy-two Tests of the Sequential Lineup Superiority Effect.” S. D.

Gronlund et al., “Robustness of the Sequential Lineup Advantage,” Journal of Experimental

Psychology: Applied 15(2): 140–152 (2009). S. D. Gronlund, J. T. Wixted, and L. Mickes,

“Evaluating Eyewitness Identiﬁcation Procedures Using ROC Analysis,” Current Directions

in Psychological Science 23(1): 3–10 (2014).

But see C. A. Carlson, S. D. Gronlund, and S. E. Clark, “Lineup Composition, Suspect

Position, and the Sequential Lineup Advantage,” Journal of Experimental Psychology-Applied

14(2): 118-128 (2008), for a counterexample. Also, Clark, Moreland, and Gronlund have

demonstrated that the accuracy advantage of sequential lineups as measured by diagnosticity

ratios has decreased over time since the original report. Reanalysis of diagnosticity data for

sequential studies showed slight, non-signiﬁcant decreases in correct identiﬁcation effects and

increases in false identiﬁcation effects, which together combine to produce a signiﬁcant de-

crease in the advantage of sequential over simultaneous lineup methods. See S. E. Clark, M. B.

Moreland, and S. D. Gronlund, “Evolution of the Empirical and Theoretical Foundations of

Eyewitness Identiﬁcation Reform,” Psychonomic Bulletin and Review 21(2): 251–267 (2014).

R. C. L. Lindsay, “Applying Applied Research: Selling the Sequential Lineup,” Applied

Cognitive Psychology 13(3): 219–225 (1999). G. L. Wells, S. M. Rydell, and E. P. Seelau,

“The Selection of Distractors for Eyewitness Lineups,” Journal of Applied Psychology 78(5):

835–844 (1993).

A recent ﬁeld-based study comparing sequential to simultaneous procedures in a limited

number of jurisdictions computed the diagnosticity ratio using ﬁller identiﬁcations as the

false alarm rate (because the innocence or guilt of the suspect is unknown in such situations).

See G. L. Wells, N. K. Steblay, J. E. Dysart, “Double-Blind Photo-Lineups Using Actual

Identifying the Culprit: Assessing Eyewitness Identification

80 IDENTIFYING THE CULPRIT

words, when using a single diagnosticity ratio as a measure of eyewitness

performance, the sequential procedure (relative to simultaneous) comes

closer to satisfying the popular criterion that those identiﬁed as guilty are

actually guilty. In light of these ﬁndings, many policy makers have advo-

cated sequential procedures, and those procedures have been adopted by

law enforcement in many jurisdictions.

While policy decisions and practice have been inﬂuenced by the afore-

mentioned studies, there are other criteria worthy of consideration when

evaluating eyewitness performance. One alternative is revealed by asking

why the diagnosticity ratio changes across lineup conditions. This ques-

tion can be addressed given a plausible model of the mechanisms underly-

ing human recognition memory. Most models of recognition memory are

based on the idea that a cue (e.g., a face in a lineup) results in the retrieval

of information stored in memory (see Chapter 4). When the retrieved in-

formation provides enough evidence to satisfy the observer, they make an

identiﬁcation—that is, they decide that the stimulus is “recognized.” Ex-

plicit in this model are two important parameters: the observer’s memory

sensitivity (that is, the “discriminability” between the strength of memory

evidence elicited by a previously encountered stimulus and that elicited by

novel stimuli), and the degree of evidence that the observer requires to make

an identiﬁcation (“response criterion” or “bias”) (see Box 5-1).

The ﬁrst of these two parameters—discriminability—is important for

evaluating eyewitness performance. It tells whether a difference in per-

formance under different task conditions reﬂects a true improvement in

memory-based discrimination, i.e., an improvement in the strength of the

observer’s retrieved memory evidence of the perpetrator.

The fact that these two measures (the likelihood that an identiﬁed

person is guilty vs. discriminability) do not assess the same thing is coun-

terintuitive—a fact that has generated controversy in the ﬁeld of eyewitness

Eyewitnesses: An Experimental Test of a Sequential versus Simultaneous Lineup Procedure,”

Law and Human Behavior, 15 June 2014, doi: 10.1037/lhb0000096. When computed in this

manner, the data revealed a modest diagnosticity ratio advantage for the sequential procedure.

However, Amendola and Wixted re-analyzed a subset of the data for which proxy measures of

ground truth were available [K. Amendola and J. T. Wixted, “Comparing the Diagnostic Accu-

racy of Suspect Identiﬁcations Made by Actual Eyewitnesses from Simultaneous and Sequential

Lineups,” accepted by Journal of Experimental Criminology (2014)]. Their analyses suggested

that identiﬁcation of innocent suspects is less likely and identiﬁcation of guilty suspects is more

likely when using the simultaneous procedures. While future ﬁeld studies are needed, these

latter ﬁndings raise the possibility that diagnosticity is higher for the simultaneous procedure.

See also Clark, Moreland, and Gronlund, who report that published diagnosticity ratios have

changed over time, reﬂecting a signiﬁcant decrease in the advantage of sequential over simul-

taneous lineup procedures. (Clark, Moreland, and Gronlund, “Evolution of the Empirical and

Theoretical Foundations of Eyewitness Identiﬁcation Reform.”)

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 81

BOX 5-1

The Influences of Discriminability and Response

Bias on Human Binary Classification Decisions

All human decisions about the classiﬁcation of objects based on memory—

including a witness’ classiﬁcations of guilt or innocence for faces in a lineup, an

individual’s decision as to whether a piece of luggage is his or her own, a bota-

nist’s recognition of a speciﬁc type of fern, a radiologist’s detection of a tumor in

a mammogram, or the determination of the sex of a newly-hatched chicken—can

be distilled down to the inﬂuence of two factors that are rooted in causal models

of recognition memory:the degree to which the relevant objects are discriminable

by the decider (the decider’s sensitivity to the difference between them), and the

decider’s criterion for making a decision (response bias, or the decider’s degree

of speciﬁcity in making choices).

There are, of course, many other variables that

will affect the outcome (e.g., levels of stress, attentional focus, potential rewards

or expectations), but all of these are believed to exert their inﬂuence over memory-

based classiﬁcation decisions by affecting discriminability and/or response bias.

To illustrate the distinction between discrimination and response bias as

applied to a real-world decision problem, consider how an audiologist conducts

a hearing test. In a hearing test, an individual might be asked to detect sounds

along a continuum of loudness and to indicate when a sound is present. The

audiologist wants to know how well someone can discriminate presence versus

absence of a sound, but that assessment is complicated by the criterion people

use when deciding to say that they heard a sound (response bias). Some people

are hesitant to respond positively, saying “I hear it” only when they are absolutely

certain (“conservative” responders). Others are more willing to respond positively,

saying “I hear it” with less information and greater uncertainty (“liberal” respond-

ers). Those with a conservative bias are less likely to report hearing a sound in

general, so they will have both fewer correct detections (“hits”) and fewer overt

mistakes (“false alarms”). By contrast, those with a liberal bias are more likely

to say that they heard a sound, so they will have more hits but also more false

alarms. Importantly, this can occur even if the conservative and liberal respond-

ers do not differ in their ability to discriminate the presence or absence of sound.

See, e.g., W. P. Banks, “Signal Detection Theory and Human Memory,” Psychological Bul-

letin 74(2): 81–99 (1970); J. P. Egan, Recognition Memory and the Operating Characteristic

(Bloomington: Indiana University Hearing and Communication Laboratory, 1958); D. M. Green

and J. A. Swets, Signal Detection Theory and Psychophysics (New York: Wiley,1966).

Identifying the Culprit: Assessing Eyewitness Identification

82 IDENTIFYING THE CULPRIT

identiﬁcation research.

Intuitively, if sequential lineups yield a higher

likelihood that an identiﬁed person is guilty (as quantiﬁed by a higher

diagnosticity ratio), then it seems as if that procedure yields objectively

better performance. The problem with this intuition is that it fails to take

into account the second of the two parameters of recognition memory

models—the response bias or degree of evidence that the observer ﬁnds ac-

ceptable to make an identiﬁcation. This parameter, which is distinct from

discriminability, reﬂects the witness’ tendency to pick or not to pick some-

one from the lineup. If a witness sets a high bar for acceptable evidence—a

conservative bias—then he or she will be unlikely to select anyone from

the lineup (low pick frequency), meaning that they will have more misses

(will be more likely to fail to select the suspect because they are less likely

to make a selection at all) and fewer false alarms.

Conversely, if a witness sets a low bar for acceptable evidence—a liberal

bias—then she or he will be more likely to make a selection from the lineup

(a high pick frequency), meaning he or she will have more hits and will

make more false identiﬁcations. Differences in pick frequency can, and gen-

erally do, lead to differences in the ratio of hit rates to false alarm rates; all

else being equal, the diagnosticity ratio will be higher for a conservative bias

than for a liberal bias.

In other words, simply by inducing a witness to

adopt a more conservative bias, it is possible to increase the likelihood that

an identiﬁed person is actually guilty. Importantly, this may be true even

if the procedure yields no better, or potentially worse, discriminability.

Despite its merits, a single diagnosticy ratio thus conﬂates the inﬂuences

of discriminability and response bias on binary classiﬁcation, which mud-

dies the determination of which procedure, if any, yields objectively better

discriminability in eyewitness performance. To overcome this problem,

some investigators have recently adopted a technique from signal detection

See, e.g., J. T. Wixted and L. Mickes, “The Field of Eyewitness Memory Should Aban-

don Probative Value and Embrace Receiver Operating Characteristic Analysis,” Perspectives

on Psychological Science 7(3): 275-278 (2012); Clark, “Costs and Beneﬁts of Eyewitness

Identiﬁcation Reform”; G. L. Wells, “Eyewitness Identiﬁcation Probative Value, Criterion

Shifts, and Policy Regarding the Sequential Lineup,” Current Directions in Psychological

Science 23(1): 11–16 (2014); and Steblay, et al. “Seventy-two Tests of the Sequential Lineup

Superiority Effect.”

The sole exception to this rule is the case in which classiﬁcations are made at chance level

of performance, i.e., when the observer exhibits no ability to discriminate.

L. Mickes, H. D. Flowe, and J. T. Wixted, “Receiver Operating Characteristic Analysis of

Eyewitness Memory: Comparing the Diagnostic Accuracy of Simultaneous vs. Sequential Line-

ups,” Journal of Experimental Psychology: Applied 18 (4): 361–376 (2012). C. A. Meissner

et al., “Eyewitness Decisions In Simultaneous and Sequential Lineups: A Dual Process Signal

Detection Theory Analysis,” Memory and Cognition 33(5): 783–792 (2005). M. A. Palmer

and N. Brewer, “Sequential Lineup Presentation Promotes Less-Biased Criterion Setting but

Does Not Improve Discriminability,” Law and Human Behavior 36(3): 247–255 (2012).

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 83

theory, which distinguishes the relative inﬂuences of discriminability and

bias on binary classiﬁcation.

This technique involves analysis of Receiver

Operating Characteristics (see Box 5-2). ROC analysis has been used ex-

tensively in multiple contexts of human decision-making, notably in basic

research on visual perception and memory and applied studies of medical

diagnostic procedures.

In essence, ROC analysis examines diagnosticity

ratios integrated over different response biases. This approach to eyewitness

research has been promoted based on the claim that it can enable lineup

procedures to be evaluated by their effect on discrimination, separate from

response bias, and—importantly—because the dimensions of analysis (dis-

criminability and response bias) correspond to the mechanistic parameters

of causal models of human recognition memory.

Use of ROC analysis to evaluate eyewitness performance requires cal-

culating the diagnosticity ratio for different response bias conditions (see

Box 5-2). Using expressed conﬁdence level (ECL) as a proxy for response

bias (see below), a small set of recent studies using ROC analysis has re-

ported that discriminability (area under the ROC curve) for simultaneous

lineups is as high, or higher, than that for sequential lineups.

In other

words, when eyewitness identiﬁcation performance is evaluated based on

a criterion of bias-free discriminability, the results differ from those based

on a single diagnosticity ratio, and they do so because the latter fails to

account for response bias.

Looking broadly at the many empirical studies that have used a single

diagnosticity ratio to evaluate eyewitness performance, as well as the more

recent ﬁndings using ROC analysis, it appears that the practical advantage

of one lineup procedure over another depends to a large degree upon the

performance criterion that one adopts. From the perspective of many,

the ideal lineup procedure would elicit a conservative bias (thus reducing

false identiﬁcations) and high discriminability (that is, optimizing memory

sensitivity). If there exists no discriminability advantage for one lineup

D. M. Green and J. A. Swets, Signal Detection Theory and Psychophysics (New York:

Wiley, 1966); D. McNicol, A Primer of Signal Detection Theory (London: George Allen and

Unwin, 1972).

J. A. Swets, “ROC Analysis Applied to the Evaluation of Medical Imaging Techniques,”

Investigative Radiology 14(2): 109–121 (1979).

Mickes, Flowe, and Wixted, “Receiver Operating Characteristic Analysis of Eyewitness

Memory." C. A. Carlson and M. A. Carlson, “An Evaluation of Lineup Presentation, Weapon

Presence, and a Distinctive Feature Using ROC Analysis,” Journal of Applied Research in

Memory and Cognition 3(2): 45–53 (2014). D. G. Dobolyi and C. S. Dodson, “Eyewitness

Conﬁdence in Simultaneous and Sequential Lineups: A Criterion Shift Account for Sequential

Mistaken Identiﬁcation Overconﬁdence,” Journal of Experimental Psychology: Applied 19

(4): 345–357 (2013). S. D. Gronlund et al., “Showups Versus Lineups: An Evaluation Using

ROC Analysis,” Journal of Applied Research in Memory and Cognition 1(4): 221–228 (2012).

Identifying the Culprit: Assessing Eyewitness Identification

84 IDENTIFYING THE CULPRIT

BOX 5-2

Analysis of Receiver Operating Characteristics (ROCs)

Binary classiﬁcation decisions by human observers are affected by both

discriminability (the observer’s sensitivity to the difference between target and

non-targets) and response bias (the observer’s degree of speciﬁcity in making a

response). Analysis of Receiver Operating Characteristics (ROCs) is a method

from signal detection theory that enables one to distinguish the relative inﬂuences

of discriminability and response bias on binary classiﬁcation decisions. ROC

analysis is performed by plotting the frequency of decisions that are hits (correctly

detecting a target) versus the frequency of decisions that are false alarms (incor-

rectly classifying a non-target as a target).

The positive diagonal in an ROC plot (see ﬁgure next page) corresponds to

response bias, moving from high speciﬁcity at the lower left corner [no detection

of targets (hit rate = 0) and no incorrect attribution of non-targets as targets (false

alarm rate = 0)], to low speciﬁcity at the upper right corner [all targets detected

(hit rate = 1.0) and all non-targets attributed as targets (false alarm rate = 1.0)].

Because all points along this positive diagonal reﬂect equal ratios of hits to false

alarms, they vary in response bias (i.e., the frequency of lineup picks, or “pick

frequency”), but they do not manifest differences in discriminability. The negative

diagonal in an ROC plot corresponds, by contrast, to discriminability, moving from

chance discriminability at the intersection with the positive diagonal, where hits

and false alarms are equally likely, to the highest discriminability in the upper left

corner, where all targets are detected (hit rate = 1.0), but no non-targets are at-

tributed as targets (false alarm rate = 0).

To see how measured hit and false alarm rates vary over different conditions

of discriminability and response bias in laboratory experiments, one can manipu-

late or estimate these conditions and record a diagnosticity ratio (HR/FAR) for

each condition. The typical result is a set of diagnosticity ratios that, when plotted

in the ROC space (represented by the dots in the ﬁgure at right), form a curve

spanning from lower left to upper right. The extent to which that curve deviates

(bows above and away) from the positive diagonal is a quantitative measure of

discriminability (assessed as the area under the curve) for which response bias

has been factored out.

ROC analysis has been used extensively in basic and applied research

on recognition memory. In these experiments, response bias is sometimes ma-

nipulated explicitly by encouraging observers to be more or less selective in

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 85

their responses. Frequently, however, “expressed conﬁdence level” (ECL)—the

conﬁdence that an observer holds in his or her classiﬁcation—is used as a proxy

for response bias, based on the assumption that more conﬁdent observers are

likely to be more speciﬁc (conservative) in their responses, whereas less conﬁdent

observers are likely to be less speciﬁc (liberal) in their responses.

Receiver Operating Characteristic (ROC) curve.

SOURCE: Courtesy of Thomas D. Albright.

False Alarm Rate

Hit Rate

1.0

Identifying the Culprit: Assessing Eyewitness Identification

86 IDENTIFYING THE CULPRIT

procedure over another,

then eyewitness performance may beneﬁt from

any procedure (such as sequential) that elicits a more conservative response

bias.

But one can only make that judgment after having applied an em-

pirical test to determine whether a procedure offers a discriminability ad-

vantage. Future research might explore the possibility that other methods

of inducing a conservative response bias (such as verbal instructions to the

witness to be cautious in making an identiﬁcation) might be combined with

procedures that improve discriminability in order to optimize eyewitness

identiﬁcation performance.

Perhaps the greatest practical beneﬁt of recent debate over the utility

of different lineup procedures is that it has opened the door to a broader

consideration of methods for evaluating and enhancing eyewitness identi-

ﬁcation performance. ROC analysis is a positive and promising step with

numerous advantages. For example, the area under the ROC curve is a

single-number index of discriminability. Moreover, this index reﬂects a

parameter-free approach to binary classiﬁcation performance; the outcome

is entirely data-dependent and thus identical across all users drawing from

The committee notes that some of the few recent reports using ROC analysis indeed claim

improved discriminability for simultaneous lineup conditions, but the reported discriminability

improvements are small.

In reality, a more conservative bias may not always be beneﬁcial, and whether it is or

not depends upon a number of factors that have an impact distinct from diagnostic accuracy

and are difﬁcult to quantify. All else being equal, the “best” response bias will be one that

maximizes the “expected value” of the outcome (Green and Swets, Signal Detection Theory

and Psychophysics; Swets, “ROC Analysis Applied to the Evaluation of Medical Imaging

Techniques”). For the problem of eyewitness identiﬁcation, the response bias that maximizes

expected value can be computed from the prevalence of guilty suspects in lineups and from

societal values or costs associated with each of the possible eyewitness decisions (errors and

correct assignments). Reliable data on prevalence are difﬁcult to come by, and value/cost

quantities are difﬁcult to assign and likely to vary signiﬁcantly across crimes and cultures.

One can nonetheless gain an intuition for how these factors might deﬁne the best response

bias conditions. Consider, for example, the consequences of decreasing the prevalence of guilty

suspects in lineups. In this case, expected value can be maximized by inducing a conservative

bias—i.e., if innocence is a priori likely, then there is value gained by being more selective in

your response. Similarly, the optimal response bias will depend upon normative costs associ-

ated with different types of eyewitness errors. Generally speaking, if a society places greater

emphasis on not identifying the innocent, relative to failing to identify the guilty, then expected

value can be increased by inducing a more conservative response bias. But the opposite would

be true if there were greater societal pressures for identifying the guilty, relative to protect-

ing the innocent. Although an understanding of the relationship between response bias and

expected value is important, expected value in this case has little to do with the diagnostic

accuracy of an eyewitness report. But it does nonetheless bear on decisions about which lineup

procedure should be employed.

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 87

the same data set.

Most importantly for its application to the problem

of evaluating eyewitness performance, the ROC approach possesses a dis-

tinct advantage because the dimensions of analysis—discriminability and

response bias—map directly onto the mechanistic parameters of causal

models of human recognition memory (see Chapter 4). In other words, the

approach affords insight into and quantiﬁcation of the sensory and cogni-

tive processes that are believed to underlie memory-based classiﬁcation

decisions (see Box 5-1), such as eyewitness identiﬁcations.

Despite these merits, as a general statistical procedure for evaluation

of binary classiﬁcation performance and as a tool for evaluation of eyewit-

ness performance, the ROC approach has some well-documented quantita-

tive shortcomings. For example, ROC analysis depends on the ability to

manipulate response bias or to estimate it from some other variable, and

in the case of eyewitness identiﬁcation that ability has been the subject of

some debate. Recent studies have used expressed conﬁdence level (ECL)—

a measure of a witness’ conﬁdence in his or her selection—as a proxy for

response bias,

based on the common-sense logic that a witness who has

high conﬁdence in their lineup selection should manifest a more conserva-

tive response bias than a witness who selected someone from the lineup

despite lacking conﬁdence in that selection (i.e., someone who made a

selection even though they were not certain—a liberal response bias). This

proxy relationship is inherently noisy within individuals, and the noisy rela-

tionship is exacerbated by the fact that the eyewitness identiﬁcation ROC is

population-based; individual data points are obtained from different people

who may scale their conﬁdence reports differently.

On the other hand,

it is empirically clear that, when scaled appropriately (within and across

individuals), different levels of expressed conﬁdence do, in fact, correspond

to different pick frequencies and response biases.

Green and Swets, Signal Detection Theory and Psychophysics. D. J. Hand, “Measuring

Classiﬁer Performance: A Coherent Alternative to the Area under the ROC Curve,” Machine

Learning 77, 103–123 (2009).

See, e.g., N. Brewer and G. L. Wells, “The Conﬁdence-Accuracy Relationship in Eyewit-

ness Identiﬁcation: Effects of Lineup Instructions, Foil Similarity, and Target-Absent Base

Rates,” Journal of Experimental Psychology: Applied 12(1): 11–30 (2012); Mickes, Flowe,

and Wixted, “Receiver Operating Characteristic Analysis of Eyewitness Memory”; and

Carlson and Carlson, “An Evaluation of Lineup Presentation.”

ECL is affected by over-conﬁdence and under-conﬁdence at the individual level, and the

current implementation of the ROC approach, combining results across subjects, does not

build this measurement error into the analysis or the comparison of empirical ROC curves.

See Appendix C.

See, e.g., Table 1 of Mickes, Flowe, and Wixted, “Receiver Operating Characteristic Anal-

ysis of Eyewitness Memory,” which summarizes conﬁdence ratings, hit rates, false alarm rates,

and diagnosticity ratios (HR/FAR) derived from data published in Brewer and Wells, “The

Conﬁdence-Accuracy Relationship in Eyewitness Identiﬁcation.” Brewer and Wells employed

Identifying the Culprit: Assessing Eyewitness Identification

88 IDENTIFYING THE CULPRIT

An additional prerequisite for the use of ECL as a measure of re-

sponse bias is that an orderly relationship exists between conﬁdence and

accuracy—that witnesses expressing greater conﬁdence are more likely to

be accurate in their identiﬁcations. Although this hypothesis conforms to

intuition,

the existence of a signiﬁcant conﬁdence–accuracy relationship

has been challenged repeatedly over the years.

Recent evidence, however,

suggests ways of improving the conﬁdence–accuracy relationship (and ob-

taining more reliable measurements of it).

While the ECL measure thus

has potential, more research on this and other possible methods of estimat-

ing or controlling response bias is warranted to support efforts to extract

a bias-free measure of discriminability.

Another technical concern raised by the use of ROC analysis to evalu-

ate eyewitness identiﬁcation performance is that it relies on a partial, rather

than full, area under the ROC curve measure (see Box 5-2) as an index of

discriminability that is separate from response bias. This is necessitated by

the fact that the highest false alarm rates in eyewitness identiﬁcation data

are commonly well below 1.0, even under the most liberal response bias

a “conﬁdence calibration” technique to normalize scaling of expressed conﬁdence across

witnesses. Both hit rates and false alarm rates declined steeply—implying an increasingly con-

servative response bias—as conﬁdence levels increased. Diagnosticity ratios increased mono-

tonically with increasing conﬁdence. An identical pattern can be seen in Table 3 of Mickes,

Flowe, and Wixted, “Receiver Operating Characteristic Analysis of Eyewitness Memory.” See

also H. L. Roediger III, J. T. Wixted, and K. A. DeSoto, “The Curious Complexity Between

Conﬁdence and Accuracy in Reports from Memory,” in Memory and Law, ed. L. Nadel and

W. Sinnott-Armstrong (Oxford: Oxford University Press, 2012), 97.

K. A. Deffenbacher and E. F. Loftus, “Do Jurors Share a Common Understanding Con-

cerning Eyewitness Behavior?,” Law and Human Behavior 6: 15–30 (1982); and G. L. Wells,

T. J. Ferguson, and R. C. L. Lindsay, “The Tractability of Eyewitness Conﬁdence and Its

Implication for Triers of Fact,” Journal of Applied Psychology 66: 688–696 (1981).

G. L. Wells and D. M. Murray, “Eyewitness Conﬁdence,” in Eyewitness Testimony: Psy-

chological Perspectives, ed. G. L. Wells and E. F. Loftus (New York: Cambridge University

Press, 1984). B. L. Cutler and S. D. Penrod, Mistaken Identiﬁcation: The Eyewitness, Psy-

chology, and the Law (Cambridge: Cambridge University Press, 1995). R. K. Bothwell, K. A.

Deffenbacher, and J.C. Brigham,“Correlation of Eyewitness Accuracy and Conﬁdence: Opti-

mality Hypothesis Revisited,” Journal of Applied Psychology 72:691–695 (1987). S. L. Sporer

et al., “Choosing, Conﬁdence, and Accuracy: A Meta-Analysis of the Conﬁdence-Accuracy

Relation in Eyewitness Identiﬁcation Studies,” Psychological Bulletin 118(3): 315–327 (1995).

T. A. Busey et al., “Accounts of the Conﬁdence-Accuracy Relation in Recognition Memory,”

Psychonomic Bulletin and Review 7(1): 26-48 (2000).

N. Brewer and G. L. Wells, “The Conﬁdence-Accuracy Relationship in Eyewitness Identi-

ﬁcation. P. Juslin, N. Olsson, and A. Winman, “Calibration and Diagnosticity of Conﬁdence

in Eyewitness Identiﬁcation: Comments on What Can Be Inferred From the Low Conﬁdence-

Accuracy Correlation,” Journal of Experimental Psychology: Learning, Memory, and Cog-

nition 22(5): 1304–1316 (September 1996). Roediger, Wixted, and DeSoto, “The Curious

Complexity between Conﬁdence and Accuracy.” Mickes, Flowe, and Wixted, “Receiver

Operating Characteristic Analysis of Eyewitness Memory.”

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 89

conditions.

In practice, partial area under the curve is computed by trun-

cating the ROC curve at the highest false alarm rate obtained. Because the

standard error of the partial area under the curve measure depends upon

the degree of truncation, accuracy of this discriminability measure can

easily vary across conditions and across studies, making the interpretation

difﬁcult.

While ROC analysis has many recognized merits for the evaluation of

binary classiﬁcation, the residual concerns associated with its typical use

for evaluating eyewitness performance merit consideration of other sta-

tistical approaches to this problem. As noted above, many methods have

been proposed—and adopted in speciﬁc applications—for evaluation of

binary classiﬁcation performance.

The committee knows of no instance

in which any of these alternative methods has been applied to the problem

of eyewitness identiﬁcation. Moreover, because they have not been vetted,

the committee is not in a position to endorse any speciﬁc statistical tool, the

committee nevertheless encourages a general exploration of these alterna-

tives. These alternatives may have their own share of unforeseen problems,

and/or the performance criteria employed by them may bear no meaningful

relationship to the sensory and cognitive processes involved in eyewitness

identiﬁcation. Nonetheless, some of these methods may provide greater

insight into the factors that affect eyewitness identiﬁcation performance

and may, in turn, suggest ways of improving performance. To illustrate this

opportunity by example, we consider the following possibilities.

It has been argued that a basic weakness of the existing ROC approach

to binary classiﬁcation performance results from the fact that, in principle

Carlson and Carlson, “An Evaluation of Lineup Presentation.” Mickes, Flowe, and

Wixted, “Receiver Operating Characteristic Analysis of Eyewitness Memory.”

Along the same lines, accuracy of discriminability measures derived from ROC studies

may be called into question when those studies do not take into account uncertainty in the

data used to construct the ROC curves; see Appendix C. An argument has also been made

that the area under the ROC curve can be a ﬂawed metric for comparing binary classiﬁcation

conditions when the costs of classiﬁcation errors are not precisely known and are different for

different conditions (Hand, “Measuring Classiﬁer Performance”). The costs of classiﬁcation

errors may be similar across some lineup comparisons and across some conditions of other

systems variables, and for others they may be different. But for the most part they are not

precisely known, and this is thus a topic that deserves greater attention given the growing use

of ROC-based evaluation of eyewitness identiﬁcation performance.

Numerous methods for the evaluation of binary classiﬁers have been developed and

applied in the ﬁeld of machine learning, which seeks to optimize autonomous classiﬁcation

devices (such as, for example, the ﬁngerprint lock access control on a smart phone, which

must quickly and reliably distinguish the ﬁnger from another). This ﬁeld has a long and rich

history, and candidate methods are summarized in several texts on statistical classiﬁcation and

machine learning, such as Hastie, Tibshirani, and Friedman, The Elements of Statistical Learn-

ing and A. Smola and S. V. N. Vishwanathan, Introduction to Machine Learning (Cambridge:

Cambridge University Press, 2009).

Identifying the Culprit: Assessing Eyewitness Identification

90 IDENTIFYING THE CULPRIT

(and in practice under certain commonly unrecognized conditions), the

area under the ROC curve is dependent on imprecise assumptions about

the costs of classiﬁcation errors across different classiﬁcation conditions.

One might suppose, for example, that the cost of a miss for a crime of

murder is greater than the cost of a miss for a stolen car. But without a

precise understanding of these relative decision costs, the area under the

ROC curve measure can be incoherent, in that it depends as much on the

classiﬁcation conditions as it does on the sensitivity of the classiﬁer. An al-

ternative method has been proposed to address this problem—derivation of

the “H measure”—that enables the performance of binary classiﬁers to be

compared using a common metric that is independent of the cost distribu-

tions for different types of classiﬁcation errors.

The committee supports

exploration of this alternative.

Another avenue for exploration emerges from the fact that the litera-

ture evaluating eyewitness identiﬁcation performance has focused exclu-

sively on the positive predictive value (PPV) of a witness’ classiﬁcation

as guilty. For a given response bias, PPV is related to the diagnosticity

ratio, in that, given equal prevalence of the culprit in two conditions (e.g.,

lineup procedures) being compared, a higher diagnosticity ratio leads to a

higher PPV. As discussed above, the diagnosticity ratio is a critical piece

of information in efforts to evaluate eyewitness performance. As for any

binary classiﬁcation, however, there is also information associated with a

negative response, which is the predictive value of a classiﬁer’s assertion

that a target is not present (in the eyewitness case, the witness’ assertion

of innocence). This negative predictive value (NPV) is related to a different

ratio of decisions, namely (1-HR)/(1-FAR),

in that, given equal prevalence

of the target in the two procedures being compared, higher values of this

ratio correspond to higher values of NPV.

While NPV is commonly used to evaluate the accuracy of human

classiﬁcation decisions, such as in medical diagnosis, and is a source of

information that may similarly be of additional value in efforts to evalu-

ate lineup procedures, it has been largely neglected in the ﬁeld of eyewit-

ness identiﬁcation.

One might hold the intuition that PPV and NPV are

monotonically related to one another—believing that the likelihood that the

See Hand, “Measuring Classiﬁer Performance.”

Ibid.

The reciprocal of this ratio is called the “negative likelihood ratio.” See, e.g., T. Hoffmann,

S. Bennett, and C. del Mar, Evidence-Based Practice Across the Health Professionals

(Chatswood: Elsevier Australia, 2009).

It seems likely that this neglect stems from the fact that the primary concern in eyewit-

ness identiﬁcation has been on incorrect assertions of guilt (i.e., false identiﬁcations) rather

than incorrect assertions of innocence. There are normative values in society that reinforce

this concern (as exempliﬁed, for example, by Blackstone’s formulation: “Better that 10 guilty

persons escape than that one innocent suffer.”)

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 91

witness will correctly identify the culprit is proportional to the likelihood

that the witness will correctly identify lineup candidates as innocent—and

thus conclude that evaluation of PPV alone is sufﬁcient. Contrary to that

intuition, however, evidence from studies of analogous binary classiﬁca-

tion problems reveals that these two predictive probabilities can vary with

respect to one another in complex ways.

In practice, NPV-related measures (quantiﬁed as negative likelihood

ratios) can be subjected to ROC analysis to account for the effects of

response bias in the same manner as PPV-related measures (quantiﬁed as

positive likelihood ratios, i.e., diagnosticity ratios)—the ROC axes in the

NPV case corresponding to 1-HR and 1-FAR. Consideration of NPV and

its relationship to PPV, by this and other means, may provide additional in-

sight into the ways in which estimator and system variables (such as lineup

procedures) inﬂuence eyewitness identiﬁcation performance.

In sum, a formal understanding of the task facing an eyewitness, in

conjunction with an appreciation of causal models of human recogni-

tion memory, has led to a potentially more comprehensive method—ROC

analysis—for evaluating eyewitness identiﬁcation performance. Despite

these advances, it is important that practitioners in this ﬁeld broadly ex-

plore the large and rich ﬁeld of statistical tools for evaluation of binary

classiﬁers. While the committee recognizes that these tools are uninvesti-

gated for this application and may possess their own share of unforeseen

problems or disadvantages, a move in this direction may be of great value

for improving the validity of eyewitness identiﬁcation.

Interactions with Eyewitnesses (Feedback)

The nature of law enforcement interactions with the eyewitness be-

fore, during, and after the identiﬁcation plays a role in the accuracy of

eyewitness identiﬁcations and in the conﬁdence expressed in the accuracy

of those identiﬁcations by witnesses.

Law enforcement’s maintenance of

neutral pre-identiﬁcation communications—relative to the identiﬁcation of

a suspect—is seen as vital to ensuring that the eyewitness is not subjected to

conscious or unconscious verbal or behavioral cues that could inﬂuence the

S-Y Shiu and C. Gatsonis, “The Predictive Receiver Operating Characteristic Curve for the

Joint Assessment of the Positive and Negative Predictive Values,” Philosophical Transactions,

Series A, Mathematical, Physical and Engineering Sciences 366 (1874): 2313–2333 (2008).

Another potentially informative analysis that combines PPV and NPV measures is known

as a PROC (predictive ROC), which affords the opportunity to see how a given system or

estimator variable may have interacting—synergistic or antagonistic—effects on assertions of

guilt and innocence. See Shiu and Gatsonis, “The Predictive Receiver Operating Characteristic

Curve.”

S. E. Clark, T. E. Marshall, and R. Rosenthal,“Lineup Administrator Inﬂuences on Eyewit-

ness Identiﬁcation Decisions,” Journal of Experimental Psychology: Applied 15(1): 63 (2009).

Identifying the Culprit: Assessing Eyewitness Identification

92 IDENTIFYING THE CULPRIT

eyewitness’ identiﬁcation (see Box 2-1).

If a witness happened to overhear

an ofﬁcer say, “We’ve got him, but before we ﬁnalize the arrest, let’s have

the witness conﬁrm it,” the witness might be biased to conﬁrm the suspect’s

identity in a showup. Furthermore, some types of law enforcement commu-

nication with a witness, after the witness has made an identiﬁcation (e.g.,

“Good work! You picked the right guy…”), can increase conﬁdence in an

identiﬁcation, regardless of whether the identiﬁcation is correct.

As discussed in Chapter 2, use of “blinded” or “double-blind” lineup

identiﬁcation procedures is an effective strategy for reducing the likeli-

hood that a witness will be exposed to cues from interactions with law

enforcement (such as feedback) that could inﬂuence identiﬁcations and/

or conﬁdence in those identiﬁcations. More generally, efforts to maintain

objectivity and eliminate potentially informative communication will help

ensure that eyewitness reports are not contaminated by knowledge or opin-

ions held by others.

RESEARCH STUDIES ON ESTIMATOR VARIABLES

The impact of estimator variables on eyewitness accuracy is harder to

measure in the ﬁeld than the impact of system variables.

Consequently,

estimator variables have been studied nearly exclusively in laboratory set-

tings. The committee’s review revealed the need for further empirical re-

search in individual studies and systematic reviews of research on these

factors.

The committee’s review focused on the most-studied estimator vari-

ables: weapon focus, stress and fear, own-race bias, exposure, and retention

interval. It is important to emphasize, however, that numerous other estima-

tor variables may affect both the reliability and the accuracy of eyewitness

identiﬁcations. Research has shown that the physical distance between the

witness and the perpetrator is an important estimator variable, as it directly

affects the ability of the eyewitness to discern visual details,

including

features of the perpetrator

(see discussion of vision in Chapter 4). Re-

Clark, Moreland, and Gronlund, “Evolution of the Empirical and Theoretical Foundations

of Eyewitness Identiﬁcation Reform”: “…the performance advantage for unbiased instruc-

tions has decreased only slightly over the past 32 years. However, none of the correlations

approached statistical signiﬁcance.” p. 258.

Douglas and Steblay, “Memory Distortion in Eyewitnesses.”

G. L. Wells, “What Do We Know about Eyewitness Identiﬁcation?” American Psycholo-

gist (May 1993): 553, 555.

B. Uttl, P. Graf, and A. L. Siegenthaler, “Inﬂuence of Object Size on Baseline Identiﬁca-

tion, Priming, and Explicit Memory: Cognition and Neurosciences,” Scandinavian Journal of

Psychology 48(4): 281–288 (2007).

C. L. Maclean et al., “Post-Identiﬁcation Feedback Effects: Investigators and Evaluators,”

Applied Cognitive Psychology 25(5): 739–752 (2011).

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 93

search has also shown that an appearance change can greatly diminish the

eyewitness’ ability to recognize the perpetrator; the eyewitness’ ability to

remember faces of his or her own age group is often superior to his or her

ability to remember faces of another age group (own-age bias); and if an

eyewitness hears information or misinformation from another person be-

fore law enforcement involvement, his or her recollection of the event and

conﬁdence in the identiﬁcation can be altered (co-witness contamination).

Interactions between and among these variables have not been addressed

systematically by researchers.

Weapon Focus

The presence of an unusual object at the scene of a crime can impair

visual perception and memory of key features of the crime event. Research

suggests that the presence of a weapon at the scene of a crime captures the

visual attention of the witness and impedes the ability of the witness to

attend to other important features of the visual scene, such as the face of

the perpetrator (see also discussion of visual attention in Chapter 4). The

ensuing lack of memory of these other key features may impair recognition

of a perpetrator in a subsequent lineup.

A 1992 analysis of weapon focus studies found that the presence of a

weapon reduced both identiﬁcation accuracy and feature accuracy (e.g., the

eyewitness’ ability to recall clothing, facial features, and more).

A more

recent analysis of the weapon focus literature concluded that the presence of

a weapon has an inconsistent effect on identiﬁcation accuracy, in that larger

effect sizes were observed in threatening scenarios than in non-threatening

ones.

As the retention interval increased, the weapon focus effect size

decreased. The analysis further indicated that the effect of a weapon on

accuracy is slight in actual crimes, slightly larger in laboratory studies, and

largest for simulations.

One possible cause of the inconsistent effects of the presence of a

weapon is suggested by a recent laboratory-based study that exposed par-

ticipants to crime videos.

These investigators used ROC analysis to inves-

tigate discriminability as a function of (1) sequential versus simultaneous

lineups; (2) the presence of a weapon; and (3) the presence of a distinctive

facial feature. Importantly for the present discussion, discriminability was

R. Zajac and N. Henderson, “Don’t It Make My Brown Eyes Blue: Co-Witness Misinfor-

mation about a Target’s Appearance Can Impair Target-Absent Lineup Performance,” Memory

17(3): 266–278 (2009).

N. K. Steblay, “A Meta-analytic Review of the Weapon Focus Effect,” Law and Human

Behavior 16(4): 413, 415–417 (1992).

Fawcett et al., “Of Guns and Geese.”

Carlson and Carlson, “An Evaluation of Lineup Presentation.”

Identifying the Culprit: Assessing Eyewitness Identification

94 IDENTIFYING THE CULPRIT

reduced when the perpetrator possessed a weapon, but only when no dis-

tinctive facial feature was present. This interaction between weapon focus

and distinctive feature highlights the importance of exploring the effects of

interactions between different estimator variables on eyewitness identiﬁca-

tion performance.

Additional questions remain as to what is the cause of reduced eyewit-

ness performance in cases where a weapon is present. Is the effect caused

by a diversion of selective attention, as is suggested by basic research on

the phenomenon of inattentional blindness (see Chapter 4)? Is stress a

signiﬁcant factor, i.e., does anxiety cause the witness to focus less on the

features of a person’s face? To what extent is the prominence of the issue an

artifact of the particular studies included in the meta-analysis? Is it possible,

for example, that the magnitude of the weapon effect depends on whether

the data are collected in a laboratory setting versus the real world? To this

latter point, some analyses of weapon focus have been conducted using

archival records of crimes involving weapons.

Unfortunately, such efforts

often encounter serious methodological difﬁculties that include a lack of

information about the crime (e.g., exposure duration) and the general lack

of “ground truth” regarding accuracy of any identiﬁcation, among other

problems.

Stress and Fear

High levels of stress or fear can affect eyewitness identiﬁcation.

71,72,73

This ﬁnding is not surprising, given the known effects of fear and stress on

vision and memory (see Chapter 4). Under conditions of high stress, a wit-

ness’ ability to identify key characteristics of an individual’s face (e.g., hair

length, hair color, eye color, shape of face, presence of facial hair) may be

signiﬁcantly impaired.

In the particular case of weapon focus, it may not be possible to suf-

ﬁciently test the effects of stress and heightened stress in the laboratory

because of limitations on human participant research that uses realistic and

heightened threats. A meta-analysis of the effect of high stress on eyewitness

See, e.g., Fawcett et al., “Of Guns and Geese.”

Deffenbacher et al., “A Meta-Analytic Review of the Effects of High Stress.”

C. A. Morgan III et al., “Accuracy of Eyewitness Memory for Persons Encountered Dur-

ing Exposure to Highly Intense Stress,” International Journal of Law and Psychiatry 27(3):

265–279 (2004).

C. A. Morgan III et al., “Accuracy of Eyewitness Identiﬁcation Is Signiﬁcantly Associated

with Performance on a Standardized Test of Recognition,” International Journal of Law and

Psychiatry 30 (3): 213–223 (2007).

C. A. Morgan III et al., “Misinformation Can Inﬂuence Memory for Recently Experienced,

Highly Stressful Events,” International Journal of Law and Psychiatry 36(1): 11–17 (2013).

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 95

memory nonetheless found some support for the notion that stress impairs

both eyewitness recall and identiﬁcation accuracy.

The study authors

noted that lineup type “moderated the effect of heightened stress on the

false alarm rate.”

They also suggested that the modest effect of stress may

be caused by the fact that the analysis included many studies that involved

modest stress-induction.

Earlier studies were more mixed but with clearer results at “high levels

of cognitive anxiety.”

The ﬁndings of an earlier study “provide a concrete

illustration of catastrophic decline” of eyewitness identiﬁcation perfor-

mance at high anxiety levels.

The correct identiﬁcation rate went from 75

percent for those with low-state anxiety to 18 percent rate for those with

high-state anxiety.

The effects of suggestion may be particularly important when the

original memory is of a highly stressful event. A recent study looked at

more than 850 active-duty military personnel participating in a mock

POW camp phase of U.S. military survival school training, which included

aggressive interrogation and physical isolation-related stress.

The study

found that misinformative details of the interrogation event (e.g., regarding

the identity of the interrogator), which were introduced after the event had

been encoded into long-term memory, affected identiﬁcation accuracy. The

study also found that memories acquired during stressful events are highly

vulnerable to modiﬁcation by exposure to post-event misinformation, even

in individuals whose level of training and experience might be considered

relatively immune to such inﬂuences.

Another recent study comparing the eyewitness accuracy of ofﬁcers

and citizens, concentrated on the effects of stress and weapon focus.

The

results of this study showed that ofﬁcers were less stressed and aroused than

Deffenbacher et al., “A Meta-Analytic Review of the Effects Of High Stress.” It should

be noted that the effect sizes for stress-induced support were small with wide conﬁdence

intervals, indicating considerable heterogeneity across studies. Although the authors assert

that 300 studies with null ﬁndings would be required to negate the small effects found in this

meta-analysis, fewer studies might be needed if they resulted in opposite effects.

Ibid, 700.

Ibid, 704.

Ibid, 689.

T. Valentine and J. Mesout, “Eyewitness Identiﬁcation Under Stress in the London Dun-

geon,” Applied Cognitive Psychology 23(2): 151–161 (2009).

K. A. Deffenbacher, “Estimating the Impact of Estimator Variables on Eyewitness Iden-

tiﬁcation: A Fruitful Marriage of Practical Problem Solving and Psychological Theorizing,”

Applied Cognitive Psychology 22(6): 822 (2008).

Morgan et al., “Misinformation Can Inﬂuence Memory.”

J. C. DeCarlo, “A Study Comparing the Eyewitness Accuracy of Police Ofﬁcers and Citi-

zens,” (PhD Diss, City University of New York, 2010).

Identifying the Culprit: Assessing Eyewitness Identification

96 IDENTIFYING THE CULPRIT

citizens, but that both police and citizens made more errors when a weapon

was inferred or present.

Own-Race Bias

The race and ethnicity of a witness as it relates to that of the perpetra-

tor is another important estimator variable. In eyewitness identiﬁcation,

own-race bias describes the phenomenon in which faces of people of races

different from that of the eyewitness are harder to discriminate (and thus

harder to identify accurately) than are faces of people of the same race as

the eyewitness.

In the laboratory, this effect is manifested by higher hit

rates and lower false alarm rates (higher diagnosticity ratio) in the recogni-

tion of an observer’s own race relative to hits and false-alarms for recogni-

tion of other races.

Own-race bias occurs in both visual discrimination

and memory tasks, in laboratory and ﬁeld studies, and across a range of

races, ethnicities, and ages. Recent analyses revealed that cross-racial (mis)

identiﬁcation was present in 42 percent of the cases in which an erroneous

eyewitness identiﬁcation was made.

A recent meta-analysis of own-race bias found an interaction between

own-race bias and the duration of viewing exposure: reducing the amount

of time allowed for viewing of each face signiﬁcantly increased the magni-

tude of the bias, largely manifested as an increase in the proportion of false

alarm responses to other-race faces.

Own-race bias also interacts with the

memory retention interval; cross-race errors of identiﬁcation were greater

when there were longer periods of time between the initial exposure and

the memory retrieval.

A recent study found that “context reinstatement,”

wherein a researcher asks an individual to mentally re-create the context in

which an incident occurred, failed to inﬂuence the identiﬁcation of other-

race faces.

Although the existence of own-race bias is generally accepted, the

causes for this effect are not fully understood. Some possible explanations

are rooted in in-group/out-group models of human behavior (e.g., favorit-

R. S. Malpass and J. Kravitz, “Recognition for Faces of Own and Other Race,” Journal

of Personality and Social Psychology 13(4): 330–334 (1969).

Meissner and Brigham, “Thirty Years of Investigating the Own-Race Bias.”

The Innocence Project, “What Wrongful Convictions Teach Us About Racial Inequality,”

available at: http://www.innocenceproject.org/Content/What_Wrongful_Convictions_Teach_

Us_About_Racial_Inequality.php.

Meissner and Brigham, “Thirty Years of Investigating the Own-Race Bias.”

Ibid.

J. R. Evans, J. L. Marcon, and C.A. Meissner, “Cross-Racial Lineup Identiﬁcation: As-

sessing the Potential Beneﬁts of Context Reinstatement,” Psychology, Crime, and Law 15 (1):

19–28 (2009).

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 97

ism in which decisions regarding members of one’s own “group” are re-

garded as having greater importance than decisions regarding members of

a different “group”) and differential perceptual expertise that results from

different degrees of exposure to and familiarity with same versus other

races.

Recent work has examined the role that stereotyping might play.

One

study suggests that, in general, cross-race identiﬁcation is further impaired

when faces are presented in a group (as opposed to one at a time).

Addi-

tional research is needed to identify procedures that may help estimate the

degree of own-race biases in individual eyewitnesses following an identiﬁ-

cation procedure. Until the scientiﬁc basis for these effects is better under-

stood, great care may be warranted when constructing lineups in instances

where the race of the suspect differs from that of the eyewitness.

Exposure Duration

Eyewitness identiﬁcation researchers have long believed that exposure

duration (e.g., time spent observing a perpetrator’s face during a crime) is

correlated with greater accuracy of eyewitness identiﬁcation. The courts

also have assumed that exposure duration has an effect on identiﬁcation

accuracy.

Meta-analyses on the effects of exposure time have found that

relatively long exposure durations produce greater accuracy

and a larger

and more stable effect size for exposure duration on eyewitness identi-

H. M. Kleider, S. E. Cavrak, and L. R. Knuycky, “Looking Like a Criminal: Stereotypical

Black Facial Features Promote Face Source Memory Error,” Memory and Cognition 40(8):

1200–1213 (2012).

K. Pezdek, M. O’Brien, and C. Wasson, “Cross-Race (but Not Same-Race) Face Identiﬁca-

tion Is Impaired by Presenting Faces in a Group Rather Than Individually,” Law and Human

Behavior 36(6): 488–495 (2012).

Manson v. Brathwaite, 432 U.S. 98, 114 (1977), for example, included as a factor for

assessing the reliability and admissibility of an identiﬁcation, “the opportunity of the witness

to view the criminal at the time of the crime” and explained that this factor includes both the

length of time and the viewing conditions.

B. H. Bornstein et al., “Effects of Exposure Time and Cognitive Operations on Facial

Identiﬁcation Accuracy: A Meta-Analysis of Two variables Associated with Initial Memory

Strength,” Psychology, Crime, and Law 18 (5): 473–490 (2012). The authors state, “We used

z as the primary effect size measure for differences between proportions correct, but we also

converted z to Pearson’s r for comparability to other meta-analyses (see Tables 1 and 2). The

rs were then normalized and averaged to obtain the overall mean effect sizes. We also report

the value of Cohen’s d associated with each mean effect size” (Bornstein et al., “Effects of

Exposure Time and Cognitive Operations).” Although not deﬁned, presumably z refers to

the usual difference in means divided by its standard error, and, from their tables, their r was

calculated as z divided by the square root of the report sample size.

Identifying the Culprit: Assessing Eyewitness Identification

98 IDENTIFYING THE CULPRIT

ﬁcation accuracy.

Longer exposures were associated with higher rates

of correct identiﬁcations and lower false alarm rates. Exposure duration

may affect, or interact with, other variables, including own-race bias and

the conﬁdence–accuracy relationship assessed immediately after the lineup

decision.

The ﬁndings and conclusions from eyewitness identiﬁcation studies of

exposure duration are in keeping with much of the basic research on visual

system function (reviewed in Chapter 4). This basic research indicates that

the additional information available from longer viewing times reduces un-

certainty and enables better detection and discrimination of visual stimuli.

Retention Interval

Retention interval, or the amount of time that passes from the initial

observation and encoding of a memory to a future time when the initial ob-

servation must be recalled from memory, can affect identiﬁcation accuracy.

Laboratory studies have demonstrated that stored memories are more likely

to be forgotten with the increasing passage of time and can easily become

“enhanced” or distorted by events that take place during this retention in-

terval (see discussion of memory in Chapter 4). The amount of time between

viewing a crime and the subsequent identiﬁcation procedure can be expected

to similarly affect the accuracy of the eyewitness identiﬁcation, either inde-

pendently or in combination with other variables.

It is difﬁcult to specify the precise relationship between retention inter-

val and the accuracy of eyewitness identiﬁcation testimony and to estimate

when a lengthy retention interval will signiﬁcantly impair the accuracy of

identiﬁcation. Although, in general, it appears that longer retention inter-

vals are associated with poorer eyewitness identiﬁcation performance, the

strength of this association appears to vary greatly across the circumstances

of the initial encounter, identiﬁcation procedures, and research method-

B. H. Bornstein, K. A. Deffenbacher, E. K. McGorty, and S. D. Penrod, “The Effect of

Cognitive Processing on Facial Identiﬁcation Accuracy: A Meta-Analysis” (Unpublished manu-

script, University of Nebraska-Lincoln, 2007).

M. A. Palmer, et al., “The Conﬁdence–Accuracy Relationship for Eyewitness Identiﬁcation

Decisions: Effects of Exposure Duration, Retention Interval, and Divided Attention,” Journal

of Experimental Psychology: Applied 19(1): 55–71 (2013).

One month is the most commonly encountered delay by British police. G. Pike, N. Brace,

and S. Kynan, The Visual Identiﬁcation of Suspects: Procedures and Practice (London: Polic-

ing and Reducing Crime Unit, 2002), cited by Deffenbacher et al., “Forgetting the Once-Seen

Face.” Law enforcement authorities may have little control over the time required to identify

a suspect and obtain the cooperation of the eyewitness to participate in an identiﬁcation

procedure. Thus, retention interval has commonly been considered an estimator variable in

eyewitness identiﬁcation studies.

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 99

ologies.

A meta-analysis of published facial recognition and eyewitness

identiﬁcation studies found, for example, that an increase in the retention

interval was associated with a decreased probability of an accurate identiﬁ-

cation of a previously seen but otherwise unfamiliar face.

This same study

also found that the rate of forgetting for an unfamiliar face is greatest soon

after the initial observation and tends to level off over time, but was unable

to specify the shape of this function.

The effect of the retention interval also is inﬂuenced by the strength

and quality of the initial memory that is encoded, which, in turn, may

be inﬂuenced by other estimator variables associated with witnessing the

crime (such as the degree of visual attention) and viewing factors (such as

distance, lighting, and exposure duration). As the retention interval be-

comes longer, the opportunity for intervening events to alter the memory

also becomes greater, and other variables may interact with the retention

interval to impair performance (see also discussion of memory in Chapter

4). During the retention interval, the ability to accurately identify faces of

other races drops off especially quickly, relative to same-race accuracy.

Also, for those eyewitnesses who initially express less conﬁdence in their

identiﬁcation, there is a greater decrease in accuracy of identiﬁcation when

the retention interval is longer.

CONCLUSION

Research on eyewitness identiﬁcation has appropriately identiﬁed the

variables that may affect an individual’s ability to make an accurate iden-

tiﬁcation. Early research ﬁndings played an important role in alerting law

enforcement, prosecutors, defense counsel, and the judiciary to factors that

See J. Dysart and R. C. L. Lindsay, “The Effects of Delay on Eyewitness Identiﬁcation Ac-

curacy: Should We Be Concerned?” in The Handbook of Eyewitness Psychology: Volume II:

Memory for People, ed. R. C. L. Lindsay, D. F. Ross, J. D. Read, and M. P. Toglia. (Mahwah:

Lawrence Erlbaum and Associates, 2006), 361–373.

Deffenbacher et al., “Forgetting the Once-Seen Face.” More than 20 of the published stud-

ies included in the meta-analysis found no signiﬁcant effect of retention interval.

J. L. Marcon et al., “Perceptual Identiﬁcation and the Cross-Race Effect,” Visual Cogni-

tion 18(5): 767–779 (2010) (ﬁnding that the cross-race effect was more pronounced when

the retention interval was lengthened). Meissner and Brigham, “Thirty Years of Investigat-

ing the Own-race Bias” [meta-analysis ﬁnding that as retention time increased “participants

increasingly adopted a more liberal response criterion when responding to other-race faces.

This liberal response criterion indicated that participants required less evidence from memory

(e.g., familiarity or memorability of the face) to respond that they had previously seen an

other-race face.”].

J. Sauer et al., “The Effect of Retention Interval on the Conﬁdence–Accuracy Relationship

for Eyewitness Identiﬁcation,” Law and Human Behavior 34: 337–347 (2010) (ﬁnding greater

overconﬁdence at lengthy retention intervals).

Identifying the Culprit: Assessing Eyewitness Identification

100 IDENTIFYING THE CULPRIT

might inﬂuence the accuracy of identiﬁcations. In some jurisdictions, eye-

witness identiﬁcation research was used to improve policies and procedures

and to educate and train ofﬁcers. However, much remains unsettled in many

areas of eyewitness identiﬁcation research.

While past research appropriately identiﬁed system and estimator vari-

ables that may affect an individual’s ability to make an accurate iden-

tiﬁcation, this research might be strengthened in several ways. Greater

collaboration between the police, courts, and researchers might lead to

increased consensus on research agendas and the conceptualization of vari-

ables to be examined. More attention to reproducibility and transparency

is needed in the selection of data collection strategies and reporting of

data. Analyses need to be reported completely, including estimates of ef-

fects, conﬁdence intervals, and signiﬁcance levels. Further, in order to be

useful to stakeholders, the statistical ﬁndings of this research need to be

translated back into terms that can be readily understood by practice and

policy decision-makers.

Further, our understanding of errors in eyewitness identiﬁcation will

beneﬁt from more effective research designs, more informative statistical

measures and analyses, more probing analyses of research ﬁndings, and

more sophisticated systematic reviews and meta-analyses. In view of the

complexity of the effects of both system and estimator variables, and their

interactions, on eyewitness identiﬁcation accuracy, better experimental de-

signs that incorporate selected combinations of these variables (e.g., pres-

ence or absence of a weapon, lighting conditions, etc.) will elucidate those

variables with meaningful inﬂuence on eyewitness performance, which can

inform law enforcement practice of eyewitness identiﬁcation procedures. To

date, the eyewitness literature has evaluated procedures mostly in terms of

a single diagnosticity ratio or an ROC curve; even if uncertainty is incorpo-

rated into the analysis, many other powerful tools for evaluating a “binary

classiﬁer” are worthy of consideration.

100

When primary studies such as those described above are available in

sufﬁcient quantities, it is important that their results are synthesized us-

ing systematic reviews that conform to current best standards.

101

These

quantitative reviews would necessarily employ transparent, reproducible

procedures for locating all relevant published and unpublished research;

employ independent, duplicate procedures for selection of studies, extrac-

tion of data, and assessment of risk of bias; use meta-analytic procedures

100

Hastie, Tibshirani, and Friedman, The Elements of Statistical Learning.

101

See, e.g., A. Liberati, et al., “The PRISMA Statement for Reporting Systematic Reviews

and Meta-Analyses of Studies That Evaluate Health Care Interventions: Explanation and

Elaboration,” PLoS Medicine 6(7): e1000100. doi:10.1371/journal.pmed.1000100 (2009) and

Institute of Medicine, Finding What Works in Health Care: Standards For Systematic Reviews

(Washington, DC: The National Academies Press, 2011).

Identifying the Culprit: Assessing Eyewitness Identification

APPLIED EYEWITNESS IDENTIFICATION RESEARCH 101

that account for the heterogeneity of outcomes both within and across stud-

ies; and interpret conﬁdence intervals around pooled effects in a way that

is readily understandable by stakeholders. These systematic reviews (which

would be regularly updated as new studies are conducted) can be used to

further reﬁne the research agenda in eyewitness identiﬁcation research and

to establish priorities for funding of additional primary research.

Identifying the Culprit: Assessing Eyewitness Identification

103

Findings and Recommendations

yewitnesses make mistakes. Our understanding of how to improve

the accuracy of eyewitness identiﬁcations is imperfect and evolving.

In the previous chapters, we described law enforcement procedures

to elicit accurate eyewitness identiﬁcations; the courts’ handling of eyewit-

ness identiﬁcation evidence; the science of visual perception and memory

as it applies to eyewitness identiﬁcations; and the contributions of scientiﬁc

research to our understanding of the variables that affect the accuracy of

identiﬁcations. On the basis of its review, the committee offers its ﬁndings

and recommendations for

• identifying and facilitating best practices in eyewitness procedures

for the law enforcement community;

• strengthening the value of eyewitness identiﬁcation evidence in

court; and

• improving the scientiﬁc foundation underpinning eyewitness

identiﬁcation.

OVERARCHING FINDINGS

The committee is conﬁdent that the law enforcement community, while

operating under considerable pressure and resource constraints, is working

to improve the accuracy of eyewitness identiﬁcations. These efforts, how-

ever, have not been uniform and often fall short as a result of insufﬁcient

training, the absence of standard operating procedures, and the continuing

Identifying the Culprit: Assessing Eyewitness Identification

104 IDENTIFYING THE CULPRIT

presence of actions and statements at the crime scene and elsewhere that

may intentionally or unintentionally inﬂuence eyewitness’ identiﬁcations.

Basic scientiﬁc research on human visual perception and memory has

provided an increasingly sophisticated understanding of how these systems

work and how they place principled limits on the accuracy of eyewitness

identiﬁcation (see Chapter 4).

Basic research alone is insufﬁcient for un-

derstanding conditions in the ﬁeld and thus has been augmented by studies

applied to such speciﬁc practical problem of eyewitness identiﬁcation (see

Chapter 5). Such applied research has identiﬁed key variables that affect

the accuracy and reliability of eyewitness identiﬁcations and has been in-

strumental in informing law enforcement, the bar, and the judiciary of the

frailties of eyewitness identiﬁcation testimony.