9800_Does contextual information improve the detection of deceptive behavioral cues

Graduate Theses and Dissertations
Iowa State University Capstones, Theses and
Dissertations
2020
Does contextual information improve the detection of deceptive
Does contextual information improve the detection of deceptive
behavioral cues?
behavioral cues?
Kristen Alicia Slapinski
Iowa State University
Follow this and additional works at: https://lib.dr.iastate.edu/etd
Recommended Citation
Recommended Citation
Slapinski, Kristen Alicia, “Does contextual information improve the detection of deceptive behavioral
cues?” (2020). Graduate Theses and Dissertations. 18231.
https://lib.dr.iastate.edu/etd/18231
This Thesis is brought to you for free and open access by the Iowa State University Capstones, Theses and
Dissertations at Iowa State University Digital Repository. It has been accepted for inclusion in Graduate Theses and
Dissertations by an authorized administrator of Iowa State University Digital Repository. For more information,
please contact digirep@iastate.edu.

Does contextual information improve the detection of deceptive behavioral cues?
by
Kristen A. Slapinski

A thesis submitted to the graduate faculty
in partial fulfillment of the requirements for the degree of
MASTER OF SCIENCE

Major: Psychology
Program of Study Committee:
Stephanie Madon, Co-major Professor
Max Guyll, Co-major Professor
Zlatan Krizan
Christian Meissner

The student author, whose presentation of the scholarship herein was approved by the program
of study committee, is solely responsible for the content of this thesis. The Graduate College will
ensure this thesis is globally accessible and will not permit alterations after a degree is conferred.

Iowa State University
Ames, Iowa
2020

Copyright © Kristen A. Slapinski, 2020. All rights reserved.
ii

TABLE OF CONTENTS

Page

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv

LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v

ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi

CHAPTER 1: GENERAL INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

Potential Cues to Deception . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Behavioral Cues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Statement Cues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Contextual Information and Deception Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Increasing Reliance on Content Cues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Providing an Opportunity to Detect Correspondence Cues . . . . . . . . . . . . . . . . . . 8

Facilitating the Accurate Interpretation of Behavioral Cues . . . . . . . . . . . . . . . . . 9

Contextual Information and Schema Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

CHAPTER 2: OVERVIEW OF PROPOSED RESEARCH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Behavioral Cues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Contextual Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Combined Effects of Behavioral Cues and Contextual Information . . . . . . . . . . . 16
Experiment Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

CHAPTER 3: PRELIMINARY STUDY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

CHAPTER 4: METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Power Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

iii

CHAPTER 5: RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

The Effects of Behavioral Cues and Contextual Information on Accuracy . . . . . . . . . . . 33

Accuracy of the Dichotomous Judgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Accuracy of the Continuous Judgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

The Effects of Ground Truth, Behavioral Cues, and Contextual Information
on Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

Accuracy of the Dichotomous Judgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

Accuracy of the Continuous Judgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Frequency of Truth or Deception Judgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Dichotomous Judgment of Truth or Deception . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Continuous Judgment of Truth or Deception . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

Effect of Including an Inconclusive Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

Use of Non-verbal Behavioral Cues vs. Verbal Information . . . . . . . . . . . . . . . . . . . . . . 47

Relationship Between Individual Difference Measures, Accuracy,
and Judgments of Deception . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

CHAPTER 6: GENERAL DISCUSSION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

Hypotheses 1 and 2: Behavioral Cues Increase or Decrease Accuracy . . . . . . . . . . . . . . 50

Hypothesis 3: Contextual Information Increases Accuracy . . . . . . . . . . . . . . . . . . . . . . . 51

Hypotheses 4 and 5: The Combined Effects of Behavioral Cues
and Contextual Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
The Effects of Ground Truth on Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
The Effects of Ground Truth, Behavioral Cues, and Contextual Information
on Judgments of Truth or Deception . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Individual Difference Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

APPENDIX A: LOGIC PROBLEMS USED IN PRELIMINARY STUDY . . . . . . . . . . . . . . . . 94
APPENDIX B: INSTRUCTIONS AND CONTEXTUAL INFORMATION . . . . . . . . . . . . . . . 96
APPENDIX C: DEMOGRAPHIC QUESTIONNAIRE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
APPENDIX D: INDIVIDUAL DIFFERENCE MEASURES . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
APPENDIX E: MEASURES OF PARTICIPANT LIE-TELLING ABILITY . . . . . . . . . . . . . . 102
APPENDIX F: INSTITUTIONAL REVIEW BOARD APPROVAL LETTER . . . . . . . . . . . . . 103
iv

LIST OF TABLES

Page
Table 1 Accuracy of Judgments Across Ground Truth, Behavioral Cues,
and Contextual Information Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

Table 2 Results of Logistic Regression Including Behavioral Cues and Contextual
Information Predicting Accuracy of Dichotomous Judgment . . . . . . . . . . . . . . . . . . 80

Table 3 Results of an ANOVA Including Behavioral Cues and Contextual
Information Predicting Accuracy of Continuous Judgment . . . . . . . . . . . . . . . . . . . . 81

Table 4 Results of Logistic Regression Including Ground Truth, Behavioral Cues,
and Contextual Information Predicting Accuracy of Dichotomous Judgment . . . . . . 82

Table 5 Results of an ANOVA Including Ground Truth, Behavioral Cues,
and Contextual Information Predicting Accuracy of Continuous Judgment . . . . . . . 83

Table 6 Results of Logistic Regression Including Ground Truth, Behavioral Cues,
and Contextual Information Predicting Dichotomous Judgments of
Truth or Deception . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

Table 7 Results of an ANOVA Including Ground Truth, Behavioral Cues,
and Contextual Information Predicting Continuous Judgments of
Truth or Deception . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

Table 8 Correlations and Descriptive Statistics for Individual Difference Measures,
Accuracy, and Judgments of Deception . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

v

LIST OF FIGURES

Page
Figure 1 Accuracy of Dichotomous Judgments by Behavioral Cues,
Contextual Information Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

Figure 2 Accuracy of Dichotomous Judgments by Ground Truth,
Behavioral Cues Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

Figure 3 Accuracy of Dichotomous Judgments by Ground Truth,
Contextual Information Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

Figure 4 Proportions of Dichotomous Judgments of Deception by Ground Truth,
Behavioral Cues, and Contextual Information Conditions . . . . . . . . . . . . . . . . . . . . 90

Figure 5 Proportions of Dichotomous Judgments of Deception by
Ground Truth Conditions (p < .001) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 Figure 6 Proportions of Dichotomous Judgments of Deception by Behavioral Cues Conditions (p = .035) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 Figure 7 Proportions of Dichotomous Judgments of Deception by Contextual Information Conditions (p < .001) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 vi ABSTRACT Influential interrogation manuals assert that investigators can detect deception from a suspect’s behaviors with high rates of accuracy when they have access to contextual information about the crime (e.g., case facts, witness statements, forensic evidence). The current study provided the first empirical test of this claim. Undergraduate participants were presented with statements made by either a liar or truth-teller who denied involvement in a real transgression and judged whether they believed that the individual was lying or telling the truth. While making this judgment, participants were either able or unable to observe the individual’s behavior and received or did not receive relevant contextual information about the alleged transgression. Results provided no evidence that the provision of behavioral cues or contextual information affected accuracy of participants’ judgments independently or in combination. Exploratory analyses revealed that behavioral cues biased judgments toward deception, while contextual information biased judgments toward truth. These findings suggest that the evaluation of behavioral cues may put innocent suspects at risk of being misclassified as deceptive, whereas the provision of contextual information may bias judgments toward truth or exacerbate individuals’ pre-existing biases. 1 CHAPTER 1. GENERAL INTRODUCTION Proponents of influential interrogation methods claim that suspects leak behavioral cues during questioning (e.g., fidgeting, hesitance, crossed-arms) that can be detected and interpreted by investigators to distinguish truth from deception with high rates of accuracy (Inbau, Reid, Buckley, & Jayne, 2013). Accordingly, the initial phase of an interrogation is often dedicated to ascertaining whether suspects, on the basis of their behavior, are being truthful or deceptive. Suspects deemed deceptive are often subjected to a coercive interrogation. Because coercive interrogations can lead innocent suspects to falsely confess, and because false confessions are a leading cause of wrongful conviction, it is critically important that innocent suspects not be misclassified as deceptive on the basis of their behavior (Kassin, 2014). However, the deception detection literature has repeatedly shown that people’s ability to detect deception from behavioral cues is little better than chance, leading researchers to conclude that the interpretation of a suspect’s behaviors is of little value for detecting deception (Bond & DePaulo, 2006). Proponents of Reid-style interviewing and interrogation methods have argued that studies showing near chance effects are not valid indicators of deception detection accuracy due to limited ecological validity. For example, proponents of these methods partially attribute the tendency of prior research to find near chance accuracy rates to the absence of relevant contextual information about the event that is typically provided to investigators, and which presumably promotes the accurate assessment of a suspect’s behavioral cues (see Inbau, Reid, Buckley, & Jayne, 2013). The objective of this research was to provide the first empirical test of the claim that the presence of contextual information improves deception detection accuracy. The current chapter addresses three issues that are directly relevant to this issue. First, it describes different sources 2 of cues that have the potential to aid in the detection of deception, including those that involve the interpretation of behavioral cues and others that involve a content analysis of an individual’s statements. Second, the chapter reviews research relevant to the claim that contextual information increases deception detection accuracy. Finally, the chapter discusses a possible psychological mechanism through which contextual information could increase the accuracy of detecting deception from behavioral cues. Background In several personal and professional domains—politics, national security, marriage, business, and police interrogation—the ability to accurately distinguish between truth and deception is potentially advantageous. Unsurprisingly, deception and its detection have garnered extensive attention in the psychological literature and generated considerable public interest. However, the psychological literature indicates that laypeople and police investigators exhibit deception detection abilities that are generally poor, and that there are no behaviors uniquely associated with truth or deception (Bond & DePaulo, 2006). In response to this finding, scholars have proposed that other cues to deception exist within the content of an individual’s verbal statements that can be detected and interpreted with higher accuracy than relying on the interpretation of behavioral cues alone. Indeed, there are multiple sources of information that individuals can rely upon when attempting to detect deception, including both the individual’s behaviors and the content of their spoken statements. The following section reviews research relevant to two classes of cues to deception that have been identified in the literature: behavioral cues and statement cues. 3 Potential Cues to Deception The literature has identified two broad classes of cues that have the potential to reveal deception: behavioral cues and statement cues. Behavioral cues include non-verbal and paraverbal behaviors. Statement cues include content cues and correspondence cues. The following sections describe behavioral cues and statement cues in more detail. Behavioral Cues In the typical deception detection paradigm, some participants (judges) evaluate the veracity of statements made by other participants (senders) on the basis of non-verbal (e.g., body language) and paraverbal (e.g., voice tone, pitch, hesitancy) behavioral cues. The interpretation of behavioral cues to deception rests on the assumption that deception is accompanied by feelings of fear, guilt, or excitement that manifest themselves in a sender’s non-verbal and paraverbal behaviors (Trovillo, 1939; Ekman, 1985). For instance, according to the theory of emotional leakage (Ekman & Friesen, 1969), behavioral cues to deception can be detected via microexpressions of excitement, fear, or guilt that flash across a sender’s face for milliseconds. Building upon this assumption, other theories have attempted to identify the conditions under which these behavioral cues are more likely to emerge. For example, some theories hypothesize that behavioral cues to deception are more likely when a liar experiences heightened emotions or cognitive load (Zuckerman, DePaulo, & Rosenthal, 1981), when deception is attempted in high stakes situations (DePaulo, Ansfield, & Bell, 1996), and in response to certain investigator behaviors (Buller & Burgoon, 1996). The interpretation of behavioral cues is recommended in police interrogation manuals (see Inbau, Reid, Buckley, & Jayne, 2013) and laypeople generally believe that behavioral cues to deception exist, such as the belief that liars avoid eye contact (Taylor & Hick, 2007; Vrij, Akehurst, & Knight, 2006). 4 However, the scientific study of deception detection indicates that deception detection accuracy is only slightly above chance. Most notably, a meta-analysis of 206 deception detection studies found that judges accurately classified lies and truths at a rate of 54%, only 4% better than what would be expected by chance alone (Bond & DePaulo, 2006). There are two related explanations that may account for this finding. First, liars and truth-tellers can experience similar feelings of fear and stress, suggesting that behavioral cues indicative of internal emotions are unreliable indicators of deception (DePaulo, Lindsay, Malone, Muhlenbruck, Charlton, & Cooper, 2003; Ofshe & Leo, 1997). Second, the actual behaviors of liars versus truth-tellers tend to be more similar than different, and the few behavioral cues that do distinguish between them tend to be faint (see DePaulo et al., 2003; Sporer & Schwandt, 2006, 2007). Consistent with these findings, there is some indication in the empirical literature that the presence of behavioral cues can be detrimental to deception detection accuracy. In particular, there is evidence that judges become less accurate at detecting deception when non-verbal behavioral cues are present than when they are absent (Anderson, DePaulo, Ansfield, Tickle, & Green, 1999; Bond & DePaulo, 2006; Mann, Vrij, & Bull, 2004). For example, when judges assessed audio or videotaped statements of senders either truthfully or deceptively confessing to crime, accuracy rates were lower for those assessing videotaped statements that included both paraverbal and non-verbal behaviors than those assessing audiotaped statements that included only paraverbal behaviors (Kassin, Meissner, & Norwick, 2005). Thus, the addition of non- verbal behaviors decreased accuracy. Similarly, while paraverbal behaviors do not appear to be particularly detrimental, they do not increase deception detection accuracy. For example, judges are no more accurate at identifying deception from audio or audiovisual presentations of a sender’s statements than they are at identifying deception from transcripts of a sender’s 5 statements (Bond & DePaulo, 2006). The findings that non-verbal behaviors are detrimental to accuracy and paraverbal behaviors are largely unhelpful raises the possibility that the interpretation of behavioral cues may generally be detrimental to deception detection accuracy. Statement Cues In response to mounting evidence that judges are not able to accurately distinguish between truth and deception through the interpretation of behavioral cues alone, some researchers have suggested that higher accuracy rates might be obtained if judges based their evaluations on statement cues – i.e., cues to deception contained within the verbal content of statements (Blair, Levine, & Shaw, 2010; Köhnken, 1987; Reinhard, Sporer, Scharmach, & Marksteiner, 2011). Importantly, the interpretation of statement cues to deception does not necessitate the observation of behavior. This is because statement cues can be available even when behavioral cues are absent, such as when judges are presented with a transcript of a sender’s statements. As briefly noted above, the literature has distinguished between two types of statement cues: content cues and correspondence cues. Content Cues Content cues refer to the verbal content of a sender’s statement, such as the amount of detail, consistency, completeness, and vividness the statement conveys (Reinhard, Sporer, Scharmach, & Marksteiner, 2011). Research shows that liars' and truth-tellers' statements differ in terms of content cues. Liars, for instance, make statements that are less detailed, consistent, and plausible than those made by truth-tellers (DePaulo et al., 2003). There is increasing evidence that relying on the analysis of content cues improves deception detection accuracy (Reinhard, Sporer, Scharmach, & Marksteiner, 2011; Reinhard, Sporer, & Scharmach, 2013), and this has led researchers to develop new techniques to aid in 6 deception detection. For example, content-based criteria analysis, which was developed specifically for cases of suspected child sexual abuse, assesses truthfulness by analyzing the content cues within a child’s statement, such as the degree to which the statements are characterized by a logical structure and unusual or superfluous details (Steller & Köhnken, 1989). Empirical research has shown that truthful versus deceptive accounts differ on these criteria, thereby supporting the efficacy of using this method to distinguish between truth and deception (Lamb et al., 1997; Vrij, 2005; Vrij, Akehurst, Soukara, & Bull, 2002). Similarly, the verifiability approach trains investigators to ask suspects to provide verifiable details about their alibi. Because deceptive suspects tend to provide fewer verifiable details than truth-tellers, assessing this content cue increases deception detection accuracy (Nahari, Vrij, & Fisher, 2012; Vrij & Nahari, 2019). Overall, the effectiveness of these methods suggest that assessing content cues, such as the number of details reported, can increase deception detection accuracy rates. Correspondence Cues By contrast, correspondence cues are cues to deception that are derived from an analysis of consistency between the verbal content of a sender’s statements and contextual information about the crime (e.g., evidence, witness statements). Presumably, truthful senders will tend to make statements that correspond to the contextual information, whereas deceptive senders will tend to make statements that do not correspond to the contextual information. For example, if the police identified a suspect’s fingerprints at a crime scene, but the suspect denies ever having been there, then the investigator can infer that the suspect is probably lying. Investigators often rely on correspondence cues to arrive at judgments of deception (Granhag & Strömwall, 1999; Granhag & Strömwall, 2002; Granhag, Strömwall, & Jonsson, 2003), and empirical research has 7 demonstrated that judges achieve superior accuracy rates when they are able to interpret correspondence cues (Blair, Levine, & Shaw, 2010; Blair, Reimer, & Levine, 2018). Contextual Information and Deception Detection The provision of crime-related contextual information – for instance, the provision of case facts, witness statements, and/or forensic evidence – has the potential to increase deception detection accuracy through three mechanisms: (1) by increasing judges’ reliance on content cues; (2) by providing judges the opportunity to detect correspondence cues; (3) by facilitating the accurate interpretation of behavioral cues to deception. These mechanisms are described in detail below. Increasing Reliance on Content Cues There is some evidence that the provision of crime-relevant contextual information can improve the interpretation of content cues to deception. For example, judges who were provided with contextual information reported relying more heavily on content cues to deception, and thereby demonstrated greater deception detection accuracy than judges who were not provided with contextual information (Reinhard, Sporer, Scharmach, & Marksteiner, 2011; Reinhard, Sporer, & Scharmach, 2013). Importantly, this effect occurred even though the contextual information contained no information that could be directly compared to the content of the suspect’s statement. This suggests that the provision of relevant contextual information may increase judges’ perceived familiarity with a situation. In turn, because judges feel more familiar with the situation, they may be more likely to attend to content cues to deception over behavioral cues. The tendency to rely more on diagnostic content cues than behavioral cues may enable judges to better distinguish between liars and truth-tellers. 8 Providing an Opportunity to Detect Correspondence Cues By definition, the interpretation of correspondence cues to deception necessitates having knowledge of crime-relevant contextual information. Accordingly, contextual information increases deception detection accuracy by creating the potential for correspondence cues to signal when the verbal content of a sender’s statement fails to correspond to known contextual information (Blair, Levine, & Shaw, 2010; Blair, Reimer, & Levine, 2018; Reinhard, Sporer, Scharmach, & Marksteiner, 2011). For example, in an alleged cheating incident, judges who were provided with contextual information indicating that a set of logic problems were so difficult that no one could get more than two correct without cheating achieved a higher deception detection accuracy rate than those who were not provided with this contextual information (Blair, Levine, & Shaw, 2010). Overall, this suggests that the interpretation of correspondence cues increases the accuracy of deception detection judgments independently of behavioral cues and content cues. Recognizing the potential utility of correspondence cues, researchers have developed several deception detection techniques that emphasize the importance of attending to correspondence cues. The strategic use of evidence technique, for example, trains judges to use contextual information to elicit correspondence cues to deception, leading to an increase in deception detection accuracy. Specifically, judges are trained to acquire relevant evidence and contextual information pertinent to the suspect and the crime prior to an interview, to ask increasingly specific questions about the withheld evidence during the interview, and only at the end of the interview to reveal the evidence they have. The objective is to keep senders unaware of known evidence in order to elicit more statement-evidence inconsistencies that are likely to be indicative of deception (see Granhag, Strömwall, Willen, & Hartwig, 2012; Hartwig, Granhag, 9 Strömwall, & Kronkvist, 2006; Hartwig, Granhag, Strömwall, & Vrij, 2005). Judges who use this technique exhibit superior accuracy compared to those who do not use the technique (Hartwig, Granhag, Strömwall, & Kronkvist, 2006). The effectiveness of using contextual information to detect correspondence cues to deception is also supported by other deception detection methods. For example, investigators using cognitive load approaches are trained to increase the suspect’s attentional demands in order to elicit contradictions between their statements and known contextual information, thereby increasing deception detection accuracy (Vrij, Fisher, Mann, & Leal, 2008). Overall, this suggests that training methods that elicit correspondence cues between the content of a sender’s statement and contextual information can improve deception detection accuracy. Facilitating the Accurate Interpretation of Behavioral Cues A third mechanism by which contextual information could increase accuracy is through the interpretation of behavioral cues. According to proponents of the Reid Technique of Interviewing and Interrogation—the most commonly used police interrogation manual in the world—providing contextual information to judges increases deception detection accuracy by facilitating the degree to which they accurately interpret suspects’ non-verbal and paraverbal behavioral cues to deception (Inbau, Reid, Buckley, & Jayne, 2013). For example, to address criticisms of their behavior-based deception detection methods, authors of the training manual (Inbau, Reid, Buckley, & Jayne, 2013) argue that the ability for judges to accurately interpret behavioral cues to deception increases when they are provided with contextual information: “When an investigator understands the context in which an interview is taking place (for example, the case facts and background information) accuracy in the assessment of a subject’s behavior symptoms greatly increases.” (p. 103). 10 Moreover, the authors claim that many deception detection studies have failed to yield high accuracy rates, in part, because: “There was little consideration given to evaluating behaviors in context. For example, identifying whether specific non-verbal behaviors are appropriate given the verbal content of the suspect’s response, identifying the consistency of the suspect’s statement across time and with known evidence, and so on.” (p. 104). Thus, while the authors do note that contextual information may increase accuracy through mechanisms such as content or correspondence cues, they additionally argue that contextual information should orient the investigator’s attention toward specific non-verbal behaviors indicative of deception. In other words, the authors of the training manual do not merely argue that contextual information improves deception detection by orienting the investigator’s attention toward content cues, or by initiating a comparison between contextual information and the content of a suspect’s statement. Instead, they argue that contextual information provides the foundation for the accurate interpretation of behavioral cues to deception. Presuming this to be true, the training manual specifically instructs police to use case- relevant contextual information when interpreting a suspect’s behavioral cues for deception. In fact, the manual includes a non-verbal lie detection method of questioning, the Behavior Analysis Interview—a non-accusatory interview conducted before an interrogation in which investigators are encouraged to interpret a suspect’s behaviors in order to determine if the suspect is being truthful or deceptive. This evaluation is then used as the basis for deciding whether or not to proceed with an accusatory interrogation. The behavioral analysis interview involves first collecting background information from the suspect in order to establish a sense of the suspects’ normative or baseline behaviors. Then, the interview proceeds by asking the suspect a series of fifteen behavior-provoking questions. 11 For example, in an alleged theft case, behavior-provoking questions might include “Do you know who stole the money?” and “Have you ever thought about stealing money?”. According to the manual, these behavior-provoking questions mobilize guilty individuals, thereby increasing physiological arousal and excitement. For example, compared to innocent suspects, guilty suspects undergoing a behavior analysis interview are expected to be less helpful and sincere and to exhibit more behaviors indicative of nervousness, such as by appearing more guarded, having a closed, static, and non-frontal body posture, and displaying less illustrative behaviors with their hands (Inbau, Reid, Buckley, & Jayne, 2013). Because these questions are intended to elicit different behavioral responses from innocent and guilty suspects, proponents of these methods argue that by conducting a behavior analysis interview, investigators are able to reliably detect and interpret behavioral cues indicative of truth or deception in a manner akin to detecting physiological arousal with a polygraph (Horvath & Jayne, 1990). Despite widespread use of this behavior-based lie detection tool, little research has been conducted that has supported its effectiveness. Initial research conducted by affiliates of the training manual reported that investigators using the behavior analysis interview achieved high deception detection accuracy rates (Blair & McCamey, 2002; Horvath, Jayne, & Buckley, 1994). However, several methodological limitations, including that the ground truth of senders’ statements could not be established, have led scholars to question the validity of the findings (Masip, Herrero, Garrido, & Barba, 2011; Vrij, Mann, & Fisher, 2006). In another study in which ground truth was established by experimentally manipulating the sender’s guilt status, innocent suspects undergoing a behavior analysis interview exhibited more nervous behaviors and were less helpful to investigators than guilty suspects, a finding that is in direct opposition of the claims made by the training manual (Vrij, Mann, & Fisher, 2006). Thus, the results from the 12 limited empirical research examining the behavior analysis interview have not conclusively established its effectiveness. In addition to the lack of empirical support, there is no research that supports or refutes the validity of the training manual’s claim that contextual information increases the accuracy of behavioral cue interpretation. Although contextual information can increase the deception detection accuracy rate, it is unclear whether the mechanisms underlying this effect are solely due to the interpretation of statement cues, or also due to the interpretation of behavioral cues. In particular, because no research has simultaneously manipulated the presence or absence of contextual information and the presence or absence of behavioral cues, it is impossible to know whether the increase in deception detection accuracy that is found when contextual information is provided could be partially attributed to an increase in the accurate interpretation of behavioral cues (Blair, Levine, & Shaw, 2010; Blair, Reimer, & Levine, 2018). Consistent with the training manual’s claims, contextual information might increase deception detection accuracy by facilitating the degree to which people correctly interpret another’s behavioral cues to deception. If this is the case, then the presence of behavioral cues should increase deception detection accuracy above and beyond any effects attributable to contextual information. Alternatively, contextual information might only increase deception detection accuracy by increasing people’s reliance on content and correspondence cues to deception, such as by leading them to compare a suspect’s verbal statements to available contextual information. If this is the case, then accuracy in deception detection judgments may actually be greatest when behavioral cues (that might otherwise mislead or distract) are absent. Distinguishing between these alternative explanations requires research that orthogonally 13 manipulates both contextual information and behavioral cues to deception, which has not previously been done. Contextual Information and Schema Theory Although authors of the training manual did not offer a theoretical explanation as to why contextual information might facilitate the accurate interpretation of a suspect’s behavioral cues, schema theory provides a mechanism that could provide support for the training manual’s claim. Specifically, contextual information could activate knowledge structures that guide how individuals encode and interpret incoming information, and in turn affect memory and comprehension (see Bransford & Johnson, 1992; Taylor & Crocker, 1981; Trope, 1986). In this way, schematic information processing may guide attention away from irrelevant perceptual details and toward more relevant information (von Hippel, Jonides, Hilton, & Narayan, 1993). With respect to police interrogation, for example, contextual information about a crime could activate a schematic representation of the criminal event. In turn, this schematic representation may organize and guide an investigator’s attention away from irrelevant behavioral cues, and conversely, toward diagnostic behavioral cues to deception, thereby increasing deception detection accuracy. Schematic information processing also simplifies cognitive processing and frees up attentional resources to attend to other information (Macrae, Milne, & Bodenhausen, 1994). Accordingly, a schema that is activated by crime-relevant contextual information may reduce the cognitive resources an investigator needs to understand the background of a case via the suspect’s spoken statement, thereby allowing the investigator to allocate more of their cognitive resources to the interpretation of behavioral cues. The availability of these cognitive resources may enable the investigator to allocate more of their information processing resources to 14 detecting and interpreting a suspect’s subtler behavioral cues to deceit, thereby improving the investigator’s ability to accurately detect deception. 15 CHAPTER 2. OVERVIEW OF PROPOSED RESEARCH Hypotheses This experiment tested several hypotheses relevant to the effects of behavioral cues and contextual information on deception detection accuracy. Behavioral Cues I tested competing, directional predictions about the effect of behavioral cues on deception detection. Hypothesis 1: Behavioral Cues Increase Accuracy One the one hand, theories of emotional leakage (e.g., Ekman & Friesen, 1969) propose that deceptive individuals show behaviors that can be detected by observers. Accordingly, this perspective predicts judges should exhibit greater deception detection accuracy when behavioral cues are present than when they are absent. Hypothesis 2: Behavioral Cues Decrease Accuracy On the other hand, empirical research has demonstrated that deception detection accuracy is greater when judges are provided with transcribed or audiotaped statements than when they are provided with videotaped statements (e.g., Kassin, Meissner, & Norwick, 2005), perhaps because visual behaviors distract judges from statement cues to deception. This view corresponds to the prediction that judges should exhibit greater deception detection accuracy when behavioral cues are absent than when they are present. Contextual Information I made a single, directional hypothesis with respect to the effect of contextual information on deception detection accuracy. 16 Hypothesis 3: Contextual Information Increases Accuracy Specifically, contextual information should increase reliance on statement cues and thereby increase deception detection accuracy (e.g., Blair, Levine, & Shaw, 2010; Blair, Reimer, & Levine, 2018). Thus, I predicted that when contextual information is present, judges should exhibit greater deception detection accuracy than when contextual information is absent. Combined Effects of Behavioral Cues and Contextual Information This experiment tested competing, directional predictions about the combined effects of behavioral cues and contextual information on deception detection accuracy. Hypothesis 4: Contextual Information Improves Interpretation of Behavioral Cues Proponents of the interviewing methods presented in the Reid training manual assert that contextual information should improve the interpretation of behavioral cues to deception, further increasing accuracy (see Inbau, Reid, Buckley, & Jayne, 2013). This perspective corresponds to an interaction, such that the improvement in accuracy created by the presence of behavioral cues is greater when contextual information is present than when contextual information is absent. However, it is not feasible to test for the presence of an interaction effect in this sample with sufficient power. For example, if I were to predict that the absence of contextual information would attenuate the effect of behavioral cues on accuracy by 50%, I would need to collect 14 times the number of participants that I would need to test for a main effect of behavioral cues (N = 448), resulting in a total sample size of 6,272 (Giner-Sorolla, 2018). Therefore, given a non-significant interaction term, I will further test the hypothesis that contextual information improves the interpretation of behavioral cues through a pairwise comparison that will assess whether accuracy rates follow the predicted pattern. Specifically, 17 this perspective would predict that when contextual information is present, accuracy will be greater when behavioral cues are present than when they are absent. By contrast, this perspective does not correspond to any clear prediction about the effect of behavioral cues on accuracy when contextual information is absent. Therefore, if accuracy follows the predicted pattern when contextual information is present, I will informally compare the sizes of the differences in accuracy when behavioral cues are absent versus present to assess whether the difference is greater when contextual information is present. Hypothesis 5: Behavioral Cues Distract from Interpretation of Contextual Information As detailed above, empirical evidence suggests that the provision of behavioral cues worsens deception detection (e.g., Kassin, Meissner, & Norwick, 2005), whereas the provision of contextual information improves deception detection (e.g., Blair, Levine, & Shaw, 2010). These results pertaining to their separate effects suggest two specific predictions about how they could operate in combination to affect deception detection accuracy. Specifically, when contextual information is provided, one would expect accuracy to be greater. However, the additional provision of behavioral cues may distract judges from statement cues and decrease accuracy. This perspective corresponds to an interaction, such that the improvement in accuracy created by the presence of contextual information is greater when behavioral cues are absent than when they are present. However, for the reasons described above, it is not feasible to test for the presence of an interaction in this sample with sufficient power. For this reason, I will use the same analytic approach described above. Namely, given a non-significant interaction term, I will further test the hypothesis that behavioral cues distract or mislead judges from accurately evaluating contextual information through two pairwise comparisons, followed by informal comparisons of the sizes of differences in accuracy. 18 In particular, this perspective would predict that when contextual information is present, accuracy will be greater when behavioral cues are absent than when they are present. It should be noted that this prediction is opposite to the prediction implied by proponents of the Reid training manual, which I articulated above. Furthermore, when behavioral cues are absent, one would expect no distraction from the beneficial effect expected by providing contextual information. Therefore, this perspective would also predict that when behavioral cues are absent, accuracy will be greater when contextual information is present than when it is absent. If accuracy follows the predicted pattern, I will informally compare the sizes of the differences in accuracy when contextual information is absent versus present to assess whether the difference is greater when behavioral cues are absent. Experiment Overview This experiment provided the first test of whether the provision of contextual information increases the accuracy with which judges can interpret behavioral cues to deception. To do so, the experiment manipulated the presence of contextual information and the presence of behavioral cues in order to isolate their unique effects. Participants were presented with statements made by a liar or truth-teller proclaiming innocence in an alleged cheating incident. These statements were either presented in the form of an audiotaped transcript narrated by a text- to-speech program in which only statement cues were present, or in the form of a videotape that communicated both the statement cues as well as any behavioral cues the liar or truth-teller might have leaked. Further, contextual information about the cheating incident was either withheld from or provided to participants.