Keywords

The purpose of this chapter is to review the available research on Human Intelligence (HUMINT) interrogations. We will argue that there has been a recent paradigm shift in the approach to HUMINT interrogations. We will describe the conceptual, methodological, and practical implications of this paradigm shift. The chapter will be structured as follows. First, we will describe the defining characteristics of HUMINT interrogations and outline the scope of our discussion. We will describe how the challenges of HUMINT interrogations may be similar as well as different from interrogations in criminal settings. Second, in order to provide a context for our claim of a paradigm shift, we provide a historical overview of practice and research on HUMINT interrogations. Third, we offer a review of the current state of knowledge about the psychology of HUMINT interrogations, with a particular focus on methods that have been shown to be effective. Finally, we will outline several challenges for future research in this domain, and discuss how research on HUMINT may proceed to fill the gap in current knowledge.

Interrogations in HUMINT and Criminal Contexts

Interviews and interrogations are major elements of intelligence gathering. The primary goal of intelligence collection from human sources is to elicit accurate and operationally useful information from another person. As straightforward as this may seem at first glance, HUMINT collection is associated with substantial difficulties, involving for example, the complexities of cross-cultural interactions, the unreliable nature of human memory, and the inherent difficulties of distinguishing between true and false statements (Evans et al. 2010). In the last decade, the use of psychologically and physically coercive interrogation techniques that aim at reducing resistance and producing compliance has received considerable attention in the media, particularly the notorious so-called enhanced interrogation techniques employed under the Bush administration (e.g., Hersh 2004). Such techniques have been met with wide criticism, based on the arguments that they are ineffective and that they may violate international laws (Costanzo and Gerrity 2009). It is outside the scope of our chapter to provide a discussion of the legality and ethics of harsh interrogation techniques (for such discussions, see Abeles 2010; Boehm 2009; Zimbardo 2007). The point is simply that HUMINT policy and practice is now a common topic in the public discourse. However, the empirical body of work on HUMINT interrogations is still relatively small—an issue we will return to later in this chapter.

Interrogations are conducted in the realm of intelligence gathering. In the criminal justice systems, interviews and interrogations of suspects are important elements of criminal investigations. Recently, research has suggested that there may be substantial problems in current interrogation practice. More specifically, a wave of DNA exonerations has demonstrated that false confessions produced during a police interrogation are significant contributors to erroneous convictions (Cutler 2012; Drizin and Leo 2004;). For example, data from the Innocence Project suggests that false confessions are involved in around 25 % of the known large number of erroneous convictions cases (http://www.innocenceproject.org; see also Sheck et al. 2000). A substantial proportion of these false confessions appear to stem from coercive and manipulative interrogation techniques (Kassin et al. 2010; Leo 2001). Partly as a function of the problem of false confessions, interrogations in criminal contexts have received considerable empirical attention in the last decades (Gudjonsson 2003; Kassin and Gudjonsson 2004).

Before we review the available research on interrogation, let us briefly outline the similarities and differences between HUMINT and criminal interrogations. First, how can interrogation be broadly defined? Evans et al. (2010) note that there is no consensus on a definition, after which they proceed to offer one definition of interrogation as “the systematic questioning of an individual perceived by investigators as noncooperative, within a custodial setting, for the purpose of obtaining reliable information in response to specific requirements” (p. 3). We rely on this definition here, with the exception that we do not consider interrogation to necessarily involve lack of cooperation on behalf of the individual being questioned.

Based on this definition, HUMINT and criminal interrogations have in common that they involve questioning of a suspect or source, with the particular aim of producing information. However, in our view, one of the critical differences between HUMINT and criminal interrogations is in the kind of information they aim at producing. Criminal interrogations typically focus on generating information about isolated, past events. Within this focus, there are some cultural differences in the focus of criminal interrogation. For example, interrogations in criminal investigations in the USA are heavily geared towards producing confessions (Kassin 1997). In many western European legal systems, the focus is instead on generating objective facts about the crime in question (Bull and Milne 2004;). Regardless of these differences, a characteristic element of criminal interrogations is that they are retrospective, closed-ended, and often confirmatory in nature, meaning that they typically attempt to produce information about single events in the past.

Compared to criminal interrogations, HUMINT interrogations may be more complex in nature. Generally speaking, HUMINT interrogations aim to map not only isolated past events, but also networks of people and events—i.e., large-scale operations. Moreover, they may be both retrospective and prospective in nature. That is, the goal of the interrogation may include not only mapping the past but also generating information about intentions, plans, and possible future events. It may be worth noting that the challenges of HUMINT are, in these respects, similar to those facing investigators of organized crime.

An additional difference between criminal and HUMINT interrogation is in the sociocultural dynamic between the interrogator and the source. In contrast to most criminal interrogations, HUMINT interrogations are frequently cross-cultural interactions. That is, the interrogator and the source may not share ethnicity, culture, or language. Because of this, HUMINT interrogations are sometimes conducted with the use of interpreters. Unfortunately, the dynamics of interviewing using interpreters have not been extensively mapped (Granhag et al. 2005).

A Historical Overview of HUMINT Interrogation Practice and Research

Before we turn to the current scientific research on HUMINT interrogations, we provide a brief historical overview of HUMINT interrogation. The purpose of this overview is to illustrate the fundamental change in the ethos of HUMINT interrogation policy, practice, and research. First, a note on the chronological scope of our review. We believe that a turning point in HUMINT policy and practice occurred in 2009, when President Obama signed Executive Order 13491, formally establishing the “United States Army Field Manual 2-22.3, Human Intelligence Collector Operations” as the norm for HUMINT interrogation (Brandon 2011). By simultaneously revoking order 13440, President Obama established that all HUMINT interrogation should conform to the Geneva convention, effectively banning the Bush regime’s program of so-called enhanced interrogation techniques. Furthermore, the executive order called for the creation of a special task force, charged with the task of studying and evaluating current interrogation practice (i.e., the effectiveness of the techniques outlined in the Army Field Manual). Subsequently, the High-Value Detainee Interrogation Group (HIG) was formed. HIG’s mission statement describes that it “will study the comparative effectiveness of interrogation approaches and techniques with the goal of identifying the most effective existing techniques and developing new lawful techniques to improve intelligence interviewing.” Thus, an official research agenda to examine HUMINT interrogation techniques was put in place. In the section below, we aim to summarize the state of the field leading up to Obama’s policy change.

Sources of Data.

How might one find information about past interrogation policy and practice? There are several sources of data. A body of descriptive literature exists chronicling the experiences of HUMINT collectors and some of the techniques they have employed over the course of a number of conflicts. In addition, official manuals used in HUMINT interrogations are widely available both from booksellers and general sites on the Internet. Journalists have also reported extensively about HUMINT interrogations and have described practices sanctioned by the military as well as methods that the US military has not officially approved. Moreover, researchers have examined the behavior of American prisoners of war (POWs) during their internment and have sought to understand under what circumstances POWs become compliant and provide their captors with information. This early empirical work, designed to develop defensive interrogation strategies, or the ability to resist interrogations, ceased after the Korean War but still represents some of the most significant research to date about interrogations in the military context. In addition, individual interrogators, from World War II through the recent conflicts in Iraq and Afghanistan, have described their activities. Official information about the policies and practices of the Central Intelligence Agency (CIA) are more difficult to find. Reports generally consist of historical documents and books that use declassified material (McCoy 2006; Otterman 2007; Weiner 2007). Some declassified historical documents from the CIA are accessible via the Internet (e.g., Pribbenow, n.d.). Retired agents have written accounts of their service but these are subject to approval by the CIA (e.g., Scheuer 2007). Most significantly, the CIA sponsored much of the early scientific research about offensive and defensive interrogation strategies until alarm over the research caused the agency to cease its sponsorship (McCoy 2006).

Accounts from POWs: World War II and the Korean War.

Accounts of the treatment of the US POWs in World War II and the Korean War suggest that the stresses of war and capture can make sources compliant, but have a negative effect on their ability to provide accurate and actionable intelligence (Biderman 1957, 1959). Isolation, sensory deprivation, and poor living conditions in prison camps were found to be negatively correlated with prisoners’ ability to provide accurate intelligence, and would sometimes induce false confessions (Biderman 1960; Biderman and Zimmer 1961). With the advent of the Cold War, communist regimes frequently used interrogations to elicit false confessions or other propagandistic statements from captured US military personnel. For example, in the Korean War, 36 American airmen confessed to “a plot to bomb civilian targets” after undergoing what was then called “touchless torture” (Margulies 2006). High-value prisoners were isolated from all human contact, except for their interrogators, and at least one airman was held in solitary confinement for 10 months. The North Koreans and the Chinese subjected these prisoners to stress positions, such as standing for hours, and prolonged interrogation sessions during which questions were repeated over and over to disorient the subject. Guards threw food at the prisoners and forced them to defecate in public. According to Margulies, after being “exhausted and demoralized”, the airmen’s resistance was overcome and they eventually confessed. All the confessions elicited under these conditions were false.

Farber et al. (1957) also examined techniques to elicit false confessions, self-denunciations, or propagandistic statements used on American prisoners of war captured during the Korean conflict (see also Carlson 2002). During their captivity, American military personnel were subjected to extreme conditions including sleep deprivation, malnutrition, isolation, poor medical care, and continual threats of death or bodily harm. Farber et al. suggested that the prisoners’ confinement included three overall elements (debility, dependency, and dread—DDD) that interacted to produce an overall, psychologically weakened effect on the prisoners. Debility was produced because prisoners were deprived of sleep, were denied consistent meals, and suffered from fatigue. In addition, prisoners were often in chronic pain resulting from untreated wounds and other medical problems that were ignored. As a result, captured personnel could not resist even minor physical abuse. Dependency was induced by the captives’ weakened physical state and the fact that they were incapable of fulfilling their most basic needs. Dread resulted because prisoners were subjected to violence and continual threats of violence, loss of control, and even their inability to satisfy the demands of their interrogators.

In stark contrast to the approaches described above are the interrogation techniques employed by the legendary interrogator Hanns Joachim Scharff, who worked for the German Luftwaffe during World War II. Scharff was famed for his charm and his ability to extract information without coercion (Shoemaker 2008; Toliver 1997). His approach was characterized by two critical elements: First, he employed a humane approach to the people he interrogated. His style was conversational rather than adversarial, and he did not press the subjects for information (Scharff 1950). Instead, he learned as much as he could about the backgrounds of his subjects, including detailed personal information, before he started an interrogation. During the interrogations, he often told elaborate stories rather than bombarding the source with requests for information (Granhag et al. in press). Second, he approached interrogation strategically, viewing its purpose as gaining reliable information. In fact, his conversational style and tendency to tell stories (and allowing the subject to correct him or to confirm certain information in his stories) was a highly deliberate strategy, aimed at what is referred to as information elicitation. The goal of information elicitation was to produce information in such a fashion that the source remains unaware of the real aim of the exchange (Brandon et al. 2011). In other words, by creating an “I already know it all” illusion, sources would inadvertently supply information that Scharff did not previously know (Granhag et al. in press).

In sum, the evidence stemming from studies of POWs suggests that HUMINT interrogation during parts of the twentieth century were, with a few notable exceptions, characterized by psychological and physical coercion. Such methods certainly had powerful effects on subjects—they produced confessions and propaganda. However, such methods did not appear to be effective in generating reliable and actionable intelligence.

US Government Sponsored Research.

The CIA directly or indirectly sponsored much of the research on offensive interrogation techniques until the 1970s (Kleinman 2006; McCoy 2005, 2006; Otterman 2007). Offensive interrogations are those designed to elicit information from a subject (in contrast to defensive interrogation techniques, which are designed to help military personnel withstand interrogation). A range of avenues were explored, including manipulation of the physical environment and psychological states through sensory deprivation isolation, and the use of psychoactive drugs. At its height, the research involved 80 institutions, 44 of them colleges or universities (Thomas 1977).

Sensory deprivation is aimed at drastically reducing the level of a person’s normal sensory stimulation for a relatively prolonged period of time (Goldberger 1982). Anecdotal evidence from people who have experienced some form of sensory deprivation, such as prisoners, explorers, and the victims of ship wrecks, uniformly describe a craving for sensual stimuli as well as changes in thinking, feeling, and perception that can be accompanied by hallucination-like visions (Kubzansky 1961). Hinkle (1961) discussed experiments conducted by Hebb, another sometime CIA-funded researcher, on sensory deprivation. He concluded that within a few hours, sensory deprivation began to exact a toll on brain function, similar to the effects of beatings, starvation, or lack of sleep. In experiments where researchers reduced levels of perceptual sensation, primarily by restricting visual input, subjects experienced a generally disorganizing effect. In studies on general sensory deprivation (with the deprivation ranging from 3 min to 6 days), researchers observed a variety of cognitive and behavioral effects, including breakdown of visual-motor coordination, loss of accuracy in tactual perception and spatial orientation, difficulty in focusing, fluctuating curvature of lines and surfaces, and a general decrease in the efficiency of perceiving relevant stimuli (Kubzansky 1961).

Researchers also studied the psychological effects of isolation. Across studies, isolation negatively affected subjects’ cognitive and decision-making capacities. Bexton et al. (1954) found that even after the experiment concluded, some subjects still experienced confusion. Overall, the predominant experiences after isolation were fatigue, drowsiness, confusion, loss of orientation, and a need to reorient oneself to the familiar aspects of reality (Kubzansky 1961). Kubzansky commented that depending on the relationship between interrogator and subject, the interrogator, in a case of isolation, might become associated with a relief from discomfort. As a result, a subject may be more willing to cooperate with the interrogator even though cognitive abilities and hence the ability to provide accurate information would be substantially impaired.

In addition, the CIA supported research on the effectiveness of psychoactive drugs to induce compliance and produce information. For example, the agency kept nine federal inmates high on LSD for 77 days to test the drug’s effects (Weiner 2007). There were also experimental studies on other hallucinogenic drugs such as mescaline, as well as barbiturates (e.g., amobarbital). The conclusion from this work was that while subjects under the influence of drugs may produce information they otherwise would not, the reliability of this information was in serious doubt (Gottschalk 1961).

In sum, the CIA research corroborated the results of the POW studies and reinforced the conclusion that sensory deprivation and isolation, even for short periods, can seriously debilitate subjects. Hence, such techniques are not conducive to effective intelligence gathering. Although we acknowledge the glaring ethical problems this research raises, we will not discuss them here (see Skerker 2012 for an excellent and extensive treatment of these issues). Instead, we want to highlight the implicit model of HUMINT interrogation inherent in this research. It seems to be based on what we may a compliance through stress model. This model has at least three assumptions. First, sources will be reluctant to cooperate during interrogation. Second, by applying psychological (or physical) stress, such resistance will be broken down. And third, when resistance has been broken down, sources will be willing (and able) to produce reliable intelligence. As our review of current research will show, this model sharply contrasts with modern views on effective interrogation.

Relevant Research on Interviewing and Interrogation

As we note above, the modern scientific literature focusing specifically on HUMINT interrogation is slim. However, there is a large body of empirical work that has bearing on the challenges of HUMINT. Here, we review some of this research, and suggest important avenues for future research. First, we discuss the available literature on interview and interrogation techniques, focusing on approaches that both promote cooperation and effectively elicit information from memory. Second, we describe research on the challenges of assessing the credibility of information elicited during an interrogation, with a focus on recent research suggesting that certain interview approaches may facilitate the detection of deception. Finally, we identify several areas in need of further research with the goal of developing ethical, evidence-based approaches to HUMINT interrogation (see also Evans et al. 2010).

Methods of Intelligence Interviewing and Interrogation.

At the heart of an effective HUMINT interview lies the methods used by an interrogator to break a subject’s resistance, solicit and maintain their cooperation, and elicit from them detailed strategic information relevant to national-security interests. Much of the current training that both criminal investigators and intelligence personnel receive is based upon customary knowledge—interview practices that have developed over time through experience, that are handed down through observational learning and storytelling, and that are ultimately codified in manuals, policies, and regulations. Unfortunately, such knowledge lacks a systematic and unbiased perspective from which to determine causality. We propose that it is ultimately scientific knowledge that will improve the effectiveness of interview and interrogation methods—a perspective that is drawn from independent observation, is theory driven and empirically derived, and is founded upon the principles of replication and peer review. In this section, we briefly review the available research literature on methods of interviewing and interrogation, identifying those methods with the potential for improving HUMINT interrogations that seek to elicit critical intelligence information.

Empirical research on the efficacy of interview and interrogation approaches was initially motivated by claims of actual innocence in the US criminal justice system (e.g., Scheck et al. 2000). Over the past two decades, researchers have sought to understand how suspects may be led to falsely confess (see Gudjonsson 2003; Kassin et al. 2010), and the process by which memory may become contaminated through the interview process, leading witnesses and victims to misremember or misinterpret their experiences (see Brainerd and Reyna 2005; Loftus 2005). Much has been learned from this research that is relevant to the challenges of collecting HUMINT as a product of interview and interrogation methods (see Loftus 2011).

First, modern interrogation approaches often rely not upon physical coercion, but rather psychological manipulation of a source to elicit a statement or confirm available information—these approaches are often referred to as accusatorial interrogation methods. As described by Kassin and Gudjonsson (2004), accusatorial interrogations often involve three phases in which (a) a source is detained in a small room and left to experience the anxiety, insecurity, and uncertainty associated with police interrogation; (b) the source is then accused or confronted (sometimes falsely) with available information, is prevented from denying his/her involvement and is warned of the potential for significant consequences associated with their affiliations or actions (referred to as maximization); and finally (c) a now sympathetic interrogator attempts to gain the source’s trust, offers face-saving excuses or justifications for their involvement, and implies more lenient consequences should the source provide information (referred to as minimization). Although customary knowledge (prior experience) leads experienced interrogators to purport that these methods are effective in producing confirmatory information from a source, scientific assessments suggest that accusatorial methods actually increase the likelihood of both true and false information when compared with a control (direct interview) condition (see Kassin et al. 2010; Meissner et al. 2012)—that is, these methods ultimately lead to confirmatory information gathering that is counterproductive to the HUMINT objective.

In contrast, several high-profile false confessions in England and Wales led to enactment of the Police and Criminal Evidence (PACE) Act of 1984 (Bull and Soukara 2010; Home Office 2003). This act prohibited the use of accusatorial methods and mandated the recording of custodial interrogations. In 1992, a new model of investigative interviewing was introduced in Great Britain (referred to as the PEACE model, see Milne and Bull 1999) that focuses on developing rapport, explaining the allegation and the seriousness of the offense, and emphasizing the importance of honesty and truth gathering. Individuals are provided the opportunity to offer their perspective without interruption, and only thereafter may interrogators engage using positive confrontation and identifying inconsistencies or contradictions. Ultimately, the goal of such information-gathering methods is one of “fact-finding” rather than that of obtaining a confirmatory statement (with an emphasis on the use of open-ended questions). In a recent systematic review of the available empirical literature on information-gathering approaches, Meissner et al. (2012) found that these approaches significantly increased truthful information from guilty individuals and significantly reduced the incidence of false information from innocent individuals when compared with accusatorial methods—in other words, information-gathering methods proved more diagnostic and productive to an investigation. Using a novel experimental paradigm that attempts to model the HUMINT interrogation context, Evans et al. (2013) recently replicated these findings and demonstrated that information-gathering approaches produced more information from guilty and innocent participants, including critical guilty knowledge that would further an intelligence investigation. Although further research and replication of these findings is warranted, the available research suggests that information-gathering approaches offer an effective alternative to modern accusatorial methods that pervade HUMINT training doctrine (i.e., customary knowledge).

Research has also highlighted the significant role of confirmatory hypothesis testing (or confirmation bias) on the elicitation of information in forensic settings. For example, compelling research by Loftus (2005, 2011) and others (e.g., Brainerd and Reyna 2005; Ceci and Bruck 1995) has detailed the ease with which human memory can become contaminated via suggestive and presumptive interview approaches, particularly for events in the distant past (Payne et al. 1994) and when the suggestive information is provided by another individual (Gabbert et al. 2003)—conditions that are relevant to the HUMINT interrogation context. Importantly, confirmation bias not only distorts the potential memory of the source, but it also has been shown to influence both perceptions of source veracity and the interrogation process itself. For example, as discussed below, Meissner and Kassin (2002) first demonstrated the role of investigator biases in judgments of truth and deception by showing that experience investigators demonstrate a proclivity to assume guilt or deception on the part of interviewees (see also Kassin et al. 2005; Meissner and Kassin 2004). This presumption of guilt has also been shown to influence the type of questions or interrogative approaches that investigators use in a suspect interview, leading to behavioral responses from the suspect that are consistent with this prior belief (including false information or confessions from innocent suspects)—a cycle of behavioral confirmation that only further increases an investigator’s confidence in his/her hypothesis (see Kassin et al. 2005; Narchet et al. 2010).

A number of structured interview protocols have been developed to address the potential role of confirmatory hypothesis testing and memory contamination in the forensic setting, including protocols aimed at improving the recall of critical information from adults and children (cf. Lamb et al. 2008; Fisher and Geiselman 1992). Generally speaking, these interview protocols emphasize the importance of establishing rapport with the source, and encouraging the interviewer to introduce very little information during the interview but to instead allow opportunities for uninterrupted, free narrative responses from the individual. One interview protocol, the cognitive interview, further encourages the use of mnemonic prompts (such as context reinstatement, changing perspectives, or affording relevant retrieval cues) that have been shown to facilitate recall in basic memory research. A recent meta-analysis of the cognitive interview literature demonstrated a large and statistically significant increase in correct recall (d = 1.20) when the cognitive interview protocol was compared with that of a standard or structured interview protocol, with no significant effect on the accuracy of information elicited (Memon et al. 2010). Given the importance of eliciting event or person information from HUMINT sources, integration of the cognitive interview with a source that has been led to be cooperative is likely to significantly improve the amount of intelligence information elicited. As described below, use of the cognitive interview and its mnemonic prompts may also produce a corollary benefit—namely, enhancing interrogators’ ability to distinguish between truth and deception.

Distinguishing between truth and deception in an interrogation.

Veracity judgments are an important component of interrogations. Interrogators may face deception in many forms: For example, sources may misrepresent their identity, conceal information, or produce deliberately false information (Buller et al. 1994). Given that the aim of interrogation is to generate reliable information, it is critical for interrogators to assess the veracity of the information that is elicited during interrogation.

Psychologists have studied interpersonal deception and its detection for roughly half a century, and there is now a substantial body of research on the issue (for a comprehensive overview, see Vrij 2008). This research focuses on three primary questions. First, are there valid and reliable cues to deception? That is, do people behave differently when they provide false statements compared to when they are telling the truth? Second, how accurate are people at distinguishing between true and false statements? Third, are there ways in which people’s ability to detect deception can be improved? Below we provide an overview of the major findings related to each of these questions.

There are hundreds of studies mapping the behavioral patterns of liars and truth tellers. Researchers have examined both verbal and nonverbal aspects of behavior in order to answer the question of whether there are cues to deception. This vast literature was synthesized in a meta-analysis by DePaulo et al. (2003), which incorporated 120 studies and 158 behavioral cues. Some cues were based on minute coding of behavior (e.g., the number of words spoken, the frequency of particular body movements), while others tapped into more global, impressionistic aspects of behavior (e.g., whether the communicator appeared positive, tense, or cooperative). Overall, the meta-analysis showed that the behavioral signs of deception are faint. That is, few of the 158 behaviors included in the meta-analysis were related to deception, and those that were actually associated with deception were only weakly linked. For example, contrary to widespread beliefs (Global Deception Research Team 2006), gaze aversion is not a reliable sign of deception, nor are frequent body movements and posture shifts (Strömwall et al. 2004). Behaviors that are most strongly linked to deception tend to be impressionistic cues—for example, liars appear more ambivalent and tense, they are perceived as less cooperative, and their speech seems less immediate and more uncertain.

How accurate are people at distinguishing between true and false statements? Researchers have examined human lie detection accuracy in a variety of ways (Frank 2005; Hartwig 2011). In the most typical approach, people are exposed to videotaped statements provided by participants in laboratory studies who generate either truthful statements or deliberately false statements about their opinions or attitudes, or about events that they have witnessed or participated in. In other studies, people are exposed to real life, high-stake lies in which others lie or tell the truth about serious crimes such as murder, rape or arson (Mann et al. 2004; Vrij and Mann 2001a, b). Summaries of this literature show that people obtain mediocre accuracy rates when facing this task—a meta-analysis of 206 lie detection studies showed an average accuracy rate of 54 %, which is far from impressive given that 50 % accuracy can be expected from chance alone (Bond and DePaulo 2006). Contrary to commonsense assumptions, meta-analytic results further show that lie detection accuracy rates are not higher when lies are told under high-stakes conditions. Instead, lie catchers are more prone to show a bias towards rendering lie judgments for statements told under high-stake conditions, suggesting that they may misinterpret the signs of pressure and motivation to convince emitted from both liars and truth tellers as signs of deception.

It is intuitively appealing to think that some people are better than others at detecting lies, due to some inherent skill, or as a function of training and “on-the job” experience (DePaulo and Pfeifer 1986, Garrido et al. 2004). Indeed, a large survey of US law enforcement officers showed a purported accuracy rate of 77 % (Kassin et al. 2007). There are two bodies of empirical evidence of relevance to this question. First, a meta-analysis of individual differences in lie detection accuracy showed that people differ no more in their ability to detect deception than what can be expected by chance (Bond and DePaulo 2008). This suggests that lie detection cannot readily be construed as a skill that some people possess to a greater extent than others (see also Leach et al. 2009). Second, comparisons of lay people and presumed lie experts such as police officers, customs officers, and prison guards show that they too obtain hit rates around the level of chance (Bond and DePaulo 2006; Hartwig et al. 2004; Kassin et al. 2005; Kraut and Poe 1980; Meissner and Kassin 2002, 2004). Thus, the self-reported accuracy rates of law enforcement officers reported by Kassin et al. (2007) appear to be substantial overestimates of their actual performance. Even though presumed lie experts achieve similar hit rates as lay people, it is worth noting that their decision making differs in critical ways: they demonstrate a bias towards rendering lie judgments, and they show overconfidence in these judgments when compared to lay people (Kassin et al. 2005; Meissner and Kassin 2002, 2004).

In sum then, a large body of empirical literature offers a rather robust conclusion that human lie detection accuracy is mediocre. This finding holds true for low- and high-stake lies and for lay people and presumed lie experts alike. Recently, Hartwig and Bond (2011) conducted a meta-analysis of the deception literature with the aim of understanding why lie detection accuracy rates are so consistently poor. Their analysis suggested the primary reason is that cues to deception are so weak—that is, lie catchers have little diagnostic material to rely on.

So far, we have painted a rather pessimistic picture of the possibilities of making accurate judgments of veracity. It seems that judgments of deception are error prone, even when made by experienced investigators who routinely assess credibility as part of their professional life. However, a wave of research produced in the last decade provides the basis for some optimism. This body of research is based on the premise that while liars might not automatically leak observable cues to deception, it may be possible to elicit cues to deception through strategic interview methods (Levine et al. 2010). That is, through strategic questioning it may be possible to magnify the behavioral differences between liars and truth tellers (Vrij and Granhag 2012), which in turn may lead to higher lie detection accuracy rates (Hartwig and Bond 2011). This line of research on interviewing to detect deception is of particular relevance for HUMINT interrogations, as it has clear practical implications for how such interrogations may be planned and carried out.

The methods to elicit cues to deception have in common that they emphasize cognitive rather than emotional differences between liars and truth tellers. That is, they assume that while liars may not necessarily be more nervous, uncomfortable, or anxious than truth tellers, they may differ in the amount of mental load they experience, and/or in the way they strategize and plan their statements (Vrij and Granhag 2012). For example, the cognitive load approach posits that lying is more mentally taxing than telling the truth, because liars face a more demanding task (for a discussion on cognitive load during lying, see Vrij et al. 2006, 2008). The cognitive load approach further suggests that by imposing additional cognitive load on liars and truth tellers, liars (who are already taxed by the mental burden of lying) may display more signs of being mentally burdened than truth tellers. Several studies support these assumptions. In brief, these studies show that when liars and truth tellers are asked to provide their statements under cognitively demanding conditions (e.g., by being asked to maintain eye contact with the interviewer, or by being asked to tell their story in reverse chronological order), liars display more cues to deception, and lie catchers are more accurate at distinguishing between true and false statements (e.g., Evans et al. 2013; Vrij et al. 2008).

Another approach to elicit cues to deception is based on the assumption that liars plan some, but not all of their responses. That is, they anticipate certain questions, and they prepare answers in response to these. This strategy makes sense—planning might make lying easier, and planned lies are generally associated with fewer cues to deception (DePaulo et al. 2003). However, the fact that liars only plan some of their responses can be exploited in order to produce cues to deception (Lancaster et al. 2013). In one study, researchers interviewed pairs of mock suspects, some of whom were telling the truth about going to lunch together, some of whom were lying about going to lunch in order to conceal a transgression. In response to the anticipated questions (e.g., regarding what they did in the restaurants), liars’ responses were consistent with each other. In contrast, when asked questions they had not anticipated (e.g., when they were asked to provide a sketch of the outline of the restaurant), discrepancies between the liars’ accounts emerged (Vrij et al. 2009). In another study, liars and truth tellers were interviewed about an upcoming trip (Warmelink et al. in press). Liars offered significantly less detailed responses to questions that were unanticipated (e.g., “What part of your trip was the most difficult to plan?”) compared to questions that were anticipated (e.g., “What is the reason for your trip? ”; for a related finding, see Sooniste et al. in press).

A third line of research, similarly anchored in the notion that liars plan and strategize prior to an interrogation is the Strategic Use of Evidence (SUE) technique (Hartwig and Granhag in press). More specifically, the SUE approach is based on the assumption that liars have different counter-interrogation strategies than truth tellers. In other words, while liars and truth tellers share the goal of convincing an interrogator of their story, they differ in the approaches they take to reach this goal (Granhag and Hartwig 2008). In particular, they differ in terms of information management strategies—in contrast to truth tellers, liars are by definition motivated to conceal certain information from an interrogator. For example, they may want to conceal information about their involvement in a transgression, or they may hold information about other people’s identities and actions that they are motivated to keep the interrogator ignorant about. Indeed, research on liars’ counter-interrogation strategies suggests that a primary focus concerns information management: their verbal strategies tend to be either avoidance (i.e., not mentioning being at a certain place) or escape/denial (i.e., denying being at a certain place when asked about it). The SUE approach exploits these information management strategies by using the available background information strategically in order to highlight liars’ avoidance and escape strategies. For example, in the first study of the SUE approach, liars stole a wallet from the briefcase, while truth tellers merely moved the briefcase in order to look for an item. The interrogator had some relevant background information in the form of witnesses placing the suspect at the scene of the crime, as well as the suspect’s fingerprints on the briefcase. When this information was strategically withheld during the interrogation, and the interrogator asked a series of questions about the suspect’s whereabouts and actions (see Hartwig et al. 2011), liars’ stories were inconsistent with the known facts (e.g., they tended to avoid mentioning being at the scene of the crime, and they often denied having seen or handled the briefcase). Furthermore, law enforcement trainees who were taught the SUE principles and tactics, obtained substantially higher hit rates in distinguishing between truths and lies than what is typically observed in the deception literature (Hartwig et al. 2006). For further discussion of the SUE approach and the tactics derived from it, see Granhag et al. (2012).

In conclusion, while detecting deception is a difficult task, recent research suggests that strategic interviewing methods may increase the chances of correct judgments of credibility. Such interview methods are based on an understanding of the challenges liars face, and the planning and strategizing they engage in order to successfully deceive. Taken together with recent empirical research on effective interview and interrogation approaches, we believe that a rather firm evidence base exists to begin moving HUMINT interrogation practice from a customary to a scientific basis.

Future Directions in HUMINT Interviewing and Interrogation

While recent research efforts in the areas of interviewing, interrogation, and credibility assessment offer an empirical perspective that can be readily applied to the HUMINT context, a number of other issues, some unique to the intelligence gathering challenge, will require further scientific inquiry. Based upon our interactions and collaborations with intelligence interrogators, we discuss below a few areas that require further focus and consideration by researchers.

Facilitating Communication Between Researchers and Practitioners: A Common Language.

A grand challenge often faced in our attempts to share important research insights with practitioners has to do with the terminology used by researchers to describe the various interview and interrogation methods that we assess. Quite simply, our research will only have impact to the extent that we (a) accurately reflect the use of certain interview and interrogation approaches and contexts in our studies and (b) facilitate understanding of key concepts and interrogative methods in our communication with practitioners. Terms such as minimization versus maximization, mnemonic cues, cognitive load, or positive confrontation have meaning to psychologists in the investigative interviewing arena, but can be quite confusing and misinterpreted when shared with practitioners. As we have noted here and elsewhere (see Meissner et al. 2010), we believe that collaboration between scientists and practitioners is key to improving the practice of interviewing and interrogation—and the foundation of any successful collaboration is a shared conceptual and linguistic understanding of the topic area. It will be important for researchers and professionals to generate a common language from which to operate. A wonderful example of this can be seen in a recent paper by Kelly et al. (2012), in which the authors (researchers and a practitioner) propose a taxonomy of interrogation methods that is grounded in both psychological theory and operational experience.

Cross-Cultural Perspectives in the Booth.

HUMINT interrogations often involve foreign national sources or high-value suspects who have knowledge of threats to the security of the government. Such interviews or interrogations present occasions that necessarily invoke cross-cultural (and intergroup) processes—including communication, person perception, and social influence among others. Unfortunately, there is only a small body of literature applying basic, cross-cultural research on such processes to the investigative interviewing context, despite the critical need for such research. Recent studies by Buene et al. (2011, 2010,2009; Giebels and Taylor 2009) provide an excellent model for addressing this challenge. Further research on the role of culture in judgments of credibility is also warranted (see Bond and Rao 2004), including the potential influence of cultural biases (see Castillo and Mallard 2012).

Interviewing Through an Interpreter: Linguistic Challenges to Communication, Persuasion, and Credibility Assessment.

A related characteristic of HUMINT interrogations is their frequent use of interpreters to facilitate communication with foreign nationals. As may be evident, the use of an interpreter presents many challenges to an already complex interaction between the source and the interrogator. Interpreters may influence the flow and accuracy of information that is communicated across the parties, and may hinder the use of certain interrogation strategies or approaches (see Lai and Mulayim in press). Interpreter expertise and prior experience/knowledge of interview and interrogation methods may moderate these effects. Further, the mode of interpretation (consecutive intervals vs. simultaneous translation) and placement of the interpreter within the interview context may further influence communication, rapport development, and the effectiveness of attempts at persuasion. Finally, the impact of language on judgments of credibility must also be considered, with recent research suggesting effects on both accuracy and response bias (see Da Silva and Leach 2011; Evans et al. 2013).

Conceptualizing and Measuring Rapport.

Professional interrogators will often describe the importance of rapport in facilitating cooperation and eliciting information from suspects (see Kleinman 2006; St. Yves 2006). Although rapport has been more widely assessed in clinical and negotiation contexts (see Tickle-Degnen and Rosenthal 1990), there is a critical need to both conceptualize and assess rapport in the interview and interrogation context. An excellent review of the relevant basic research literature has been offered by Abbe and Brandon (2013), as the authors highlight the importance of developing our understanding of rapport in the investigative interviewing context. Recent empirical studies by Bull and Soukara (2010), Driskell et al. (2013), and Vallano and Schreiber Compo (2011) have demonstrated the importance of rapport in police interviews on the elicitation of quality information from sources or witnesses (see also Walsh and Bull 2012). Further research that provides a strong theoretical basis for conceptualizing and measuring the behavioral manifestations of rapport in both the interviewee and interviewer, including the influence of pseudo rapport (see DePaulo and Bell 1990), appears warranted.

Moving From the Laboratory to the Field: Empirically Assessing the Effectiveness of Evidence-Based Methods in the Training Academy and in the Field.

Finally, generating data from experimental research will undoubtedly lead to a strong evidence base from which to inform training and practice; however, we should be careful about simply “handing off” our research findings with the assumption that our findings will generalize beyond the laboratory. It will be important for researchers to partner with training academies and agencies to support experimental evaluations of the methods we develop in the laboratory in comparison with those methods currently trained and used in the field. Over the past few years we have begun such a process of partnering with trainers to develop training assessments—experimental tests of our methods versus that trained by the academy. These studies help to inform trainers and operators about the scientific process and provide an empirical basis for a potential policy change in the academy’s curriculum. Further, it will be important to monitor the effectiveness of the training materials that are developed, and to assess the effectiveness of our novel approaches when deployed into the operational environment.

Conclusions

In the current chapter, we have sought to provide a historical overview of HUMINT interrogation practices and to describe the contemporary challenges faced by interrogators in this context. We have also reviewed research in the investigative interviewing realm that can offer a scientific perspective to the current use of customary knowledge. We believe that ethical, evidence-based approaches can be developed that offer an effective alternative to the use of physical or psychologically coercive methods. Ultimately, we hope that other researchers will join us in addressing the unique challenges offered by the HUMINT context.