9.7.3 The Concealed Information Test
The second type of polygraph test is the Concealed Information Test (also referred to as the Guilty Knowledge Test (GKT) (Furedy & Heslegrave, 1991; Lykken, 1959, 1960; Verschuere & Meijer, 2014). The basic idea behind the CIT is straightforward: it aims to detect concealed knowledge that only the guilty suspect has. This is done by presenting a question together with a number of answer alternatives, one of which is correct (e.g. “What weapon was used to kill Mr. Sylvester? Was it a knife? A dagger? A pair of scissors?”). The assumption is that a guilty suspect, who will recognise the correct answer, will experience more physiological arousal when the correct alternative is presented due to what is called an orienting reflex (Meijer & Verschuere, 2015), compared to the incorrect alternatives. In contrast, an innocent suspect will react on average similarly to all alternatives, since they lack the so-called guilty knowledge (MacLaren, 2001).
9.7.3.1 Validity of the CIT
The same methodological problems with field and laboratory studies with the CQT also apply to studies evaluating the validity of the CIT, and will not be repeated. In contrast to the CQT, the CIT seems to be slightly more accurate in classifying innocent than guilty. In an overview by Vrij (2008), well over 90% of innocent suspects and around 80% of guilty suspects were correctly classified in laboratory studies. A review of 20 such studies showed somewhat lower figures, with 83% of innocent suspects and 76% of the guilty being correctly classified (MacLaren, 2001). Results from real-life studies (c.f. Elaad, 1990; Elaad, Ginton, & Jungman, 1992) support the asymmetric pattern, and in a summary of field studies (Vrij, 2008), 96% of innocent suspects were found to be cleared by the test, while only 59% of the guilty suspects failed it.
9.7.3.2 Problems with the CIT
A survey showed that 75% of a sample of both general psychologists and psychophysiologists considered the CIT to be based on scientifically sound principles, while only 33% considered this to be true for the CQT (Iacono & Lykken, 1997). Despite this, there are problems with the CIT: first, the validity of the test can be seriously challenged if the correct alternative stands out in any way. Such transparency could possibly make the reaction pattern of innocent suspects look like that of guilty ones. The second concerns the applicability of the CIT. If the test is to be used, innocent suspects must not know the correct alternatives, otherwise an innocent suspect might give “guilty” responses because of having been exposed to the critical information, for example, via the media. Moreover, the guilty suspect must know the answer. If a guilty person failed to perceive a certain detail at the crime scene, the guilty person might give “innocent” responses to questions about the detail because of lack of knowledge. These conditions drastically affect the applicability of the technique, which has been estimated to be lower than 10% of FBI cases (Podlesny, 2003, quoted in Honts, 2004).
9.7.4 Countermeasures
There has been serious debate in the research field concerning to what extent it is possible for subjects to influence the outcome of polygraph tests. Any attempts with such a purpose are called countermeasures. The aim would be to (during a CQT) to purposely enhance reactions to control questions or (during a CIT) purposely provide similar responses to all the alternatives. It has consistently been shown that people’s spontaneous countermeasures are not effective in producing false negative outcomes (Honts & Amato, 2002). However, coaching subjects in countermeasures before the test occasion is a more serious threat to the polygraph. In experimental research, subjects who received training in countermeasures managed to impair the accuracy of the CQT. The countermeasures were both physical, such as biting the tongue or pressing the toes to the floor (Honts, Hodes, & Raskin, 1985), and cognitive, such as counting backwards (Honts, Raskin & Kircher, 1994). For a classification and discussion on different types of countermeasures, see Honts and Amato (2002).
9.8 STRATEGIC INTERVIEWING IN ORDER TO ELICIT AND ENHANCE CUES TO DECEPTION
Research suggests that lack of valid cues is a main reason for the typically mediocre lie detection accuracy rates (DePaulo et al., 2003; Hartwig & Bond, 2011). It is therefore promising to note a new wave of research where the aim is to elicit and enhance diagnostic cues to deception by actively interviewing in strategic ways (Vrij & Granhag, 2012). This line of research can be viewed in contrast to the passive set-up used in the paradigmatic deception study (assessing short video clips in which there are few, if any, cues to deception and truth). It is possible to identify several different approaches within this new wave of research, and below we will discuss three of these: (1) the cognitive load approach, (2) the unanticipated questions approach, and (3) the verifiability approach.
CASE STUDY 9.1 PASSIVE VS. ACTIVE DECEPTION DETECTION
As described in this chapter, there has been a paradigm shift in the study of deception. For several decades, lie-catchers were typically confined to simply watching a videotape (or, at times, listening to an audiotape or reading a written transcript) of a person either lying or telling the truth. The underlying premise of this approach is that there is some degree of leakage of cues to deception, which may be detectable to an observer. An example of a study using this passive approach was conducted by Mann, Vrij, Fisher and Robinson (2008). In this study, lie-catchers were exposed to either deceptive or truthful statements, which were presented in different modalities. Some saw the suspect on a muted videotape (visual condition), others heard the suspect on an audiotape (audio condition), and a third group of lie-catchers both saw and heard the suspect (audiovisual condition). Based on the finding that nonverbal cues tend to be weaker indicators of deception than verbal cues, the authors predicted that when lie-catchers only had access to nonverbal behaviour (i.e. in the visual condition), they would achieve a lower hit rate than they would in the audio or audiovisual condition. Indeed, these predictions were supported: accuracy rates were lowest in the visual-only condition, where judgments were also characterised by a more pronounced tendency toward a lie bias (a greater tendency to label true accounts as lies, rather than vice versa).
In contrast to this passive approach, the recent decade has seen a wave of research focusing on active approaches to detecting deception. By that term, we mean attempts to elicit cues to deception using various interventions – typically, interviewing methods that elicit different responses from liars and truth tellers. One of the earliest studies using such an approach was the study by Vrij and colleagues (2008). This study is based on the cognitive load theory discussed earlier; that is, the notion that lying is more mentally demanding than telling the truth. Vrij and colleagues extended this idea by suggesting that if lying is indeed more cognitively demanding, liars ought to have a more difficult time coping with additional load. Expressed differently, if faced with an additional task that requires cognitive resources, liars may show more signs of being cognitively taxed compared to truth tellers. In their study, Vrij and colleagues operationalised cognitive load as providing the statement in reverse chronological order (which is indeed cognitively demanding). In short, they had liars and truth tellers provide a statement denying a mock crime, and they were instructed to do so either in normal chronological order or in reverse order. When senders provided their statements in normal order, there were barely any signs of deception. However, when they were asked to tell the story backwards, there were far more cues to deception. Moreover, when lie-catchers viewed the tapes where senders had been asked to provide the statement in reverse order, their accuracy rates were higher compared to when they saw statements told in normal order. This study suggests that cues to deception might not appear automatically, and that in order for lie detection to be possible, an active approach to create behavioural differences between liars and truth tellers is a more viable option.
9.8.1 The Cognitive Load Approach
As already discussed there is strong evidence that lying is often more cognitive demanding than telling the truth (e.g. Vrij, Fisher, Mann & Leal, 2006). The core assumption of the cogni
tive load approach is that by placing targets under additional cognitive load, it might be possible to elicit new and/or enhance already existing cues to deceit. Put differently, it is assumed that liars would be more hampered by additional cognitive tasks, as their resources are already preoccupied with the cognitive challenge of lying. One example is the use of reverse-order recall described in Case Study 9.1 by Vrij et al., 2008. In another study, targets were instructed to maintain eye contact, which previous research has found to be mentally taxing (Beattie, 1981). Critically, cues to deception were more pronounced in this condition compared to a control condition in which no instruction to maintain eye contact was given (Vrij, Mann, Leal, & Fisher, 2010).
9.8.2 The Unanticipated Questions Approach
Similar to the cognitive load approach, this approach uses a consistent research finding as the point of departure: liars tend to prepare themselves before an interview; and they predict the questions they will be asked and rehearse answers to those questions (e.g., Granhag, Andersson, Strömwall, & Hartwig 2004; Hartwig, Granhag, & Strömwall, 2007). This strategy is a good one as planning makes lying easier, and planned lies typically contain fewer cues to deceit than spontaneous lies (DePaulo et al., 2003). However, investigators can exploit this by asking unanticipated questions. Before reviewing some research it is important to understand that this approach is not about asking any kind of unanticipated question (“What was your grandmother’s shoe size?”), and then expecting liars’ and truth tellers’ answers to systematically differ. The approach is about asking unanticipated questions that truth tellers are in a comparatively better position to answer; unanticipated questions to which truth tellers will answer differently from liars.
In one of the first studies within this approach Vrij and colleagues (2009) used a set-up where pairs of liars and pairs of truth tellers were interviewed individually about an alleged visit to a restaurant. The opening questions (e.g. “What did you do in the restaurant?”) were anticipated, whereas a request to sketch the layout of the restaurant was not. On the basis of the answers to the conventional opening questions the pairs were classified at chance level; however, based on the similarity in the two pair members’ drawings, 80% of the liars and truth tellers were classified correctly (drawings were less alike for the pairs of liars than the pairs of truth tellers). Interestingly, pairs of truth tellers showed the same degree of overlap in their answers to the anticipated and unanticipated questions, whereas this was not the case for liars. As predicted, liars showed significantly more overlap in their answers to the anticipated questions than in their answers to the unanticipated questions. In essence, had the interviewers posed conventional questions only, it would have been very difficult (if not impossible) to separate lying from truth telling pairs.
There is a clear link between the unanticipated questions approach and the cognitive load approach. For liars, anticipated questions are easier to answer than unanticipated questions, due to the fact that they can evoke and use their rehearsed answers, but for the unanticipated questions they are forced to fabricate answers on the spot. The difference liars will experience in terms of cognitive load while answering anticipated and unanticipated questions will become evident in their verbal responses. In contrast, for truth tellers, the levels of cognitive load experienced while answering anticipated and unanticipated questions are not too different from each other, and they will therefore produce more comparable answers to the anticipated and unanticipated questions than liars.
9.8.3 The Verifiability Approach
This approach is one of the newest within the “strategic interviewing” strand of research, Basically, the verifiability approach rests on two assumptions: (1) liars often include less detail in their account than truth tellers, although liars are aware that the level of detail is important for being assessed as credible; and (2) liars prefer to not mention details that can be checked. Key to the approach is that these two assumptions put liars in an information management dilemma: on the one hand they know they need to provide details to be assessed as honest, on the other hand they realise that there is a risk in providing details that can be checked. The verifiability approach revolves around how liars’ navigate this dilemma: by providing details that cannot be verified (e.g. “On my way back home a black car passed by, the left front light socket was broken”). Critically, a series of studies by Vrij and Nahari on this innovative approach show a consistent finding: liars report fewer details that can be checked than truth tellers (e.g., Nahari & Vrij, 2014; Nahari, Vrij, & Fisher, 2014a and Case Study 9.2).
CASE STUDY 9.2 WHEN REVEALING THE LIE DETECTION TECHNIQUE MAKES THE TECHNIQUE MORE EFFECTIVE
Recently Galit Nahari and her colleagues (Nahari, Vrij, & Fisher, 2014b) published a paper on deception detection with an unusual outcome. Most people would think that informing suspects about the method of lie detection to be used before an interview would only help the suspect, and be damaging to the technique. Indeed, this intuitive notion is supported by research, for example informing participants about Criteria-Based Content Analysis (CBCA) and reality monitoring (RM) results in the liars providing stories that are richer in CBCA and RM criteria compared to uninformed liars (e.g., Caso, Vrij, Mann, & DeLeo, 2006). In short, the discriminative value of the technique is reduced because informed liars will employ effective countermeasures. Nahari and colleagues were able to show the opposite. That is, the effectiveness of the verfiability approach became stronger when the participants were informed that that approach was about to be used for detecting deceit. For their study truth tellers wrote down their activities during the last half hour, whereas liars fabricated an account. Half of the liars and half of the truth tellers were informed about the lie detection method to be used and that their statements would be checked for verifiable details. The findings were intriguing. First, informed truth tellers included more verifiable details into their accounts than uniformed truth tellers. Second, informed liars did not include more verifiable details than uninformed liars. The accuracy rate for the verifiability approach was 61% for uninformed suspects, and 67% for uninformed suspects. Put differently, the informed truth tellers searched their memory more effectively for verifiable details than uninformed truth tellers, whereas being informed about the lie technique to be used did not assist the liars. Simply put, liars could not risk adding verifiable details to their accounts, just because these details could be checked. The fact that the difference between truth tellers and liars was magnified by informing them about the lie detection method to be used should not be interpreted as that the technique is completely resistant to countermeasures, but it is definitely a promising approach.
Yet another approach with respect to actively interviewing in order to elicit cues to deceit is the extensive research programme on the strategic use of evidence or SUE-technique (for recent overviews see Granhag & Hartwig, 2015; Hartwig, Granhag, & Luke, 2014). The SUE-technique is applicable to situations where the interviewer holds some sort of critical background information (evidence) pointing to the suspect’s guilt, and the technique is discussed in chapter 8.
9.9 NEW DIRECTIONS IN DECEPTION DETECTION RESEARCH
9.9.1 fMRI as a Deception Detection Tool
There are many methods for scanning the human brain, and one of the more common is called functional Magnetic Resonance Imaging (fMRI). This method enables monitoring of neural activity during, for example, cognitive operations. During recent years fMRI has been used for many different purposes, among them the study of brain activity taking place during deception. The findings from this research indicate that the neural activity during lying may be different compared to the activity during telling the truth (Spence, Hunter, Farrow, Greed, Leung, & Hughes, 2004). Specifically, fMRI studies show an increased activity in the prefrontal cortex during deception (i.e. the “executive” part of the brain, supporting complex human behaviour such as speech and problem solving). These studies support the theoretical perspective that lying is more cognitively dem
anding than telling the truth (Vrij, Fisher, Mann, & Leal, 2006; Vrij, 2015b).
PHOTO 9.3 fMRI has been used recently to study the brain activity taking place during deception.
Source: © James Steidl/Shutterstock
The use of the fMRI as a deception detection tool has received extensive media coverage. Indeed, the studies on neural correlates of deception are intriguing, and the findings reported look promising. Importantly, though, this does not mean that the deception enigma is solved once and for all. fMRI equipment is extremely expensive, immobile, and demands the suspect remain still. Hence, the few studies conducted so far are limited in external validity. It still remains to be shown whether the fMRI is diagnostic for deception in situations of high forensic relevance. It is even more controversial whether fMRI produces evidence of sufficient reliability to be admissible in court (Langleben & Moriarty, 2013). It is probably safe to assume that studies using fMRI will shed some light on the cognitive processes taking place during deception, but its future applications to forensic settings appears problematic.
Forensic Psychology Page 44