Handbook of Psychology of Investigative Interviewing: Current Developments and Future Directions

by Ray Bull, Tim Valentine, Dr Tom Williamson

  one ’ s motivation not only defi nes the interaction (truth or lie), but also dictates

  how it will reveal itself, both qualitatively and quantitively.

  For the purpose of the present chapter, a lie is defi ned as the deliberate

  intention to deceive another person without prior notifi cation (Ekman, 1992 ).

  For example, a fi nancial adviser who provides poor investment advice is viewed

  as lying if she or he knows their advice is poor yet represents it as good, but

  is not seen as lying if the advice is well intended and proves to be poor. As

  another example, a woman with a bona fi de paranoid delusion, who states that

  she is Mary Magdalene, is not lying, while a woman deliberately feigning a

  delusion during a psychological examination is lying.

  In order to detect lies, one needs to understand the nature of truth - telling,

  which adds to the complexity of evaluation. That is, one needs to gain knowl-

  edge and skills in two distinct but related areas: how to identify truths when

  they are present and how to detect lies when they are present. Expertise in

  only one of these areas will undoubtedly lead to many errors, with the expert

  Evaluating Truthfulness

  303

  truth - seeker missing many lies (i.e., false negatives) and the expert lie - catcher

  missing many truths (i.e., false positives). For this reason, we have moved away

  from the popular nomenclature of ‘ lie detection ’ or ‘ credibility assessment ’ ,

  choosing instead to describe the differentiation of truths and lies as ‘ evaluating

  truthfulness ’ .

  The practice of evaluating truthfulness is inherently complex as it never

  occurs in a vacuum, in contrast to the vast majority of laboratory research on

  the topic. The focus of most research and practice in assessing truthfulness is

  usually in the context of some formal or quasi - formal assessment. This can be

  a police interview of a witness or suspect, a customs agent interviewing an

  incoming passenger, a salesperson talking to a potential client, a lawyer con-

  ducting a discovery or a mental health professional doing a forensic examina-

  tion. Evaluating truthfulness in the context of an assessment involves

  multi - tasking as, in addition to evaluating truthfulness, the assessor is always

  involved in other tasks (e.g., forming the next question, listening to the inter-

  viewee, monitoring professional tasks). Multi

  - tasking inevitably makes the

  evaluation of truthfulness more diffi cult. Indeed, evaluating truthfulness is a

  diffi cult task by itself, something that is exacerbated whenever there are any

  distracters. Adding to its diffi culty is the reality that evaluating truthfulness is

  dynamic in nature. That is, the task will change during a single interview, as

  well as across interviews: an interviewee may lie about a particular topic at one

  point but not at another, or may display a lie about a particular topic differ-

  ently at different times (e.g., verbally at fi rst but non - verbally subsequently).

  As is emphasized below, research indicates that there is no lie response and,

  therefore, truthfulness must always be inferred. Sometimes, what appears to

  be an indication of a lie may turn out to be in indication of something else;

  therefore, the accurate evaluation of truthfulness requires repeated reassess-

  ment of one ’ s hypotheses and conclusions (see below).

  The complexity of evaluating truthfulness is enhanced by the presence of

  both individual and cultural differences. For example, individuals differ in their

  motivations for lying and telling the truth (Ekman, 1992 ; Cooper & Yuille,

  2006 ; Spidel et al., 2003 ), as well as their ability to deceive and/or detect

  deception (Ekman & O ’ Sullivan, 1991 ; Porter, Woodworth & Birt, 2000 ). In

  addition, although some of the clues to deception that are reviewed below are

  cross - cultural, others are culturally specifi c. For example, the facial expression

  of anger appears to be universal (Ekman, 2003 ), however, the triggers that

  cause anger are thought to be, at least in part, culturally determined. To com-

  plicate matters further, individual difference variables must always be inter-

  preted in terms of context and a host of other factors. For example, just

  because a forensic assessor knows that she or he is dealing with an interperson-

  ally gifted psychopath with a penchant for lying does not mean that everything

  that the psychopath says during the interview(s) is deceptive.

  Clearly, the complex nature of conducting evaluations of truthfulness poses

  considerable challenges for both practice and research. That said, it is impor-

  tant to note that these challenges are not insurmountable obstacles but rather

  roadblocks to be carefully navigated. Before turning our attention to the

  304

  Handbook of Psychology of Investigative Interviewing

  impact of these obstacles, as well as ways around them, we briefl y review a

  number of contemporary methods available for evaluating truthfulness.

  Approaches to evaluating truthfulness

  There is a long history associated with discriminating truth from lies. For

  example, Ancient Egyptian papyri, as well as records of classical Chinese courts,

  included hints or recommendations on how to discriminate a truth - teller from

  a liar (Ford, 2006 ). However, the twentieth century witnessed an explosion

  of both theory and technology related to evaluating truthfulness. The

  approaches that have developed can be classifi ed into two basic types: those

  that are technology - based and those that are skill - based. Technology - based

  techniques for discriminating truth from lies can be classifi ed as either psycho -

  physiological or neuropsychological in nature. The best known of the psycho -

  physiological techniques is the polygraph, which measures heart rate, skin

  conductance and respiration while a person is answering a number of ques-

  tions. The polygraph is often mislabeled a lie detector test. The polygraph does

  not detect lies; it detects stress. Perhaps the most effective aspects of the use

  of the polygraph are the polygraphers, who are often excellent interviewers,

  and the fact that the polygraph detects change, a core aspect to our proposed

  approach to evaluating truthfulness (see below). The polygraph is a useful tool,

  but it has a focused use (e.g., criminal suspect investigations and national

  security) and can produce both false - positive and false - negative errors (National

  Research Council, 1996 ).

  In terms of more recent technological advances, a number of companies

  have been promoting and selling voice stress analysers as lie detectors (for a

  review, see Vrij & Granhag, 2007 ). Such devices detect changes in the pitch

  and tension of the voice and there is no question that detecting change is an

  important aspect in evaluating truthfulness. However, although changes in the

  voice can be a clue to deception, voice characteristics are unreliable as a single

  basis for evaluating truthfulness (Vrij & Granhag, 2007 ), largely due to the

  fact that vocal changes can occur for a variety of reasons (see below). A more

  promising approach relies on thermal imaging, which measures temperature

  changes in the body. Not o
nly has the technology evolved to allow for mea-

  sures to be taken covertly at a distance, recent research suggests that there may

  be reliable thermal changes (e.g., on the face, particularly around the eyes)

  when a person is being deceptive (Vrij & Granhag, 2007 ). Research is also

  currently being conducted on the value of a functional magnetic resonance

  imaging ( f MRI) and other techniques for assessing brain activity as a method

  for discriminating truth from lies (for a review, see Spence

  et al .,

  2004 ).

  Although promising, it is important to note that these technological advances

  are in their infancy and require equipment that is intrusive, non - mobile and

  expensive. Accordingly, it would be premature to draw fi rm conclusions

  regarding the applied utility of such techniques.

  Evaluating Truthfulness

  305

  Although technology - based approaches to evaluating truthfulness have their

  merits, there are a number of advantages of skill - based techniques. The main

  advantage of skill - based techniques is that they are extremely portable and

  unobtrusive. Skill - based assessment techniques typically fall into two general

  categories: those that rely on the assessment of verbal behaviour and those

  that focus on non - verbal aspects of behaviour. Verbal clues to truth and decep-

  tion include the content of speech, the style of speech and voice characteristics

  (Horowitz, 1991 ; Porter & Yuille, 1996 ). Nonverbal clues are generally sepa-

  rated into those related to the face and those related to the rest of the body

  (Ekman, 1992 ; Ekman, O ’ Sullivan, Friesen & Scherer, 1991 ). The approach

  to skill - based assessment of truthfulness introduced in this chapter involves the

  integration of all of these channels of information, both verbal and nonverbal

  (see below).

  As with technology - based approaches, there is no single verbal or nonverbal

  channel that clearly communicates deception. Rather, research and clinical -

  forensic experience suggest that it is the change in a particular channel and/

  or inconsistencies across channels that are particularly revealing (for a review,

  see Griesel & Yuille, 2007 ). Viewed in this context, research is beginning to

  demonstrate that skill - based approaches parallel technology - based approaches

  in terms of reliability and validity, without, however, the pitfalls associated with

  reliance on technology. Even proponents of technology

  - based approaches

  (e.g., polygraphers) understand the merit of skill - based methods.

  Research on evaluating truthfulness

  As noted above, the approach to evaluating truthfulness introduced in this

  chapter is research - based. Before this approach is outlined, it is important to

  discuss certain conceptual and methodological limitations inherent in this line

  of research. The basic diffi culty in conducting research on evaluating truthful-

  ness stems from the complexity of the topic itself (as defi ned above). In fact,

  we argue that current research methodology and associated statistical proce-

  dures cannot do the topic justice in terms of identifying and assessing clues to

  lies/truths and identifying how people evaluate truthfulness in the real world.

  As with other areas in psychology, such as the fi eld of eyewitness memory,

  most of the research that has been done on evaluating truthfulness is labora-

  tory - based (for reviews, see DePaulo, Lindsay, Malone, Muhlenbruck, Charlton

  & Cooper, 2003 ; Griesel & Yuille, 2007 ). In a typical study, undergraduate

  volunteers are asked either to tell the truth or lie in highly controlled condi-

  tions. Often the motive for lying is weak (e.g., course credit, small monetary

  reward, praise) and the controls so stringent as to render the context psycho-

  logically sterile; thus, the generalizability of the fi ndings to other contexts is

  limited. Indeed, one characteristic that discriminates relatively useful labora-

  tory research from less useful research is the effort the researcher has put into

  developing effective or strong motivation for the participants in the laboratory

  306

  Handbook of Psychology of Investigative Interviewing

  study. Another factor is the multifaceted level attained by the mock design:

  the addition of variables often renders studies relatively more realistic. To

  approximate real - world scenarios more closely, we believe that research in this

  area should also attempt to vary the level of the participants

  ’ motivation

  according to the presence or absence of certain infl uencing variables. Evaluating

  truthfulness is both complex and dynamic; thus the research to support its

  techniques should be similarly complex and dynamic.

  Field research (e.g., studies using tapes from criminal investigations; tapes

  from offenders discussing their crimes; tapes of people being interviewed at

  immigration entry points) generally does not have the motivational limitations

  of laboratory - based research (e.g., Cooper, Ternes, Griesel, Viljoen & Yuille

  2007 ; Ternes, Cooper & Yuille, 2007 ). However, unlike laboratory studies,

  fi eld research, although high in external validity, often lacks ground truth.

  Moreover, there is considerable variability in fi eld research in terms of the

  manner in which ground truth is examined and measured. In other words, the

  nature and quality of the information determining ground truth is a major

  factor discriminating the scientifi c contribution of fi eld research.

  Irrespective of whether the research on evaluating truthfulness is laboratory -

  or fi eld - based, research in this area has suffered from poor adherence to assess-

  ment training protocols. For example, one technique that is described in more

  detail below involves the assessment of the content of statements, i.e., Criteria

  Based Content Analysis (CBCA; Steller, 1989 ; Steller & Koehnken, 1989 ). It

  turns out that the response to training in this method is quite variable. In our

  experience, some trainees can learn this method and apply it reliably after two

  days of training, but others require weeks of training and practice before they

  are able to obtain the same degree of reliability. Some trainees, however, seem

  unable to acquire the methodology at all and research studies on this method

  of statement analysis have rarely taken this variability into account. Thus,

  researchers often end up with a mixed group of assessors rendering the study

  ineffective for evaluating the usefulness of the technique.

  This area of enquiry is further limited by researchers ’ bias for quantitative

  research paradigms, often to the exclusion of qualitative approaches. To con-

  tinue with the example of the CBCA, this approach to statement analysis is

  of a qualitative nature (Griesel & Yuille, 2007 ); however, researchers have

  shown a clear preference for statistical cut - off scores. Consequently, they often

  impose a quantitative structure on this qualitative assessment procedure,

  resulting in a distortion that often misrepresents research outcomes. As an

  aside, the same appears to be the case with structured clinical guidelines

  for the assessment of risk for recidivism. That is, even though it is the asses-


  sor ’ s decision of the offender ’ s risk level based on an overall evaluation of the

  risk factors examined that matters (Cooper, Griesel & Yuille, 2007 ), research-

  ers have a preference to use numbers and cut

  - off scores to indicate low

  - ,

  medium - and high - risk levels, which distorts the spirit of structured clinical

  judgement.

  Evaluating Truthfulness

  307

  In addition to the methodological limitations reviewed above, this area of

  study is constrained by the limits of available statistical procedures. One of the

  main problems with applying traditional statistics to research on evaluating

  truthfulness is that statistics impose limits on the quality of the questions being

  answered. Indeed, while the practice of evaluating truthfulness is unique to

  the individual being assessed, it is often the case that researchers use group -

  based statistics that dilute these all

  - important individual differences. For

  example, some research suggests that examining body language has no or little

  valid role in helping evaluate truthfulness (DePaulo et al ., 2003 ; Vrij, Mann

  & Fisher, 2006 ). However, such research fails to consider the role of different

  types of body language (e.g., illustrators vs. manipulators vs. emblems), each

  having been found to relate to truthfulness differently (Ekman, Friesen &

  Scherer, 1978 ; Ekman, O ’ Sullivan, Friesen & Scherer, 1991 ). In fact, most

  research does not take into consideration the reality that, while certain types

  of body movements may increase in one person when he or she is lying, the

  same type of movement may decrease in another person when he or she is

  lying. Moreover, researchers build methodologies and thereafter rely on sta-

  tistical procedures that assume that evaluating truthfulness is static in nature

  (i.e., is revealed at one point and/or consistently across lies/time) when in

  fact it is dynamic, changing within and across people and time. This raises

  another important point: although statistics appear to provide the context of

  objectivity and scientifi c integrity, the fact remains that the quality of the data

  that go into the analysis determines the quality of the results. No matter how

  sophisticated the statistical procedure employed, the above noted method-