A Sequence Analysis of Nonverbal Behaviour and Deception

A Sequence Analysis of Nonverbal Behaviour and Deception J Police Crim Psych (2018) 33:109–117 DOI 10.1007/s11896-017-9238-9 1 2 3 1 A Marono & DD Clarke & J Navarro & DA Keatley Published online: 7 June 2017 The Author(s) 2017. This article is an open access publication Abstract The ability to correctly interpret nonverbal commu- Keywords Nonverbal communication Behaviour sequence . . . nication (NVC) is an important ability in everyday interac- analysis Credibility assessment Deception Suspect tions, which may use NVC techniques to identify the conceal- behaviour ment of information. In the present study, a novel approach was used to understand NVC. Behaviour sequence analysis identified specific sequences of behaviours that indicate psy- Introduction chological distress caused by deception. The study involved the analysis of 55 videos of real criminals and high-power Understanding nonverbal communication (NVC) is important individuals that were filmed fabricating statements, which in a range of situations and settings, from daily relationships to were later exposed as being untruthful at the time of being interrogations (Krauss et al. 1996). Research has shown that filmed. In addition, 53 clips of criminals making truthful state- NVC includes both reflexive and nonreflexive movements of ments were also analysed as a contrast group. Results indicat- the body, which communicate an emotional message to others ed clear differences between honest and deceptive responses, (Tamietto and De Gelder 2010). Correct interpretation of in- such as furrowing of eyebrows in the deceptive sequences dividuals’ emotional states and intentions is obviously impor- occurring more often than honest statements. In addition, se- tant for effective communication; however, a particularly im- quences of behaviours were shown in the present data set, portant area for NVC is in police and legal investigations. which could indicate a new method for analysing NVC and During a police interview or interrogation, investigators may detecting psychological distress caused by deception. The implicitly or explicitly use NVC reading techniques to identify possible implications and applications for police and forensic the concealment of emotions, which may indicate an intent to investigation are also outlined. deceive or a line of questioning worth pursuing (Hartwig and Bond 2011;Mann et al. 2004). Behavioural cues of mental effort, memories and emotions are used by security personnel Electronic supplementary material The online version of this article to identify behavioural ‘hot spots’ that indicate that a topic is (doi:10.1007/s11896-017-9238-9) contains supplementary material, worth considering and investigating further (Frank et al. which is available to authorized users. 2008). Typically, research has focused on behaviours related to particular emotions or concealment of information (Ekman * DA Keatley DKeatley@lincoln.ac.uk and O’Sullivan 1991). Sporer and Schwandt (2007)outline several reliable indicators of deception, such as nodding, J Navarro http://www.jnforensics.com movement of feet and legs and hand movements; however, they showed no systematic relationship between avoidance of Researchers in Behaviour Sequence Analysis, Psychology eye contact and deception. However, researchers and profes- Department, University of Lincoln, Lincoln LN67TS, UK sional practitioners in the field have suggested that clusters of Researchers in Behaviour Sequence Analysis, Psychology behaviours are better indicators of changes in emotion and Department, University of Nottingham, Nottingham NG72RD, UK possible psychological discomfort,which may arise from JN Forensics, LCC, Tampa, FL, USA attempting to deceive someone (DePaulo et al. 2003; 110 J Police Crim Psych (2018) 33:109–117 Navarro and Karlins 2008,Navarro 2011;Vrijet al. 2004). Burgoon et al. (2014, 2015). In their seminal research, The present research continues a more recent trend in NVC by patterns of nonverbal behaviour, across time, were investigating the possibility that it is not only discrete behav- analysed, showing clear differences between deceivers iours or clusters of movements that can be used to indicate and truthful participants. Burgoon and colleagues’ research emotion or distress but also the sequence that these move- provides a clear foundation to begin future development in ments occur in (Burgoon et al. 2014; Burgoon et al. 2015). investigating patterns and sequences of behaviours, rather Previous research that has focused on sequences of behaviour than individual behaviours or clusters. It is to this direction has used a software programme that detects patterns in time- the present research builds on, by using real-world decep- ordered data. These T-patterns (see Casarrubea et al. 2015)are tive statements, rather than mock crime experiments. a multivariate approach to identifying temporal structure in behaviour. Burgoon et al. (2014), therefore, used a very sim- ilar statistical approach to the one taken in the current study; Behaviour Sequence Analysis however, their series of studies focused on mock theft, cheating game behaviours and group discussions and decep- Behaviour sequence analysis (BSA), also referred to as lag tion. The current study uses a similar statistical approach, but sequence analysis (LSA), is a useful method for understanding with real-world deception. the dynamic relationship between progressions of behaviours Understanding and interpreting NVC typically start by and social interactions occurring over time (Beune et al. 2010; identifying discrete, single behaviours, referred to as tells Keatley et al. 2016; Taylor et al. 2008). Sequence analysis (Collett 2003; Ekman et al. 1991; Navarro and Karlins typically involves three key stages (Clarke and Crossland 2008). Hartwig and Bond (2011) outlined the leakage 1985). First, unitisation involves taking a person’s whole re- hypothesis as a means by which individuals leak informative sponse to a question and cutting it into discrete behaviours or tells while trying to conceal the truth. For instance, poker events. Second, classification involves placing the behaviours players who have become successful at reading NVC have or events into distinct categories that are functionally similar. the ability to identify particular behaviours of their opponents. Finally, analysis involves the statistical measurement of This line of research, however, has limitations in terms of whether transitions between behaviours are occurring signifi- individual differences in the ability to mask emotion and de- cantly above the level of chance. For example, if we imagine a ception (Blanck et al. 1981). simple chain of behaviours, question asked (category a), re- In contrast to looking at single behaviours or micro re- spondent shakes head (category b), respondent taps finger actions, researchers have suggested that ‘clusters’ of be- (category c) and respondent looks up and right (category d), haviours are better indicators of changes in emotion and first-order or lag-one behaviour sequence analysis tests wheth- psychological distress, which may reflect attempts to de- er ‘a-b’, ‘a-c’,b-c’ and ‘c-d’ pairs, for instance, are more or ceive (DePaulo et al. 2003; Hartwig and Bond 2011; less likely to occur than by chance alone. Therefore, sequence Navarro and Karlins 2008;Vrijetal. 2004). Therefore, analysis begins by first finding stimuli (e.g., video clips of simply recognising one single behaviour as an indication individuals caught lying or telling the truth) that can be of emotions or lying is not completely reliable, though analysed. A coding scheme is then developed so that each Hartwig and Bond (2011) highlight that while multiple individual behaviour shown in the stimuli can be categorised cues may be better than single cues, individual behaviours (e.g., ‘head-nod’, ‘eyes-looks down’). When an exhaustive still contribute a lot to the detection of deception. The clus- and mutually exclusive list is developed, all stimuli are then ter approach, which is based on more applied settings, out- coded. The sequence analysis then measures transitions be- lines that NVC involves the changes of behaviours across tween pairs of behaviours, to see if pairs occur above chance. the entire body (i.e., movement of feet, hands, face, etc.). BSA is underpinned by Markov models, which involve This research has gained support from applied fields of studying transitions between behaviour pairs (Ivanouw forensic and investigative psychology and police work 2007). The first event in a pairing is the antecedent (e.g., nods (Furnham and Taylor 2011). While the cluster approach head); the second behaviour in a pair is the sequitur (e.g., taps has proven effective, there may be an additional step in finger). In the simplest form of BSA, the analysis determines terms of understanding the sequence of body movements. whether the antecedent causes the sequitur to be more likely to For instance, it may be possible that body movements oc- occur, than expected by chance. Sequence analysis has been cur in sequence (i.e., the feet change movement first, used in a variety of social interactions and behaviours, such as followed by the face). Research has shown that if an indi- marital conflict (Gottman 1979), violent episodes between vidual attempts to mask one behaviour, they may leak be- people (Beale et al. 1998; Turner and Clarke 2009) and rape haviours on other parts of their body (Ekman and Friesen cases, in relation to physical interactions (Fossi et al. 2005)as 1969). Looking at patterns of behaviours in deceptive and well as verbal interactions and strategies (Lawrence et al. truthful interactions has been previously researched by 2010) between attacker and victim. J Police Crim Psych (2018) 33:109–117 Present Study cases. For the truthful response group, 12 criminals were analysed (11 males, 1 female). For the deception group, a The main aim of the current research is to provide a new total of 51 clips were used. Length of clips ranged from 4 method, behaviour sequence analysis, for understanding to 58 s (M = 16.05, SD = 11.64). For the truthful group, a and researching NVC. Previous research has shown that total of 49 clips were used. Length of clip ranged from 6 to the accuracy rates of trained professionals are modest, in 57 s (M = 18.6, SD = 10.2). Clips were obtained through terms of detecting deception through NVC techniques cross-referencing well-known cases of individuals caught (Burgoon et al. 2014; Burgoon et al. 2015;Frank and lying with existing recordings of incidents in which they Ekman 1997; Vauch et al. 2016). However, many NVC are clearly shown on tape lying. For both groups, lying studies lack ecological validity (Richardson et al. 2000; about committing murder was the most frequent video clip Scriba et al. 1999). Although participants in lab-based ex- used (n = 10 in dishonest group, n = 6 in honest group). periments may be motivated towards an outcome, they are Inclusion criteria for each group were that the person was not at risk of major negative consequences for being ex- unequivocally exposed as lying or telling the truth by later posed, as a criminal would be in a real investigation. investigation and evidence. Everyone in the study had no in- Indeed, the absence of such high stakes might not produce dication of any medical condition that would affect their body a sufficient amount of anxiety or motivation for the person movements. The study was approved by the University of in question to display valid facial cues or NVC (Miller and [omitted] Research Ethics Committee. Stiff 1993). Therefore, the current study used real-world recorded videos of people, rather than in a laboratory. Coding Procedure The sample chosen were criminals, and/or people of high power, who were later unequivocally exposed as being Behaviours were recorded in relation to a question and answer guilty at the time of making the statements of innocence pattern. A question was asked; then once the participant began that were recorded and analysed. Owing to the nature of to answer with a response, behaviours were recorded until the the current study, no formal hypotheses were made. end of the response. When a new question was asked, a new However, it is likely that previous research into NVC will sequence of behaviours was analysed. Initially, a list of possi- be supported as elements of larger sequences. There are ble behaviours was developed from existing literature also likely to be differences in the chains of behaviours (Burgoon et al. 2014, 2015). This provided a baseline of be- exhibited between deceptive and truthful statements. haviours that might occur when viewing the clips. If any Based on the research by Hartwig and Bond (2011)and additional behaviours became apparent when viewing the Sporer and Schmidt (2007), behaviours related to head clips, these behaviours were added to the list. Clips were movements and hand movements are likely to be different slowed down and viewed frame by frame, in order to get a between honest and dishonest statements, and clusters of more accurate analysis of the sequence of NVC. The coding behaviours are likely to occur, but with certain individual categories were mutually exclusive and exhaustive, which is a behaviours occurring more frequently in honest and dis- prerequisite of sequence analysis (Bakeman and Quera 2011). honest statements. Each clip was analysed by two separate researchers involved in the study to ensure inter-rater reliability. Finally, the coding scheme was given to an expert in the field to assess and ap- Methods prove. Given the nature of the coding and behaviour sequence analysis, future research can be directly added to the current Sample data to build larger data sets for analyses. Video-recorded interview clips of politicians, criminals Statistical Analysis and people of high interest were obtained through various media sources. Content of these clips were real-life exam- After videos were coded into chains of discrete categories, ples of deception or honesty. Each clip contained state- data were input into the statistical software R (R Core Team ments made by participants in response to a question, For descriptions of individuals used in the analysis, topic of statement, and which were later proven to be a lie or the truth. Clips came length of clip, see supplementary material S1. It should be noted that different from popular online video websites and included scenes individuals were used for honest and deception groups, owing to availability of from press conferences through to police interrogations. materials. However, both groups contained criminals—to allow better com- parison. Sampling was therefore purposeful, focusing on only those videos A sample of videos of 19 individuals (15 male, 4 female) that were of high quality and of individuals who were dishonest or honest. between the ages of 19–48 was collected via online As video clips only showed clear movements from the chest-upwards for the websites and archive footage documents, for both groups. majority of people, lower limbs were not put into the sequence analysis. See Multiple clips were used for the same individuals, in some Supplementary material S2 for coding list. 112 J Police Crim Psych (2018) 33:109–117 Table 1 Frequencies of Behaviour Frequency behaviours—deception statements Head-shake head 66 Eyes-looks down 58 Head-nod head 53 Face-furrow eyebrows (pull brows down) 44 Face-raised eyebrows 39 Mouth-pressing lips together 22 Head-tilts head sideways 21 Eyes-irregular blinking 20 Eyes-looks to the side 17 Head-tilt head forward; eyes-looks up 14 Mouth-full mouth smile 13 Head-tilts head down 12 Eyes-avert eyes 10 Body-half shrug; mouth-licking lips; mouth-harsh swallow 9 Face-tightening jaw; hands-clenching fists; hands-slams hands down; mouth-pouting (push lips 8 forward and together); head-circle head Eyes-pupil dilation 7 Hands-self-touch (rubs hand or fingers) 6 Body-tapping foot; hands-raises hands; mouth-half smile (from one side of the mouth); 5 body-pointing (away); body-self-touch (face); mouth-opens mouth; head-tilts head up; eyes-look towards exit Body-shrugging; hands-palms outwards 4 Face-keep a ‘frozen’ face; face-flair nostrils; hands-reach hand out 3 Face-self-touch (head); body-slouching; body-straightening up; body-steps backwards; 2 body-fidget; body-stiffen shoulders; body-disconnects Eyes-widen eyes; body-crossing arms; face-scratching; mouth-biting lip; hands-twitch hand; 1 body-creating physical barrier; body-twitch leg; body-point part of body towards exit; body-tilts forwards; body-self-grooming; head-shake head; eyes-looks up Behaviours are coded by body location (e.g., face, eyes, body) followed by what the behaviour/movement was that occurred (e.g., twitch, look away, open mouth). Behaviours with the same frequencies are grouped together 2013) and analysed using a behaviour sequence analysis pro- The next, main stage of sequence analysis is to calculate gramme that the researchers wrote. The programme calculated transition frequencies between antecedents and sequiturs. A frequencies of individual behaviours, transitional frequencies, state transition diagram was made on the data (see Fig. 1). chi-squared (χ ) statistics and standardised residuals. The first thing to note with the state transition diagram is that a first-order, also known as lag one, sequence analysis, was conducted. Therefore, only links between behaviour pairs are analysed. These pairs then form longer chains. The cor- Results rect way to read the diagram is moving from one behaviour in single steps to a following behaviour, linked via an arrow. Deception Sequences Whereas the diagram shows a full map from start to end, which is not to say any/all participants followed the same Analyses were conducted on 55 video sequences of a partic- routes entirely, numbers beside the arrows indicate likelihood ular question-answer episode in which participants were ex- and frequency of transitions. Second, all transitions in the posed to be lying. Frequencies of behaviours that participants diagram are statistically significant (p < 0.05). Klonek and showed were first calculated (see Table 1). The most frequent- ly occurring behaviours were participants shaking their head Transition matrices are typically extremely large and are therefore available (n = 66), looking down (n = 58), nodding their head (n =53) as Supplementary material S3. and furrowing their eyebrows (pulling their eyebrows down) It would be possible to conduct ‘higher order’ sequence analysis (e.g., (n = 44). These results support previous findings on deception AB➔C, BC➔D chains); however, this typically leads to over-fitting of data related to NVC behaviours (Mann et al. 2002)(Table 2). to particular stimuli, and typically does not offer clearer analyses. J Police Crim Psych (2018) 33:109–117 Table 2 Frequencies of Behaviour Frequency behaviours—truthful statements Eyes-looks up 39 Head-nod head 35 Face-raised eyebrows 27 Eyes-looks to the side 26 Face-keep a ‘frozen’ face 20 Head-tilts head up 19 Face-furrow eyebrows (pull brows down) 15 Eyes-looks down 14 Mouth-harsh swallow; head-tilts head sideways 13 Eyes-widen eyes; mouth-pouting (push lips forward and together); mouth-pressing lips together 12 Body-slouching; head-tilt head forward 11 Hands-palms outwards 10 Face-flair nostrils 9 Face-tightening jaw 8 Mouth-licking lips; body-creating physical barrier 6 Face-self-touch (eyes); eyes-irregular blinking; body-straightening up; hands-reach hand out; 5 body-pointing (away); mouth-opens mouth Eyes-avert eyes; mouth-full mouth smile; head-tilts head down 4 Body-fidget; body-tilts forwards; body-slopping shoulders 3 Face-scratching; hands-twitch hand; head-shake head; body-lean away from you 2 Face-self-touch (mouth); body-shrugging; hands-clenching fists; body-crossing arms; 1 body-tapping foot; body-self-touch (other/body); head-circle head; body-puffed chest; eyes-look towards exit Behaviours are coded by body location (e.g., face, eyes, body) followed by what the behaviour/movement was that occurred (e.g., twitch, look away, open mouth). Behaviours with the same frequencies are grouped together colleagues suggest a cut-off criteria of transitional frequencies Figure 1 shows that overall, the first significant behaviour below 5 being omitted from analyses for large data sets displayed after a question has been asked, is the furrowing of (Klonek et al. 2015); however, given the smaller data set of eyebrow (SR = 1.2, n = 6). This behaviour leads to several the current sample, a frequency cut-off of only showing tran- following behaviours: head tilt sideways (SR = 1.2, n =3), sitions above frequency of 3 was used, to reduce complexity slam hands down (SR = 3.2, n = 3), tilt head forward and only include meaningfully high transitions. (SR = 2.0, n = 3), shake head (SR = 2.0, n = 9) and eyes look The diagram should be read one step at a time, between down (SR = 1.4, n = 7). Head shaking was the most frequently behaviours. For instance, after being asked a question, occurring behaviour and had a direct link to end of the se- furrowing the eyebrows has a standardised residual (SR) quence (i.e., end of answering) (SR = 3.6, n =14). This link of 1.2, and six individuals made that response, in the cur- was the most frequently occurring link in the diagram. This rent data set. The n values in the diagram refer to the rep- suggests that the final gesture individuals make before resentation of transitions across the video clips, rather than finishing a lying sequence was to shake their head. The be- individual people. After eyebrows have been furrowed, haviour ‘eyes-looking down’ was the next most frequently several different behaviours followed, for different individ- occurring behaviour, but was not directly linked to the end uals. For instance, after furrowing the eyebrows, head of the answer sequence. Instead, several individuals ended shaking is two times more likely to occur than chance, their answer sequence by nodding their head (SR = 1.2, and nine individuals showed that link. These links illus- n =7)orpressinglipstogether(SR = 3.1, n = 6), with slightly trate the pairings of behaviours, rather than long (multiple fewer tightening their jaw (SR = 4.1, n = 4) or simply tilting behaviour) sequences. Of the original six people who head forward (perhaps in a half nod gesture) (SR = 1.2, n =7). showed furrowing eyebrows, several may have shown oth- er behaviours next, not directly moving on to head shaking. Truthful Sequences Therefore, a first-order, or lag-one state transition, diagram shows only links between behaviour pairs, not longer Analyses were also conducted on truthful responses. Frequencies of behaviours exhibited by honest respondents were chains. 114 J Police Crim Psych (2018) 33:109–117 Fig. 1 State transition diagram of behaviours exhibited during question shown in brackets. Position of boxes does not indicate temporal order— response for deception sequences. All transitions shown are significant. arrows indicate sequence of behaviours Standardised residuals are given above or beside arrows; frequencies are calculated first (see Table 1). The most frequently occurring Discussion behaviours were eyes-looks up (n = 39) and head-nods (n = 35). The next main stage of sequence analysis, to allow a con- The aim of the current research was to show the benefits of a trast between deceptive and honest movements, was to con- behaviour sequence analysis approach to understanding NVC duct a sequence analysis on the honest group. The first notable in relation to deception. The first part of the BSA, which shows difference between the two sequences is that the honest se- frequencies of individual behaviours, indicates support for pre- quence state transition diagram has fewer behaviour exhibited. vious findings in the literature (Mann et al. 2002). The frequen- Within the sequences, there is no clear path between start and cy analyses also indicate that deceptive individuals repeatedly end of response; however, head tiling forward (SR = 4.1, demonstrated one behaviour particular to them, aka their ‘tell’, n = 6) and head tilting sideways (SR = 2.0, n =4) were the which they tended to repeat regularly when making a lying first significant behaviours exhibited by honest respondents, statement; this is why head shake, for instance, occurs so fre- which is in contrast to dishonest responses, wherein furrowing quently. In contrast, honest responses typically exhibited fewer of the eyebrows preceded both behaviours (Fig. 2). behaviours overall, which supports previous findings in the J Police Crim Psych (2018) 33:109–117 Fig. 2 State transition diagram of behaviours exhibited during question response for truthful sequences. All transitions shown are significant. Standardised residuals are given above or beside arrows; frequencies are shown in brackets. Position of boxes does not indicate temporal order—arrows indicate sequence of behaviours literature (Watson et al. 2016). In this way, the frequencies in In contrast to people shaking their heads, it may be that the current research are analogous to previous research that has some individuals are aware of this tell and deliberately nodded focused on individual tells. However, as Hartwig and Bond instead, to mask their knowledge. This finding indicates that (2011) suggest, multiple cues may be better indicators of de- nodding or shaking of the head alone are not conclusive indi- ception and honesty. Therefore, the next stage of the research cators of deceit, which supports previous literature in the area was to investigate the sequence of behaviour pairs. of NVC (Mann et al. 2002). The role of eye movements was The state transition diagram for the deception sequences not clearly supported in the current findings, which supports shows that the first behaviour following being asked a question the research of Sporer and Schwandt (2007). It should be was furrowing of the eyebrows. This may be an indicator of noted that the current sample involved people who may be individuals needing to stop and think about their answer aware and deliberately masking body movements. Indeed, a (Schmidt and Cohn 2001). This is also in contrast to the honest strength of the cluster approach and sequence analysis ap- sequence diagram, in which furrowing of the eyebrows did not proach to NVC is that is capitalises on tracking and highlight- occur at all. In the deception sequences, several behaviours ing multiple behaviours individuals might accidently make or followed, which indicate further consideration and possible re- leak when concentrating too much on controlling a particular flection. For instance, the head tilt sideways linked directly body part. back to furrowing of the eyebrows and may be the case that There are several limitations and areas of future research these behaviours show an internal conflict requiring attention stemming from the present study. First, the current analysis and consideration (Schmidt and Cohn 2001). This shows that used a behaviour sequence approach to understand the pro- head tilting sideways is not, in itself, an indicator of deception. gression of behaviours, which did not account for time, per se. 116 J Police Crim Psych (2018) 33:109–117 Compliance with Ethical Standards Future research could focus on time-interval sequences, which might clarify the changes between micro tells (occurring very Funding (If any): No funding was received for the current research. early in the sequence) compared to deliberative body move- ments (occurring after a few seconds have elapsed). A further Conflict of Interest The authors declare that they have no conflict of limitation was the use of behaviours above chest height (i.e., interest. mainly the face). Also, baseline measures were not an option in the current recordings, which were clips of only the lying Ethical Approval Ethics was approved by the School of Psychology answer. A further limitation was that multiple clips from the Research Ethics Committee at the University of [removed for anonymous review]. same person have been used in the current data set, which could lead to over-representation of particular cues. Future Informed Consent No participants were used in the current study, as research should build on the current findings with different all materials were freely accessibly from online media sites. If ownership individuals, to reduce the potential impact of this issue. of videos was in dispute, consent for the use would have been given; Future research should also include more women and men however, as all materials were available on multiple online free sites, it was deemed appropriate to use for analysis by the Ethics Committee. so that gender differences can be analysed. Future research could investigate a ‘pre-question’ sequence of behaviours Open Access This article is distributed under the terms of the Creative and analyse whether there are significant changes in behav- Commons Attribution 4.0 International License (http:// iours, which could indicate changes in emotion or deceit. A creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give major strength of the current research, beyond the new method appropriate credit to the original author(s) and the source, provide a link introduced, was the use of real-world data. The majority of to the Creative Commons license, and indicate if changes were made. NVC and deceit research is conducted in laboratory settings; the current method allows for real-world behaviours to be analysed. This also provides an opportunity for future research References to compare sequences of actual behaviours caught on film compared to lab study behaviours. The outcome of this com- Bakeman R, Quera V (2011). Sequential analysis and observational parison might show that lab studies show very different se- methods for the behavioral sciences. Cambridge University Press, quences of behaviours, in contrast to real-world cases. Finally, Cambridge future research could focus on verbal alongside nonverbal Beale D, Cox T, Clarke DD, Lawrence C, Leather P (1998) Temporal architecture of violent incidents. J Occup Health Psychol 3:65–82 communication. Research using BSA has previously focused Beune K, Giebels E, Taylor PJ (2010) Patterns of interaction in police on verbal strategies (Lawrence et al. 2010), however, not in interviews: the role of cultural dependency. Crim Justice Behav 37: relation to deceitfulness of the statements. Therefore, a much 904–925 needed progression in the literature is to investigate the role of Blanck PD, Rosenthal R, Snodgrass SE, DePaulo BM, Zuckerman M verbal and nonverbal lies. (1981) Sex differences in eavesdropping on nonverbal cues: devel- opmental changes. J Pers Soc Psychol 41:391–396 Burgoon JK, Schuetzler R, Wilson D (2015) Kinesic patterning in decep- tive and truthful interactions. J Nonverbal Behav 39:1–24 Conclusions Burgoon JK, Proudfoot JG, Schuetzler R, Wilson D (2014) Patterns of nonverbal behavior associated with truth and deception: illustrations The current research will not only help to understand what from three experiments. J Nonverbal Behav 38:325–354 Casarrubea M, Jonsson GK, Faulisi F, Sorbera F, Di Giovanni G, Benigno behaviours are produced when a deceptive statement is con- A, Crescimanno G, Magnusson MS (2015) T-pattern analysis for the structed, but it may also determine the sequences of produc- study of temporal structure of animal and human behavior: a com- tion and deduce potential critical behaviours within larger se- prehensive review. J Neurosci Methods 239:34–46 quences of nonverbal chains. The sequence analysis approach Clarke, D. D., & Crossland, J. (1985). Action systems: an introduction to allows us to see very clearly, changes in behaviours and NVC, the analysis of complex behaviour, London: Methuen. Collett P (2003) The book of tells. Bantam, New York which are an indication of changes within the individual as a DePaulo BM, Lindsay JJ, Malone BE, Muhlenbruck L, Charlton K, result of the question that has been asked. This approach is a Cooper H (2003) Cues to deception. Psychol Bull 129:74–118 novel way of statistically analysing changes in behaviour and Ekman P, Friesen WV (1969) Nonverbal leakage and clues to deception. indicating what this may mean, for example, changes in emo- Psychiatry 32:88–106 tion or reasons for emotional changes. Investigating temporal Ekman P, O’Sullivan M (1991) Who can catch a liar? Am Psychol 46:913 changes in NVC could be a useful investigation tool for inter- Ekman P, O’Sullivan M, Friesen WV, Scherer KR (1991) Face, voice, and body in detecting deception. J Nonverbal Behav 15:125–135 viewers or interrogators to use when analysing responses. Fossi JJ, Clarke DD, Lawrence C (2005) Bedroom rape sequences of Sequences of behaviours could be used to identify changes sexual behavior in stranger assaults. Journal of interpersonal vio- in behaviour as a reaction to questions and therefore indicate lence 20:1444–1466 where to focus future further questioning, to investigate the Frank MG, Ekman P (1997) The ability to detect deceit generalizes across different types of high-stake lies. J Pers Soc Psychol 72:1429 cause of the behaviours. J Police Crim Psych (2018) 33:109–117 117 Frank MG, Menasco MA, O’Sullivan M (2008) Human behavior and R Core Team. (2013). R: a language and environment for statistical com- puting. R Foundation for Statistical Computing, Vienne, Austria. deception detection. In: Voeller JG (ed) Handbook of science and technology for homeland security, vol 5. John Wiley & Sons, Inc., ISBN 3–900051–07-0, URL http://www.R-project.org/. Hoboken Richardson CK, Bowers D, Bauer RM, Heilman KM, Leonard CM Furnham A, Taylor J (2011) Bad apples: identify, prevent & manage (2000) Digitizing the moving face during dynamic displays of emo- negative behavior at work. Basingstoke, Palgrave Macmillan tion. Neuropsychologia 38:1028–1039 Gottman JM (1979) Marital interaction: experimental investigations. Schmidt KL, Cohn JF (2001) Human facial expressions as adaptations: Academic Press, New York evolutionary questions in facial expression research. Am J Phys Hartwig M, Bond CF Jr (2011) Why do lie-catchers fail? A lens model Anthropol 116:3–24 meta-analysis of human lie judgements. Psychol Bull 137:643–659 Scriba H, Stoeckli SJ, Veraguth D, Pollak A, Fisch U (1999) Objective Ivanouw J (2007) Sequence analysis as a method for psychological re- evaluation of normal facial function. Annals of Otology, Rhinology search. Nordic Psychology 59:251–267 & Laryngology 108:641–644 Keatley, D. A., Barsky, A. D., & Clarke, D. D. (2016). Drink driving in Sporer SL, Schwandt B (2007) Moderators on non-verbal indicators of the UK: a sequence analysis approach. In: British Society for the deception: a meta-analytic synthesis. Psychol Public Policy law 13: Psychology of Individual Differences, 8 April 2016, Nottingham 1–34 Trent University. Tamietto M, De Gelder B (2010) Neural bases of the non-conscious Klonek FE, Quera V, Kauffeld S (2015) Coding interactions in motiva- perception of emotional signals. Nat rev Neurosci 11:697–709 tional interviewing with computer-software. Comput hum Behav Taylor PJ, Jacques K, Giebels E, Levine M, Best R, Winter J, Rossi G 44(C):284–292 (2008) Analysing forensic processes: taking time into account. Krauss RM, Chen Y, Chawla P (1996) Nonverbal behavior and nonverbal Issues in Forensic Psychology 8:45–57.57 communication: what do conversational hand gestures tell us? Adv Turner K, Clarke DD (2009) Aggression in intellectual disability—anew Exp Soc Psychol 28:389–450 approach. Ment Health rev J 14:28–36 Lawrence C, Fossi J, Clarke D (2010) A sequential examination of of- Vauch V, Sporer SL, Michael SW, Meissner CA (2016) Does training fenders’ verbal strategies during stranger rapes: the influence of improve the detection of deception? A meta-analysis. Commun res location. Psychol Crime law 16:381–400 43:283–343 Mann S, Vrij A, Bull R (2002) Suspects, lies, and videotape: an analysis Vrij A, Akehurst L, Soukara S, Bull R (2004) Detecting deceit via anal- of authentic high-stake liars. Law hum Behav 26:365–376 yses of verbal and nonverbal behavior in children and adults. Hum Mann S, Vrij A, Bull R (2004) Detecting true lies: police officers’ ability Commun res 30:8–41 to detect suspects’ lies. J Appl Psychol 89:137 Chicago Miller GR, Stiff JB (1993) Deceptive communication, vol 14. Sage Watson, S. J., Conchie, S. M., Taylor, P. J., & Poppe, R. W. (2016). Publications, Inc., Newbury Park Utilising motion capture technology to identify trusted testimony th Navarro J (2011) Clues to deceit: a practical list. A & A Printing, Tampa in military encounters. Poster presented at the BPS 5 Military Navarro J, Karlins M (2008) What every BODY is saying, vol 27. Harper Conference, Basingstoke, UK. Collins, New York, p 90 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Police and Criminal Psychology Springer Journals

A Sequence Analysis of Nonverbal Behaviour and Deception

Free
9 pages

Loading next page...
 
/lp/springer_journal/a-sequence-analysis-of-nonverbal-behaviour-and-deception-j02elFs3yQ
Publisher
Springer Journals
Copyright
Copyright © 2017 by The Author(s)
Subject
Psychology; Psychology, general; Law and Psychology; Criminology and Criminal Justice, general
ISSN
0882-0783
eISSN
1936-6469
D.O.I.
10.1007/s11896-017-9238-9
Publisher site
See Article on Publisher Site

Abstract

J Police Crim Psych (2018) 33:109–117 DOI 10.1007/s11896-017-9238-9 1 2 3 1 A Marono & DD Clarke & J Navarro & DA Keatley Published online: 7 June 2017 The Author(s) 2017. This article is an open access publication Abstract The ability to correctly interpret nonverbal commu- Keywords Nonverbal communication Behaviour sequence . . . nication (NVC) is an important ability in everyday interac- analysis Credibility assessment Deception Suspect tions, which may use NVC techniques to identify the conceal- behaviour ment of information. In the present study, a novel approach was used to understand NVC. Behaviour sequence analysis identified specific sequences of behaviours that indicate psy- Introduction chological distress caused by deception. The study involved the analysis of 55 videos of real criminals and high-power Understanding nonverbal communication (NVC) is important individuals that were filmed fabricating statements, which in a range of situations and settings, from daily relationships to were later exposed as being untruthful at the time of being interrogations (Krauss et al. 1996). Research has shown that filmed. In addition, 53 clips of criminals making truthful state- NVC includes both reflexive and nonreflexive movements of ments were also analysed as a contrast group. Results indicat- the body, which communicate an emotional message to others ed clear differences between honest and deceptive responses, (Tamietto and De Gelder 2010). Correct interpretation of in- such as furrowing of eyebrows in the deceptive sequences dividuals’ emotional states and intentions is obviously impor- occurring more often than honest statements. In addition, se- tant for effective communication; however, a particularly im- quences of behaviours were shown in the present data set, portant area for NVC is in police and legal investigations. which could indicate a new method for analysing NVC and During a police interview or interrogation, investigators may detecting psychological distress caused by deception. The implicitly or explicitly use NVC reading techniques to identify possible implications and applications for police and forensic the concealment of emotions, which may indicate an intent to investigation are also outlined. deceive or a line of questioning worth pursuing (Hartwig and Bond 2011;Mann et al. 2004). Behavioural cues of mental effort, memories and emotions are used by security personnel Electronic supplementary material The online version of this article to identify behavioural ‘hot spots’ that indicate that a topic is (doi:10.1007/s11896-017-9238-9) contains supplementary material, worth considering and investigating further (Frank et al. which is available to authorized users. 2008). Typically, research has focused on behaviours related to particular emotions or concealment of information (Ekman * DA Keatley DKeatley@lincoln.ac.uk and O’Sullivan 1991). Sporer and Schwandt (2007)outline several reliable indicators of deception, such as nodding, J Navarro http://www.jnforensics.com movement of feet and legs and hand movements; however, they showed no systematic relationship between avoidance of Researchers in Behaviour Sequence Analysis, Psychology eye contact and deception. However, researchers and profes- Department, University of Lincoln, Lincoln LN67TS, UK sional practitioners in the field have suggested that clusters of Researchers in Behaviour Sequence Analysis, Psychology behaviours are better indicators of changes in emotion and Department, University of Nottingham, Nottingham NG72RD, UK possible psychological discomfort,which may arise from JN Forensics, LCC, Tampa, FL, USA attempting to deceive someone (DePaulo et al. 2003; 110 J Police Crim Psych (2018) 33:109–117 Navarro and Karlins 2008,Navarro 2011;Vrijet al. 2004). Burgoon et al. (2014, 2015). In their seminal research, The present research continues a more recent trend in NVC by patterns of nonverbal behaviour, across time, were investigating the possibility that it is not only discrete behav- analysed, showing clear differences between deceivers iours or clusters of movements that can be used to indicate and truthful participants. Burgoon and colleagues’ research emotion or distress but also the sequence that these move- provides a clear foundation to begin future development in ments occur in (Burgoon et al. 2014; Burgoon et al. 2015). investigating patterns and sequences of behaviours, rather Previous research that has focused on sequences of behaviour than individual behaviours or clusters. It is to this direction has used a software programme that detects patterns in time- the present research builds on, by using real-world decep- ordered data. These T-patterns (see Casarrubea et al. 2015)are tive statements, rather than mock crime experiments. a multivariate approach to identifying temporal structure in behaviour. Burgoon et al. (2014), therefore, used a very sim- ilar statistical approach to the one taken in the current study; Behaviour Sequence Analysis however, their series of studies focused on mock theft, cheating game behaviours and group discussions and decep- Behaviour sequence analysis (BSA), also referred to as lag tion. The current study uses a similar statistical approach, but sequence analysis (LSA), is a useful method for understanding with real-world deception. the dynamic relationship between progressions of behaviours Understanding and interpreting NVC typically start by and social interactions occurring over time (Beune et al. 2010; identifying discrete, single behaviours, referred to as tells Keatley et al. 2016; Taylor et al. 2008). Sequence analysis (Collett 2003; Ekman et al. 1991; Navarro and Karlins typically involves three key stages (Clarke and Crossland 2008). Hartwig and Bond (2011) outlined the leakage 1985). First, unitisation involves taking a person’s whole re- hypothesis as a means by which individuals leak informative sponse to a question and cutting it into discrete behaviours or tells while trying to conceal the truth. For instance, poker events. Second, classification involves placing the behaviours players who have become successful at reading NVC have or events into distinct categories that are functionally similar. the ability to identify particular behaviours of their opponents. Finally, analysis involves the statistical measurement of This line of research, however, has limitations in terms of whether transitions between behaviours are occurring signifi- individual differences in the ability to mask emotion and de- cantly above the level of chance. For example, if we imagine a ception (Blanck et al. 1981). simple chain of behaviours, question asked (category a), re- In contrast to looking at single behaviours or micro re- spondent shakes head (category b), respondent taps finger actions, researchers have suggested that ‘clusters’ of be- (category c) and respondent looks up and right (category d), haviours are better indicators of changes in emotion and first-order or lag-one behaviour sequence analysis tests wheth- psychological distress, which may reflect attempts to de- er ‘a-b’, ‘a-c’,b-c’ and ‘c-d’ pairs, for instance, are more or ceive (DePaulo et al. 2003; Hartwig and Bond 2011; less likely to occur than by chance alone. Therefore, sequence Navarro and Karlins 2008;Vrijetal. 2004). Therefore, analysis begins by first finding stimuli (e.g., video clips of simply recognising one single behaviour as an indication individuals caught lying or telling the truth) that can be of emotions or lying is not completely reliable, though analysed. A coding scheme is then developed so that each Hartwig and Bond (2011) highlight that while multiple individual behaviour shown in the stimuli can be categorised cues may be better than single cues, individual behaviours (e.g., ‘head-nod’, ‘eyes-looks down’). When an exhaustive still contribute a lot to the detection of deception. The clus- and mutually exclusive list is developed, all stimuli are then ter approach, which is based on more applied settings, out- coded. The sequence analysis then measures transitions be- lines that NVC involves the changes of behaviours across tween pairs of behaviours, to see if pairs occur above chance. the entire body (i.e., movement of feet, hands, face, etc.). BSA is underpinned by Markov models, which involve This research has gained support from applied fields of studying transitions between behaviour pairs (Ivanouw forensic and investigative psychology and police work 2007). The first event in a pairing is the antecedent (e.g., nods (Furnham and Taylor 2011). While the cluster approach head); the second behaviour in a pair is the sequitur (e.g., taps has proven effective, there may be an additional step in finger). In the simplest form of BSA, the analysis determines terms of understanding the sequence of body movements. whether the antecedent causes the sequitur to be more likely to For instance, it may be possible that body movements oc- occur, than expected by chance. Sequence analysis has been cur in sequence (i.e., the feet change movement first, used in a variety of social interactions and behaviours, such as followed by the face). Research has shown that if an indi- marital conflict (Gottman 1979), violent episodes between vidual attempts to mask one behaviour, they may leak be- people (Beale et al. 1998; Turner and Clarke 2009) and rape haviours on other parts of their body (Ekman and Friesen cases, in relation to physical interactions (Fossi et al. 2005)as 1969). Looking at patterns of behaviours in deceptive and well as verbal interactions and strategies (Lawrence et al. truthful interactions has been previously researched by 2010) between attacker and victim. J Police Crim Psych (2018) 33:109–117 Present Study cases. For the truthful response group, 12 criminals were analysed (11 males, 1 female). For the deception group, a The main aim of the current research is to provide a new total of 51 clips were used. Length of clips ranged from 4 method, behaviour sequence analysis, for understanding to 58 s (M = 16.05, SD = 11.64). For the truthful group, a and researching NVC. Previous research has shown that total of 49 clips were used. Length of clip ranged from 6 to the accuracy rates of trained professionals are modest, in 57 s (M = 18.6, SD = 10.2). Clips were obtained through terms of detecting deception through NVC techniques cross-referencing well-known cases of individuals caught (Burgoon et al. 2014; Burgoon et al. 2015;Frank and lying with existing recordings of incidents in which they Ekman 1997; Vauch et al. 2016). However, many NVC are clearly shown on tape lying. For both groups, lying studies lack ecological validity (Richardson et al. 2000; about committing murder was the most frequent video clip Scriba et al. 1999). Although participants in lab-based ex- used (n = 10 in dishonest group, n = 6 in honest group). periments may be motivated towards an outcome, they are Inclusion criteria for each group were that the person was not at risk of major negative consequences for being ex- unequivocally exposed as lying or telling the truth by later posed, as a criminal would be in a real investigation. investigation and evidence. Everyone in the study had no in- Indeed, the absence of such high stakes might not produce dication of any medical condition that would affect their body a sufficient amount of anxiety or motivation for the person movements. The study was approved by the University of in question to display valid facial cues or NVC (Miller and [omitted] Research Ethics Committee. Stiff 1993). Therefore, the current study used real-world recorded videos of people, rather than in a laboratory. Coding Procedure The sample chosen were criminals, and/or people of high power, who were later unequivocally exposed as being Behaviours were recorded in relation to a question and answer guilty at the time of making the statements of innocence pattern. A question was asked; then once the participant began that were recorded and analysed. Owing to the nature of to answer with a response, behaviours were recorded until the the current study, no formal hypotheses were made. end of the response. When a new question was asked, a new However, it is likely that previous research into NVC will sequence of behaviours was analysed. Initially, a list of possi- be supported as elements of larger sequences. There are ble behaviours was developed from existing literature also likely to be differences in the chains of behaviours (Burgoon et al. 2014, 2015). This provided a baseline of be- exhibited between deceptive and truthful statements. haviours that might occur when viewing the clips. If any Based on the research by Hartwig and Bond (2011)and additional behaviours became apparent when viewing the Sporer and Schmidt (2007), behaviours related to head clips, these behaviours were added to the list. Clips were movements and hand movements are likely to be different slowed down and viewed frame by frame, in order to get a between honest and dishonest statements, and clusters of more accurate analysis of the sequence of NVC. The coding behaviours are likely to occur, but with certain individual categories were mutually exclusive and exhaustive, which is a behaviours occurring more frequently in honest and dis- prerequisite of sequence analysis (Bakeman and Quera 2011). honest statements. Each clip was analysed by two separate researchers involved in the study to ensure inter-rater reliability. Finally, the coding scheme was given to an expert in the field to assess and ap- Methods prove. Given the nature of the coding and behaviour sequence analysis, future research can be directly added to the current Sample data to build larger data sets for analyses. Video-recorded interview clips of politicians, criminals Statistical Analysis and people of high interest were obtained through various media sources. Content of these clips were real-life exam- After videos were coded into chains of discrete categories, ples of deception or honesty. Each clip contained state- data were input into the statistical software R (R Core Team ments made by participants in response to a question, For descriptions of individuals used in the analysis, topic of statement, and which were later proven to be a lie or the truth. Clips came length of clip, see supplementary material S1. It should be noted that different from popular online video websites and included scenes individuals were used for honest and deception groups, owing to availability of from press conferences through to police interrogations. materials. However, both groups contained criminals—to allow better com- parison. Sampling was therefore purposeful, focusing on only those videos A sample of videos of 19 individuals (15 male, 4 female) that were of high quality and of individuals who were dishonest or honest. between the ages of 19–48 was collected via online As video clips only showed clear movements from the chest-upwards for the websites and archive footage documents, for both groups. majority of people, lower limbs were not put into the sequence analysis. See Multiple clips were used for the same individuals, in some Supplementary material S2 for coding list. 112 J Police Crim Psych (2018) 33:109–117 Table 1 Frequencies of Behaviour Frequency behaviours—deception statements Head-shake head 66 Eyes-looks down 58 Head-nod head 53 Face-furrow eyebrows (pull brows down) 44 Face-raised eyebrows 39 Mouth-pressing lips together 22 Head-tilts head sideways 21 Eyes-irregular blinking 20 Eyes-looks to the side 17 Head-tilt head forward; eyes-looks up 14 Mouth-full mouth smile 13 Head-tilts head down 12 Eyes-avert eyes 10 Body-half shrug; mouth-licking lips; mouth-harsh swallow 9 Face-tightening jaw; hands-clenching fists; hands-slams hands down; mouth-pouting (push lips 8 forward and together); head-circle head Eyes-pupil dilation 7 Hands-self-touch (rubs hand or fingers) 6 Body-tapping foot; hands-raises hands; mouth-half smile (from one side of the mouth); 5 body-pointing (away); body-self-touch (face); mouth-opens mouth; head-tilts head up; eyes-look towards exit Body-shrugging; hands-palms outwards 4 Face-keep a ‘frozen’ face; face-flair nostrils; hands-reach hand out 3 Face-self-touch (head); body-slouching; body-straightening up; body-steps backwards; 2 body-fidget; body-stiffen shoulders; body-disconnects Eyes-widen eyes; body-crossing arms; face-scratching; mouth-biting lip; hands-twitch hand; 1 body-creating physical barrier; body-twitch leg; body-point part of body towards exit; body-tilts forwards; body-self-grooming; head-shake head; eyes-looks up Behaviours are coded by body location (e.g., face, eyes, body) followed by what the behaviour/movement was that occurred (e.g., twitch, look away, open mouth). Behaviours with the same frequencies are grouped together 2013) and analysed using a behaviour sequence analysis pro- The next, main stage of sequence analysis is to calculate gramme that the researchers wrote. The programme calculated transition frequencies between antecedents and sequiturs. A frequencies of individual behaviours, transitional frequencies, state transition diagram was made on the data (see Fig. 1). chi-squared (χ ) statistics and standardised residuals. The first thing to note with the state transition diagram is that a first-order, also known as lag one, sequence analysis, was conducted. Therefore, only links between behaviour pairs are analysed. These pairs then form longer chains. The cor- Results rect way to read the diagram is moving from one behaviour in single steps to a following behaviour, linked via an arrow. Deception Sequences Whereas the diagram shows a full map from start to end, which is not to say any/all participants followed the same Analyses were conducted on 55 video sequences of a partic- routes entirely, numbers beside the arrows indicate likelihood ular question-answer episode in which participants were ex- and frequency of transitions. Second, all transitions in the posed to be lying. Frequencies of behaviours that participants diagram are statistically significant (p < 0.05). Klonek and showed were first calculated (see Table 1). The most frequent- ly occurring behaviours were participants shaking their head Transition matrices are typically extremely large and are therefore available (n = 66), looking down (n = 58), nodding their head (n =53) as Supplementary material S3. and furrowing their eyebrows (pulling their eyebrows down) It would be possible to conduct ‘higher order’ sequence analysis (e.g., (n = 44). These results support previous findings on deception AB➔C, BC➔D chains); however, this typically leads to over-fitting of data related to NVC behaviours (Mann et al. 2002)(Table 2). to particular stimuli, and typically does not offer clearer analyses. J Police Crim Psych (2018) 33:109–117 Table 2 Frequencies of Behaviour Frequency behaviours—truthful statements Eyes-looks up 39 Head-nod head 35 Face-raised eyebrows 27 Eyes-looks to the side 26 Face-keep a ‘frozen’ face 20 Head-tilts head up 19 Face-furrow eyebrows (pull brows down) 15 Eyes-looks down 14 Mouth-harsh swallow; head-tilts head sideways 13 Eyes-widen eyes; mouth-pouting (push lips forward and together); mouth-pressing lips together 12 Body-slouching; head-tilt head forward 11 Hands-palms outwards 10 Face-flair nostrils 9 Face-tightening jaw 8 Mouth-licking lips; body-creating physical barrier 6 Face-self-touch (eyes); eyes-irregular blinking; body-straightening up; hands-reach hand out; 5 body-pointing (away); mouth-opens mouth Eyes-avert eyes; mouth-full mouth smile; head-tilts head down 4 Body-fidget; body-tilts forwards; body-slopping shoulders 3 Face-scratching; hands-twitch hand; head-shake head; body-lean away from you 2 Face-self-touch (mouth); body-shrugging; hands-clenching fists; body-crossing arms; 1 body-tapping foot; body-self-touch (other/body); head-circle head; body-puffed chest; eyes-look towards exit Behaviours are coded by body location (e.g., face, eyes, body) followed by what the behaviour/movement was that occurred (e.g., twitch, look away, open mouth). Behaviours with the same frequencies are grouped together colleagues suggest a cut-off criteria of transitional frequencies Figure 1 shows that overall, the first significant behaviour below 5 being omitted from analyses for large data sets displayed after a question has been asked, is the furrowing of (Klonek et al. 2015); however, given the smaller data set of eyebrow (SR = 1.2, n = 6). This behaviour leads to several the current sample, a frequency cut-off of only showing tran- following behaviours: head tilt sideways (SR = 1.2, n =3), sitions above frequency of 3 was used, to reduce complexity slam hands down (SR = 3.2, n = 3), tilt head forward and only include meaningfully high transitions. (SR = 2.0, n = 3), shake head (SR = 2.0, n = 9) and eyes look The diagram should be read one step at a time, between down (SR = 1.4, n = 7). Head shaking was the most frequently behaviours. For instance, after being asked a question, occurring behaviour and had a direct link to end of the se- furrowing the eyebrows has a standardised residual (SR) quence (i.e., end of answering) (SR = 3.6, n =14). This link of 1.2, and six individuals made that response, in the cur- was the most frequently occurring link in the diagram. This rent data set. The n values in the diagram refer to the rep- suggests that the final gesture individuals make before resentation of transitions across the video clips, rather than finishing a lying sequence was to shake their head. The be- individual people. After eyebrows have been furrowed, haviour ‘eyes-looking down’ was the next most frequently several different behaviours followed, for different individ- occurring behaviour, but was not directly linked to the end uals. For instance, after furrowing the eyebrows, head of the answer sequence. Instead, several individuals ended shaking is two times more likely to occur than chance, their answer sequence by nodding their head (SR = 1.2, and nine individuals showed that link. These links illus- n =7)orpressinglipstogether(SR = 3.1, n = 6), with slightly trate the pairings of behaviours, rather than long (multiple fewer tightening their jaw (SR = 4.1, n = 4) or simply tilting behaviour) sequences. Of the original six people who head forward (perhaps in a half nod gesture) (SR = 1.2, n =7). showed furrowing eyebrows, several may have shown oth- er behaviours next, not directly moving on to head shaking. Truthful Sequences Therefore, a first-order, or lag-one state transition, diagram shows only links between behaviour pairs, not longer Analyses were also conducted on truthful responses. Frequencies of behaviours exhibited by honest respondents were chains. 114 J Police Crim Psych (2018) 33:109–117 Fig. 1 State transition diagram of behaviours exhibited during question shown in brackets. Position of boxes does not indicate temporal order— response for deception sequences. All transitions shown are significant. arrows indicate sequence of behaviours Standardised residuals are given above or beside arrows; frequencies are calculated first (see Table 1). The most frequently occurring Discussion behaviours were eyes-looks up (n = 39) and head-nods (n = 35). The next main stage of sequence analysis, to allow a con- The aim of the current research was to show the benefits of a trast between deceptive and honest movements, was to con- behaviour sequence analysis approach to understanding NVC duct a sequence analysis on the honest group. The first notable in relation to deception. The first part of the BSA, which shows difference between the two sequences is that the honest se- frequencies of individual behaviours, indicates support for pre- quence state transition diagram has fewer behaviour exhibited. vious findings in the literature (Mann et al. 2002). The frequen- Within the sequences, there is no clear path between start and cy analyses also indicate that deceptive individuals repeatedly end of response; however, head tiling forward (SR = 4.1, demonstrated one behaviour particular to them, aka their ‘tell’, n = 6) and head tilting sideways (SR = 2.0, n =4) were the which they tended to repeat regularly when making a lying first significant behaviours exhibited by honest respondents, statement; this is why head shake, for instance, occurs so fre- which is in contrast to dishonest responses, wherein furrowing quently. In contrast, honest responses typically exhibited fewer of the eyebrows preceded both behaviours (Fig. 2). behaviours overall, which supports previous findings in the J Police Crim Psych (2018) 33:109–117 Fig. 2 State transition diagram of behaviours exhibited during question response for truthful sequences. All transitions shown are significant. Standardised residuals are given above or beside arrows; frequencies are shown in brackets. Position of boxes does not indicate temporal order—arrows indicate sequence of behaviours literature (Watson et al. 2016). In this way, the frequencies in In contrast to people shaking their heads, it may be that the current research are analogous to previous research that has some individuals are aware of this tell and deliberately nodded focused on individual tells. However, as Hartwig and Bond instead, to mask their knowledge. This finding indicates that (2011) suggest, multiple cues may be better indicators of de- nodding or shaking of the head alone are not conclusive indi- ception and honesty. Therefore, the next stage of the research cators of deceit, which supports previous literature in the area was to investigate the sequence of behaviour pairs. of NVC (Mann et al. 2002). The role of eye movements was The state transition diagram for the deception sequences not clearly supported in the current findings, which supports shows that the first behaviour following being asked a question the research of Sporer and Schwandt (2007). It should be was furrowing of the eyebrows. This may be an indicator of noted that the current sample involved people who may be individuals needing to stop and think about their answer aware and deliberately masking body movements. Indeed, a (Schmidt and Cohn 2001). This is also in contrast to the honest strength of the cluster approach and sequence analysis ap- sequence diagram, in which furrowing of the eyebrows did not proach to NVC is that is capitalises on tracking and highlight- occur at all. In the deception sequences, several behaviours ing multiple behaviours individuals might accidently make or followed, which indicate further consideration and possible re- leak when concentrating too much on controlling a particular flection. For instance, the head tilt sideways linked directly body part. back to furrowing of the eyebrows and may be the case that There are several limitations and areas of future research these behaviours show an internal conflict requiring attention stemming from the present study. First, the current analysis and consideration (Schmidt and Cohn 2001). This shows that used a behaviour sequence approach to understand the pro- head tilting sideways is not, in itself, an indicator of deception. gression of behaviours, which did not account for time, per se. 116 J Police Crim Psych (2018) 33:109–117 Compliance with Ethical Standards Future research could focus on time-interval sequences, which might clarify the changes between micro tells (occurring very Funding (If any): No funding was received for the current research. early in the sequence) compared to deliberative body move- ments (occurring after a few seconds have elapsed). A further Conflict of Interest The authors declare that they have no conflict of limitation was the use of behaviours above chest height (i.e., interest. mainly the face). Also, baseline measures were not an option in the current recordings, which were clips of only the lying Ethical Approval Ethics was approved by the School of Psychology answer. A further limitation was that multiple clips from the Research Ethics Committee at the University of [removed for anonymous review]. same person have been used in the current data set, which could lead to over-representation of particular cues. Future Informed Consent No participants were used in the current study, as research should build on the current findings with different all materials were freely accessibly from online media sites. If ownership individuals, to reduce the potential impact of this issue. of videos was in dispute, consent for the use would have been given; Future research should also include more women and men however, as all materials were available on multiple online free sites, it was deemed appropriate to use for analysis by the Ethics Committee. so that gender differences can be analysed. Future research could investigate a ‘pre-question’ sequence of behaviours Open Access This article is distributed under the terms of the Creative and analyse whether there are significant changes in behav- Commons Attribution 4.0 International License (http:// iours, which could indicate changes in emotion or deceit. A creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give major strength of the current research, beyond the new method appropriate credit to the original author(s) and the source, provide a link introduced, was the use of real-world data. The majority of to the Creative Commons license, and indicate if changes were made. NVC and deceit research is conducted in laboratory settings; the current method allows for real-world behaviours to be analysed. This also provides an opportunity for future research References to compare sequences of actual behaviours caught on film compared to lab study behaviours. The outcome of this com- Bakeman R, Quera V (2011). Sequential analysis and observational parison might show that lab studies show very different se- methods for the behavioral sciences. Cambridge University Press, quences of behaviours, in contrast to real-world cases. Finally, Cambridge future research could focus on verbal alongside nonverbal Beale D, Cox T, Clarke DD, Lawrence C, Leather P (1998) Temporal architecture of violent incidents. J Occup Health Psychol 3:65–82 communication. Research using BSA has previously focused Beune K, Giebels E, Taylor PJ (2010) Patterns of interaction in police on verbal strategies (Lawrence et al. 2010), however, not in interviews: the role of cultural dependency. Crim Justice Behav 37: relation to deceitfulness of the statements. Therefore, a much 904–925 needed progression in the literature is to investigate the role of Blanck PD, Rosenthal R, Snodgrass SE, DePaulo BM, Zuckerman M verbal and nonverbal lies. (1981) Sex differences in eavesdropping on nonverbal cues: devel- opmental changes. J Pers Soc Psychol 41:391–396 Burgoon JK, Schuetzler R, Wilson D (2015) Kinesic patterning in decep- tive and truthful interactions. J Nonverbal Behav 39:1–24 Conclusions Burgoon JK, Proudfoot JG, Schuetzler R, Wilson D (2014) Patterns of nonverbal behavior associated with truth and deception: illustrations The current research will not only help to understand what from three experiments. J Nonverbal Behav 38:325–354 Casarrubea M, Jonsson GK, Faulisi F, Sorbera F, Di Giovanni G, Benigno behaviours are produced when a deceptive statement is con- A, Crescimanno G, Magnusson MS (2015) T-pattern analysis for the structed, but it may also determine the sequences of produc- study of temporal structure of animal and human behavior: a com- tion and deduce potential critical behaviours within larger se- prehensive review. J Neurosci Methods 239:34–46 quences of nonverbal chains. The sequence analysis approach Clarke, D. D., & Crossland, J. (1985). Action systems: an introduction to allows us to see very clearly, changes in behaviours and NVC, the analysis of complex behaviour, London: Methuen. Collett P (2003) The book of tells. Bantam, New York which are an indication of changes within the individual as a DePaulo BM, Lindsay JJ, Malone BE, Muhlenbruck L, Charlton K, result of the question that has been asked. This approach is a Cooper H (2003) Cues to deception. Psychol Bull 129:74–118 novel way of statistically analysing changes in behaviour and Ekman P, Friesen WV (1969) Nonverbal leakage and clues to deception. indicating what this may mean, for example, changes in emo- Psychiatry 32:88–106 tion or reasons for emotional changes. Investigating temporal Ekman P, O’Sullivan M (1991) Who can catch a liar? Am Psychol 46:913 changes in NVC could be a useful investigation tool for inter- Ekman P, O’Sullivan M, Friesen WV, Scherer KR (1991) Face, voice, and body in detecting deception. J Nonverbal Behav 15:125–135 viewers or interrogators to use when analysing responses. Fossi JJ, Clarke DD, Lawrence C (2005) Bedroom rape sequences of Sequences of behaviours could be used to identify changes sexual behavior in stranger assaults. Journal of interpersonal vio- in behaviour as a reaction to questions and therefore indicate lence 20:1444–1466 where to focus future further questioning, to investigate the Frank MG, Ekman P (1997) The ability to detect deceit generalizes across different types of high-stake lies. J Pers Soc Psychol 72:1429 cause of the behaviours. J Police Crim Psych (2018) 33:109–117 117 Frank MG, Menasco MA, O’Sullivan M (2008) Human behavior and R Core Team. (2013). R: a language and environment for statistical com- puting. R Foundation for Statistical Computing, Vienne, Austria. deception detection. In: Voeller JG (ed) Handbook of science and technology for homeland security, vol 5. John Wiley & Sons, Inc., ISBN 3–900051–07-0, URL http://www.R-project.org/. Hoboken Richardson CK, Bowers D, Bauer RM, Heilman KM, Leonard CM Furnham A, Taylor J (2011) Bad apples: identify, prevent & manage (2000) Digitizing the moving face during dynamic displays of emo- negative behavior at work. Basingstoke, Palgrave Macmillan tion. Neuropsychologia 38:1028–1039 Gottman JM (1979) Marital interaction: experimental investigations. Schmidt KL, Cohn JF (2001) Human facial expressions as adaptations: Academic Press, New York evolutionary questions in facial expression research. Am J Phys Hartwig M, Bond CF Jr (2011) Why do lie-catchers fail? A lens model Anthropol 116:3–24 meta-analysis of human lie judgements. Psychol Bull 137:643–659 Scriba H, Stoeckli SJ, Veraguth D, Pollak A, Fisch U (1999) Objective Ivanouw J (2007) Sequence analysis as a method for psychological re- evaluation of normal facial function. Annals of Otology, Rhinology search. Nordic Psychology 59:251–267 & Laryngology 108:641–644 Keatley, D. A., Barsky, A. D., & Clarke, D. D. (2016). Drink driving in Sporer SL, Schwandt B (2007) Moderators on non-verbal indicators of the UK: a sequence analysis approach. In: British Society for the deception: a meta-analytic synthesis. Psychol Public Policy law 13: Psychology of Individual Differences, 8 April 2016, Nottingham 1–34 Trent University. Tamietto M, De Gelder B (2010) Neural bases of the non-conscious Klonek FE, Quera V, Kauffeld S (2015) Coding interactions in motiva- perception of emotional signals. Nat rev Neurosci 11:697–709 tional interviewing with computer-software. Comput hum Behav Taylor PJ, Jacques K, Giebels E, Levine M, Best R, Winter J, Rossi G 44(C):284–292 (2008) Analysing forensic processes: taking time into account. Krauss RM, Chen Y, Chawla P (1996) Nonverbal behavior and nonverbal Issues in Forensic Psychology 8:45–57.57 communication: what do conversational hand gestures tell us? Adv Turner K, Clarke DD (2009) Aggression in intellectual disability—anew Exp Soc Psychol 28:389–450 approach. Ment Health rev J 14:28–36 Lawrence C, Fossi J, Clarke D (2010) A sequential examination of of- Vauch V, Sporer SL, Michael SW, Meissner CA (2016) Does training fenders’ verbal strategies during stranger rapes: the influence of improve the detection of deception? A meta-analysis. Commun res location. Psychol Crime law 16:381–400 43:283–343 Mann S, Vrij A, Bull R (2002) Suspects, lies, and videotape: an analysis Vrij A, Akehurst L, Soukara S, Bull R (2004) Detecting deceit via anal- of authentic high-stake liars. Law hum Behav 26:365–376 yses of verbal and nonverbal behavior in children and adults. Hum Mann S, Vrij A, Bull R (2004) Detecting true lies: police officers’ ability Commun res 30:8–41 to detect suspects’ lies. J Appl Psychol 89:137 Chicago Miller GR, Stiff JB (1993) Deceptive communication, vol 14. Sage Watson, S. J., Conchie, S. M., Taylor, P. J., & Poppe, R. W. (2016). Publications, Inc., Newbury Park Utilising motion capture technology to identify trusted testimony th Navarro J (2011) Clues to deceit: a practical list. A & A Printing, Tampa in military encounters. Poster presented at the BPS 5 Military Navarro J, Karlins M (2008) What every BODY is saying, vol 27. Harper Conference, Basingstoke, UK. Collins, New York, p 90

Journal

Journal of Police and Criminal PsychologySpringer Journals

Published: Jun 22, 2017

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off