Group Decision and Negoti{lfion 13: 61-79,2004
© 2004 Kluwer Academic Publishers. Pri11ted in the. Nerherlwuls
Detecting Deception: The Benefit of Looking at a
Combination of Behavioral, Auditory and Speech
Content Related Cues in a Systematic Manner
ALDERT VRJJ AND SAMANTHA MANN
Psychology Departme/11, University of Ponsmouth, King Henry Building, King Henry I Street, Ponsmouth,
P012DY. UK
Abstract
People com.municate with each o ther across distance in a variety of ways , for example. via the telephone. via
e lectro nic forms of communication (the wriuen word) or via video!inks. In this article. behavioral. auditory and
speech content related cues which research has s hown d iscriminate (to some extent) between liars a nd truth tellers will be d iscussed. Although research has indicated that people are generally poor at detecting deceit, some
recent studies suggest that looking at nonverbal , auditory and speech content related responses in an objective
and systematic manner will lead to more accurate classifications of liars and truth tellers. These studies will be
discussed. Apart from examining the available cues objectively and systematically, lie detection might further
improve if the lie detector employs certain communication techniques. In the final part of this article we will
discuss some of these techniques.
People communicate with each other in a variety of ways, for example, face to face, via
the telephone, via electronic forms of communication (the written word) or via videolinks.
In theory, lie detection might be possible in each communication form as long as a person's behavior, voice or speech conte nt contains signs of deceit. In this a rticle, behavioral
cues (i.e. body related cues, such as gaze behavior, s miles, and movements), auditory cues
(i.e. vocal cues such as pitch of voice, speech rate, a nd speech disturbances) and speech
content related cues (i.e. verbal criteria such as the amount of details mentioned) which
might be worth examining in order to detect deceit, will be presented.
People can detect lies (we will use the words 'lying' and 'deception' interchangeably
throughout this article) on the basis of these cues in different ways. Lie detection can be
conducted in an objective and systematic manner. That is, the freq ue ncy of occurrence of
the available cues are calculated and truth/lie decisions are made on the basis of the frequency of occurrence of these c ues using multiple c ue decision models which are based
upon our current knowledge of cues to deceit. However, this is not the way that lie detection typically occurs. In most daily life situations, lie detection takes place in a more subjective and unsyste matic manner. That is, instead of calcul ating the frequency of occurrence
of the available cues, people form global impressions of the frequency of occurrence of
the available cues. Instead of using clearly defined multiple c ue decision models, their
impress ion of the available information is then translated into t.ruth/lie decisions in a more
subjective and less systematic way, using private decision models which somet.imes rake
invalid cues to deception into account. When people detect deceit in the latter, more sub-
TSA 15-00014 - 000905
62
VRJJ AND MANN
jcctivc and mo re unsystematic manne r, hit rates (percentage of correct a nswers) arc typically low. They usually vary between 45 and 60%, when a 50% hit rate would be obtained
by tossing a coin (DePaulo, Stone, and Lassiter 1985; Kraut 1980; Vrij 2000). In this a rticle, however, we will provide evide nce that higher hit rates could be ac hieved by employing an objective and systema tic lie detection me thod.
L ie telling and detection often occurs in situatio ns in which interviewer and inte rviewee
interact with each othe r. We believe that, apart from us ing mo re objective lie detectio n
decision mode ls, lie detectio n mig ht be further facilitated when the lie detecto r employs
certain communicatio n techniques. A few such techniques wi ll also be discussed.
Theoretical reasons for cues to deceit
Four decades of deceptio n research in which more tha n I 00 s tudies have been published
have revealed one major finding. The mere fact that someone lies will not affec t his behavior,
voice or speech, a nd therefore typical deceptive responses such as 'Pinocchio's growing
nose' do no t exis t (DePaulo, Lindsay, Malone, Muhlenbruck, Charlton, and Cooper 2003;
Vrij 2000).
However, some responses are more likely to occur during deception than others, because
lia rs may e xpe rience three different processes during deceptio n, called 'emotional', 'content complexity' and 'attempted behavioral control (impression management)' processes,
and each of these processes may influence someone's response (Vrij 2000). 1 In other words,
a liar's behavior, voice o r speech might be affected, not because the pe rson lies, but because, for example, s/he experiences certa in emotions when s/he is lying. Each process e mphasizes a different aspect of deception and deceptive responses. However, the d istinction
be tween them is artificial. Lies may well feature a ll three aspec ts, and the three processes
should no t be conside red as opposing camps.
Three p rocesses
Te lling a lie mig ht evoke three different e motions: Fear, guilt or duping delight (Ekma n
1985/200 I ). A liar mig ht feel guilty because (s)he is lying, might be afraid of getting caught,
or might be excited about having the opportunity to fool someone. The strength of these
emotions depe nds o n the personality of the liar and on the circ umstances under whic h the
lie takes place (Ekman 1985/200 I ; Vrij 2000).
Sometimes liars find it difficult to lie, as they have to think of plausible answers, avoid
contradic ting themselves and tell a lie that is consistent with everything whic h the observer
knows or might find out, whilst avoid making s lips of the to ngue. Moreover, they have to
remember what they have said, so that they can say the same things again when asked to
repeat their sto ry (Burgoon, Bulle r, and Guerrero 1995; Vrij 2000) .
Liars may won·y that several cues will give their lies away, and there fore w ill try to
suppress suc h signs, e ngaging in impression manageme nt in order to avoid getting caug ht
(Bulle r and Burgoon 1996; Burgoo n and Buller 1994; Burgoon, Buller, Floyd, and Grandpre
TSA 15-00014 - 000906
A SYTSTEMATIC MANNER TO DETECT DECEIT
63
1996; Burgoon, Buller, White, Afifi, and Bus lig 1999; Krauss 198 1), that is, trying to make
a convincing impressio n. However, this is not easy. T hey sho uld s uppress their nervousness effectively, sho uld mask evidence that they have to think ha rd, sho uld know how they
no rmally respo nd in o rder to make an honest and convinc ing impression, and should show
the responses they want to show. Tt may well be the case that, when controlling their be havior,
lia rs may ' over-control' their behavior, resulting in exhibiting a pattern of behavior that
will appear pla nned, rehearsed, and lacking in spontane ity. For example, liars may believe
that movements will give their lies away (Hocking and Leathers 1980; Vrij and Semin 1996),
and will therefore move very deliberately and tend to avoid any moveme nts that are not
stric tly essentia l. This will result in an unusual degree of rigidity and inhibition, because
people do normally make movements which are not essential (DePaulo and Kirkendol 1989).
Tt is also like ly that liars will think that the use o f speech hesitatio ns and speech errors
sound dubio us (Hocking and Le athers 1980; V1ij and Semin 1996). The refore they will try
to avoid making suc h no ntluenc ies. This, however, may result in a speech pattern which
sounds unus ually smooth, as it is normal for most people to make some erro rs in speech
(Vrij and Heaven 1999). Impression management mig ht also result in liars being reluctant
to say certain things (for example, they mig ht be reluctant to spontaneously admit that what
they previo usly said was incorrect) as they w ill fear that this will make their stories appear
less convincing.
Another possible c ue that may result from inadequate control of behavior is that performa nces may look tlat due to a lack of involvement (Burgoon and Buller 1994; Burgoon
et al. 1995; DePaulo e t al. 2003, but see Burgoon e t al. 1999; White and Burgoon 2001, for
poss ible c hanges in this pattern over time). An artist who applies fo r a job as salesperson
because he needs the mo ney may not look convinc ingly e nthusiastic during the selection
interview. A mothe r who punishes her child for wrongdoing might not look sincere if s he,
in fact, was a mused by the c hild 's behavior.
All three processes may occur at the sam e time. That is, liars could be ne rvous, having
to think hard, and trying to control themselves a ll simultaneously. Which of these processes is most prevale nt depends o n the type o f lie. Liars will be mo re ne rvous w hen the
stakes (negative consequences of getting caught and positive conseque nces o f succeeding)
are high, hence ne rvous responses a re more likely to occur in high sta ke lies. Liars have to
think harder when the lie is complicate d, therefore indicators of increased cognitive load
are more likely to occur in complicated lies than in easy lies. Liars who are motivated to
avoid getting caught may try harder to make an honest impression tha n those w ho are less
mo tivated. Therefore, attempts to control behavior, voice and speech may especially occur
in motiva te d liars .
Be fo re discussing the o utcomes of reviews about how liars respond, it should be e mphasized that the approaches only suggest that the presence of s igns of emotio ns, content
complexity and impression management may be indicative of deceptio n. None of these
approaches claim that the presence of these signs necessarily indicates deception. Truth
tellers might e xperience exactly the same processes. For example, innocent (truthful) suspects mig ht a lso be anxious if they worry that they will not be believed by a police officer
(Ofshe and Leo 1997). Because of that fear, they may show the same nervous reactions as
guilty liars who are afraid o f being caught (Bo nd and Fahey 1987). T his puts the lie detec-
TSA 15-00014 - 000907
64
VRJJ AND MANN
tor in a diffic ult position. S ho uld the sig ns of fear be inte rpreted as signs of guilt o r of innocence? T he behavior doesn't provide the answer. Ekman ( 1985/200 I ) labe led the fa lse
accusatio n of a truth te ller o n the basis of the behavio rs slhe displays the Othello error,
after S hakespeare's play. Othe llo falsely accused Desdemo na (his wife) of be trayal. He told
her to confess since he was going to kill her for her treachery. Desdemona asks Cassio (her
alleged lover) to be called so that he can testify he r innocence. Othello tells her that he has
already murdered Cassio. Realizing that she cannot prove her innocence, Desdemona reacts
with an emotiona l o utburst. Othello mis interprets this outburst as a s ig n of he r infide lity.
C ues to deceit
Nonverbal and auditory cues
Recent reviews about nonverbal and auditory c ues to deceit, including more than 120 stud
ies, have revealed a limited subset o f nonverbal and a udito ry c ues that are re liably associated with deceptio n (De Paulo e t a l. 2003; Vrij 2000).
L iars tend to have a highe r pitched voice than truth-tellers, which is pro bably caused by
arousal (Ekm an, Friesen, a nd Scherer 1976). The results concerning speech errors (word
and/or sentence repetition, sentence c hange, sente nce incompletions, s lips of the tongue,
and so on) and speech hesitations (use o f speech fi llers such as "ah", " urn", "er" and so on)
show a conflicting pattern. In some studies an increase in such errors (partic ularly word
and phrase repetitions) and hesitations d uring deception have been fou nd, whereas other
studies have revealed the opposite pattern. There is some evidence that variations of lie
complexity are responsible fo r these conflicting findings (Vrij and Heaven 1999). Lies which
are diffic ult to tell result in an increase in speech erro rs and speech hesitatio ns (in line with
the content complexity approach), whereas lies whic h are easy to tell result in a decrease
in speech hesitatio ns and speech erro rs (in line with the attempted control approach) .
Moreover, com ared to truth te lle rs, liars tend to sound vocally less expressive, more
passive, more uncertain and less involved (which might be the result of an overcontrol of
behavio r) and more tense (which mig ht be the result of arousal They a lso come across as
be ing less cooperative and their face: a ear le.u_ Pf.e asant. Both mig_ht be caused b):: a
negative e motion fe lt by the liar.
Finally, liars tend to make fewe r illustrators (hand and arm movements designed to
modify and/or s upple ment what is be ing said verba lly) and fewer hand and finger movements (no n-functional movements of hands and fingers without moving the arms) than truth
tellers. The decrease in these movements mig ht be the result of bo th lie complexity (Ekman
and Friesen 1972) and attempted control.
I n pa rtic ula r, Ekman's ( 1985/200 I ) work has shown tha t lies m ay also result in fraudulent facia l emotional expressions, so-called "micro-expressions". These are facial expressio ns that are displayed for o nly a fraction of a second but clearly reveal the liar's true
feel ings before being q uickly covered with a fal se expressio n.
Perhaps a striking finding of the literature is that liars do no t seem to show c lear patterns of nervous behaviors such as gaze aversion and fidgeting. However, this mig ht be
TSA 15-00014 - 000908
A SYTSTEMATIC MANNER TO DETECT DECEIT
65
the result of a n artefact. Deceptio n researc h has almost exclusively been conducted in
univers ity laboratories whe re people (mostly college students) te ll the truth or lie for the
sake of the experiment. Perhaps in experime nta l laboratory studies the stakes are not hig h
eno ugh for the liar to e licit c lear deceptive c ues to deception (Miller and Stiff 1993).
In the most comprehensive study of real life high sta ke lies to date, Mann (200 I ), examined, amo ngst othe rs, the behavio r displayed by 13 male suspects during their police
interviews (see also Mann, Vrij, a nd Bull 2002; Vrij and Mann 2003). The suspects were
a ll being inte rviewed in connection with serious c rimes s uch as murder, rape and arson.
Truthful and deceptive behaviors were coded and compared (for each suspect both truthful a nd deceptive fragments were available ). T he results revealed that the suspects in these
hig h s take situatio ns also did not show c lear ne rvous behavior w he n they were lying. In
fact, they showed a decrease in limb moveme nts when they lied w hich is mo re in line with
the content complexity and atte mpted control approaches than with the emotio nal approach.
The strongest evidence that conte nt complexity more than ne rvousness has affected suspects' behavior was the finding regarding eyeblinks. S uspects made fewer blinks when they
lied. Research has shown tha t nervousness results in an increase in eyeblinking (Harrigan
and O 'Connell 1996; Tecce 199 2), whereas inc reased cognitive load results in a dec rease
(Wa llbo tt and Scherer 1991 ).
The predominance of cogniti ve load processes compared to e motional processes in
suspect interviews is pe rhaps not surp i; sing. Many suspects have had prior regular contact
with the po lice. In that respect they are pro bably familiar with police interviews which might
decrease their ne rvousness during those interviews. H owever, suspects in police interviews
are often not very brig ht, and are, on average, less intelligent than the average person
(Gudjonsson 1992). There is evide nce that less inte lligent people will pa rtic ularly have
diffic ulty in inventing convincing stoi;es (Ekman and Frank 1993).
Also othe r high-stake lie investigatio ns, suc h as the examination of the behavior of a
convicted murderer during his police interview (Vrij and Mann, 200 I ) and the examinatio n of Sadda m Hussein's behavior dui; ng a CNN television interview during the Gulf War
(Davis and Hadiks 1995) did not show sig ns of ne rvousness. In fact, the studies reveale d
that the convicted murderer a nd Saddam Hussein both showed a decrease in hand and arm
moveme nts when they were lying. In other words, these high stake investigatio ns show
findings s imilar to those of studies with college students.
Verbal cues to deceit: Criteria-based content analysis
For decades, psychologists have looked at verbal criteria whic h might distinguish truths
from lies. Statement Validity Assessment (SVA) is the most popular technique to date for
measuring the veracity of verbal statements (Vrij 2000). The techniq ue was developed in
Germany to determine the c redibility of c hild witnesses' testimonies in trials for sexua l
offe nces. To date, SVA assessments are accepted as evidence in criminal courts in several
countries, such as Germany, the Netherlands, and Sweden (Vrij 2000).
The SVA method was developed through German psycho logists' clinical experience
(rather than via expe rimenta l research) since the 1930s. The first comprehensive descrip-
TSA 15-00014 - 000909
66
VRJJ AND MANN
tion of SVA was published by Undcutsch ( 1967) in German. SVA was further developed
and refined by Steller and Ko hnke n who published an English (and final) versio n in 1989.
One part of SVA is what is known as C riteria-Based Content Analysis (CBCA), the systematic assessment of the credib ility of a verbal statement. In o rder to extract a statement,
children are interviewed following a 'structured interview' procedure. Strict guide lines for
interviewing childre n have been laid down in recent years (Bull 1998). They include the
necessity to build a good rapport w ith the child, and endeavoring for as full a free narrative
regarding the event as possible (particularly important for SVA). Finally, questioning should
begin with very open ended questions becoming increasingly narrowed to obtain more
specific details, bearing in mind that each state ment must stand up in a court of law and
therefore sho uld contain a minimal amount of leading questions (Bull 1998).
These interviews are a udiotaped and then transcribed for CBCA coding. The assessment
takes place on the basis of these written transcripts, with 19 criteria used in the CBCA assessment. Trained evaluators examine the statement and judge the presence o r absence of
each o f the 19 criteria. The presence of each c rite rio n in the statement e nhances the quality
of the statement and strengthens the hypothes is that the account is based on genuine personal experience. This idea is originally stated by Undeutsch ( 1967), and therefore known
as the Undeutsch Hypothesis (Steller 1989).
Vrij (2000) gives a de tailed overview of the 19 criteria used in CBCA assessmen ts, of
which some will be descri bed here. First of all, observers look at the logical structure of
the statement. This criterion is concerned with whether the statement fi ts togethe r. A second criterion is unstructured production. Liars tend to tell their stories more in chronological order (this happened first, and then this, and then that, and so on) , than truth tellers,
who te nd to give their account in unstructured and incoherent ways, particularly when talking about emotional events. A third criterion is the number ofdetails mentioned in a stateme nt. It is hypothesized that liars include fewer detai ls in their accounts than truth te llers
do. The type of d etails CBCA experts are looking fo r include: contextual embedding (does
the statement contain details about times and locatio ns) , reproduction of speech (did the
interviewee recall literally w hat has been said during the event), unusual details (are there
any details mentioned which are 'od d ' but not unrealistic), accounts of subjective mental
state (does the statement include details about how the interv iewee actually felt d uring the
event). 2
The criteria mentioned so far might d iffer between truth tellers and liars because they
are believed to be too diffic ult for people to fab1icate (Kohnke n 1999; Steller 1989). T his
is similar to the cognitive complexity approach described earlier. Other criteria d iffe r between liars and truth tellers for motivational reasons (Steller 1989). This is related to the
attempted con trol approach desc1ibed earlier. Liars will try to construct a report which they
believe will make a credible impression o n others, and will omit information which, in their
v iew, will damage their image of being a s incere person (Ko hnken 1999). Motivation-based
criteria include: spontaneous correcTions (when the person spontaneously admits that the
previous description was incorrect and modifies that description), admittin lack o memoD)
s ontaneous admission of no t re membe rin sol]le. (crucial) details), J!pd raisjr,g__.ifm.tbts
about one:~ own testimony s o ntaneousl admittino that the de~cri tion sounds odd o r
im lausible).
TSA 15-00014 - 000910
A SYTSTEMATIC MANNER TO DETECT D ECEIT
67
In o rder to test whether the CBCA approach actually works a nd can discriminate between truthful and fabricated accounts, both fie ld studies and laboratory studies have been
conducted. In fie ld studies, researchers have evaluated C BCA assessments made in actua l
sexua l abuse cases. The problem w ith field studies is that it is often diffic ult to establish
the exact facts of the case, and to esta blish w ith certa inty w hether the state ment was actually truthful or not. O bvio usly, to know the "gro und truth" (to know whether the stateme nt
was actua lly truth o r false) is essential to test the accuracy of CBCA assessments. In m any
fie ld studies, however, the gwund truth is no t always assured, and the results of field studies are therefore often diffic ult to inte rpre t (Vrij 2000). Laboratory studies are problematic
as well. In real life, CBCA assessments are made sole ly o n state ments given by a lleged child
victims o f sexua l abuse. In other words, this involves state ments describing highly e motio nal e ve nts. Obvio us ly, la boratory studies can never simulate those types of experie nces.
Many CBCA experts therefo re believe that laboratory studies are of little use in testing the
accuracy of SVA assessme nts (Vrij 2000).
So me autho rs descri be CBCA as a technique solely to e va luate statements o f childre n
who a re alleged victims in sexua l abuse cases, as the technique is developed fo r this purpose (Raskin a nd Esplin 199 1). Othe rs have ad vocated the additiona l use o f the techniq ue
to evaluate the testimo nies o f suspects or adult w itnesses who ta lk about issues other than
sexual abuse. They pointed out that the underlying U ndeutsch hypothesis is neither restricted
to c hildre n, witnesses and vic tims no r to sexual abuse (Ko hnken, Schimossek, Asche rma nn,
and Hofe r 1995).
Vrij (2000) reviewed 17 studies re la ted to CBC A, most of them (12) were laboratory
studies, a nd in most of these (9) the statements whic h were assessed were given by adults.
As was the case with the rela tions hip between nonverbal and audito ry cues a nd deception,
there is no typical speech content related deceptive response. That is, not all liars say or
avoid saying specific things. However, the criteria disc ussed above (with the ex e tio n of
admitting lac k o f me mory) a re often found to be resent more in truthful statements than
in dece tive statements (in both adults a nd children), supporting the Undeutsch-H.)I othesis.
Verbal cues to deceit: Reality monitoring
Recently, Reality Monitoring (RM) has been used as an a lte rnative method to measure verbal
differences between respo nses believed to be true and false (Alo nso-Quecuty 1992, 1996;
He rnandez-Ferna ud and Alonso-Quecuty 1997; HOfer, A kehurs t, a nd Metzger 1996;
Manzanero a nd Diges 1996; Roberts, La mb, Zale, and Randall 1998; S po rer 1997; Vrij,
Edward, Roberts, and Bull 2000; Vrij, Ed ward, and Bull 200 I a,c; Vrij, Akehurst, Soukara,
and B ull in press). At the core of Reality Monitoring is the claim that memories of experienced events d iffer in q uality from memories of imagined events. Memories of real experiences a re o btained thro ugh perceptual processes and a re therefore likely to contain,
amo ngst other things, perceptual information (visual details (descriptions of what has been
seen: "I walke d in to the room", contains three visual details), and auditory details (details
abo ut what has been heard: "She said to s it down" conta ins one a udito ry detail)), spatial
TSA 15-00014 - 000911
68
VRJJ AND MANN
details (details about where the event took place ("I walke d in to the room"), deta ils about
how o bjects a nd people were situa ted in relatio nship to each othe r ("he stood behind me")),
and temporal details (de ta ils about time o rde r of the events ("first he switc hed on the videorecorde r and the n the TV"), and details about duration o f events ("we were playing for a
while")).
Accounts of imag ined events are de rived from an inte rnal source and are the refo re likely
to conta in cognitive operations, suc h as tho ughts a nd reasonings ('Tmust have had my coat
o n, as it was very cold that night' ; 'she was quite tall for a girl') (Johnson, Hashtroud i, a nd
Lindsay 1993; Jo hnson and R aye 1981, 1998). One might argue that 'expe rienced events'
re flect truth te lling whereas ' imagined events ' reflect deceptio n. The re fore, d iffere nces
be tween truth te llers a nd liars could be expecte d in Reality Monitoring crite ria. Tn s upport
o f this assumption, studies have revealed that truth te lle rs do include mo re visual , aud ito ry, spatial a nd te mpora l deta ils than liars (see Vrij 2000, for a review). Tn a recent study
Vri" e t al. if!.Press) it was also found that liars. on the o the r hand, a re likely to include more
cognitive o erations in the ir statements than truth tellers.
Detecting deceit via a systematic analysis of nonverbal, auditory a nd speech
content r elated cues
Tn recent years we have conducted two stud ies in w hich we atte mpted to de tect deceit in an
objective and systematic manner. That is, in both s tudies we ca lculated the freque ncy of
occurrence o f various nonverbal, aud itory and speech conten t re lated c ues whic h, according to previous researc h, are " re liable" 3 cues to deception. We then made truthn ie d ecisions
uti lizing multiple cue models w hic h were based upon these reliable cues.
Tn Vrij et al.'s (2000) experim ent, 7 3 nurs ing students e ithe r to ld the truth o r lied about
a film they had just seen. Co mpared to truth te llers, lia rs showed fewer illustrators, fewer
hand and finger movements, more speech hesitations, and a lo nger respo nse latency (period o f silence between question and a nswer). Liars a lso had a lower total CBCA score
(CBCA crite ria combined ) and lower Rea lity M o nito ring score (R M c rite ria combined
except cognitive operatio ns).
Tn Vrij et al. 's (in press) study, 52 undergradua te students participated in 'playing a game
of Connect 4 and ru bbing the blackboard' event. Tn a subseque nt inte rview they told the
truth or lied about the event. This study revealed fewer cues to deceit, but m ost of the cues
that were found were the same c ues fo und in Vrij e t a l. (2000). Compare d to truth te llers,
liars showed fewer hand and finger movements, and obtained lower CBCA- and Reality
Monitoring scores. Fina lly, liars included more cognitive operations in the ir sta tements than
truth te llers.
Tn order to examine to what extent liars and truth te llers could be correctly c lassified o n
the basis of such c ues, discriminant a na lyses were conducted in both studies. Ta ble I provides a summary of the o utcomes o f these ana lyses.
Tn these ana lyses, the objective truth status was the classifying variable, and, in Vrij et
al. 's (2000) e xperime nt, total CB CA score, tota l Reality Monitoring score, response late ncy,
speech erro rs, speech hesitatio ns, speech rate, illustra to rs and hand and finger moveme nts
TSA 15-00014 - 000912
69
A SYTSTEMATIC MANNER TO DETECT DECEIT
Table I . Discriminant analyses with nonverbal behavior, criteria-based content analysis and reality monito ring
Hit rates
De tection technique
Yrij et al. (2000)
CBCA + RM + nonverbaUauditory cues
Lie
85%
Truth
77%
Total
8 1%
Eigenvalue
0.77
Lambda
0.57
df
6
X2
38.79**
Yrij et al. (in press)
CBCA + RM ' + nonverbaUauditory cues
89%
88%
88%
1.09
0.48
7
34.29**
Replication o f Yrij et al. (2000)
with Yrij et al. (in press) data
CBCA + RM + nonverbaUauditory cues
81%
77%
79%
0.64
0.61
4
23.83**
'In Vrij et al. (in press) RM represents two variables: the total Reality Mo nitoring score and the cognitive operations variable *p < 0.05, np < 0.01.
were the depe ndent variables. D ifferen t discriminant analyses were conducted, for nonverbal/auditory responses and speec h content related responses separately and for the nonverbal/auditory and speech content related responses combined. The combined analyses
revealed the most accurate classification of (jars and truth teiJers w ith a total hit rate of 8 1%
(85% Iie detection hit rate a nd 77% truth detection hit rate). CBCA score, response latency,
hand and fi nger movements, speech hesitations, illustrators and speech rate contr ibuted to
the discriminant function.
I n Vrij et al. 's (in press) experiment, total CBCA score, total R eality M onitoring score,
cognitive operations, gaze aversion, limb movements (illustrators, self manipulations, hand
and finger movements and foot and leg movements combined), response latency, pauses,
speech disturbances (speech errors and speech hesitations combined), and speech rate were
included as depe ndent variables. Again, the combined analysis revealed most accurate hit
rates, with a total hit rate of 88% (88% lie detection and 89% truth detection). Reality
Monitoring total score, frequency of pauses, cognitive operations, late ncy period, limb
movements, CBCA total score and gaze aversion contributed to the discrimi nant function.
Although there was some overlap between the two discriminant functions (total CBCA
score and response latency emerged in both functions), they were not ide ntical. DePaulo,
Anderson, and Cooper (1999) correc tly pointed out that this lack of replication is problematic, as it does not provide a set of variables on whi ch someone can reliably build upon
in lie detection. However, it is not possible to compare both discriminant functions, as in
both discriminant analyses d ifferent dependent variables were ente red.
In order to seek replication of Vrij et al . 's (2000) fmdings, an additional discriminant
analysis was carried out with the data reported in Vrij et al. (in press). This time the dependent variables which were included in Vrij et al. 's (2000) study (total CBCA score, total
Reality Monitoring score, la tency period, speech errors, speech hesitations, speech rate,
illustrators a nd hand/finger movements) were included. The hit rates obtained with the Vrij
et at. (in press) data were very similar to the fi ndings obtained w ith the Vrij et al. (2000)
data: 8 1% hit rate for lie detection, 77% hit rate for truth detection and 79% total hit rate.
In the Vrij e t al. (in press) data fou r variables contributed to the discriminant function: Ia-
TSA 15-00014 - 000913
70
VRJJ AND MANN
tcncy period, total Reality Mo nito ring score, hand and finger movements and speech erro rs . In o the r words, there was an overla p of two c ues between the Vrij e t a l. (2000) and
Vrij et al. (in press) data sets (late ncy perio d and hand a nd finger movements). In both
analyses deceptio n was associated w ith an inc rease in la te ncy perio d and a decrease in hand
and fi nger mo ve me nts.
Guidelines to detect deception
Cues to examine
In our view, the findings presented in this article illus trate the benefit of looking at certain
cues in a systematic manner in orde r to detect deceit. Which nonverbal, auditory and speech
conte nt related cues lie de tecto rs should partic ularly focus upo n is hard to say. Typical
deceptive response s do not exist and responses mig ht differ per situation and per individual.
However, there is g rowing evidence that CBCA scores, Reality Monitoring scores and some
no nverba l cues, pa rtic ula rly illustrato rs and ha nd and finger moveme nts are use ful to loo k
at. It sounds reasonable to suggest that the more these cues occur simultaneously in a person's respo nse, the more like ly it is that the person is lying. Our own study (Vrij et a l., 2000)
showed tha t lie de tectio n with each of the cues individually d id not result in high hit rates.
In o ther words, it is essential to work w ith multiple c ue models.•
Using Reality Monitoring assessme nts to disting uis h between liars a nd truth tellers is
relatively new but in o ur view promising. H it rates of correct classific ations of truth te llers
and liars are around 70% (Sporer 1997; V1ij e t al. 2000, in press) whic h are comparable
with CBCA hit rates, whic h are typically a bout 70% as we ll (see Vrij 2000, for a review).
The benefit of Reality Monitoring assessments, compared to C BCA assessments, is that it
is easie r to teach coders the Reality Monitoring me thod. There is an overlap between the
two verbal methods, as both C BCA and Reality Monitoring coding occurs o n the basis of
written transcripts. Ho wever, applying the Reality M onito ring method is less time -consuming as the re are fewer criteria to examine. Moreove r, the individual c riteria are more clearly
defined and, as a result, interrate r re liabilities between coders are usually higher for Reality M o nito ring assessments than for CBCA assessments (Sporer 1997; Vrij et a l. 2000, in
press).
Baseline method
The fact that people don' t have standard reactio ns when they lie, pro bably makes a baseline comparison method useful. Baseline comparison refers to comparing the respo nse under
investigatio n with someone's natural truthful respo nse. A deviation from that response might
indicate that a person is lying . However, a lie detector sho uld keep in mind that a deviatio n
does no t necessarily indicate d eception (see for example the Othello-error). Research has
cons istently indicated that people become bette r at de tecting truths and lies w he n they are
familiar with the truthful behavio r of the pe rson they ha ve to j udge (Brandt, Miller, and
TSA 15-00014 - 000914
A SYTSTEMATIC MANNER TO DETECT DECEIT
71
Hocking 1980a,b, 1982; Feeley, dcTurc k, a nd Yo ung 1995; Seager 2001 ). For example,
Feeley and his colleagues exposed some observe rs to baseline inte rviews of the people they
had to assess in the lie de tectio n task. Their hit rates were significantly hig her (72%) than
the hit rates of observers (56%) who were not e xposed to baseline inte rviews prior to the
lie de tec tion task.
However, the baseline me thod will o nly work if the me thod is applied correctly. Crucial
in the use of the baseline technique is that the correct parts of the interview are compared.
Suppose a colleague with whom you are o n friendly te rms initiates a casual c hat. After a
while he te lls you that people at work have come to the conclusio n that yo u are responsible
for the breakdown of expensive equipme nt yesterday, whic h made your boss extreme ly
angry ! Yo u didn' t to uc h the mac hine, so you know that the accusatio n is false, and that is
what you immediate ly say to your colleague. However, the false accusation clearly upsets
you a nd makes you react nervously. What ma kes the m suspect you? Your colleague notices your nervous reactions and subsequently accuses you of lying. Your colleague makes
a serious (but common) mistake. Indeed, you a re nervous, but it is the accusation itself that
makes you nervous ! The situation you are in at the moment is totally different fro m the
situation before the accusatio n. Comparing yo ur c urrent reactions with your reactio ns before the accusation is no t legitimate. The situations before and after the accusation are not
comparable, and c hanges in your behavior caused by the accusatio n can say nothing about
whether o r not you a re lying. In o ther words, o ne sho uld not compare apples with oranges,
but sho uld compare the respo nse under investigation with a comparable truth, that is, truthful
response exhibited in a similar situa tio n. U nfortunately, in police interviews comparisons
between someone's response d uring s ma ll ta lk and their reaction during an actual interrogation (apple - o range comparison) is common practice (Mosto n and Engelbe rg 1993). It
is probably unfa ir to blame police officers for making this m ista ke. T hey are advised to
establish a baseline in this way in popular and frequently read police manuals, such as the
American manual written by Inbau, Re id, and Buc kley ( 1986). (A new edition of this manual,
Inbau, Re id, Buckley, and Jayne (2001 ), has recently been p ublished). See Vrij and Mann
(2001 ) how during inte rviews fai r truth - lie comparisons can be made.
Information-gathering style
Lie detection is probably easie r w he n lie detec to rs e mploy an info rmation-gathering communication style (where they encourage people to ta lk or write) compared to an accusatory communication style (whe re they confront people with circumstantial evidence or are
simply accused of lying). Cognitive difficulties liars might face, as described in the content complexity approach, partic ularly occur when liars are encouraged to ta lk o r write.
Throug ho ut the ir lies, liars have to think of p laus ible answe rs, sho uld not contradict the mselves, should report a lie that is consistent with everything whic h the observer knows or
might find out, sho uld avoid ma king s lips of the tongue, and should try to make an honest
(convincing) impression. Moreover, they have to remember w hat they have previo usly said
or written, so that their story re mains consistent when someone asks them to repeat it. This
is by no means an easy task and the re fore it sounds plausible that the mo re a liar ta lks or
TSA 15-00014 - 000915
72
VRJJ AND MANN
writes, the more likely it is that he o r she (fina lly) will g ive his or her lies away via speech
content or no nverbal/auditory c ues (as they continuous ly have to m o nitor both speech
content a nd nonve rbal/auditory behavior).
Inbau and his colle ag ues (Inbau et a l. 1986/2001) advise (criminal) investigators to confront suspects with pieces of evidence they have already gathered. They hereby try to show
suspects that it is meaningless to remain silent and that it might be bette r for them to talk.
This interview style will hamper lie detection. One of the difficulties for liars is that they
do not know what the observer knows. They therefore do no t know what they can say
without running the tisk of being caught out. By disclosing to suspec ts the facts they kno w,
criminal investigators reduce the uncertainty of lying fo r suspects and make it easier for
them to lie.
Also accusing people of lying by referring to their behavior: " You are lying, I can see
that in your e yes!" (Inbau et al. 1986, 200 I) is unfortunate, as it gives people the ideal opportunity to "escape" from the interview s ituation. They might te ll interviewers that they
no longer w ish to continue co operating with the investigatio n, as further interviewing would
no t make sense given the fact that they (the interviewers) do not believe them (the intervie wee) anyway. Also, it sho uld be kept in mind that accusing someone in itself mig ht e lic it
response changes (Othello-etTOr).
The first contact
Lie detection is perhaps easier during the first contact than during subsequent interactio ns.
During the first contact, interviewees may not yet be full y prepared and may no t have worked
out their respo nse strategy. For example, members of the resistance during the Second World
War who were caught by the Gem1ans and were subsequently interviewed by German police
were often poo rly prepared for those inte rviews (J. Vrij 2001 ). All their precautio n activities were focused on a voiding getting caught by the German police, and the equally important aspect of what to say if they were captured was ig nored.
In additio n, after the initial phase of an interaction, interview d ynamics are probably
going to play an impo rta nt ro le which makes interpreting the responses of the a lleged liar
more complex. For example, the inte raction may create suspic io n in the inte rviewer and
this suspicion, in turn, may affect the interviewee's reaction (Bond and Fa hey 1987; Bulle r,
Stiff, and Burgoon 1996; Burgoon, Buller, Dillman, and Walthe r 1995; Burgoon et a l. 1996).
Also, it is known tha t whe n people communicate with each o ther, matching a nd sync hrony
takes place (Akehurst and Vrij 1999; Burgoon e t al. 1995, 1996, 1999; Chartrand and Bargh
1999). People may mirror each other's posture or they may converge in how quickly and
how loudly they speak. They may also reciprocate each o ther' s gazing, nodding a nd smiling be havio r, and even accents can converge (DePaulo and Friedman 1998). This e ffect,
recently labelled " the c hameleon effect" (Chartrand a nd Bargh 1999) e me rges even whe n
total strangers interact with each other and it happens w ithin a few minutes (Akehurst and
Vrij 1999; C hartrand and Bargh 1999). Akehurst and Vrij ( 1999) found that w hen in a simulated police inte rview the police detective made g rooming gestures, the " suspects" mimicked these mannerisms within one minute . As a result, as the communication progresses
TSA 15-00014 - 000916
A SYTSTEMATIC MANNER TO DETECT DECEIT
73
the responses of the interviewee mig ht become influenced by the responses of the interviewer.
Implicit lie detection
W hic h behavioral and auditory c ues do people think are associated with deceptio n? In
general terms, people probably find those behavio rs susp ic io us that deviate from a no rma l
or expected pattern (Bas ket a nd Freedle 1974; Bo nd, Omar, Pitre, Lashley, S kaggs, and
Kirk 1992; Burgoon, Buller, Ebesu, White, a nd Rockwell 1996). However, surveys have
indicated that people, both laypersons a nd police o ffi cers, particularly associate gaze aversion a nd fidgeting with deception (Akehurst, Kohnken, Vrij, and Bull 1996; Mann 200 I;
Vrij and Semin 1996). For example, around 75% of police officers be lieve that liars look
away (Mann 200 1; Vrij a nd Semin 1996). One explana tion of why po lice officers e ndo rse
this view is that police manuals typically promote the idea that liars look away and fidget
(Gordon and Fleisher 2002; Hess 1997; Inbau et al. 1986, 2001). As already discussed above,
there is no empirical evide nce to suggest that liars actually exhibit such behavio rs. Unfortunately, the manuals which make suc h claims also do not present research data to support
these suggestio ns.
S uch inaccurate views about c ues to deceptio n hampers lie detection. In two recent studies
where we e xamined the re lationship between c ues which people think are associated with
deception and their ability to detect deceit, we found that police officers who e ndorse the
views that liars fidget and look away a re the worst lie detectors (Mann, Vrij, and Bull in
press; Vrij and Mann 200 I). T his may well expla in why, contrary to what seems reasonable, people who have access to the full picture of a liar often have more diffic ulty in
de tecting deceit (De Pa ulo e t al. 1985; Wiseman 1995). When watc hing the full pic ture,
powerful but misleading c ues suc h as gaze a nd fi dgeting are available, whereas such cues
are unava ilable when reading a text or listening to a voice.
An innovative me thod of letting people focus o n mo re valid c ues to deception is the
implicit lie detection method. In Vrij, Edward, and Bull's (200 1b) s tudy, police officers
watched videotapes o f truth te llers and liars. Some participants were asked whethe r each
of the people were lying (explic it lie detection), others were asked to indicate for each person
whether that person "had to think hard" (implicit lie detection, they were not info rmed that
some people were actua lly lying). Police officers could distinguish between truths and lies,
but o nly by using the implicit method. An explana tion for this f inding is that, when detecting deceit explic itly, po lice officers re ly o n their stereotypical views about deceptive
behavior, suc h as " liars look away" and " lia rs make many movements". As me ntio ned previously, these stereotypical views are mostly incorrect. While de tecting deceit in an implicit way, however, the participants were implic itly directed to look at more valid cues.
Indeed, the findings showed that only in the indirect method did the police officers pay
attention to the c ues which actua lly discriminated between truth te llers a nd liars on the
videotape, such as a decrease in hand and finger moveme nts. See De Paulo ( 1994) a nd A.
Vrij (200 1) for overviews of implic it lie detection studies.
TSA 15-00014 - 000917
74
VRJJ AND MANN
Discu ssion
Major conclusions
In this a rticle we have argued that lie detection improves when a variety of verba l and
no nverbal/auditory cues are systematically examined. We have a lso argued tha t, apart from
looking at c ues in a mo re o bjective and syste matic manner, lie detectio n might imp rove if
certain communicatio n techniques a re e mployed.
Implications for automatic deception detection
For the purpose o f auto matic detection of deceptio n, computer software programs could
be developed to calculate the freque ncy of occurrence of CBCA and Reality Monitoring
criteria in electronic communic atio n. To o ur knowledge no publications are available re
ga rd ing this issue but we believe that developing s uch programs is possible. At the very
least, it should be possible to make good estimates of the frequency of occurrence of certain criteria via computer software programs. For example, as mentioned in footnote 2, Lamb
et al. ( 1997) fo und a very high (0.86) corre latio n between the number of details present in
a state me nt (one of the CBCA c rite ria) and leng th of the state ment (calculated via number
of words). In o the r words, calculating the number of words (already available with computer software progra ms) could provide a good estimate o f the number of deta ils present
in a state me nt.
Limitations and advantages
As arg ued in this article, we be lie ve tha t looking at a combination of behaviora l, auditory
and speech conte nt relate d c ues in a syste matic manner has c lear pote ntial for lie detectio n, a nd this po tential would be fu lly exploited if a software program, c apable o f automatically identifying some o f the valid c ues (see previous sectio n) could be developed.
We base this positive message regarding po tential lie detection on deception research.
However, a lmost a ll deception research to date has some important lim itatio ns. Fo r example, in deceptio n research, lying typically takes place in a dyad setting (the liar communicates w ith j ust o ne person) and not in a group setting (in which a liar communicates with
a group o f people). We do n' t be lieve this is a serio us constra int. The theoretical reasons
why c ues to deception mig ht occur are the same when people lie in a dyad or in a group.
A lso, the communication techniques we have discussed a re not restricted to a dyad setting
e ither. We therefore believe that the info rmatio n provided in this article are applicable to
group settings as well. The potential to genera lise the find ings presented in this article m ight
be limited by the consideratio n that in deceptio n stud ies, liars and truth tellers typically give
verbal sta te me nts and not written state ments. Even in assessing verbal c ues to deception
with methods suc h as C riteria-Based Conte nt Analysis and Reality Monitoring, researchers o btain verbal statements that are then transcribed. We simply do not know whe ther the
TSA 15-00014 - 000918
A SYTSTEMATIC MANNER TO DETECT DECEIT
75
verbal c ues w hich appear in such transcripts will appear in written stateme nts as well, and
we are reluctant to speculate about this. Instea d, we recomme nd further research into this
issue.
Concluding remarks
Although typical deceptive responses such as "Pinocchio's growing nose" do not exist, some
cues are, to some extent, re lated to deception. We have argued that looking at such c ues in
a systematic ma nner has clear potential for lie detection. We also believe that software
programs could be developed, to auto matically identify at least some of these c ues, paving
the way for automatic lie detection.
Acknow ledg ements
The studies presented in this article were sponsored by grants from the Economic and Social
Researc h Council (R000222820) and the Leverhulme Trust (F/678/E). Correspondence
concerning this article s hould be addressed to: Aldert Vrij, U niversity of Portsm outh, Psychology Departme nt, King Henry Building, King Henry I Street, Portsmouth PO I 2DY,
United Kingdo m or via e mail: a ldert.vrij @port.ac.uk.
Notes
I.
2.
3.
4.
Zuckerman, DePaulo, and Rosenthal (198 1). who introduced these three factors, also included a fourth factor
in their theoretical model. labeled "arousal'' . We le ft this factor o ut as, in o ur view, it shows an overlap
with the emotion factor. Zuckem1an eta!.. ( 1981, p. 9). themselves already. suggested this by finishing their
arousal factor paragraph with the following statement: "II is possible, however, that the general autonomic
responsivity to deception reflects specific emotions. lf so, cues to deception may be accounted for by lhe
particular affects that are involved rather than by general arousaL"
The fact that truthful statements contain more details than deceptive statements s uggests that truthful statements are likely to be longer than deceptive statements. Indeed, Lamb. Sternberg. Esplin, Hershkowitz.
and Orbach ( 1997) found a very strong correlation (r = 0.86) between number of details included in the
statement and le ngth of the statement (measured via the number of words), and DePaulo et al. 's (2003)
review s howed that truthful messages were s ignificantly longer than deceptive messages.
The tetm "reliable cues to deception" always needs to be used with caution. because, as was discussed before.
truly reliable c ues to deceptio n (Pinocchio's growing nose) do not exis t.
C BCA scores and Reality Monitoring scores themselves are already multiple cue models.
References
Akehurst, L. and A. Vrij. ( 1999). "Creating Suspects in Police Interviews," Jounral ofApplied Social Psychology 29, 192- 210.
Akehurst, L., G Kohnken, A. Vrij, and R. Bull. (1996). "Lay Perso ns' and Police Officers' Beliefs Regarding
Deceptive Behaviour," Applied Cognirive Psychology 10.461-471.
TSA 15-00014 - 000919
76
VRJJ AND MANN
Alonso-Quecuty, M . L . ( 1992) . "Deception D etection and Reality Monitoring: A New Answer to a n O ld Question?"" in F. Losel, D . Bende r, and T. Bliesener (eds.), Psycho logy and Law: International Perspec tives. Ber-
lin, Germa ny: Wa lter de Gruyter, 328-332.
Alonso-Quecuty, M . L. ( 1996). "Detecting Fact from Fallacy in Child and Adult Witness Accounts," in G. Davies,
S. Lloyd-Bostock, M . McMurran, and C. Wilson (eds.), Psychology, Law, mul Criminlll Justice: l ntemlllional
Developments in Research and Practice. Berlin, Ge rmany: Walter de Gruyter, 74-80.
Alonso-Quecuty, M. L., E. Hem a ndez-Ferna ud, and L. Campos. ( 1997). "Child Witnesses: Lying About Something Heard," inS. Redondo, V. Garrido, J. Perez, and R. Barbaret (eds.). Advances in Psyclrologyand Law.
Berlin, Ge rmany: Walte r de Gruyter, 129-135.
Basken, G. D. and R. 0. Freedle.. ( I974) .. "Aspects of Language Pragmatics and the Social Pe rception of Lying,"
l ou mal of Psyclwlinguistic R esearch 3, l I 7- 131.
Bond, C. F. and W. E. Fahey. (1987). ''False Suspicion a nd tbe Misperception of Deceit." British Journal ofSocial
Psychology 26, 4 1-46.
Bo nd, C. F., A. Omar, U. Pitre, B. R. Lashley, L. M. Skaggs. and C. T. Kirk . ( 1992). "F ish y-Looking L iars:
Deception Judgmen t from Expectancy Violatio n,'' l o u mal of Persona lity tmd Social Psychology 63, 969977.
Brandt. D. R .• G R. M iller, and J. E. Hocking . ( I980a). "The.Truth-Deception Attribution: Effects o f Fa miliarity
o n the Abi lity of Observers to Dctc(;l D eception ." /Iuman Communication Re search 6 , 99-1 10.
Brandt, D. R .• G. R. Miller, a nd J . E. Hocking. (1980b). "Effect~ o f Self-Monitoring and Fami lia rity on Deception Detection:· Comnumication Quarterly 28,3- 10.
Bull, R. ( 1998). "Ob taining Information From Child Witnesses," in A . Memon, A . Vrij , and R. Bull (eds.), Psychology and l.Lnv: Tmthfulness, Accuracy and Credibility. Maidenhead, England:. McGraw-Hill. 188-210.
Buller, D. B . and J . K. Burgoon. ( 1996). "Interpersonal Deception Theory," Communication Theory 6, 203- 242.
Bulle r. D. B .. J . B . Stiff. and J. K. Burgoon. (1996). ·'Behavioral Adaptation in Deceptive Transactions: Fact or
Fictio n: Reply to Levine and McComack.'~ Human Communication Research 22 . 589- 603.
Burgoon, J. K . a nd D . B . Bulle r. ( 1994). " Interpersonal Deception: Ill. Effects of Deceit o n Perceived Communication and Nonverbal Beha vio r Dy namics," lou mal of Non verbal Behavior 18, 155- 184.
Burgoon, J . K., D .. B. Buller. and L. K. Guerrero. (1995). " lnterper.;ona l Deception IX:. Effects of Socia l S kill
and Non verbal Commu nication on Deception Success a nd De tection Accuracy," Joumal of l.Lmguage and
Social Psychology 14, 289- 311.
Burgoon, J. K. , D . B . Buller, L. Dillman, a nd J. B. Wa lther. ( 1995). "Interpersonal Deception IV: Effects of
Suspicion on Perceived Communication and Nonverbal Behavior D ynamics." Human Communication Reset~rch 2 2, 163-196.
Burgoon, J .. K .• D. B. Buller, A. S. Ebesu, C . H . White, a nd P. A . Rockwell..( I996) .. "Testing Interpersonal Deceplion T heory: E ffects of Suspicionon Conulmnication Behaviors and Perception,·· Communication Theory
6, 243- 267.
Burgoon, J. K., D . B. B uller, K. Floyd, and J. Grandpre. ( 1996). "Deceptive Realities: Sender, Receiver, and
Observer Perspectives in Deceptive Conversations," Commuflication Research 23,724---748.
Burgoon, J. K., D. B. Buller, C. H. White, W. Afifi, and A. L. S. Buslig. ( 1999). "T he Role of Conversation Involvement in Deceptive Interper.;onal Interactions," Personality and Social Psychology B ulletin 25, 669-685.
Chartrand. T. L. and J. A . Bargh. ( 1999). "T he Chameleon Effect: The Perception-Behavior Link and Social
Interaction," Joumal of Personality and Social Psychology 76, 893- 910.
Davis, M. and D. Hadiks. ( 1995). "Demeanor and Credibility,'' Semiotica I 06. 5- 54.
De Paulo, B. M . ( 1994). ·'Spotting L ies: Can Humans Learn to do Better~·· Current Directions in Psychological
Science 3, 83- 86.
De Paulo. B. M. a nd S. E. Kirkendol. ( 1989). "The Motivational Impairment Effect in the Communication of
Deception," in J. C . Yuille (ed.), Credibility Assessme/11. Dordrecht, the Netherlands: Kluwer, 5 1-70.
DePaulo, B . M. and H . S . Friedman. (1998). ' 'Nonverbal Communicatio n." in D. T. Gilbert. S . T. Fis ke, and G.
Lindzey (eds.), The Handbook of Social Psychology. Boston, MA: McG ra w-Hill, 3-40.
De Paulo , B . M., D. E. Anderson, a nd H. Cooper. ( 1999, October). Explic it and Implicit Deception Detection.
Paper presented at the Society of Experimental Social Psychologists, St. Lo uis.
De Paulo , B. M., J . J. Lindsay, B. E. Malone, L. Muhlenbruc k. K. Charlton, and H. Cooper. (2003). "Cue& to
Deception," Psyc/wlogical Bulletin 129, 74- 118.
TSA 15-00014 - 000920
ASYTSTEMATIC MANNER TO DETECT DECEIT
77
De Paulo. B. M .~ J. L. Stone, and G. D . Lassiter. (1985). "Deceiving and Detec ting Deceit;· in B. R. Schenkler
(ed.), The Self and Social Ufe, New York, NY: McGraw-Hill, 323-370.
Ekman, P. ( 19851200 I). Telling U es. New York: W. W. Norton.
Ekman, P. and W. V. Friesen. ( 1972). "Hand Movements," J ournal of Communication 22, 353- 374.
Ekman, P. and M. G Frank. ( 1993). "Lies that Fail," in M. Lewis and C. Saami (eds.). Lying and Deception in
Everyday Life. New York, NY: Guildford Press, 184-201.
Ekman, P. a nd W. V. Friesen. ( 1974). " Detecting Deceptio n from the Body or Face," Journal of Personality and
Social Psychology 29 , 288- 298.
Ekman, P., W. V. Friesen, and K . R. Scherer. ( 1976). " Body Movement and Voice Pitc h in Deceptive Interaction,". Semiotica 16, 23-27.
Feeley, T. H., M ..A.deTurck, and M. J . Young. ( 1995). "Baseline Familiarity in Lie Detection," Communication
Research Reports 12. 160-169.
Go rdon, N. J. and W. L. Fleisher. (2002). Effective Interviewing and lmerrogatian Techniques. San Diego, CA:
Academic Press.
Gudjonsson, G H. ( 1992). The Psychology oflmerrogations, Confessions and Testimony. Chichester, England:
Wiley & Sons.
Harrigan, J. A. and D. M. O'Connell. ( 1996). "Facial Moveme nts During Anxie ty. States," Personality and Individual Differmces 21,205-212.
Hess, J. E. (1997)./nterviewing and flllerrogationfor Law Enforcement. Reading. United Kingdom: Anderson
Publishing Co.
Hocking, J. E. and D . G Leathers. ( 1980). "Nonverbal Indicators of Deception: A New Theoretical Perspective,''
Communication Monographs 41, 119- 131.
HOfer, E., L. Akehurst, and G Metzger.. ( 1996, August). Reality Monitorilzg: A Chance for Further Development ofCBCA? Paper presented at the Annual meeting of the European Association on Psychology and Law
in Sienna, Italy.
lnbau, F. E., J. E. Reid, and J. P. Buckley. (1986). Criminallmerrogation and Confessions, 3rd Ed. Baltimore,
Md.: Williams & Wilkins.
Inbau, F. E .• J . E. Reid, J.P. Buckley, and B. C. Jayne. (200 1). Crimina/ Interrogation and Confessions, 4th Ed.
Gaithersburg, Maryland: Aspen Publishers.
Johnson, M. K. and C. L. Raye. (1981). ''Reality Monitoring,'' Psychological Review, 88.67- 85.
Johnso n, M. K. and C. L. Raye, C. L. ( 1998). '·False Memories and Co nfabulation," Trends in Cognitive ScieJKes 2, 137- 145.
Johnso n, M. K ., S. Hashtroudi. and D . S . Lindsay. ( 1993). "Source Monitoring," Psyc/wlogica/.Bul/etin 114,329.
Kohnken, G. ( !999, July). Statement Validity Assessment. Paper presented at the. pre-conference programme of
applied courses· Assessing cred.i bility' organised by the European Association of Psychology and Law, Dublin, Ireland.
Kohnken, G., E. Schimossek, E. Aschermann, and E. Hofer. ( 1995). "The Cognitive Interview and the Assessment of the Credibility of Adult's Statements," J ournal of Applied Psychology 80, 671-684.
Krauss, R. M. ( 1981). "Impression Formation, Impression Management, and Nonverbal Behaviors,'~ in E. T.
Higgins, C. P. Herman, and M . P. Zanna (eds.), Social Cognition: The Ontario Symposium, vol I. Hillsdale,
NJ: Erlbaum, 323-341.
Kraut, R. E. ( 1980). "Humans as Lie Detectors: Some Second Thoughts,'' Journal of Communication 30, 209216.
Lamb, M. E., K. J. Sternberg, P. W. Esplin, I. Hershkowi tz, andY. Orbach. ( 1997). "Assessing the Credibility of
Children's Allegations of Sexual Abuse: A Survey of Recent Research," Leaming and Individual Differences
9, 175- 194.
Mann, S. (2001). Suspects, Lies and Videotape: An Investigation into Telling and Detecting Lies in Police/Suspe<·tlnterviews. Unpublis hed PhD thes is. University of Po rtsmouth (United Kingdom), Psychology Department.
Mann, S., A. Vrij, and R. Bull. (in press). "Detecting True Lie: Police Officers' Ability to Detect Suspects' Lies,''
Journal ofApplied Psychology.
TSA 15-00014 - 000921
78
VRJJ AND MANN
Mann, S .. A . Vrij, and R. Bull. (2002). "Suspect, Lies and Videotapes: An Analysis o f Authentic High-Stake Liars,•·
Law and Human Behavior 26, 365-376.
Manzanero, A. L. and M. Diges. ( 1996). "Effects of Preparation on lntcmal and External Memories," in G. Davies,
S. Lloyd-Bostock, M . McMurran, and C. Wilson (eds.), Psychology. Law, mul Criminlll Justice: lntemmional
Developments in Research tmtl Practice. Berlin, Germany: Walter de Gruyter, 56-63.
Miller, G. R. and J. B. Stiff. (1993). Deceptive Communication. Newbury Park, CA: Sage.
Moston, S . J. and T. Engelberg. (1993). " Police Questioning Techniques in Tape Recorded Interviews with Crimina l Suspects," Po licing and Society 3, 223-237.
Moston, S. J., G. M. Stephenson, and T. M. Williamson. ( 1992). "The Effects of Case Characteristics on Suspect
Behaviour D uring Po lice. Questioning," British Joumal of Criminology 32,23-39.
Ofshe, R. J. and R. A . Leo. ( 1997). ''The Decision to Confess Falsely:. Rational Choice and Irrational Ac tion,"
Denver University Law Review 74,979-11 12.
Raskin, D. C. and P. W. Esplin. (1991). "Statement Validity Assessment: Interview Procedures and Conte nt
Analysis o f Children' s Statements of Sexual Abuse," Behavioral Assessment 13, 265-29 1.
Roberts, K. P., M. E. Lamb, J. L. Zale, and D. W. Randall. (1998). Qualitative Diffe rences in Children s Accoums
of Confirmed and Ultconfirmed lncidems of Sexual Abuse. Paper presented at the biennial meeting of the
American Psychology-Law Society, Redondo Beach, March 5- 7.
Seager. P. (2001). Improving the Ability of Individuals to Detect Ues. Unpublished PhD-thesis . University of
Hertfordshire (United Kingdom), Psycho logy Department.
Sporer, S. L ( 1997) . ''The Less Travelled Road to T ruth: Verbal Cues in Deception Detectio n in Accounts of
Fabricated and Self-Expe rienced Events," Applied Cognitive Psychology II , 373-397.
Steller, M. ( 1989). "Recent Developments in Statement Analysis," in J .. C. Yuille (ed.), Credibility Assessme11t.
Deventer, the. Netherlands: Kluwer, 13 5- 154.
Steller, M. and G. Kohnken. (1989). "Criteria-Based Content Analysis." in D. C. Raskin (ed.). Psychological
Methods in Crimina/ Investigation and Evidence. New York. NY: Springer-Verlag, 217- 245.
Tecce, J. J. ( 1992). "Psychology. Physio logy and Experimental," in McGraw-Hill Yearbook ofScience and Technology. New York: McGraw-Hill, 375- 377.
Undeutsch, U. ( 1967). " Beurteilung der. G laubhaftigkeit vo11 Aussagen," in U. Undeutsch (ed.), Handbttch der
Psychologie Vol. II: Forensische Psychologie. Gouingen, Germany: Hog refe, 26-181.
Vrij, A. (2000). Detecting Lies and Deceit: The Psychology of Lying and the Implications for Professional Practice. Chichester: Wiley and Sons.
Vrij, A. (2001). " Implicit Lie Detec tion,'' The Psychologist 14, 58-{iO.
Vrij, A. and G. R. Semin. ( 1996). "Lie Experts Beliefs About Nonverbal Indicators of Deception,'' Jo umal. of
Nonverbal Behavio r 20, 65- 80.
Vrij,A. and S. Heaven. ( 1999) .. "Vocal and Verbal Indicators of Dece ption as a Function of Lie Co mplexity, ..
Psychology, Crime. & Law 4. 401--413.
Vrij, A. and S. Mann. (200 1). 'Telling and Detecting Lies in a High-Stake Situation: The Case of a Convicted
Murdere r,'' Applied Cognitive Psychology 15, 187-203.
Vrij, A. and S. Mann. (2003). " Deceptive Respo nses and Detecting Deceit," in P. W. Halligan, C. Bass, and D.
Oakley (cds.), Malingering and 1/lness Deception: Clinical and Theoretical Perspectives. Oxford: University Press, 67-88.
Vrij, A .. K. Edward. and R. Bull. (200 1a). "People's Insight into their Own Behaviour and Speech Content While
Lyi ng,'' British lou mal of Psychology 92. 373-389.
Vrij, A .• K. Edward, and R. Bull. (2001b). "Police Officers' Ability to Detect Deceit: The Benefit of Indirect
Deception Measures," Legal. and Criminologica/.Psyc/wlogy 6, 185- 196.
Vrij, A., K. Edward, and R. Bull. (200 Ic). "Stereotypical Verbal and Nonverbal Responses While. Deceiving
Others," Personality and Social Psychology Bulletin 27, 899- 909.
Vrij, A. , K. Edward. K. P. Roberts, a nd R. Bull. (2000). "Detec ting Deceit Via Analysis of Verbal and Nonverbal
Beha vio r,'' lou mal of Nonverbal Behavior 24 (4), 239- 263.
Vrij, A., L. Akehurst, S . Soukara, and R. Bull. (in press). "Detecting Deceit Via Analyses o f Ve rbal and Nonverbal Behaviour in Adulls and Children," Human Communictllion Research.
Vrij, J. (2001). "Verzet tcgen angst,". Kontakt door Aantreden 56 (5), 4.
TSA 15-00014 - 000922
A SYTSTEMATIC MANNER TO DETECT DECEIT
79
Wallbott, H . G. and K. R. Sc he re r. (199 1). "Stress Specificities : Di ffere nti al Effects of Coping Style, Gende r,
and Type o f Stressor on Automatic Arousal, Facia l Expression , a nd Subjective Feeling ," Joumal of Personality and Social Psyc/wlogy 6 1, 147-156.
While, C. H _ and J . K .. Burgoon .. (2001 ). "Adaptalion and Communicmive. Design: Pa tterns. of lmeracliotl in
Truthful and Deccplive Conversations," Hum.an Communication Research 21, 9- 37.
Wiseman. R. (1995). ·'The Mega lab Truth Test," Nature 373.391.
Zuckerman, M ., B. M. DePa ulo, and R. Rosenthal. ( I 98 1). " Verbal and Nonverbal Communication of Deception," in L. Berkowilz (ed.), Advances in Experimental Social Psychology, vol. 14. New York, NY: Academic
Press , 1-57.
TSA 15-00014 - 000923
TSA 15-00014 - 000924