Content uploaded by Bruno Verschuere
Author content
All content in this area was uploaded by Bruno Verschuere on Nov 15, 2018
Content may be subject to copyright.
The role of deception in P300 memorydetection
Bruno Verschuere
1
*, J. Peter Rosenfeld
2
,Michael R. Winograd
2
,
Elena Labkovsky
2
and Roeljan Wiersema
1
1
Department of Psychology,Ghent University,Henri Dunantlaan, Ghent, Belgium
2
Department of Psychology,Northwestern University,Evanston, IL, USA
Purpose.P300 memorydetection test is aneuroscientific procedure to assess
memories stored in the brain. P300 memorydetection can and is currently applied to
assess criminal suspects on recognition of critical crime information. Contrasting
memorydetection with lie detection, researchers have argued that P300 memory
detection does not involvedeception. We empirically investigated this argument by
manipulating deception between groups.
Methods. Thirty-four community volunteers participated in aP300 memory
detection test, answering either deceptively (deceptivecondition) or truthfully (truth
condition) to their own name.
Results. P300 memorydetection was significant in the truth condition, indicating
that deceptiveresponding is not aprerequisite for valid P300 memorydetection.
However, there wereclear indications that deceptiveresponding improvedmemory
detection.
Conclusions. Deception seems involved in the P300 memorydetection test; and
deceptiveresponding mayadd to test accuracy.
The P300 memory detection test is aneuroscientific procedure to assess memories
stored in the brain. Memory detection has promising forensic potential. Amurderer can
be assessed on recognition of the murderweapon. Veracity of memoryfailure canbe
assessedin-patients whoare suspected of malingering.Recognitionofsecret
membership details can be assessed in terrorism suspects. Several laboratorystudies
found P300 memorydetection to be highlyaccurate in detecting crime memoryand
malingering (for areview see Rosenfeld, 2002). Memorydetection has often been
contrasted with lie detection. Agreat disadvantageoflie detection is that one cannot
measure lies directly.Insteadone measures the presumed physiological arousal that
accompanies stressoranxiety whenlying. Memorydetection, on the other hand, is
based upon recognition rather than deception. Pushing the lie detection versus memory
detection dichotomytothe limit, its was argued that deception is not involved in
*Correspondence should be addressed to Dr Bruno Verschuere,Department of Psychology,Ghent University,H.Dunantlaan 2,
B-9000 Ghent, Belgium (e-mail: bruno.verschuere@ugent.be).
The
British
Psychological
Society
253
Legal and Criminological Psychology(2009), 14, 253–262
q2009 The British Psychological Society
www.bpsjournals.co.uk
DOI:10.1348/135532508X384184
memorydetectionand that it ‘can (and should)bemeasured withoutdishonest
responding (Meegan, 2008, p. 18)’. Instead, we argue that deception may be involved in
memorydetection, and perhaps that it should be involved to obtain high accuracy.
P300 memory detection is based uponthe Concealed Information Te st (CIT;Lykken,
1959).The concealed Information Te st assesses memorybypresenting the examinee
with aseries of questions, each having one correct and several incorrect alternatives.
Autonomic nervous system (ANS) activitytothe alternatives is measured. Stronger
physiological responding to the correct items than to the incorrect items indicates
recognition. Amurder suspect, forexample, could be tested on recognition of crime
details such as the murderweapon: ‘Was it :::Agun? Abaseballbat? Arope? Aplastic
bag? Aknife?’. The Concealed Information Test relies uponasound theory and
demonstratedaccuracy (Ben-Shakhar &Elaad, 2003; Ve rschuere, Crombez, De Clercq, &
Koster,2004).Inthe eighties, neuroscientists have modified the Concealed Information
Test to allow examination of event-related potential (ERP). ERP studies use an oddball
paradigm, asking participants to classify stimuli in two categories: Target (1/6 of the
stimuli) versus non-target (5/6 of the stimuli). The target items are defined prior to the
task, during which theyrequire auniqueresponse. Compared to the non-target items,
the target items are infrequent and salient. Theyare expected to evoke aP300, apositive
ERP that peaks about400–900 ms after the onset of an infrequent and/orsalient
stimulus. The crucial manipulation is that to-be-assessed (‘probe’) items are embedded
within the non-target category. AP300 memory detection test in the murdercase above,
forexample, could involveasking the participant to press one keyfor ‘scissors’ and
another keyfor all non-target items (e.g.gun, baseballbat, rope, plastic bag, and knife).
Amongthe non-target items is the actual murderweapon (e.g.plastic bag).For the
innocent, the plastic bag will be anon-targetitem amongstthe others. The guilty,
however,will recognize the plastic bag as the actual murder weapon. Forhim/her,this
probe item is infrequent and salient, and thus will elicit alarger P300 amplitude and
longer reaction times (RTs) compared to the non-target items. At the individual level,
P300 memorydetectionoften reaches accuracy levels around 85–95% (Allen, Iacono, &
Danielson, 1992; Farwell &Donchin, 1991; Rosenfeld, Biroschak, &Furedy,2006;
Rosenfeld, Shue, &Singer,2007).
P300 memorydetection is based upon recognition rather than deception. There is,
however,noconsensus on the precise role of deception in P300memorydetection.
Reviewing the literature, we found three differenttheoretical positions on the role of
deception.
The first possibility is that deception is irrelevant and does not affect P300memory
detection. In their Psychological Bulletin review article, Bashore and Rapp (1993) take
such aposition whenstating that ‘it will not matter :::if the subject is told to be
deceptive or not (1993; pp. 12)’. Farwell and Smith (2001; pp. 8), who commercialized
P300 memory detection as ‘brainfingerprinting’, also argued that ‘the subject neither
lies nor tells the truth during the procedure, and in fact the results (:::)are exactly the
same whether the subject lies or tells the truth at any time’. However,itmay be
premature to conclude that deception is not involved. Classifying probes as irrelevant
items may be considered alie. Moreover,the main task of the participant is concealing
probe recognition, clearly involving deception. Thisanalysis suggests that deception is
involved in P300 memory detection.
The second possibility is that deception increases task difficulty,thereby decreasing
P300 probe amplitude. Indeed, anumber of studies have shown that P300 amplitude in
asecondarytask decreases with increasing workload of the primarytask (for areview
254 B. Verschuere et al.
see Kok, 2001).Toillustrate, Allison and Polich (2008) presented participants with
1000 Hz tones while playing afirstpersonshooter computer game. Comparedto
viewing the game setup screen, P300 amplitude to the tones was decreased when
activelyengaged in the game. Common interpretation of such finding is that the more
attention is allocated to the (primary) screen, the less attentionisallocated to the
(secondary) tones. Similarly, deception may increase workload, diverging resources
away from the oddball task and reducingthe oddball evoked P300. Thus, researchers
have arguedthat deception is detrimentalfor P300memorydetection (Meegan, 2008;
Rosenfeld, Biroschak, &Furedy,2006). Although deception may increase workload,it
may also increase attention to the probes therebyincreasing probe P300 amplitude
(Polich &Kok, 1995).
The third possibility is that deception may increase P300amplitudebyincreasing
probe saliency(see e.g. Allen, Iacono, &Danielson, 1992). One should keep in mind
that the primarytask of the participant is to classify stimuli as target versus non-target.
Whether stimuli are probes is irrelevant forthis task.Thus, the investigator assumes that
‘subjectsare sensitive to such seemingly irrelevant items if these items are distinctive in
some dimension that is important to the subject, even though it is not relevant to the
task the subjectisperforming (Farwell &Donchin, 1991; pp. 545)’. The more salient the
probesare, the greater the chance that the participant will attend to them and display a
P300. Some studies support the idea that deception may improve P300 memory
detection. First, ameta-analysis on 80 ANS memory detectionstudies explored factors
that moderated test accuracy (Ben-Shakhar &Elaad, 2003). It was found that overtly
answering ‘no’ to the stimuli (thereby lying to the probes) increased accuracy compared
to silentanswer tests. These data indicate that deceptive responding adds to the ANS-
based Concealed Information Test, and one could assume that this maygeneralize to the
P300-based modification of the Concealed Information Te st. Second, arecent study by
Meijer and colleagues (2007) is of relevance here. Theseauthorsperformed two studies
using facesasprobes. In afirstexperiment, ahit rate of 92% wasobtained when
participants wererequired to give deceptive responses to faces of friends/siblings.Ina
secondexperiment, participants were not explicitly instructed to give deceptive
responses. Ahit rate of only 17% was found forfaces of teachers. Results suggest that
deceptive responding adds to P300memorydetection accuracy.However,because the
two studies also differ in probe saliency (highly salient friend/sibling faces versus low
salientteacher faces) no firmconclusions can be drawn with regard to the role of
deceptive responding.
In sum, there are three differentpositions with regard to the role of deception in
P300 memorydetection. The first positionisthat deception is irrelevant and it is
predicted that deception does not affect P300 memorydetection(Bashore &Rapp,
1993; Farwell &Smith, 2001). The second position is that by increasing task difficulty
deception will decrease P3 amplitude, predicting worse P300 memory detection
performance when answering deceptivelycompared to answering truthfully (Meegan,
2008; Rosenfeld, Biroschak, &Furedy,2006). However,note that the latter prediction
is not that straightforward. Increasing workload may reduce P300 amplitude to
all stimuli, but does notnecessarily reducethe crucial probe-irrelevant difference.
The third position is that deception will increase probe saliency thereby increasing P300
amplitudetothe probes. Compared to truthfulresponding, deceptive responding
is predictedtoresult in better P300memorydetection performance by this view.
In the present study,weexaminethe role of deception by instructing half of the
participants to answer truthfully (truth condition), and half to answer deceptively
Deception and P300 255
(deceptive condition). This manipulationallows us to investigate whether and how
deceptive responding affects P300 memorydetection accuracy.Wepredicted better
probe versus irrelevant differentiationinthe deceptive condition than in the truth
condition.
Method
Participants
Forty-fourparticipants were recruitedthroughanagencyand also through
advertisements and flyers.
Apparatus
The master PC, running WINEEG(MitsarCorp.), recorded the EEG and RTs, and
triggered aslave PC running PSYTASK software (MitsarCorp.), to present stimuli to the
participants on the monitor.
Procedure
Participants were askedtotake partina‘Brain Wa ves and Deception’ study.Written
informed consent was obtained from all participants. Participants werepresented with
the irrelevant items,which were common North American names consisting of 1to3
syllables. Items that resembled names that were of high relevance to the participant
(e.g. name of relative) were changed. Participants wereinformedthat theywere
required to watch names on the computer screen and to execute atask while doingso.
Participants were asked to press the right mouse button with the right middle finger
whenever theysaw the target name (CASEY) and to press the left mouse button with the
right indexfinger forall other items.
The crucial manipulationisthe way that the meaning of the buttons was explained.
Participants in the deceptive conditionwere told that: ‘You have been randomly
assigned to the deceptive condition. Yo uhave to trytoconcealrecognition of
meaningful information,inthis case your own first name. Yo uhave to pretend that
there’snothing special about this name. Yo ualso have to execute atask: Wheneveryou
see the target name, CASEY,you press the right buttonasfast as possible. Pressingthis
button means YES, Irecognize this name.For all other names, you press the left button
as fast as possible. Pressingthis buttonmeans NO,Idon’t recognize this name.You can
press YES only to the target name, CASEY.Thatmeans that you are obliged to press NO
to your own name. Clearly,this is alie, since youpress abutton that says youdon’t
recognize the name when in fact, youdo’. Participants in the truth condition were told
that: ‘You have been randomly assigned to the control condition. While watching the
computer screenyou will see several names.One of these will be the target name,
CASEY.Whenever you recognize the target name, CASEY,you press the right button as
fast as possible. Pressingthis buttonmeans YES,Irecognize this as the target name.For
all othernames,you will press the leftbutton as fast as possible. Pressingthis button
means NO,this is not the targetname.Weare using the names of all the participants in
this study,soyou may also see you own name, occasionally.Itisnot the target name, so
you also press NO to your name along with the other non-target names’.
The oddball task consistedof360 trials, with 60 repetitions of the participants’first
name, 60 repetitions of the target name, and 60 repetitions of each of 4irrelevant item
256 B. Verschuere et al.
(totaling 240 irrelevant names). Presentation was random with the exception that there
could be no consecutive trials of one name. The stimuli, about 1cmtall, were presented
as words in white capital lettersinArial font on ablack display monitor about 1mfrom
the subject’seyes. Stimuli were presented for300 ms, with the next stimulus appearing
2700 ms after offset of the previous stimulus. After the test, participants were paid and
debriefed.
Recording and scoring of the data
EEG was recorded with silver electrodes filled with adhesive paste attached to sites
Fz, Cz, and Pz. The scalp electrodes werereferenced to linked mastoids. EOG was
recorded with silver electrodes above and below the right eye, but laterally offset
from one another by 1cm. Theywere placed diagonallytopick up both vertical and
horizontal eye-movements. Impedance was kept below 10 k Vforall electrodes. The
EEG electrodes were referentially recordedbut the EOG electrodes were differentially
amplified. The forehead was grounded. Amplifier output was passed to an A/D
converter sampling at 500 Hz. Signals first passed an on line hardware filter (0.3–
30 Hz),and later an additional offline filter (0–6 Hz). The artifact rejectioncriterion
forEOG was 50mV. Trials were averaged per stimulus type. We used the peak-to-peak
method to calculateP300 amplitude. The algorithm first searches forthe positive
peak, which is the maximum 100 ms mean positivity in the 350–900 ms period post
stimulusonset. Next, the negative peak (or NEG) is determined, which is the the
maximum 100 ms mean negativity in the 1300 ms following the positive peak. P300
amplitudeiscalculated by subtracting the amplitudeofthe negative peak from that
of the positive peak.Ithas been repeatedly shown that this peak-to-peak method
is more sensitive forindividual classification than the baseline to peak method
(e.g. Soskins et al.,2001).
Results
Data from participants with too high anumber of behavioral errorsorEEG artifacts,
were excluded from further analyses ( n¼10).The final sample consisted of 16
participants (8 women) in the deceptive condition, and 18 participants (10 women) in
the truth condition. There werenosignificant group differences in gender,ethnicity,
handedness, time of last meal, length of name or medication use. The deceptive group
had alower mean agethan the truthgroup (Deceptive: Mage¼21 years [SD ¼3];
Truth: Mage¼27 years [SD ¼9]), t(32) ¼2.33,p,.05. Becauseage did not correlate
with any of the dependent variables,wedid not control forthis factor.
Separate 2(Condition: deceptive versus truth) by 2(Stimulus: probe
versus irrelevant) analyses of variance (ANOVA)were runonthe mean scorefor
each dependent variable (RTs, error rates, and the P300). We reportpartial eta squared
( h
p2
¼SS
effect
/(SS
effect
þSS
error
)) as an estimate of effect size.
To determine how manyparticipants showed areliable probe .irrelevant P300
difference, we usedthe bootstrap method(Wasserman &Bockenholt, 1989).For each
individual one has available only one averageresponsefor each stimulus type. To allow
statistical testingofprobe versus irrelevant differences within an individual, a
distribution is generated foreach stimulus type. First, aset of responses is drawn at
random,with replacementfrom the probe items.Mean P300 amplitude is calculated
from the drawn sweeps using the peak-to-peak method.The sameisdone forthe
Deception and P300 257
irrelevant items.The number of responses is the actual number of accepted trials for
that stimulus type forthat individual. Next, adifference score is obtained by subtracting
the irrelevant P300 from the probe P300. This process is repeated 500 times,resulting in
adistribution of 500 differences scores. If the mean difference score minus 1.29 times
the standard deviation is greater than zero, it can be concluded with 90% confidence
that the probe P300isgreater than the irrelevant P300. Asimilar procedure was used for
reactiontimes.
Errors
The 2(condition: deceptive versus truthful) by 2(stimulus: probeversus irrelevant)
ANOVA revealed asignificant main effect of stimulus, F (1, 32) ¼8.45, p,.01, h
p2
¼.21,
indicating ahigher error percentageonprobe compared to the irrelevant items,see
Table 1. No other effect reached significance, F’s ,1.7.
RTs
Error trials were excluded from RT analyses, as were RTs smaller than 150 ms or
greater than 800ms(,2% of all trials). The 2(condition: deceptive versus truthful) by
2(stimulus: probeversus irrelevant) ANOVA revealed amain effect of stimulus,
F(1, 32) ¼33.53, p,.001,h
p2
¼.51, indicating greater RTs to probe compared to
irrelevant items,see Table 1. There was amain effect of condition, F (1, 32) ¼9.93,
p,.01, h
p2
¼.24, indicating that participants in the deceptive conditionwere slower
than participants in the truth condition. Finally, there wasatrend towards stronger
reactiontime slowing on probe versus irrelevant items stimuli in the deceptive
condition compared to the truth condition, F(1, 32) ¼2.92, p¼.10, h
p2
¼.08. Hit rates
were similar in the deceptive (9 out of 16 participants or 56%) and the truth condition
(9 out of 18 participants or 50%),x
2
(1,34) ,1.
P300 amplitude
The 2(condition: deceptive versus truthful) by 2(stimulus: probeversus irrelevant)
ANOVA revealed amain effect of stimulus, F(1, 32) ¼21.38, p,.001, h
p2
¼.40,
indicating larger P300amplitudes to probe compared to irrelevant items, see Ta ble1
and Figure 1. The main effect of condition wasnot significant, F,1. Although, the
stimulus£condition effect was not significant in the ANOVA ,F,1.7, h
p2
¼.05, there
was aclear trend towardsbetter detection in the deceptive (12 out of 16 participants or
75%) than in the truth condition (8 out of 18 participants or 44%), x
2
(1,34) ¼3.26,
p¼.07.
Discussion
Exploring the role of deception in P300 memorydetection, participants either
answered truthfully or deceptivelytotheir own name. Twoimportant findings result
from this study: (1) That P300 memorydetection is valid in the truth condition, and (2)
that deceptive responding seems to add to P300 memory detection accuracy.
First, given that P300 memory detection was significant when participants answered
truthfully; deceptive responding seems not aprerequisitefor P300memorydetection.
At first, this finding seems to contrast with Meijer et al. (2007; Experiment 2) who found
258 B. Verschuere et al.
Ta ble 1. Mean percentage errors, reaction times and P300 amplitude for target, probe, and irrelevant itemsinthe deceptiveand the truthful condition
DeceptiveTruthful
Ta rget M(SD)Probe M(SD)
Irrelevant
M(SD)
Probe-irrelevant
difference
Ta rget
M(SD)
Probe
M(SD)
Irrelevant
M(SD)
Probe-irrelevant
difference
Errors (%) 7.08 (6.28) 1.04 (1.91) 0.03 (0.10) p¼.05 9.54 (10.41) 0.46 (0.96) 0.07 (0.21) p¼.05
RTs(ms) 541 (52) 483 (69) 451 (55) p,.001 473 (61) 418 (51) 401 (44) p,.004
P300 (m V) 17.69 (9.78) 14.72 (9.73) 8.64 (4.43) p,.008 16.55 (5.67) 12.64 (5.24) 9.44 (5.07) p,.002
Note.The probe-irrelevant difference arethe result from planned paired sample ttests.
Deception and P300 259
no significant memorydetection in students fortheir teacher’sfaces. An important
difference betweenthe present study and Meijer et al is that, we used highlysalient
probes. Basic P300research(Polich &Kok, 1995) and applied P300 memory detection
research(Rosenfeld, Biroschak, &Furedy,2006) has shown that probe saliency
increases P300amplitude. Ta ken together,our data indicate that recognitionofan
infrequent significantevent is necessaryand sufficient fordifferential responding in
P300 memory detection. Thesefindings are informative on when and in whom P300
memorydetection will work. Let’sconsider the Harrington case, whereP300 memory
detection was used to tryand exonerate aconvicted murderer (http://www.
brainwavescience.com). More than 20 years after the crime, Harrington was tested on
recognition of crime details such as ‘weeds and grasses’ on the escaperoute. The
examiner, Larry Farwell, concluded that ‘with a99.9% statistical confidencelevel that
the recordstoredinHarrington’sbrain does not match the crime scene’. In terms of
signal detection theory,Farwell concluded that lack of significant probe –irrelevant
discrimination was atruenegative test result. However,one could question how likely it
is that someone would remember apparenttrivial crime details formorethan 20 years.
The validityofthe P300memorydetection test is questionable because Harrington may
have forgotten crime details after 20 years. Moreover,probe items such as ‘weeds and
grasses’ may lack the saliency needed to elicit areliable P300. Stated differently, our
analysis of the necessaryconditions forsuccessful P300 memorydetectionindicates
that Harrington’stest outcome may have been afalse negative test result.
Second, our data indicate that deceptive responding may add to P300 memory
detection accuracy: the probe –irrelevant difference forRTs tendedtobegreater
with deceptive responding (32 ms) compared to truthful responding (17 ms). For
P300, there was aclear tendency towards higher hit rates with deceptive responding
(75%) compared to truthfulresponding (44%). Caution is warranted because statistical
tests didnot reach conventional levels of statistical significance.Wethink that our
findings are reliable because of three reasons.First, effects were found fortwo
different measures: RTs and P300 amplitude. Second, the currentfindings are in line
Figure1.P300 amplitude grand averages (expressed in m V) for target, probe,and irrelevant items in
the truthful condition (left) and the deceptivecondition (right). The first vertical line (at 100 ms)
indicates stimulus onset, the second vertical line (at 400 ms) indicates stimulus offset.
260 B. Verschuere et al.
with findings from ANS-based memorydetection where it was found that deceptive
responding adds to memorydetection accuracy (Ben-Shakhar &Elaad, 2003).Third, a
recent study by Nittono and Kubo (2008) also indicates that deception enhances P300
memorydetectionvalidity: P300amplitude was enhanced whenparticipants were
asked to hide recognition of the probes (deception) compared to acontrol condition
withoutdeception.
Taken together,our data indicate that P300 memorydetectionbenefits from
deceptive responding. Instructions in P300 memorydetectiontests should stress
deception. Thiscan be easily done by instructing the participant that pressing NO
means non-recognition, hence deception if one does recognize aparticular stimulus.
From an applied perspective, it is sufficient to know that explicit deception instructions
enhance differential probe-irrelevant responding. From atheoretical perspective it is
important to understand why deception affects P300 memorydetection. The workload
hypothesis predicted that P300 amplitude would be decreased by the greater task
difficulty when responding deceptively. Participants in the deceptive conditionwere
generally slower than participants in the truth condition, indicatingincreased task
difficulty.Still, P300was increased rather than decreased. We hypothesized that
deceptive responding increases probe saliency,hence, P300 amplitude. Future research
should investigate whether deceptive responding affectsP300 memorydetection
troughprobe saliencyorother (e.g.emotional or motivational) factors.
Acknowledgements
This researchwas supported by agrant from the ResearchFoundation –Flandersawarded to the
first author (Grant number K.2.144.07.N.01) and by agrant from the US Department of Defense
Polygraph Institute awarded to the second author (Grant Number DODP198-P-0001). We thank
Gershon Ben-Shakhar forhis valuable comments on an earlier version of this manuscript.
References
Allen, J. J.,Iacono, W. G.,&Danielson, K. D. (1992). The identification of concealed memories
using the event-related potential and implicit behavioral measures –amethodology for
prediction in the face of individual-differences. Psychophysiology,29(5), 504–522.
Allison, B. Z., &Polich, J. (2008). Workload assessment of computer gaming using asingle-stimulus
event-related potential paradigm. Biological Psychology ,77(3), 277–283.
Bashore, T. R., &Rapp, P. E. (1993). Are there alternatives to traditional polygraph procedures.
Psychological Bulletin,113(1), 3–22.
Ben-Shakhar,G., &Elaad, E. (2003). The validity of psycho physiological detection of information
with the Guilty knowledgetest: Ameta-analytic review.Journal of Applied Psychology ,88,
131–151.
Farwell, L. A., &Donchin, E. (1991). The truth will out –interrogative polygraphy(lie detection)
with event-related brain potentials. Psychophysiology,28(5), 531–547.
Farwell, L. A., &Smith, S. S. (2001). Using brain MERMER testing to detect knowledgedespite
effortstoconceal. Journal of Forensic Sciences,46(1), 135–143.
Kok, A. (2001). On the utility of P3 amplitude as ameasureofprocessing capacity.
Psychophysiology,38,557–577.
Lykken, D. T. (1959). The GSR in the detection of guilt. Journal of Applied Psychology ,43,
385–388.
Meegan, D. V. (2008). Neuroimaging techniques formemorydetection: Scientific, ethical, and
legal issues. TheAmerican Journal of Bioethics,8(1), 9–20.
Deception and P300 261
Meijer,E.H., Smulders, F. T. Y. ,Merckelbach, H. L. G. J.,&Wo lf, A. G. (2007). The P300 is sensative
to concealed face recognition. International Journal of Psychophysiology ,66,231–237.
Nittono, H., &Kubo, K. (2008). The effect of intentional concealment on the event-related
potentials in aconcealed information test. International Journal of Psychophysiology ,69(3),
149–150.
Polich, J.,&Kok, A. (1995). Cognitive and biological determinants of P300 –anintegrativereview.
Biological Psychology ,41(2), 103–146.
Rosenfeld, J. P. (2002). Event-related potentials in the detection of deception, malignering, and
false memories. In K. Murray (Ed.), Handbook of Polygrahtesting (pp. 265–286). UK:
Academic Press.
Rosenfeld, J. P. ,Biroschak, J. R., &Furedy, J. J. (2006). P300-based detection of concealed
autobiographical versus incidentallyacquired information in target and non-target paradigms.
International Journal of Psychophysiology,60,251–259.
Rosenfeld, J. P. ,Shue, E., &Singer,E.(2007). Single versus multiple probe blocks of P300-based
concealed information tests forself-referring versus incidentallyobtained information.
Biological Psychology ,74(3), 396–404.
Soskins, M., Rosenfeld, J. P. ,&Niendam, T. (2001). The case forpeak-to-peak measurement of P300
recorded at .3 hz high pass filter settings in detection of deception. International Journal of
Psychophysiology,40,173–180.
Verschuere, B., Crombez, G.,DeClercq, A., &Koster,E.H.W.(2004). Autonomic and behavioral
responding to concealed information: Differentiating orienting and defensive responses.
Psychophysiology,41,461–466.
Wasserman, S., &Bockenholt, U. (1989). Bootstrapping: applications to psychophysiology.
Psychophysiology,26,208–221.
Received 14 August 2008; revised version received 8October 2008
262 B. Verschuere et al.
