Original article

‘Munchausen’s syndrome by proxy’ or a ‘miscarriage of justice’?An initial application of functional neuroimaging to the question

of guilt versus innocence

Sean A. Spence*, Catherine J. Kaylor-Hughes, Martin L. Brook,Sudheer T. Lankappa, Iain D. Wilkinson

Academic Clinical Psychiatry, University of Sheffield, The Longley Centre, Norwood Grange Drive, Sheffield S5 7JT, UK

Received 22 June 2007; received in revised form 30 August 2007; accepted 2 September 2007

Abstract

‘Munchausen’s syndrome by proxy’ characteristically describes women alleged to have fabricated or induced illnesses in children under theircare, purportedly to attract attention. Where conclusive evidence exists the condition’s aetiology remains speculative, where such evidence islacking diagnosis hinges upon denial of wrong-doing (conduct also compatible with innocence). How might investigators obtain objective ev-idence of guilt or innocence? Here, we examine the case of a woman convicted of poisoning a child. She served a prison sentence but continuesto profess her innocence. Using a modified fMRI protocol (previously published in 2001) we scanned the subject while she affirmed her accountof events and that of her accusers. We hypothesized that she would exhibit longer response times in association with greater activation of ven-trolateral prefrontal and anterior cingulate cortices when endorsing those statements she believed to be false (i.e., when she ‘lied’). The subjectwas scanned 4 times at 3 Tesla. Results revealed significantly longer response times and relatively greater activation of ventrolateral prefrontaland anterior cingulate cortices when she endorsed her accusers’ version of events. Hence, while we have not ‘proven’ that this subject is inno-cent, we demonstrate that her behavioural and functional anatomical parameters behave as if she were.! 2007 Elsevier Masson SAS. All rights reserved.

Keywords:Munchausen syndrome by proxy; Fabricated or induced illness; Miscarriage of justice; fMRI; Ventrolateral prefrontal cortex; Anterior cingulate cortex;Response time; Factitious or induced illness; Lies; Truth

1. Introduction

Many areas of psychiatric practice require the psychiatristto consider, whether implicitly or explicitly, the veracity oftheir informant [30]. Sometimes, deceptive conduct is centralto the diagnosis considered (e.g., antisocial personality, psy-chopathic and conduct disorders, ‘instrumental psychoses’);alternatively it may be a consequence of another condition(e.g., the addictions) [11]. However, the attribution of

deception to others is rarely straightforward [37] and thereare several psychiatric conditions where the subject’s aware-ness of their ‘deceit’ is either contested or unknown (e.g.,some neurological accounts of ‘hysteria’ [28,29], Ganserstates [10], Munchausen’s syndrome [2]). Still more difficultare cases in which the facts themselves are unconfirmed sothat not only is the subject’s behaviour disputed but also theirmotivations are susceptible to radically different interpreta-tions. Munchausen’s syndrome by proxy (MSBP; factitiousor induced illness) is one such example: characteristically,the female carer of a child is alleged to have caused symptomsof illness in that child, purportedly as a way of attracting atten-tion (to herself) [3,19,20,27]. Not only are the reasons for suchbehaviour unknown (see Ref. [27]) but also the response which

* Corresponding author. Tel.: !44 (0) 114 22 61519; fax: !44 (0) 114 2261522.

E-mail address: s.a.spence@sheffield.ac.uk (S.A. Spence).

0924-9338/$ - see front matter ! 2007 Elsevier Masson SAS. All rights reserved.doi:10.1016/j.eurpsy.2007.09.001

ARTICLE IN PRESS

Please cite this article in press as: Sean A. Spence et al., ‘Munchausen’s syndrome by proxy’ or a ‘miscarriage of justice’? An initial application of functionalneuroimaging to the question of guilt versus innocence, European Psychiatry (2007), doi:10.1016/j.eurpsy.2007.09.001

European Psychiatry xx (2007) 1e6http://france.elsevier.com/direct/EURPSY/

+ MODEL

the woman exhibits following accusation offers little clarifica-tion: if she admits wrong-doing she is guilty, if she denieswrong-doing her denial is considered symptomatic ofMSBP; but of course, her denial might also reflect herinnocence.

‘‘If the mother appears calm or distressed, charming orhostile, distant or over-involved, either appearance can bedescribed as characteristic of mothers with MSBP.’’ [24,p. 91]

While such cases might be readily diagnosed if there wereobjective, clear-cut evidence of wrong-doing, matters are farmore complex when the nature of the child’s illness is itselfgenuinely difficult to diagnose. In such cases the extent ofan endogenous contribution to the illness may be so obscurethat accurately apportioning (exogenous) blame is perilous[24].

Nevertheless, the behaviour that constitutes MSBP is a se-vere form of child abuse and potentially lethal. The actions in-volved, such as tampering with medical samples or equipmentcan be so goal-directed as to render implausible the proposi-tion that the subject ‘did not know what she was doing’. Ifharm has occurred, much would seem to hinge upon differen-tiating a carer who has made genuine errors, or is mentally ill,from one who sets out, intentionally, to harm the child underher care. How might clinical neuroscience inform such diffi-cult evaluations, where the result of a decision might literallymean life or death for a dependent child? In this report weposit one possible approach: the use of functional neuroimag-ing to determine whether an informant is lying when shedenies having deliberately harmed a child. In essence, this isa modified form of ‘lie detection’, utilising what is knownabout the cognitive neurobiology of deception [34].

The validity and reliability of methods of lie detection havelong been causes for debate [11,37]. However, recent advancesin functional neuroimaging technology have provided the op-portunity of studying the neural correlates of lying, while sub-jects answer questions in the scanner; yielding data consistentwith the view that deception constitutes an executive function[30]. Elements of deceptive neurocognition include: responsesuppression (withholding the truth), implicating ventrolateralprefrontal cortices (Brodmann area, BA 47 [1,8,12e14,17,23,26,32,34]), response monitoring, involving anteriorcingulate cortices (BA32 [1,9,13,14,16,17,23,32]) and re-sponse manipulation, implicating parietal regions (e.g.,BA40 [4,8,16e18,26,32,34]). Also consistent with this (exec-utive) hypothesis is the finding of longer response times (RTs)during lying than truth-telling [9,11,23,32,33]. Hence, decep-tion appears to incur a cognitive processing cost (as reflectedin longer RTs).

Such findings have prompted speculation concerning theirpotential application in the forensic arena. However, althoughthese developments have been promising, there remain con-cerns regarding their validity and reliability [5,25,31,34]. Sofar, functional neuroimaging studies of lying have examinedonly healthy, predominantly right-handed subjects, often re-cruited from academic environments, where the lies told and

the consequences of their detection are altogether minor[1,4,8,9,11e18,23,26,32e34]. A systematic critique of thecurrent literature exposes the failure of most investigators toprecisely replicate their own findings [34]; hence, there isa need for empirical replication and cautious interpretationat this stage of the science [34].

This report describes a radical departure from earlier stud-ies: the application of functional magnetic resonance imaging(fMRI) to the case of a subject who has been convicted of a se-rious offence but who wished to demonstrate her innocence byanswering questions in the scanner.

2. Subject

X is a 42-year-old married woman who was convicted ofa crime against a child in her care (4 years prior to the study).Found guilty by a majority verdict (in another jurisdiction) sheserved a longer prison term than she might have as she contin-ued to profess her innocence. Her family and friends believedher and there is currently [2007] a campaign to clear her name.Medical opinion obtained by the court initially described a psy-chologically normal subject [pre-verdict], then features ofMSBP [post-verdict]. The grounds for MSBP comprised X’scontinued denial of guilt. X is right-handed, with a projectedverbal IQ of 109 [22]. Prior to her conviction she had nohistory of neuropsychiatric, forensic or substance misusedisorders, exhibiting a stable work and family background(remarked upon in the judge’s summation of her case). Thechild concerned had an undisputed, though complex, congen-ital condition (precise diagnosis unknown), necessitatingrepeated hospital admissions throughout early life. An olderchild in X’s care had never been a focus of concern.

At the time of the study X was euthymic although she hadbeen treated for depression during the period of her prison sen-tence and remained on prophylactic antidepressant medica-tion: venlafaxine 225 mg/day. She also received gabapentin400 mg b.d. to treat peripheral neuralgia, secondary to an as-sault sustained while imprisoned.

Prior to the study X was interviewed and the details of hercase examined in depth to allow the composition of questionsthat would be used to probe her account (questions that couldbe answered with ‘yes’/‘no’ responses, where answering yes orno would specifically endorse either X’s account or that of heraccusers). This method was adapted from previous work ingroups of healthy subjects studied in our unit [11,32,33]. Es-sentially, our method relies upon the subject answering thesame question with both an affirmation and a denial, on sepa-rate occasions, over the course of repeated scans [32].

X provided full written consent after the procedure hadbeen explained to her (and members of her family); the studywas approved by the ethics committee of the University ofSheffield, UK.

3. Method

Imaging data were acquired using an MR system operatingat 3 Tesla (Achieva 3.0T, Philips Medical Systems, Best,

2 S.A. Spence et al. / European Psychiatry xx (2007) 1e6

ARTICLE IN PRESS

Please cite this article in press as: Sean A. Spence et al., ‘Munchausen’s syndrome by proxy’ or a ‘miscarriage of justice’? An initial application of functionalneuroimaging to the question of guilt versus innocence, European Psychiatry (2007), doi:10.1016/j.eurpsy.2007.09.001

Holland). The subject underwent 4 separate functional ‘runs’.Each of these lasted 180 s, during which 60 sets of imageswere acquired at a rate of 1 set (giving complete brain cover-age) every 3 s. Blood oxygen-level dependent (BOLD), andT2*-weighted functional images were acquired in the trans-axial plane parallel to a line bisecting the anterior and poste-rior commissures, using a gradient-recalled, single-shot echoplanar imaging (EPI) technique (TE" 35 ms; TR" 3000 ms;SENSE factor" 1.5 in the anterioreposterior direction). Auto-mated volume selective shimming was performed prior to thefirst functional scan. At each imaging time point 32 contigu-ous, 4 mm thick slices were acquired having a 230 mm in-plane field of view, such that the acquisition voxel size was2.88# 2.93# 4 mm3 interpolated to 1.8# 1.8# 4 mm3.Hence, each functional scanning run comprised 6, 30 s epochs,during which truth and lie responses alternated in a standardcounterbalanced between-sessions box-car design (over3 min). (Each scan was preceded by the visual presentationof a contextual stem over 6 s, which would frame the questionspresented during the scan, below.) Routine anatomical imag-ing consisting of 2-dimensional, T2-weighted and 3-dimen-sional T1-weighted datasets were also acquired to excludethe presence of intracranial pathology.

During each scan 36 questions, specific to X’s case, werepresented visually via an MR-compatible, and radiofrequencyhead-coil mounted computer screen (Eloquence, InVivo Corp,Orlando, FL, USA). Questions could be negative (accusatory)or positive (exculpatory) in valence. Examples were thefollowing:

You deliberately harmed her [the child]You placed salt in her syringeYou were innocent of the chargesYou told your husband the truthYou lied to nursing staffYou told medical staff the truth

X was required to answer ‘yes’ or ‘no’ to each question us-ing an MR-compatible keyboard (Eloquence), responding withwhat she regarded as the ‘truth’ or a ‘lie’ according to a colourrule (alternating over 6 epochs, each containing 6 questions;with the same questions repeated later under the opposite con-dition in a counterbalanced design: ABCCBA. No series of 6questions would have elicited 6 consecutive ‘yes’ or ‘no’ re-sponses). Runs were balanced so that each question was askedtwice per run and the colour rule alternated so that each ques-tion was answered once with a putative ‘lie’ (i.e., X endorsedher accusers’ version of events) and once with the putative‘truth’ (i.e., X answered according to her own account ofevents). The stimulus questions presented during runs 3 and4 comprised inverted repetitions of those administered duringruns 1 and 2, respectively. The lie/truth colour rule presentedduring runs 2 and 4 was an inversion of that presented duringruns 1 and 3, respectively. (It should be noted that although Xwas instructed in the method of the scans and administereda ‘practice run’ outside the scanner, just prior to scanning,the questions posed during practice were unique and X hadnot seen the questions posed within the scanner until scanning

actually occurred.) X was directly observed and filmed duringthe scanning process.

Images were analysed using statistical parametric mapping(SPM)-2 (www.fil.ion.ucl.ac.uk), implemented in MATLAB(Mathworks Inc., Sherborn, MA, USA) on a PC. SPM com-bines the general linear model and Gaussian field theory todraw statistical inferences from functional data regarding devi-ations from the null hypothesis in 3-dimensional brain space[7]. The 3-dimensional co-ordinates given by SPM were con-verted from those of the Montreal Neurological Institute(MNI) to Talairach and Tournoux co-ordinates using the pro-gram MNI2TAL (http://imaging.mrc-cbu.cam.ac.uk/downloads/MNI2tal/mni2tal.m). Anatomical localizations were derivedfrom the atlas of Talairach and Tournoux [35]. The datawere motion-corrected, normalized (to the EPI template pro-vided in SPM: 2# 2# 2 mm voxels) and spatially smoothedusing a Gaussian kernel (full-width half-maximum" 12 mm).The responses to stimuli were modelled by a box-car wave-form convolved with a canonical haemodynamic responsefunction (HRF). A fixed-effects analysis was performed andresults of the subtraction of each condition (e.g., ‘lie minustruth’) displayed at a pre-determined statistical threshold ofp< 0.05 corrected, and extent threshold" 0. We adoptedthis conservative threshold to reduce the possibility of false-positive results (type 1 errors). Given that our subject actsas her own ‘control’ fixed-effects analyses are justified inthis case.

4. Results

Mean response time data demonstrated that X took signif-icantly longer to respond to questions requiring her to endorsethe account of her accusers (Wilcoxon’s signed ranks test:Z"$5.15, p< 0.001; we used non-parametric statistics inview of greater variance of RT during ‘lie responses’;Fig. 1). Neuroimaging data revealed significant activation inbilateral ventrolateral prefrontal (BA 47) and anterior cingu-late cortices (BA 32) (together with other executive regions)during the generation of such responses (i.e., when X endorsedher accusers’ version of events) (Fig. 2, right). In the oppositesubtraction (‘X’s version minus that of her accusers’) she ex-hibited no activation in prefrontal cortices (Fig. 2, left); therewas a small focus of activation in left lingual gyrus (BA 19).

5. Discussion

We sought to examine the possibility that functional neuro-imaging might be used to investigate whether a woman con-victed of poisoning a child in her care, who continues toprofess her innocence, is telling the truth or lying when she en-dorses her account and that of her accusers. We used fMRI toexamine her brain activity when she endorsed items underthese 2 conditions. If her own account had comprised a seriesof lies then we should have expected her to exhibit longer re-sponse times and greater activation within ventrolateral pre-frontal (VLPFC) and anterior cingulate cortices (ACC) whenshe answered accordingly (i.e., when she claimed to be

3S.A. Spence et al. / European Psychiatry xx (2007) 1e6

ARTICLE IN PRESS

Please cite this article in press as: Sean A. Spence et al., ‘Munchausen’s syndrome by proxy’ or a ‘miscarriage of justice’? An initial application of functionalneuroimaging to the question of guilt versus innocence, European Psychiatry (2007), doi:10.1016/j.eurpsy.2007.09.001

innocent) [11,32e34]. Instead, we found that these behaviou-ral and functional anatomical responses occurred when she en-dorsed the accounts of her accusers, i.e., when she ‘admittedguilt’. In other words, and bearing in mind our foregoingwork [11,32e34], and the hypotheses which we specified a pri-ori (above), X’s data are consistent with (though they do notprove) the thesis that she is telling the truth when she denieshaving harmed the child in her care. To our knowledge, thisis the first case described where fMRI or any other form offunctional neuroimaging has been used to study truths andlies derived from a genuine ‘real-life’ scenario, where theevents described pertain to a serious forensic case. All themore reason then for us to remain especially cautious while in-terpreting our findings and to ensure that we make explicit

their limitations: the weaknesses of our approach. It is impor-tant to consider whether we might have obtained the ‘wrong’result.

The caveats encountered may be divided into those attribut-able to the experimental procedure and those implicating thesubject.

5.1. Experimental procedural caveats

First of all, is it possible that the subject might have becomeso used to the details of her case that her responses are ‘auto-mated’, so that they do not require the engagement of thecognitive executive to execute deception [30]? This is conceiv-able, following years of cross-examination and counterargu-ment, but would such automation not have occurred underboth experimental conditions (i.e., ‘truth’ and ‘lie’), therebypromoting a type 2 error? It is a feature of our experimentaldesign that the subject answers the same questions truthfullyand dishonestly, so that the same material is probed underboth conditions; crucially, we are interested in the contrast bet-ween response pairs. Whether or not X’s understanding of hercase has become automated, it is not clear why the finalcognitive subtraction should have revealed highly significantresults (despite being obtained from a single subject, they sur-vived statistical correction for multiple comparisons) in the‘direction’ consistent with her account (i.e., with longer res-ponse times and greater VLPFC and ACC activity when sheendorsed the accusations against her). Hence, even allowingfor automation, the data still seem to favour X.

Nevertheless, it might be argued that the response signatureseen is one connected with the emotive nature of the materialbeing probed: X activates more when she admits the unaccept-able (child abuse). This is a harder problem to resolve. Wehave seen in our previous work that people answering

Fig. 1. Box plot of response times (in milliseconds) exhibited by subject X inresponse to questions concerning accusations against her [‘truth’"when sheaffirmed her own account of events; ‘lie’"when she affirmed her accusers’account of events]. Results show a greater variation in response times and a sig-nificantly longer mean response time during ‘lying’.

Fig. 2. Rendered images of X’s brain showing areas of significantly greater activity when she endorsed her accusers’ version of events as opposed to her own (right)and vice versa (left). Prefrontal activity is significantly greater during her endorsement of accusations against her, i.e., when she says she is ‘lying’ ( p< 0.05,corrected for multiple comparisons).

4 S.A. Spence et al. / European Psychiatry xx (2007) 1e6

ARTICLE IN PRESS

Please cite this article in press as: Sean A. Spence et al., ‘Munchausen’s syndrome by proxy’ or a ‘miscarriage of justice’? An initial application of functionalneuroimaging to the question of guilt versus innocence, European Psychiatry (2007), doi:10.1016/j.eurpsy.2007.09.001

questions concerning shameful or embarrassing episodes intheir lives still evince greater activity within VLPFC whenthey lie as opposed to telling the truth regarding the same ma-terial (see Ref. [34]), but we also have to acknowledge thatnone of the material probed in that study was comparable tothe poisoning of a child. So, we cannot fully counter this ob-jection without further empirical research.

Our subject was receiving medication at the time of thestudy; might this explain our findings? Given that she is being‘compared with herself’ in this design, it is not apparent whythe administration of venlafaxine or gabapentin should ‘fa-vour’ one account over another (i.e., why greater prefrontal ac-tivation should occur while the subject endorses her accusers’account). But again, we require further empirical studies toclarify this issue (and to replicate our findings).

5.2. Subject-related caveats

Given the serious nature of the material addressed in thisstudy, the severe illness of a child and the conviction of a carer,we must consider the possibility that our subject influencedher results through some means other than ‘simply’ tellingtruths or lies. Is it possible that she might have ‘cheated’?

One possibility could have been the use of movement, to in-troduce ‘noise’ during data acquisition. If X hadmoved her headconstantly, or in time with experimental conditions, then thismight have obscured the true findings or added spurious signal.However, we have key observations addressing this issue. Xwasdirectly observed (and filmed) throughout the procedure, so weknow that she did not exhibit noticeablemovements of her limbsor torso during scanning. With respect to her head movement,we should already have discarded scan datasets containinghead movement in excess of 2 mm in any direction (and thisdid not occur). Also, when we performed post hoc analyses ofmovement data, acquired under truth and lie conditions, exam-ining them for systematic changes in head position across eachof 6 dimensions (x, y, z planes, togetherwith pitch, roll and yaw),to the level of 10$5 mm,wewere unable to demonstrate any sys-tematic movement between conditions (in paired t-tests, all p-values" 0.2e0.7). Hence, insofar aswe could observe,measureor analyse movement, empirically, we were unable to find anyevidence that X had manipulated our findings in this way.

However, a second possibility might be that given the rele-vance of response times to this experiment X had deliberatelyslowed her responses during putative ‘lie’ scans, in contrast toputative ‘truthful’ responses. This is feasible, although de-manding to execute, and we know that there was greater RTvariance under the ‘lie’ condition (Fig. 1). However, we per-formed further post hoc analyses to discover whether therewas any correlation between the duration of RT during lyingand the magnitude of BOLD response in those brain regionsmaximally activated during this condition. If X had cheatedwe should have expected a positive correlation between theseobservations. As it was we found evidence of a negative cor-relation between lie RTs and BOLD response in left VLPFC(i.e., the opposite of that predicted, if she had cheated) andno relationship in our other regions of interest (right VLPFC

and bilateral ACC). Hence, deliberate slowing of ‘lie re-sponses’ does not appear to account for X’s fMRI data.

Thirdly, we must consider whether X utilised an alternativecognitive strategy to influence her data. Might she, for instance,have ‘thought about’ another child she had cared for, in preferenceto the index child, in order to generate an ‘honest’ dataset? This isconceivable, and impossible to exclude empirically. It would haverequired considerable mental flexibility and might also have ren-dered some of our probe questions meaningless (e.g., with respectto adding salt to a syringe). Hence, we return to the question ofwhether such an evasive strategy would have yielded a highly sig-nificant result, in favour of X’s thesis, or simply a false negative.We cannot answer this questionwithout further empirical enquiry.

Finally, might it be that the subject has ‘convinced herself’ ofher innocence and that she answered sincerely though ‘incor-rectly’? Might this not be the very essence of MSBP? Undersuch conditions, X would not be lying but ‘self-deceiving’.Have we merely imaged ‘self-deception’? Again, this is a diffi-cult objection to counter, not least because so little is resolvedregarding the plausibility or neurobiology of ‘self-deception’as a phenomenon [6,21,29,36]. However, if the neurology of in-trapersonal (self) deception was similar to that of interpersonaldeception then our results would still favour X’s version ofevents (i.e., if greater prefrontal activation accompanied self-deception then this would implicate those scans where X en-dorsed the account of her accusers; Fig. 2). Nevertheless, furtherwork is needed to examine the neural correlates of responses totruth and lie conditions when a subject is genuinely mistaken(e.g., through ‘self-deception’, ‘denial’, psychogenic disorderor delusion). One issue would be that a subject is not guilty ofdeception if they answer with what they sincerely believe to bethe truth; theymay bemistaken but they cannot be said to be dis-honest. There is clearly a need for careful work in this area,which extends to include the veracity (and strength of convic-tion) of delusions, ‘denial’ and indeed the ‘psychiatric defence’itself (where that defence can comprise solely the reported pos-session of a false belief). Our work is admittedly preliminary inthis regard but it constitutes an important first step nonetheless.

5.3. Conclusion

X’s data replicate each of the findings reported previously(and hence predicted a priori) in groups of healthy subjects en-gaged in lying on the index paradigm [32]. More importantly,and specifically, they replicate these findings during the condi-tion when X endorsed the accusations of her accusers and notwhen she endorsed her own account. The most obvious fea-tures are pronounced activation of VLPFC and ACC, in asso-ciation with significantly longer response times during such‘lies’. In other words, while we cannot prove that X is innocentwe have found that her behavioural and functional anatomicalparameters behave as if she were.

Acknowledgements

This study was funded by Quickfire Media in associationwith Channel Four Television (in the UK). The process of

5S.A. Spence et al. / European Psychiatry xx (2007) 1e6

ARTICLE IN PRESS

Please cite this article in press as: Sean A. Spence et al., ‘Munchausen’s syndrome by proxy’ or a ‘miscarriage of justice’? An initial application of functionalneuroimaging to the question of guilt versus innocence, European Psychiatry (2007), doi:10.1016/j.eurpsy.2007.09.001

the study and a brief exposition of its findings were broadcaston British television in June 2007, but the data have not beenpublished before. The avoidance of the child’s name in the ex-perimental stimuli was a legal requirement for the study toproceed. We are grateful to X and her family for their timeand cooperation during participation, for allowing us to submitthese findings for publication, and to colleagues in AcademicRadiology at the Royal Hallamshire Hospital for their invalu-able assistance. We also thank Professor Caroline Kennedy-Pipe for helpful discussions prior to ethical approval, and those3 anonymous reviewers whose comments considerably im-proved this manuscript during its revision.

References

[1] Abe N, Suzuki M, Tsukiura T, Mori E, Yamaguchi K, Itoh M, et al. Dis-sociable roles of prefrontal and anterior cingulate cortices in deception.Cerebral Cortex 2006;16:192e9.

[2] Asher R. Munchausen’s syndrome. Lancet 1951;1:339e41.[3] Bass C, Jones DPH. Fabricated or induced illness. Psychiatry 2006;

5(2):60e5.[4] Davatzikos C, Ruparel K, Fan Y, Shen DG, Acharyya M, Loughead JW,

et al. Classifying spatial patterns of brain activity with machine learningmethods: application to lie detection. Neuroimage 2005;28:663e8.

[5] A lying matter. Science 2005;310:227. Editorial.[6] Fingarette H. Self-deception. 2nd edn. Berkeley: University of California

Press; 2000 (originally published 1969).[7] Friston KJ, Holmes AP, Worsley KJ, Poline J-B, Frith CD,

Frackowiak RSJ. Statistical parametric maps in functional imaging:a general linear approach. Human Brain Mapping 1995;2:189e210.

[8] Gamer M, Bauermann T, Stoeter P, Vossel G. Covariations among fMRI,skin conductance and behavioural data during processing of concealedinformation. Human Brain Mapping 2007 Feb 8 [Epub ahead of print].

[9] Ganis G, Kosslyn SM, Stose S, Thompson WL, Yurgelun-Todd DA.Neural correlates of different types of deception: an fMRI investigation.Cerebral Cortex 2003;13:830e6.

[10] Ganser SJM. A peculiar hysterical state. In: Hirsch SR, Shepherd M, ed-itors. Themes and variations in European psychiatry. Virginia: UniversityPress of Virginia; 1897. p. 67e77. 1974, first published 1897.

[11] Hughes CJ, Farrow TFD, Hopwood M-C, Pratt A, Hunter MD,Spence SA. Recent developments in deception research. Current Psychi-atry Reviews 2005;1:271e9.

[12] Kozel FA, Johnson KA, Mu Q, Grenesko EL, Laken SJ, George MS. De-tecting deception using functional magnetic resonance imaging. Biolog-ical Psychiatry 2005;58:605e13.

[13] Kozel FA, Padgett TM, George MS. A replication study of the neural cor-relates of deception. Behavioral Neuroscience 2004;118:852e6.

[14] Kozel FA, Revell LJ, Lorberbaum JP, Shastri A, Elhai JD, Horner MD,et al. A pilot study of functional magnetic resonance imaging brain cor-relates of deception in healthy young men. The Journal of Neuropsychi-atry and Clinical Neurosciences 2004;16:295e305.

[15] Langleben DD, Loughead JW, Bilker WB, Ruparel K, Childress AR,Busch SI, et al. Telling truth from lie in individual subjects with fastevent-related fMRI. Human Brain Mapping 2005;26:262e72.

[16] Langleben DD, Schroeder L, Maldjian JA, Gur RC, McDonald S,Ragland JD, et al. Brain activity during simulated deception: an event-related functional magnetic resonance study. Neuroimage 2002;15:727e32.

[17] Lee TMC, Liu H-L, Chan CCH, Ng Y-B, Fox PT, Gao J-H. Neural cor-relates of feigned memory impairment. Neuroimage 2005;28:305e13.

[18] Lee TMC, Liu H-L, Tan L-H, Chan CCH, Mahankali S, Feng C-M, et al.Lie detection by functional magnetic resonance imaging. Human BrainMapping 2002;15:157e64.

[19] Meadow R. Munchausen syndrome by proxy: the hinterland of childabuse. Lancet 1977;2:343e5.

[20] Meadow R. Munchausen syndrome by proxy. Archives of Disease inChildhood 1982;57:92e8.

[21] Mele AR. Real self-deception. The Behavioral and Brain Sciences1997;20:91e136.

[22] Nelson HE, O’Connell A. Dementia: the estimation of pre-morbid intel-ligence levels using the national adult reading test. Cortex 1978;14:234e44.

[23] Nunez JM, Casey BJ, Egner T, Hare T, Hirsch J. Intentional false re-sponding shares neural substrates with response conflict and cognitivecontrol. Neuroimage 2005;25:267e77.

[24] Pankratz L. Persistent problems with the Munchausen syndrome byproxy label. The Journal of the American Academy of Psychiatry andthe Law 2006;34:90e5.

[25] Pearson H. Lure of lie detectors spooks ethicists. Nature 2006;441:918e9.

[26] Phan KL, Magalhaes A, Ziemlewicz TJ, Fitzgerald DA, Green C,Smith W. Neural correlates of telling lies: a functional magnetic reso-nance imaging study at 4 Tesla. Academic Radiology 2005;12:164e72.

[27] Rogers R. Diagnostic, explanatory, and detection models of Munchausenby proxy: extrapolation from malingering and deception. Child Abuse &Neglect 2004;28:225e39.

[28] Sanoo M. Abductor sign: a reliable new sign to detect unilateral non-organic paresis of the lower limb. Journal of Neurology, Neurosurgeryand Psychiatry 2004;75:121e5.

[29] Spence SA. Hysterical paralyses as disorders of action. Cognitive Neuro-psychiatry 1999;4:203e26.

[30] Spence SA. The deceptive brain. Journal of the Royal Society of Medi-cine 2004;97:6e9.

[31] Spence SA. Playing devil’s advocate: the case against fMRI lie detection.Legal and Criminological Psychology, in press.

[32] Spence SA, Farrow TFD, Herford AE, Wilkinson ID, Zheng Y,Woodruff PW. Behavioural and functional anatomical correlates of de-ception in humans. Neuroreport 2001;12:2849e53.

[33] Spence S, Farrow T, Leung D, Shah S, Reilly B, Rahman A, et al. Lyingas an executive function. In: Halligan PW, Bass C, Oakley DA, editors.Malingering and illness deception. Oxford: Oxford University Press;2003. p. 255e66.

[34] Spence SA, Hunter MD, Farrow TFD, Green RD, Leung DH, Hughes CJ,et al. A cognitive neurobiological account of deception: evidence fromfunctional neuroimaging. Philosophical Transactions of the Royal Soci-ety of London, Series B 2005;359:1755e62.

[35] Talairach P, Tournoux JA. A stereotactic co-planar atlas of the humanbrain. Stuttgart: Thieme; 1988.

[36] Trivers R. The elements of a scientific theory of self-deception. Annals ofthe New York Academy of Sciences 2000;907:114e31.

[37] Vrij A. Detecting lies and deceit: the psychology of lying and the impli-cations for professional practice. Chichester: Wiley; 2000.

6 S.A. Spence et al. / European Psychiatry xx (2007) 1e6

ARTICLE IN PRESS

Please cite this article in press as: Sean A. Spence et al., ‘Munchausen’s syndrome by proxy’ or a ‘miscarriage of justice’? An initial application of functionalneuroimaging to the question of guilt versus innocence, European Psychiatry (2007), doi:10.1016/j.eurpsy.2007.09.001