fMRI in Patients With Motor Conversion Symptoms and Controls WithSimulated WeaknessJON STONE, PHD, ADAM ZEMAN, MD, ENRICO SIMONOTTO, PHD, MARTIN MEYER, PHD, RAYNA AZUMA, PHD,SUSANNA FLETT, MSC, AND MICHAEL SHARPE, MD

Background: Conversion disorder (motor type) describes weakness that is not due to recognized disease or conscious simulationbut instead is thought to be a “psychogenic” phenomenon. It is a common clinical problem in neurology but its neural correlatesremain poorly understood. Objective: To compare the neural correlates of unilateral functional weakness in conversion disorderwith those in healthy controls asked to simulate unilateral weakness. Methods: Functional magnetic resonance imaging (fMRI) wasused to examine whole brain activations during ankle plantarflexion in four patients with unilateral ankle weakness due toconversion disorder and four healthy controls simulating unilateral weakness. Group data were analyzed separately for patients andcontrols. Results: Both patients and controls activated the motor cortex (paracentral lobule) contralateral to the “weak” limb lessstrongly and more diffusely than the motor cortex contralateral to the normally moving leg. Patients with conversion disorderactivated a network of areas including the putamen and lingual gyri bilaterally, left inferior frontal gyrus, left insula, and deactivatedright middle frontal and orbitofrontal cortices. Controls simulating weakness, but not cases, activated the contralateral supplemen-tary motor area. Conclusions: Unilateral weakness in established conversion disorder is associated with a distinctive pattern ofactivation, which overlaps with but is different from the activation pattern associated with simulated weakness. The overall patternsuggests more complex mental activity in patients with conversion disorder than in controls. Key words: conversion disorder,fMRI, simulation, paralysis, weakness.

BA � Brodmann area; DLPFC � dorsolateral prefrontal cortex;DSM-IV � Diagnostic and Statistical Manual of Diseases, fourthrevision; EPI � echo planar imaging; fMRI � functional magneticresonance imaging; ICD-10 � International Classification of Dis-eases, version 10; MNI � Montreal Neurological Institute; MRI �magnetic resonance imaging; NCS � nerve conduction studies;PET � positron emission tomography; SMA � supplementary mo-tor area; SPECT � single photon emission computed tomography;SPM99 � statistical parametric mapping software.

INTRODUCTION

A pproximately one third of patients seen in neurologyclinics have symptoms that are only somewhat or not at

all explained by disease (1,2). A proportion of these patientshave weakness or paralysis of a limb that cannot be explainedby organic neurological disease. Estimates of the incidence ofthis clinical problem suggest that it is as common as multiplesclerosis (3,4). Functional paralysis of this kind is clinicallyrecognizable by the presence of positive signs of inconsis-tency on examination such as Hoover’s sign (5). In the Diag-nostic and Statistical Manual of Diseases, fourth revision(DSM-IV) psychiatric classification system, symptoms of pa-ralysis unexplained by disease form the basis of the psychi-atric diagnosis of conversion disorder whereas in InternationalClassification of Diseases, version 10 (ICD-10) they fall intothe category of dissociative disorder (motor type).

Several small studies have used functional brain imaging toexamine the neural basis of conversion disorder (6–12) (Table1). This preliminary evidence is conflicting and has not yetprovided clear answers to the following questions: Does con-version disorder have consistent neural correlates? How dothese differ from the neural correlates of deliberately feignedor simulated weakness? At what stage in motor execution isthe failure to produce movement occurring? Is the processprimarily an inhibitory one (8), a failure of initiation (9), orcan a number of neural mechanisms give rise to the sameclinical phenomenon?

This is the first study to apply functional magnetic resonanceimaging (fMRI) to the problem of motor conversion disorder. Weplace our findings in the context of the other published studiesand explore the implications for theories of symptom generationin conversion disorder and for future research.

METHODSSelection of CasesFour right-handed patients with functional weakness were recruited from

a larger prospective study of patients with weakness unexplained by diseasein Southeast Scotland all of whom had undergone assessment by a consultantneurologist and a detailed research psychiatric assessment and life eventsinterview. Four patients were selected who met the following criteria: a)DSM-IV diagnosis of conversion disorder (all patients had positive evidenceof a conversion paralysis, e.g., Hoover’s sign (5) and evidence of psycholog-ical stressors including work related conflict, marital and financial difficul-ties); b) no pain in the limb; c) duration of weakness longer than 9 months;and d) not taking psychotropic medication. We excluded patients with severemajor depression or other severe comorbid psychiatric disorder but not thosewith mild to moderate emotional disorder (generalized anxiety, panic disor-der, or depression according to a structured clinical interview for DSM-IV(13)). The patients varied in the severity of their weakness—two had no anklemovement or sensation in the plantar aspect of the foot. The other two hadvariably reduced power at the affected ankle and reduced sensation over theaffected foot. The duration of weakness and a locally very low misdiagnosisrate (14), as well as evidence from a systematic review (15) suggests a highlevel of confidence in the diagnosis.

Selection of ControlsHealthy controls volunteered from hospital secretarial, medical, and para-

clinical staff, were right handed, and of comparable age and sex to the

From the Division of Clinical Neurosciences (J.S., A.Z.), School of Mo-lecular and Clinical Medicine, University of Edinburgh, Western GeneralHospital, Edinburgh, UK; Division of Psychiatry (E.S., M.S.), School ofMolecular and Clinical Medicine, University of Edinburgh, Royal EdinburghHospital, Morningside Park, Edinburgh, UK; Department of Neuropsychol-ogy (M.M.), Institute for Psychology, University of Zurich, Zurich, Switzer-land; Division of Psychological Medicine (R.A.), Institute of PsychologicalMedicine, Institute of Psychiatry, King’s College, London, UK; Departmentof Theoretical and Applied Linguistics (S.F.), School of Philosophy, Psychol-ogy and Language Sciences, University of Edinburgh, Edinburgh, UK.

Address correspondence and reprint requests to Dr Jon Stone, Dept ClinicalNeurosciences, Western General Hospital, Edinburgh EH4 2XU, UK. E-mail:Jon.Stone@ed.ac.uk

Received for publication January 3, 2007; revision received July 27, 2007.The first two authors contributed equally to the study.DOI: 10.1097/PSY.0b013e31815b6c14

961Psychosomatic Medicine 69:961–969 (2007)0033-3174/07/6909-0961Copyright © 2007 by the American Psychosomatic Society and the American Psychiatric Association

patients. Clinical details of the cases and controls are shown in Table 2. TheLothian research ethics committee provided ethical approval for the study.

Imaging ProcedureIn 2002, all cases and controls had an initial normal structural T1 and T2

weighted MRI of their brain. Before functional imaging, all cases and controlswere given a single training session lasting approximately 30 minutes on anMRI simulator to familiarize them with the paradigm. Subjects lay supine in

the scanner with supportive cushions under their knees and ankles allowingfree movement at the ankle but only very limited movement at the hip or knee.Both arms were strapped with cushioned splints to minimize elbow, wrist, andfinger movements. Instructions were presented on an LCD (liquid crystaldisplay) monitor reflected in a mirror 50 cm from the patients’ eyes. Subjectscould not see their ankles. All subjects wore headphones.

Functional MRI scans were carried out using a combined event-relatedand block design. During the task, subjects were asked to plantarflex one or

TABLE 1. Studies of Functional Brain Imaging in Motor and Sensory Conversion Disorder and Hypnotic Correlates

First Author, Year Subjects Paradigm Activations and Deactivations Seen

Conversion DisorderTiihonen, 1995 (6) 1 subject with left-sided

paralysis and sensorydisturbance

SPECT pre and postsymptoms—stimulation ofleft median nerve

Increased perfusion of right frontal lobe andhypoperfusion in right parietal lobe duringsymptomatic state compared with during recovery

Marshall, 1997 (8) 1 subject withlongstanding left-sided hemiplegia

PET. Attempted movementof leg against resistance

Activation of right anterior cingulate and orbitofrontalcortex

Yazici, 1998 (7) 5 subjects withsymmetrical gaitsymptoms

SPECT Predominantly left sided temporal (4/5) and parietal(2/5) hypoactivation

Spence, 2000 (9) 3 subjects with paralysisof an arm (2 left,1right); 3 controlsfeigning weakness

PET. A hand joystick task ofaffected limb

Hysterical paralysis associated with deactivations inthe left dorsolateral prefrontal cortex duringmovements but not at rest. Feigners showeddeactivation in the right anterior frontal cortexregardless of laterality of weakness

Vuilleumier, 2001 (10) 7 subjects with unilaterall sensory disturbanceand/or weakness

SPECT. Buzzers applied toall four limbs. Fourpatients postrecoveryscans

Reduced cerebral blood flow in the contralateralthalamus and basal ganglia. Less severehypoactivation predicted recovery

Mailis-Gagnon, 2003(11)

4 subjects with chronicpain and“nondermatomal”sensory abnormalities

fMRI. Sensory stimulation When stimulus not perceived–anterior cingulateactivation. Deactivation in somatosensory,prefrontal, inferior frontal, and parietal cortex.Failure to activate thalamus and posterior cingulate

Werring, 2004 (19) 5 patients withmedically unexplainedvisual loss; 7 controlvolunteers

fMRI. 8 Hz visual stimulation Reduced activation of visual cortices, with increasedactivation of left inferior frontal cortex, left insula-claustrum, bilateral striatum and thalami, left limbicstructures, and left posterior cingulate cortex

Burgma, 2006 (18) 4 patients withhemiparesis (3 left, 1right); 7 controls

fMRI. Movementobservation andexecution

No difference between cases and controls inmovement execution. During movementobservation lower or no activation of contralateralmotor cortex

Ghaffar, 2006 (12) 3 subjects with sensoryconversion disorder

fMRI. Unilateral bilateralvibrotactile stimulation

No activation of contralateral primary somatosensorycortex (S1) with unilateral stimulation of affectedside. Normal activation of S1 bilaterally duringbilateral stimulation. Inconsistent activation inanterior cingulate, orbitofrontal cortex, S2,striatum, and thalamus

Kanaan, 2007 (31) 1 subject with right-sided paralysis

fMRI. Auditory probesrelating to prior events

Cued recall of repressed event associated with rightamygdala, right inferior frontal lobe, and rightparietal lobe activity and decreased activity in thecontralateral motor cortex

Hypnotically induced paralysisHalligan, 2000 (23) 1 normal subject with

paralyzed left legPET. Attempted movement

of paralyzed legRight anterior cingulate and right medial orbitofrontal

cortex activation similar to that seen in the singlepatient described by Marshall et al. (8)

Ward, 2003 (21) 12 normal subjectstested with the left leghypnotically paralyzedthen feigned left legparalysis

PET. Attempted movementof paralyzed leg

Hypnotic paralysis: Activations in the putamenbilaterally, left thalamus, right orbitofrontal cortex,left cerebellum, and left supplementary motor area.Feigned paralysis: Activations in the right parietaloperculum, left inferior frontal sulcus, right SMAright ventral premotor cortex, bilateral cerebellum,and left inferior parietal cortex

STONE et al.

962 Psychosomatic Medicine 69:961–969 (2007)

the other ankle repetitively. We chose plantarflexion of the ankle in prefer-ence to knee extension to minimize movement artifact and also becauseweakness of plantarflexion is unusual in neurological disease but common infunctional weakness. The basic task consisted of 20 blocks of 5 trials. At thebeginning of each block the word “PAUSE” was presented (a blank screen forthe first block only) in the center of the screen for 4 seconds. This was toindicate the break between the blocks and give the subjects the chance to rest.Each trial started with the presentation of instruction word “LEFT” or“RIGHT” for 2 seconds, immediately replaced by a fixation cross (“�”) for4 seconds, also in the center of the screen. The subjects were instructed topoint the ankle down while the fixation cross was on the screen. The blocksof five movements on each side (one movement every 6 seconds) wererandomly interleaved over an 11-minute period (e.g., RRRRRLLLLLR-RRRRRRRRRLLLL and so on) resulting in balanced numbers of left andright movement.

Healthy controls were asked to carry out the same task but to simulateweakness of one ankle (side matched to cases). To remind them which anklewas supposed to be “too weak and heavy to move,” the cue word “(heavy)”was added to the instruction (i.e., (heavy)/LEFT or (heavy)/RIGHT).

All four cases reported (and were observed to be) contracting other legmuscles during attempted ankle plantarflexion with zero or minimal anklemovement. They described a combination of mental effort and of a sensationof movement in the affected leg. We therefore instructed the controls toreproduce this combination of mental and physical effort when trying to movethe feigned weak ankle but not to actually make a movement. Cooperationwith this instruction was monitored visually and by debriefing.

After brain imaging, all subjects were debriefed to ask how comfortable theyhad been in the scanner and whether they had understood the instructions.Subjects were asked to describe to what extent they felt they had been movingtheir ankles.

Brain ImagingScanning was performed at the Scottish Higher Education Funding Coun-

cil Brain Imaging Research Centre for Scotland, Western General Hospital,Edinburgh on a 1.5 T General Electrics (General Electric, Milwaukee, WI) in2002. fMRI data were acquired with a gradient echo planar imaging (EPI)sequence (Echo Time (TE) � 40 ms, Repetition Time (TR) � 2.5 s, Field of

Volume (FOV) � 24 � 24 cm2, matrix � 64 � 64, slice thickness � 5 mm,no slice gap, 30 slices per volume).

AnalysisThe data were analyzed with statistical parametric mapping software

(SPM99). Images were reconstructed into “Analyze” format, realigned, nor-malized to the SPM EPI template, and smoothed using a Gaussian filter.Estimates for movement were included. Comparisons were made using amodel that combined a transient and steady regressor. The results weredisplayed on a t-score brain map as either uncorrected (thresholded at p �.001 (two tailed) with a minimum cluster size of 100 to minimize thepossibility of type 1 errors) or corrected data (thresholded at p � .05 butcorrected for multiple comparisons and with an analysis cluster size of zero).For corrected data, tables of the areas with the highest t score and cluster sizesof greater than 10 were prepared using Tailarach coordinates in MontrealNeurological Institute (MNI) space.

We carried out the following analyses: a) Corrected group analysis ofcases and group analysis of controls comparing movement of the weak anklewith that of the normal ankle. For these group analyses data from two patientswith left-sided weakness in each group were left-right flipped to correspondwith the patients with right-sided weakness; b) Uncorrected group analysis asabove provided for the purpose of hypothesis generation; c) Illustrativeuncorrected intra-subject single case comparisons of left ankle movementrelative to right ankle movement. Images were shown with the left hemisphereon the right side of the picture.

RESULTSBehavioral Data

All subjects tolerated the procedure well. All cases reporteda sense of mental effort, which was being variably transmittedtoward the weak leg with a feeling that the “message was notgetting through.” All cases reported “tensing their whole legwith effort” in the appropriate leg. Two patients (Cases 1 and4) were unable to move their ankles at all whereas two patients(Cases 2 and 3) had minimal ankle movement. The controls

TABLE 2. Clinical Characteristics of the Cases and Controls in the Study

Sex AgeDuration ofSymptoms

WeaknessSeveritya

SensorySeveritya

LegAffected

orFeigned

NormalInvestigationsb

Depressionand Anxiety;HADS Scorec

Pain inLimb?

PsychotropicMedication?

ObservedMovementsin Affected

Footd

Motor and sensory conversion symptoms1. Case 1 F 41 30 months Arm�Hip/

Ankle������ Left MRI brain, MRI

Spine. NCSNo; No; 5; 4 No No 0; AA

2. Case 2 F 36 12 months Arm��Hip/Ankle��

�� Left MRI brain D; A; 11; 11 No No 1; AA

3. Case 3 F 41 14 months Arm��Hip/Ankle��

�� Right MRI brain D; A; 10,13 No No 1; AA

4. Case 4 M 31 9 months Arm�Hip��Ankle���

��� Right MRI brain, MRISpine. NCS

D; A; 10, 10 No No 0; AA

Healthy controls5. Control 1 F 40 — ��� — Left MRI brain No; No; 0; 4 No No 0; AA6. Control 2 F 37 — ��� — Left MRI brain No; No; 0; 4 No No 0; AA7. Control 3 F 37 — ��� — Right MRI brain No; No; 3; 7 No No 1; AA8. Control 4 M 32 — ��� — Right MRI brain No; No; 0; 2 No No 1; AA

a For controls, severity indicates simulated weakness severity. � � mild weakness or sensory disturbance; �� � moderate weakness or sensory disturbance;��� � severe weakness or anesthetic.b NCS � nerve conduction studies.c D � DSM-IV diagnosis of “current major depression”; A � DSM-IV diagnosis of “generalized anxiety disorder”; HADS � Hospital Anxiety and DepressionScore – first number is Depression subscore, second number is Anxiety subscore.d 0 � no movements seen; 1 � intermittent movements; 2 � impaired movements; AA � agonist/antagonist contraction reported by subject.

fMRI IN MOTOR CONVERSION DISORDER

963Psychosomatic Medicine 69:961–969 (2007)

reported a similar experience during simulated weakness—that is of mental effort combined with some contraction of legmuscles without resulting movement (3 cases) or with mini-mal movement (1 case) at the ankle. Overall, therefore, therewas little reported or observable difference in the degree ofmovement seen in cases and controls while they were “tensingthe leg” during the task. Movements on the “good” side werenormal in all cases and controls.

Imaging DataGroup Comparisons

Figures 1 and 2, Supplementary Figures 2E to 5E (online) andTables 3 and 4 show the results of the corrected and uncorrectedgroup analyses. Our principle findings are as follows:

1) Reduced and Diffuse Activation of Motor Corticesin Cases and Controls

There was less intense and more diffuse activation contralat-eral to the weak limb in patients with conversion disorder and incontrols simulating weakness than contralateral to the normallymoving limb (in a subsequent task, data available from authors,we demonstrated symmetrical activations in controls movingnormally).

2) Activation of Basal Ganglia, Insula, Lingual Gyri,and Inferior Frontal Cortex in Cases

Cases, but not controls, activated regions of the basal ganglia,insula, lingual gyri, and inferior frontal cortex.

3) Hypoactivation of Middle Frontaland Orbitofrontal Cortex in Cases

There was relative hypoactivation of the right middle frontalgyrus and orbitofrontal cortex in cases but not in controls on

attempted movement of the weak ankle. The uncorrected data inFigure 2 suggests that this hypoactivation is probably bilateral.Our data are also compatible with the alternative interpretationthat these areas are relatively hyperactivated on movement of thenormal ankle. This ambiguity arises because we compared acti-vations associated with right-limb movements with those associ-ated with left-limb movements in the absence of comparisonwith rest.

4) Activation of Contralateral Supplementary MotorArea in Controls Feigning Weakness

Controls feigning weakness, but not cases, activated the con-tralateral (left) supplementary motor area moving the weak anklecompared with moving the normal ankle.

Intrasubject Comparisons

Supplementary Figure 1E with an accompanying descrip-tion (available as a supplementary Web file) illustrates someof the findings of intrasubject comparisons.

DISCUSSIONThis preliminary functional imaging study of motor con-

version disorder reveals both similarities and differences be-tween patients with motor conversion disorder and healthycontrols simulating weakness. We shall discuss our principalfindings in turn, compare our results with those obtained inprevious functional imaging studies (Table 1) and consider theimplications of existing work for definitive future studies.

1) Reduced and Diffuse Activation of Motor Corticesin Cases and Controls

The extent of activation of the motor cortex, Brodmannarea 4, and of the supplementary motor area is known to be

Figure 1. Corrected group analysis of task. Cases with conversion disorder versus controls simulating weakness. Images have been flipped to correspond withweakness of the RIGHT ankle. Activations in RED � areas more active when moving the right leg relative to the left. Activations in BLUE � areas more activewhen moving the left leg relative to the right. The left hemisphere is on the right of each image. Thresholded at p � .05, corrected, k � 0. The images havebeen corrected for multiple comparisons. Significance is indicated by the t-score bars.

STONE et al.

964 Psychosomatic Medicine 69:961–969 (2007)

proportional to the force generated by the correspondingmovement (16). This relationship probably explains the re-duced activation of motor cortex contralateral to the “weak”limb in cases and controls. The more diffuse activations notedcontralateral to the weak limb, which are particularly apparent

in the uncorrected group analyses, may reflect the more wide-spread recruitment of agonists and antagonists in the “weak”leg. This hypothesis is supported by the verbal reports of thecases and controls who described “tensing their legs witheffort” on the weak side while on the normal side onlymuscles required to move the ankle were recruited.

Two previous studies (8,17) found decreased motor cortexactivation contralateral to the paralysis during movement ex-ecution but another study by Burgmer et al. did not (18).Burgmer et al. suggest that in their patients, simultaneousagonist and antagonist contraction represented a failure ofcoordination of movement rather than absence of movement.

TABLE 3. Coordinates and Magnitude of Activations andDeactivations in Cases With Conversion Disorder (Group Data)

Area (Brodmann area, BA) KMNI Coordinate

(x, y, z)Voxel TValue

Areas more active when moving weak right ankle compared withmoving normal left ankle

L Paracentral Lobe (BA 4) 286 �6, �20, 70 7.53R Putamen 42 20, 12, �4 6.65L posterior Insula 36 �38, �24, 20 6.61L anterior insula 11 �24, 20, 14 5.52L Inferior Frontal Gyrus

(pars orbitalis) (BA 47)55 �40, 26, �8 6.52

L Inferior Frontal Gyrus(pars triangularis) (BA 45)

13 �46, 38, �6 5.5014 �44, 16, 6 5.50

L Lingual (BA 18) 37 �12, �78, 2 6.18R Lingual (BA 18) 42 10, �68, 2 6.12L Putamen 53 �26, 14, �2 6.12R Inferior frontal gyrus (pars

orbitalis)14 40, 20, �10 5.73

L Superior ParietalLobule/Precuneus (BA 7)

11 �14, �60, 62 5.47

Areas more active when moving normal left ankle compared withmoving weak right ankle

R Paracentral Lobe (BA 4) 362 8, �44, 68 10.46R middle frontal gyrus (BA 46) 12 48, 46, 18 5.97R middle frontal gyrus (BA 10) 12 38, 62, 2 5.67R orbitofrontal gyrus (BA 10/11) 20 12, 50, �6 5.40

Thresholded at p � .05, corrected, k � 10. Data for patients with weak leftankle has been “flipped.”MNI � Montreal Neurological Institute.

TABLE 4. Coordinates and Magnitude of Activations andDeactivations in Controls Simulating Weakness (Group Data)

Area KMNI Coordinate

(x, y, z)Voxel

T Value

Areas more active when moving weak right ankle comparedwith moving normal left ankle

L Paracentral Lobe (BA 4) 248 �8, �30, 66 8.88R Cerebellum 113 34, �52, �32 7.28L Supplementary Motor

Area (BA 6)15 �10, 4, 54 5.67

Areas more active when moving normal left ankle comparedwith moving weak right ankle

R Paracentral Lobe (BA 4) 885 6, �32, 68 14.30L Cerebellum 96 �12, �44, �28 7.00R Parahippocampal (BA 29) 55 6, �46, 8 6.49R mid occipital gyri (BA 19/39) 26 46, �76, 32 5.96L Superior Parietal Lobule/

Precuneus (BA 7)18 �6, �62, 36 5.28

Thresholded at p � .05, corrected, k � 10. Data for patients with weak leftankle has been “flipped.”MNI � Montreal Neurological Institute.

Figure 2. Uncorrected group analysis of task. Cases with conversion disorder versus controls simulating weakness. Images have been flipped to correspond withweakness of the RIGHT ankle. Activations in RED � areas more active when moving the right leg relative to the left. Activations in BLUE � areas more active whenmoving the left leg relative to the right. The left hemisphere is on the right of each image. Thresholded at p � .001, uncorrected, k � 100. Significance is indicated bythe t-score bars.

fMRI IN MOTOR CONVERSION DISORDER

965Psychosomatic Medicine 69:961–969 (2007)

They put forward their finding of reduced or absent contralat-eral motor cortex activation in their four patients during move-ment observation as support for this hypothesis.

2) Activation of Basal Ganglia, Insula, Lingual Gyri,and Inferior Frontal Cortex In Cases

Overall, cases were distinguished from controls by thepresence of a more complex set of activations. Cases showedmore activation when trying to move their weak ankle com-pared with their “good” ankle, bilaterally in the putamen andlingual gyri and in the left inferior frontal cortex and leftinsula. Although we refer to “activations” here, the design ofour experiment allows us to be confident only of relativedifferences between activations associated with movement ofthe normal and affected limbs. In other words, the “activa-tions” just detailed, associated with movement of the weaklimb, could theoretically reflect areas of hypoactivation asso-ciated with movement of the normal limb (although this seemsless likely).

These regions have been implicated in other work onconversion disorder and hypnosis, a possible model for con-version disorder. Werring et al. (19) have reported findingsrather similar to ours in the context of visual loss unexplainedby disease: compared with controls, during visual stimulation,patients showed increased activation in the left inferior frontalcortex, left insula-claustrum, and bilaterally in the striatum.Werring et al. (19) draw attention to the comparable activationof these areas in Sahraie’s fMRI study of blindsight (20),associated with the “unaware” mode of visual processing (20).Relating these brain areas to function is fraught with diffi-culty, but the inferior frontal cortex has an executive role inselecting, comparing, or deciding on information in short-termmemory. It is therefore likely to be activated when subjects areespecially mindful of their task. Ward et al. (21), studyingvolunteers with subjectively experienced paralysis (inducedby hypnosis), found the strongest activations bilaterally in theputamen. These activations were ascribed to the role of theseand other areas in movement preparation. It is interesting inour study that the cases but not controls had activation in theseareas. This may be evidence of genuine movement preparationin our patient group. In the same study by Ward et al. (21) leftinferior frontal activation was seen in patients with feignedparalysis but not in those with subjectively experienced pa-ralysis. This contrasts with our finding of left inferior frontalactivation in cases, but not in controls deliberately feigningweakness.

Vuilleumier et al. (10), using single photon emission com-puted tomography (SPECT), detected a decrease in regionalblood flow in the caudate and putamen contralateral to theweak limb in subjects with recent onset of “hysterical” weak-ness: this resolved as they recovered. Their “within subject,”“before and after” design together with evidence of a corre-lation between the degree of hypoactivation and the degree ofrecovery provides the most compelling data yet published inthis area.

Other evidence links activation of the insula and lingualgyri to attentional and effortful emotional processing. Forexample, Critchley et al. found that activity in these areascovaried with the sympathetic skin conductance response in-duced by rewards and punishments in a decision-making task(22). The insular activation was attributed to its role in medi-ating cardiovascular sympathetic arousal while the lingualactivation was attributed to enhancement of activity in extra-striate visual areas when subjects process visual cues in statesof high arousal.

Finally, the finding of activation of the superior parietalregion/precuneus in cases may once again reflect increasedmental effort or the greater emotional demands of the task inthe cases. These brain regions tend to be activated along withfrontal regions in working memory and other tasks demandingattention. Their role in imagery and spatial attention could berelevant to patients struggling with movement of a weak limb.

Thus, our findings contribute to the evidence that a networkof areas including the insula, inferior frontal cortex, the basalganglia, and the lingual gyri is implicated in motor conversiondisorder, though whether these activations reflect causes, con-sequences or compensatory mechanisms, and the relative im-portance of motor, cognitive, and emotional processes, arecurrently unclear.

3) Hypoactivation of Middle Frontaland Orbitofrontal Cortex in Cases

Figure 2 and Table 3 suggest that regions of the rightmiddle frontal and orbitofrontal cortex were underactive dur-ing movement of the weak ankle in the cases with conversiondisorder. Inspection of the uncorrected data in Figure 4Esuggests that this hypoactivation is probably bilateral. Onceagain, our design leaves open an alternative interpretation interms of relative hyperactivation associated with attemptedmovement of the normal ankle. The initial interpretation ismore plausible, however, as these areas are not normallyactivated by actual or intended movement. The orbitofrontalarea we have identified is close to the area identified byMarshall et al. in their study of a patient with hystericalparalysis (8); however, in Marshall’s patient this area wasactivated on attempted movement of the paralyzed limb. Theauthors suggested that this area was the locus of “unconsciousinhibition” of movement.

4) Activation of Contralateral Supplementary MotorArea (SMA) in Controls Feigning Weakness

The SMA has an important role in the selection, prepara-tion, and sequencing of voluntary real and imagined move-ments. States of SMA hypoactivity have been associated withimpairment in initiation of voluntary movement (for example,Parkinson’s disease). Its contralateral activation in controlsfeigning weakness suggests an excess of movement-planningactivity in comparison to movement of the normal limb. Incontrast, the absence of a similar activation in cases withconversion disorder may suggest impairment of voluntarymovement planning that is perhaps less emotionally neutral.

STONE et al.

966 Psychosomatic Medicine 69:961–969 (2007)

Interestingly, contralateral SMA was also activated in patientswith feigned paralysis, but not hypnotic subjective paralysis,in the study by Ward et al. (21).

Attempting a Synthesis of Previous Studies

The studies summarized in Table 2 have examined theneural correlates of conversion disorder. No wholly consistentfindings have yet emerged. The current data are compatiblewith two competing hypotheses: a) that the primary neuralmechanism of conversion disorder is excessive inhibition ofnormal movement; and b) that there is a failure to activatemovement normally.

The first hypothesis, suggested initially by Marshall et al.(8) and supported by their subsequent data from a singlepatient with hypnotic paralysis (23), suggests that frontal andcingulate activations inhibit movement of the weak limb.Findings suggesting that the underlying problem is increasedbrain activity are appealing because they are in keeping withclinical observations of transient improvement during sedation(24), hypnosis (25), or sleep (26) repeatedly described inpatients with conversion disorder. Activation of these frontalareas might also be in keeping with a resting state of relativethalamic and basal ganglia hypoactivation as seen in the studyby Vuilleumier et al. (10). Hypoactivation of the contralateralthalamus during symptomatic states is intriguing because itmight explain why so many patients with unilateral conver-sion symptoms feel “split down the middle.” Similar contralat-eral thalamic hypoperfusion has also been shown in patientswith complex regional pain (27) who often have nonderma-tomal (28) and placebo-responsive sensory abnormalities (29).

In keeping with the second hypothesis, of failure to activatemovement normally, Spence et al. (9) found hypoactivation ofleft dorsolateral prefrontal cortex (DLPFC) in three subjectswith hysterical paralysis, while subjects feigning weaknessshowed right DLPFC deactivation, regardless of the side ofthe affected limb. These findings were interpreted as evidenceof functional pathology in a region known to be active innormal individuals choosing actions and to be dysfunctional inpatients with disorders of volition such as schizophrenia.

More recent studies suggest, perhaps unsurprisingly, thatthings might be more complex than these earlier studies indi-cated. Mailis-Gagnon et al. (11) studied patients with “func-tional” sensory loss associated with chronic pain. They founda combination of both activations (notably in the anteriorcingulate and prefrontal cortex (Brodmann area 10)) and de-activations (in primary somatosensory cortex, other areas ofthe prefrontal cortex (Brodmann area 9) and inferior frontalgyrus). A variety of possible plausible explanations wereproposed, including functional deafferentation, attention topain, and artifactual deactivation caused by tonically raisedbaseline levels of activity. The studies by Werring et al. (19)and Ward et al. (21), summarized in Table 2, have alsoimplicated a network of areas, particularly prefrontal, cingu-lated, and thalamic, which overlaps with those just mentionedand with those identified in this study. The study by Ghaffaret al. (12) describes the absence of contralateral primary

somatosensory cortex activation in three patients with sensoryconversion symptoms, even though this was activated nor-mally when a bilateral stimulus is applied. This study alsoimplicates a network of regions including the anterior cingu-late, orbitofrontal cortex, thalamus, and striatum, although theresults were inconsistent across patients.

Limitations of This Study

Our own study has shortcomings in the number of patientsand controls studied, the use of patients with weakness ofdifferent sides (creating a need to “flip” images to obtaingroup data), and the lack of a rest condition. The third limi-tation implies that although we can implicate certain brainregions in functional weakness, we remain uncertain whetherthey were hyperactive or hypoactive. The design of our studydid not provide for direct statistical comparison of patient andcontrol activations, a desirable analysis in future work.

There was some heterogeneity in the subjects. First, co-morbid anxiety or depression are very common in motorconversion disorder: three of our cases had such comorbidity.We felt that exclusion of patients with mild to moderateemotional disorder would lead to the selection of a group ofpatients unrepresentative of motor conversion disorder. Sec-ond, the severity of weakness also varied among our cases:two were able to make reduced but visible movements withtheir feet whereas two others were not able to move theirankles at all (although all could move their legs to somedegree). However, as both cases and controls reported a sensationof tensing their leg while carrying out the task on the affectedside, it is unlikely that the differences in imaging findings relateto differences in actual movement. Third, there is always thepossibility that the group findings have been influenced by largeactivations or deactivations within one subject. This does notseem to be the case looking at the base data but cannot be ruledout. Finally, in comparing brain activation of moving the “weak”side with activations of moving the “normal” side in patients withconversion disorder, we are making an assumption that theirnormal leg is indeed normal.

The Difficulties of Functional Imaging of MotorConversion Disorder

Further work is clearly required to resolve the conflictingfindings from the studies published to date. However, futureresearch on these phenomena must address several challenges.First, the concept of conversion disorder is itself controversial(30). Although patients can be diagnosed with relative reli-ability as having a symptom such as paralysis that is notexplained by disease (15), there is considerable debate aboutthe psychological mechanisms underlying these symptoms.The psychodynamic “conversion” hypothesis remains domi-nant in DSM-IV, whereas in ICD-10 the hypothesis that thesymptoms are a consequence of dissociation is given primacy.Others have considered patients with motor conversion disor-der to have a “disorder of action” (17). These are not mutuallyexclusive hypotheses but one has to consider the possibilitythat patients who share the same outward symptoms and signs

fMRI IN MOTOR CONVERSION DISORDER

967Psychosomatic Medicine 69:961–969 (2007)

of paralysis could have markedly different psychologicalmechanisms. This not only makes it difficult to know what thestarting point for a functional imaging hypothesis should bebut also raises the possibility that individual differences inpsychological mechanisms could be obscured in the analysisof group data. Second, patients with paralysis due to conver-sion disorder may vary from one another in factors includingthe duration and severity of symptoms, the co-existence ofmood disturbance, the use of psychotropic drugs, the presenceof pain and other sensory symptoms, the degree of unfitness,and probably their degree of conscious control over the symp-tom. Third, control subjects, asked to imagine or simulateweakness, may adopt different strategies to achieve this, eitherattempting to eliminate effort to move altogether (“imagineyour leg has become so weak you cannot move a muscle . . .”) orattempting to move against insuperable, imaginary, resistance(“imagine that however hard you try you cannot raise yourleg . . .”). For either of these strategies, controls may regard theirtask as being merely to imagine weakness or as a challenge todeceive an observer. The permutations of these differing strate-gies are likely to give rise to different patterns of activation. It isplausible that the neural mechanisms of weakness in patients withmotor conversion disorder are similarly heterogenous. In somepatients, the underlying process may be negative, a failure toinitiate movement, whereas in others, positive inhibitory pro-cesses may be to the fore. Detailed individual case studies com-paring patterns of activations on several occasions during thesymptomatic state (to establish intrasubject reliability) and afterrecovery (to control for individual differences) offer one ap-proach to meeting these difficulties.

CONCLUSIONSThis study suggests that fMRI offers an informative ap-

proach to studying the neural basis of conversion disorder. Wefound consistent reductions in the activation of the motorcortex in cases with conversion disorder and control subjectssimulating weakness. Cases but not controls showed activa-tions in the basal ganglia, insula, lingual gyri, and inferiorfrontal cortex in association with movement of the weak limb.We tentatively interpret these activations as evidence that thecases were attempting to move with greater resulting mentaleffort than occurred in controls. Failure to activate supplemen-tary motor areas and deactivation of middle frontal and or-bitofrontal cortex suggests disorganization in the executivecontrol in movement. It would be going too far on the basis ofthe data to suggest that fMRI can enable us to tell whether apatient with conversion disorder is feigning, but it is certainlyintriguing that overall there was evidence of a more intenseand complex pattern of activation in the patients than in thecontrols who were feigning weakness.

Future work should address the many potential sources ofvariability in this patient group and explore the likely heteroge-neity in the neural mechanisms of motor conversion disorder.

The Chief Scientist’s Office (Scotland) funded this work. The re-search was conducted at the SHEFC Brain Imaging Research Centrefor Scotland, which is supported with a Joint Research Equipment

Initiative grant from the UK’s Medical Research Council and theScottish Higher Education Funding Council with industrial collab-oration from GE Medical Systems, Boehringer Ingelheim, Novartis,and Schering. A.Z. was supported by The Health Foundation (pre-viously The PPP Foundation). We thank Charles Warlow, Ian Deary,Steven Laureys, Elvina Goutona, Ian Marshall, and Joanna Wardlawfor contributions.

REFERENCES1. Carson AJ, Ringbauer B, Stone J, McKenzie L, Warlow C, Sharpe M. Do

medically unexplained symptoms matter? A prospective cohort study of300 new referrals to neurology outpatient clinics. J Neurol NeurosurgPsychiatry 2000;68:207–10.

2. Nimnuan C, Hotopf M, Wessely S. Medically unexplained symptoms: anepidemiological study in seven specialities. J Psychosom Res 2001;51:361–67.

3. Binzer M, Andersen PM, Kullgren G. Clinical characteristics of patientswith motor disability due to conversion disorder: a prospective controlgroup study. J Neurol Neurosurg Psychiatry 1997;63:83–8.

4. Rief W, Hessel A, Braehler E. Somatization symptoms and hypochon-driacal features in the general population. Psychosom Med 2001;63:595–602.

5. Stone J, Zeman A, Sharpe M. Functional weakness and sensory distur-bance. J Neurol Neurosurg Psychiatry 2002;73:241–5.

6. Tiihonen J, Kuikka J, Viinamaki H, Lehtonen J, Partanen J. Alteredcerebral blood flow during hysterical paresthesia. Biol Psychiatry 1995;37:134–5.

7. Yazici KM, Kostakoglu L. Cerebral blood flow changes in patients withconversion disorder. Psychiatry Res 1998;83:163–8.

8. Marshall JC, Halligan PW, Fink GR, Wade DT, Frackowiak RS. Thefunctional anatomy of a hysterical paralysis. Cognition 1997;64:B1–B8.

9. Spence SA, Crimlisk HL, Cope H, Ron MA, Grasby PM. Discreteneurophysiological correlates in prefrontal cortex during hysterical andfeigned disorder of movement. Lancet 2000;355:1243–4.

10. Vuilleumier P, Chicherio C, Assal F, Schwartz S, Skusman D, Landis T.Functional neuroanatomical correlates of hysterical sensorimotor loss.Brain 2001;124:1077–90.

11. Mailis-Gagnon A, Giannoylis I, Downar J, Kwan CL, Mikulis DJ,Crawley AP, Nicholson K, Davis KD. Altered central somatosensoryprocessing in chronic pain patients with “hysterical” anesthesia. Neurol-ogy 2003;60:1501–7.

12. Ghaffar O, Staines WR, Feinstein A. Unexplained neurologic symptoms:an fMRI study of sensory conversion disorder. Neurology 2006;67:2036–8.

13. First MB, Spitzer RL, Gibbon M, Williams JBW. Structured ClinicalInterview for DSM-IV-TR Axis I Disorders, Research Version, PatientEdition (SCID-I/P). New York: Biometrics Research, New York StatePsychiatric Institute; 2001.

14. Stone J, Sharpe M, Rothwell PM, Warlow CP. The 12 year prognosis ofunilateral functional weakness and sensory disturbance. J Neurol Neuro-surg Psychiatry 2003;74:591–6.

15. Stone J, Smyth R, Carson A, Lewis S, Prescott R, Warlow C, Sharpe M.Systematic review of misdiagnosis of conversion symptoms and “hyste-ria”. BMJ 2005;331:989.

16. Passingham RE. The Frontal Lobes and Voluntary Action. Oxford: Ox-ford University Press; 1993.

17. Spence SA. Hysterical paralyses as disorders of action. Cognit Neuro-psychiatry 1999;4:203–26.

18. Burgmer M, Konrad C, Jansen A, Kugel H, Sommer J, Heindel W,Ringelstein EB, Heuft G, Knecht S. Abnormal brain activation duringmovement observation in patients with conversion paralysis. Neuroimage2006;29:1336–43.

19. Werring DJ, Weston L, Bullmore ET, Plant GT, Ron MA. Functionalmagnetic resonance imaging of the cerebral response to visual stimulationin medically unexplained visual loss. Psychol Med 2004;34:583–9.

20. Sahraie A, Weiskrantz L, Barbur JL, Simmons A, Williams SC, BrammerMJ. Pattern of neuronal activity associated with conscious and uncon-scious processing of visual signals. Proc Natl Acad Sci U S A 1997;94:9406–11.

21. Ward NS, Oakley DA, Frackowiak RS, Halligan PW. Differential brainactivations during intentionally simulated and subjectively experiencedparalysis. Cognit Neuropsychiatry 2003;8:295–312.

STONE et al.

968 Psychosomatic Medicine 69:961–969 (2007)

22. Critchley HD, Elliott R, Mathias CJ, Dolan RJ. Neural activity relating togeneration and representation of galvanic skin conductance responses: afunctional magnetic resonance imaging study. J Neurosci 2000;20:3033–40.

23. Halligan PW, Athwal BS, Oakley DA, Frackowiak RS. Imaging hyp-notic paralysis: implications for conversion hysteria. Lancet 2000;355:986 –7.

24. White A, Corbin DO, Coope B. The use of thiopentone in the treatmentof non-organic locomotor disorders. J Psychosom Res 1988;32:249–53.

25. Oakley DA. Hypnosis and conversion hysteria: a unifying model. CognitNeuropsychiatry 1999;4:243–65.

26. Lauerma H. Nocturnal limb movements in conversion paralysis. J NervMent Dis 1993;181:707–8.

27. Fukumoto M, Ushida T, Zinchuk VS, Yamamoto H, Yoshida S. Con-

tralateral thalamic perfusion in patients with reflex sympathetic dystrophysyndrome. Lancet 1999;354:1790–1.

28. Rommel O, Gehling M, Dertwinkel R, Witscher K, Zenz M, Malin JP,Janig W. Hemisensory impairment in patients with complex regional painsyndrome. Pain 1999;80:95–101.

29. Verdugo RJ, Ochoa JL. Reversal of hypoaesthesia by nerve block, orplacebo: a psychologically mediated sign in chronic pseudoneuropathicpain patients. J Neurol Neurosurg Psychiatry 1998;65:196–203.

30. Halligan P, Bass C, Marshall JC. Contemporary approaches to the scienceof hysteria: clinical and theoretical perspectives. Oxford: Oxford Univer-sity Press; 2001.

31. Kanaan RA, Craig TK, Wessely SC, David AS. Imaging repressedmemories in motor conversion disorder. Psychosom Med 2007;69:202–5.

fMRI IN MOTOR CONVERSION DISORDER

969Psychosomatic Medicine 69:961–969 (2007)