Localisation of increased prefrontal white matter in pathological liars

We examined white matter volumes in four prefrontal subregionsusing structural magnetic resonance imagingin 10 pathologicalliars,14 antisocial controls, and 20 normal controls. Liars showeda relatively widespread increase in white matter (23–36%)in orbitofrontal, middle and inferior, but not superior, frontalgyri compared with antisocial and normal controls. This whitematter increase may predispose some individuals to pathological lying.

INTRODUCTION

TOP

INTRODUCTION

Using a symptom-based approach, we found that pathological liarshave abnormally increased prefrontal white matter (Yang et al, 2005). It was suggested that this increase may represent a predispositionto lying. In this study it is hypothesised that pathologicalliars will show structural abnormalities particularly in thedorsolateral, prefrontal and orbitofrontal cortex.

METHOD

TOP

METHOD

Participants were taken from a total sample of 108 communityvolunteers drawn from five temporary employment agencies inLos Angeles (Raine et al, 2000). The three groups consistedof 10 people with a history of repeated lying (‘liars’),20 normal controls who had neither antisocial personality disordernor a history of pathological lying, and 14 ‘antisocial’controls matched for antisocial behaviours but with no historyof pathological lying. Participants were defined as pathological liars if they fulfilled criteria for: pathological lying onthe Psychopathy Checklist–Revised (PCL–R; Hare, 1991);or conning/manipulative behaviour on the PCL–R; or deceitfulnessfor DSM–IV antisocial personality disorder (American Psychiatric Association, 1994),or malingering (for details see Yang et al, 2005). Of the10 liars in this study, 5 were classified as malingerers. Fullinformed, written consent was obtained from all participantsin accordance with institutional review board procedures. Fivebrain volumes from the original sample (Yang et al, 2005)could not be segmented owing to irretrievable corruption on data storage. Missing data were relatively evenly distributedacross groups, with two from the liar group, two from the antisocialcontrol group and one from the normal control group.

Structural MRI was carried out on a 1.5-Telsa Philips S15/ACS(Selton, Connecticut, USA) scanner using three-dimensionalT₁-weighted gradient-echo scans (for detals see Yang et al, 2005).All image data-sets were processed with a series of preparatorysteps before manual delineation of prefrontal subregions (Sowellet al, 1999, 2002). First, all images were anonymised to excludepersonal information. Second, non-brain tissue and the cerebellumwere removed from the brain volume, and signal intensity inhomogeneitieswere corrected (Sled & Pike, 1998). Third, fully automatedtissue segmentation was applied and brain voxels were automaticallyclassified as gray matter, white matter, or cerebrospinal fluidusing a validated partial volume correction method (Shattuck et al, 2001).Finally, a spherical mesh surface was created using a three-dimensionalactive surface algorithm to facilitate identification of anatomicalboundaries (MacDonald et al, 1994).

The parcellation of the prefrontal lobe into four subregionsfor each hemisphere followed the methods of Ballmaier et al (2004). A three-dimensional shape representation and coronaltwo-dimensional MRI scan of the segmentation of the prefrontalcortex of one of the participants are shown in the data supplementto the online version of this paper. All anatomical delineations were conducted by two research assistants trained by Y.Y. Unlikegray matter subregions, which are clearly defined by sulcallandmarks, white matter delineations are arbitrary and thesegmentation results should be viewed as estimated volumes.To assess interrater reliability, all anatomical regions weredelineated on ten randomly chosen image data-sets; intraclasscorrelation coefficients ranged between 0.90 and 0.97 for graymatter and white matter in all four frontal subregions. Eachof the eight subregional volumes was divided by total intracranialvolume to account for potential differences in individual brainsize. Since there was a lack of hemisphere effect, white mattervolumes from two hemispheres were averaged to create a meanregional volume.

RESULTS

TOP

RESULTS

A multivariate analysis of variance (ANOVA) showed a main groupeffect for whole-brain-corrected white matter volume in prefrontalregions (i.e. inferior frontal, middle frontal, orbitofrontaland superior frontal cortices); F(8, 78) 4.19, P=0.001, ${eta}$ ²=0.29).Groups differed in the volume of white matter in the inferior(F(2, 41)=11.09, P=0.001), middle (F(2, 41)=7.05, P=0.003)and orbitofrontal cortex (F(2, 41)=6.87, P=0.001), with increasedwhite matter in liars. However, a trend towards lower whitematter volume was found in the superior frontal cortices forliars (F(2, 41)=0.42, P=0.66). Liars showed significantly increasedwhite matter in inferior, middle and orbitofrontal cortex comparedwith both antisocial controls (P=0.001, P=0.004, and P=0.006,respectively) and normal controls (P=0.001, P=0.005, and P=0.001respectively; Fig. 1). No difference was found for gray mattervolume in the four subregions (F(8, 78) =0.54, P=0.82).

View larger version (15K):
[in this window]
[in a new window]
[as a PowerPoint slide]
Fig. 1 Whole-brain-corrected prefrontal white matter volumes in the inferior frontal cortex (IFC), middle frontal cortex (MFC), orbitofrontal cortex (OFC), and superior frontal cortex (SFC) in pathological liars ( ${blacksquare}$ ), normal controls ( ${square}$ ) and antisocial controls (APD, $blk12$ ). Vertical lines indicate the standard error of the mean.

DISCUSSION

TOP

DISCUSSION

Following our previous finding of a prefrontal white matterincrease in people who lie, cheat and manipulate others (Yang et al, 2005), this study found pathological liars to have increased whitematter volumes in some prefrontal subregions, particularlyorbitofrontal cortex (22–26% increase), inferior frontalcortex (32–36% increase) and middle frontal cortex (28–32%increase) compared with both antisocial and normal controls.An important exception was that no white matter increase wasfound for the superior frontal cortices. Such an increase mightbe expected based on findings of an fMRI study in which activationof superior frontal cortices was found during a deception taskinvolving motor responses (Langleben et al, 2002). In contrast,one study using a potentially more realistic lying task involvinga verbal response found prefrontal activation specificallyin ventrolateral and orbitofrontal cortex, but not superior frontal cortices (Spence et al, 2004). Moreover, these nonsuperiorfrontal regions are most frequently shown to be activated bydeception tasks (e.g. Spence et al, 2004; Langleben et al, 2005; Phan et al, 2005). This may in part explain why we observedwhite matter increases in the ventral (orbitofrontal cortex),ventrolateral (inferior frontal cortex) and inferior aspectof dorsolateral (middle frontal cortex), but not superior dorsolateral (superior frontal cortices), frontal regions in the liar group.

One interpretation of the white matter increases in the ventraland lateral nonsuperior frontal regions could be that a pre-existingvariation in prefrontal structure may predispose individualsto engage in pathological lying. Alternatively, several studieshave argued that long-term training may induce regional increasesin white matter volume (Schmithorst & Wilke, 2002; Bengtsson et al, 2005).In the case of lying, it is conceivable that excessive lyingrepeatedly activates the prefrontal circuit underlying lying,resulting in permanent changes in brain morphology. This ‘Pinocchio’s nose’ hypothesis of pathological lying could be comparedwith the competing predispositional hypothesis using a prospectivelongitudinal study assessing both white matter volume and degreeof lying from childhood to adulthood.

The engagement of ventral and lateral prefrontal regions inlying may be anticipated from fMRI studies, several of whichhave associated these regions with executive functions crucialto successful deception, including decision-making, moral reasoning,rule maintenance/retrieval and response inhibition (Bunge, 2004). Although some studies showed partial activation in the superiorfrontal cortex when lying involved a non-vocal motor response(Langleben et al, 2002, 2005), this region is more associatedwith functions less directly linked to deception, such as spatial information processing, attention reorientation and noveltydetection (Gomot et al, 2006). Conversely, gains in whitematter volume in these prefrontal regions (in the absence ofgray matter reduction) may lead to faster sharing of information within frontal–cortical circuits in pathological liars.Thus, increased white matter in these subregions of the prefrontalcortex in liars may predispose to maintaining a lifestyle ofpathological lying and malingering. The use of advanced imagingtechniques such as diffusion tensor imaging to assess neuralconnectivity (Nakamura et al, 2005) may allow more thoroughinvestigation of the subtle abnormalities responsible for pathologicallying.

ACKNOWLEDGMENTS

TOP

ACKNOWLEDGMENTS

We thank Samantha Henry, Elizabeth Culley, Donna Kha, ReimarMacaranas, Henry Wu, Lydia Lee, Sum-yan Ng and Sridhar Chadalavadafor data collection and scoring. This study was supported bygrants to A.R. from NIMH and the Wacker Foundation, grantsfrom the National Center for Research Resources the NIH Roadmapfor Medical Research and a NARSAD Young Investigator Award for K.L.R. and A.W.T.

REFERENCES

TOP