Response inhibition and impulsivity: an fMRI study

Neuropsychologia 41 (2003) 1959–1966

Response inhibition and impulsivity: an fMRI study N.R. Horn a,∗ , M. Dolan a , R. Elliott a , J.F.W. Deakin a , P.W.R. Woodruff a,b b

a Neuroscience and Psychiatry Unit, University of Manchester, Stopford Building, Oxford Road, Manchester M13 9PT, UK SCANlab, Academic Department of Psychiatry, The Longley Centre, University of Sheffield, Norwood Grange Road, Sheffield S5 7JT, UK

Received 14 August 2001; received in revised form 6 March 2003; accepted 11 March 2003

Abstract Aggressive, suicidal and violent behaviour have been associated with impulsive personality and difficulty in inhibiting responses. We used functional magnetic resonance imaging (fMRI) of the whole brain to examine the neural correlates of response inhibition in 19 normal subjects as they performed a Go/NoGo task. Subjects completed Eysenck’s Impulsivity Scale, Barratt’s Impulsivity Scale (BIS) and behavioural impulsivity tasks. Associations between blood oxygen level dependent (BOLD) response, trait impulsivity, task performance and National Adult Reading Test (NART) IQ were investigated. Neural response during response inhibition was most prominent in the right lateral orbitofrontal cortex. Responses were also seen in superior temporal gyrus, medial orbitofrontal cortex, cingulate gyrus, and inferior parietal lobule, predominantly on the right side. Subjects with greater scores on impulsivity scales and who made more errors had greater activation of paralimbic areas during response inhibition, while less impulsive individuals and those with least errors activated higher order association areas. Exploratory factor analysis of orbital activations, personality measures and errors of commission did not reveal a unitary dimension of impulsivity. However, the strong association between posterior orbital activation and Eysenck’s impulsivity score on a single factor suggests that greater engagement of right orbitofrontal cortex was needed to maintain behavioural inhibition in impulsive individuals. Lower IQ was more important than impulsivity scores in determining errors of commission during the task. Neuroimaging of brain activity during the Go/NoGo task may be useful in understanding the functional neuroanatomy and associated neurochemistry of response inhibition. It may also allow study of the effects of physical and psychological interventions on response inhibition in clinical conditions such as antisocial personality disorder. © 2003 Elsevier Ltd. All rights reserved. Keywords: Go/NoGo; Impulsivity; Functional magnetic resonance imaging; Orbitofrontal cortex; Frontal lobe; Antisocial personality disorder

1. Introduction Impulsivity is a multidimensional concept that incorporates failure of response inhibition, rapid processing of information, novelty seeking, and inability to delay gratification (Barratt, 1985, 1994). Impulsivity is one of the defining characteristics of a number of psychiatric diagnoses, particularly borderline and antisocial personality disorders (Stein, Hollander, & Liebowit 1995; Stein, Towney, & Hollander, 1995). Poor impulse control correlates significantly with suicidal, violent and aggressive behaviour (Plutchik & Van Praag, 1989, 1995) and is an increasingly important aspect of risk assessment in a variety of clinical situations, including assessment of dangerousness (Monahan et al., 2000). Most attempts to measure impulsivity rely on psychometric self-report trait measures. Some ∗ Corresponding author. Department of Psychiatry, Royal Liverpool University Hospital, Liverpool L69 3GA, UK. Tel.: +44-151-7065151; fax: +44-151-7093765. E-mail address: [email protected] (N.R. Horn).

0028-3932/$ – see front matter © 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0028-3932(03)00077-0

behavioural impulsivity tasks have been developed, measuring preference for a smaller more immediate reward over a delayed larger reward, and impaired motor inhibition, however, psychometric and behavioural impulsivity measures do not correlate well with each other (Barratt & Patton, 1983; Barratt, Stanford, Kent, & Felthous, 1997). Behavioural impulsivity tasks tend to have low test–retest reliability, apart from the Go/NoGo task, which has reasonable temporal stability (Kindlon, Mezzacappa, & Earls, 1995). Impulsivity is a feature of damage to the frontal lobe and an “acquired sociopathic” syndrome has been described following ventromedial frontal lobe lesions (Damasio, Tranel, & Damasio, 1990; Grafman et al., 1996; Paradiso, Chemerinski, Yazici, Tartaro, & Robinson, 1999). This has lead to suggestions that impaired ventromedial frontal lobe function may contribute to poor impulse control in antisocial personality disorders (Damasio, 2000). In support of this notion a variety of neuropsychological deficits have been reported in antisocial populations (Morgan & Lilienfeld, 2000). Neuroimaging studies in this population report a reduction in prefrontal metabolism (Raine, Buchsbaum,

1960

N.R. Horn et al. / Neuropsychologia 41 (2003) 1959–1966

& LaCasse, 1997), and reduced prefrontal grey matter volumes (Raine, Lencz, Bihrle, La Casse, & Colletti, 2000). Positron emission tomography (PET) (Kawashima et al., 1996; Krams, Rushworth, Deiber, Frackowiak, & Passingham, 1998; Nobre, Coull, Frith, & Mesulam, 1999) and functional Magnetic Resonance Imaging (fMRI) studies (Casey et al., 1997; Garavan, Ross, & Stein, 1999; Hager et al., 1998; Konishi, Nakajima, Uchida, Sekihara, & Miyashita, 1998) of response inhibition and processing novel stimuli suggest a role for the prefrontal cortex, especially the right lateral frontal cortex, and a network of associated regions in response inhibition. It is not clear to what extent these inhibition-related activations may depend on subjects’ impulsivity. Garavan, Ross, Murphy, Roche, & Stein (2002) have reported greater reliance on anterior cingulate cortex during Go/NoGo in subjects who are more absent-minded, using a cognitive measure that they have found to be correlated with Barrett’s Impulsivity Scale (BIS). The key novel aspect of the present study is the direct investigation of the relationship between trait impulsivity measures and blood oxygen level dependent (BOLD) response. We hypothesized that: (1) orbitofrontal cortex would be activated during response inhibition, (2) low scores on psychometric (trait) impulsivity measures would be associated with greater activation in orbitofrontal cortex during response inhibition, (3) low error-rates on the Go/NoGo response inhibition task would be associated with greater activation in the orbitofrontal cortex during response inhibition, (4) psychometric (trait) impulsivity scores would correlate positively with error-rate on response inhibition tasks. 2. Methods 2.1. Subjects Twenty-one subjects were recruited by advertisement at their work-places. All subjects were male, right-handed, aged 18–50 years, and employed by the National Health Service or the University of Manchester. The Local Research Ethics Committee granted ethical approval. Subjects participated in a Structured Clinical Interview (SCID-NP) (Spitzer, Williams, Gibbon, & First, 1990) administered by a psychiatrist. Exclusion criteria were: current DSM III-R psychiatric axis I disorder, neurological disorder, currently (within 72 h) taking psychotropic medication or illicit substances, previous serious head injury, and any contraindication to MRI. 2.2. Assessment of impulsivity and neuropsychological test performance Trait impulsivity was assessed using the impulsivity subscale of Eysenck’s impulsivity, venturesomeness and empathy inventory (IVE-I) (Eysenck & Eysenck, 1991), and the Barratt’s Impulsivity Scale (BIS) (Barratt, 1994). Both were used as they differ in their conceptual basis (for a review see

Parker & Bagby, 1997). The computerised behavioural tests of impulsivity were adaptations of the Go/NoGo task used by Casey et al. (1997) (see below) and Newman’s card playing task (CPT) (Newman, Patterson, & Kosson, 1987). CPT was scored on the amount of money won and number of cards played. Intelligence was estimated using the National Adult Reading Test (NART) (Nelson & Willison, 1991). Data were analysed using SPSS v10.0 (SPSS Inc., Chicago, IL). Correlations between psychometric and behavioural measures of impulsivity were examined using Spearmann’s rank-order correlations. Partial correlations were computed controlling for NART scores. 2.3. Scanning procedure Images were acquired on a 1.5-T Philips Gyroscan ACS NT (Philips, Hamburg) scanner with Powertrack 6000 gradients operating at a software level of 6.1.2 using a quadrature headcoil as a radiofrequency receiver. Functional images were acquired using a single shot echo planar imaging (EPI) sequence (relaxation time = 3100 ms, echo time = 50 ms, field of view = 230 mm2 , flip angle = 90◦ , voxel dimension = 1.8 mm × 1.8 mm × 7 mm and 128 × 64 matrix) to prescribe the functional slice locations. Seventy-two volumes were acquired, each comprising 14 contiguous transverse slices (7 mm thick with no slice skip, in plane resolution = 1.8 mm) aligned with the corpus callosum, covering the entire brain. (The corpus callosum was defined on scout images in the sagittal plane before EPI acquisition.) The T-2 weighted images depicted blood oxygen level dependent contrast. To facilitate later registration of individual fMRI data sets in standard space, a T-1 weighted EPI data set (relaxation time = 6850 ms, echo time = 18 ms, field of view = 230 mm2 , voxel dimension = 0.89 mm × 0.89 mm × 3.5 mm) was acquired in the same session in 28 contiguous transverse planes (3.5 mm thick, no slice skip, in plane resolution = 0.89 mm), parallel to the corpus callosum. 2.4. Activation paradigm A simple Go/NoGo paradigm was developed to probe response inhibition without unduly loading working memory. It was based on a modification of the task described by Casey et al. (1997). Subjects were shown a continuous series of 120 letters and instructed to respond, by pressing a pneumatic bulb with their right hand, to any letter except V (V was used as the non-target instead of X because X may be associated with meanings such as “Stop!”). Seventy-five percent of the trials were targets (i.e. letters other than V). There were two conditions: Block A—the Go condition had 20 targets, and Block B—the NoGo condition had 10 targets and 10 non-targets. Because it is not possible to control for the effects of manual response frequency in a simple Go/NoGo analysis, Casey et al. used a motor control block. They found that data from the motor control block did not alter their

N.R. Horn et al. / Neuropsychologia 41 (2003) 1959–1966

Go/NoGo findings and in view of this we considered that a motor control block would not be helpful. Within each block targets were presented in pseudo-randomised order. Blocks were presented in the order ABBABA, rather than ABABAB in order to reduce the predictability of the task. Stimulus presentation duration was 0.5 s and inter-stimulus interval was 1.395 s, resulting in a total of 120 stimuli presented in a 227 s period. Subjects were not given an indication of their performance during the task. The task was presented using a Macintosh PowerBook computer and projected using a rear projection screen at the subject’s feet, viewed with the aid of prism glasses attached to the inside of the headcoil. Reaction times (RT) were measured from the beginning of stimulus presentation, recorded using a pressure-sensitive switch attached with a narrow gauge rubber tube to a rubber bulb squeezed by the subject. (We considered that the sensitivity of this measurement was not accurate enough to use reaction time data in further analyses.) Errors of commission were defined as a response to a “V” stimulus prior to subsequent stimulus presentation and the total was termed “error-rate”. 2.5. fMRI data analyses Data processing was conducted using Statistical Parametric Mapping (SPM99, Wellcome Department of Cognitive Neurology, London, UK) implemented in MATLAB (v5.0 Mathworks Inc., Sherborn, MA, USA) and run on a SPARC workstation (Sun Microsystems Inc., Surrey, UK). Scans were realigned using the first scan as a reference. Realigned scans were normalised, by transformation into a standard space corresponding to the stereotactic atlas of Talairach & Tournoux (1988) using MNI templates (Montreal Neurological Institute). Normalised images were smoothed with an 8 mm FWHM isotropic Gaussian kernel. Analysis was carried out using the general linear model with a delayed boxcar waveform. Subject-specific low-frequency drift in signal was removed by modelling with low-frequency sine and cosine waves and global changes were removed by proportional scaling (Holmes, Josephs, Buechel, & Friston, 1980). Effects at each voxel were estimated, and regionally specific effects were compared using linear contrasts. The resulting set of voxel values for each contrast constituted a statistical parametric map of the t statistic (SPM [T]) which was transformed into the unit distribution, SPM [Z]. Statistical inferences were based on the theory of random Gaussian fields (Friston et al., 1995). A random effects model was used. At the first level, mean images for each subject were created, depicting the subtraction of BOLD response during Block A from BOLD response during Block B (NoGo/Go) and vice versa (Go/NoGo). At the second level, these mean images were then combined in one-sample t-tests to assess the significant group effects. We used a threshold of P < 0.001 uncorrected, rather than the more rigorous P < 0.05 corrected for the entire brain volume, because previous studies of response inhibition provide a rationale for hypothesis-

1961

ing activation in certain brain regions (Casey et al., 1997; Garavan et al., 1999; Hager et al., 1998; Konishi et al., 1998). As a convenient shorthand, the statistical difference in BOLD response is referred to as “activation”. To investigate which of the regions activated by the task correlated with behavioural/personality measures and the possibility that the group mean effect was confounded by NART, analyses of covariance were performed using the group mean effect in the NoGo/Go analysis as the variable with covariates of mean corrected IVE-I, BIS, NART, and error-rate. We used an inclusive mask of the group mean effect (NoGo/Go) activation at a threshold of P < 0.05 in this analysis. The stereotactic coordinates of Talairach & Tournoux (1988) were used to report activations that covaried with the personality, IQ and performance measures. Factor analysis was used to explore whether activations of orbitofrontal cortex might underlie a unitary personality and performance dimension of impulsivity. The following variables were entered: NoGo/Go brain activations (Z-scores; Table 2) at left and right, anterior and posterior ventral frontal foci, trait impulsivity (BIS and IVE-I), errors of commission and NART. Principal components analysis accounted for 60–95% of the variance in each variable. Four principal components had eigenvalues greater than unity and they were rotated to simple structure using Varimax rotation.

3. Results Of 21 subjects recruited, 1 subject withdrew from the study, and 20 subjects completed the psychometric and behavioural impulsivity assessments. Although subjects were not explicitly informed of their performance during the Go/NoGo task it was observed the training session that subjects knew immediately when they had made errors. One further subject did not attend the scanning session, so we analysed complete imaging data for 19 subjects. One subject’s performance data in the scanned Go/NoGo task was not captured due to a failure of the pressure-sensitive switch and the analysis of variance was carried out with 18 subjects’ data. Data for key variables are summarised in Table 1. BIS and IVE-I scores correlated positively with each other (r = 0.45, P = 0.05). NART scores correlated Table 1 Personality trait and behavioural measures of impulsivity for 18 subjects Variable

Mean (S.D.)

Newman’s Play score Newman’s Win score Errors of commission in Go/NoGo task Reaction time in NoGo trials (ms) Barrett’s Impulsivity Scale Eysenck’s Impulsivity Scale NART score

45.0 (36.0) 87.0 (43.0) 7.67 (4.24) 453.0 (41.0) 41.9 (11.6) 6.00 (2.58) 115.0 (7.0)

1962

N.R. Horn et al. / Neuropsychologia 41 (2003) 1959–1966

Table 2 Foci of clusters where evoked activity was detected in the NoGo compared to Go analysisa Brain region

Left/right

Brodmann area

Talairach coordinates of maximum activity x

y

Cluster size (voxels)

Z value

z

42 −42

48 54

−6 −12

74 44

4.75 3.71

9 8

3 −3

60 39

27 45

299 15

3.81 3.16

R L

9 8

21 −24

51 39

30 39

17 43

3.22 3.14

Inferior frontal gyrus

R

45

57

21

12

26

2.99∗

Posterior lateral orbitofrontal cortex

R L

47 47

42 −45

24 27

−24 −21

b

3.12 3.45

Cingulate gyrus/precuneus

Midline

−6

−42

45

963

4.02

Anterior lateral orbitofrontal cortex

R L

Anterior medial prefrontal

R L

Dorsolateral prefrontal

10/11 11

7/31

Cuneus

Midline

18

0

−93

21

11

3.11

Superior temporal gyrus

R L

38 38

42 −36

18 9

−24 −18

576 214

4.11 3.79

Inferior parietal lobule

R

40

42

−60

45

613

4.03

a

Brain regions and Brodmann areas as described in Talairach and Tournoux’s atlas (Talairach & Tournoux, 1988). Talairach co-ordinates for the voxel of maximum activity in each cluster are given. The number of voxels activated in each cluster to a significance of P ≤ 0.001 is shown. b No data—a focus continuous with the L superior temporal gyrus cluster. ∗ P > 0.001—included to illustrate bilateral pattern of activation.

negatively with error-rate (r = −0.65, P = 0.003). BIS and IVE-I scores did not significantly correlate with error-rate or Newman’s Win or Play scores. In the NoGo/Go analysis several regions of evoked activity were identified. The foci of these activations are described in Table 2. Activation was more pronounced on the right side as indicated by higher Z-scores. The most intensely activated region was deep to the inferior frontal sulcus as shown in Fig. 1. Although activation was detected in subcortical structures in the NoGo/Go analysis this was continuous with cortical activation (Fig. 1). In the Go compared to NoGo analysis, activation was maximal in the left post-central gyrus (BA1). Correlations between neural response during response inhibition and psychometric trait impulsivity scores, error-rate and NART are summarised in Table 3. BIS and errors of commission both correlated with the NoGo/Go superior temporal gyrus activations but IVE-I did not. IVE impulsivity was associated with the inferior frontal—insula foci. Negative correlations between BIS and IVE-I scores and neural response are illustrated in Fig. 2—this figure also shows a negative correlation between activation and both trait impulsivity measures in the anterior medial superior frontal gyrus, but these activations were at lower levels of significance (P = 0.002 and 0.01, respectively). Factor analysis of lateral orbitofrontal activations, trait impulsivity, error-rate and NART generated four orthogonal factors accounting for 75% of the variance (Table 4). The first factor accounted for 28% of the variance and had high loadings on orbitofrontal activations except for the right pos-

terior region. The latter showed a unique association with IVE-I on factor 3 which accounted for 18% of the variance. Factor 2 showed the inverse association between NART and error-rate and that this did not relate to orbitofrontal activations. Factor 4 had a high loading on BIS score, a weaker loading on the IVE-I but none on errors of commission.

4. Discussion 4.1. Psychometric (trait) and behavioural measures of impulsivity We found no correlations between trait impulsivity measures and errors of commission on the Go/NoGo task. Instead we found that lower IQ was strongly associated with more errors of commission in both the correlation and the factor analysis (see factor 2 Table 4). This suggests that intelligence rather than impulsivity may be the main source of variation in a simple Go/NoGo task. 4.2. Brain activation during response inhibition The most significant neural responses during the NoGo condition were seen in right anterior lateral anterior orbitofrontal cortex (BA10/11). The voxel of maximum activation in this study,1 was in a similar location to that seen 1

Talairach and Tournoux co-ordinates x = 42, y = 48, z = −6.

N.R. Horn et al. / Neuropsychologia 41 (2003) 1959–1966

1963

Fig. 1. Neural substrates of response inhibition in 19 healthy subjects: NoGo minus Go conditions. Three orthogonal views, sagittal at top left, coronal at top right, transverse at bottom. Statistical parametric maps of the fMRI data, showing regions that are significantly activated, are superimposed on a normalised MR image in standard space corresponding to the stereotactic atlas of Talairach and Tournoux. Z-scores are shown for areas where all contiguous voxels exceed the predefined threshold for statistical significance (T = 2.55). Crosshairs indicate the location of the slices and intersect at the voxel of maximum activation (42, 48, −6) in the right ventro-lateral orbitofrontal cortex. Activation in the left anterior and bilaterally in the posterior orbitofrontal cortex; the temporal poles and right parietal lobe is also shown.

Table 3 Areas where activation significantly covaried (P ≤ 0.001) with trait impulsivity scores, error-rate and NARTa Covariable brain region

Left/right

Brodmann area

Talairach coordinates of maximum activity x

Barrett’s Impulsivity Scale Superior frontal gyrus (medial) Superior temporal gyrus

Midline L

8 22

Eysenck’s Impulsivity Scale Inferior parietal lobule Inferior frontal gyrus Insula (anterior)

R R R

40 44/45

Error-rate Middle temporal gyrus Superior temporal gyrus

R L

NART Precentral gyrus Middle frontal gyrus

R L

y

Cluster size (voxels)

Positive/negative association

z

−3 −51

39 −6

45 −6

63 65

Negative Positive

57 60 39

−30 15 0

51 6 −6

45 25 45

Negative Positive Positive

21 22

60 −42

−18 0

−15 −12

35 108

Negative Positive

6 9

45 −36

3 45

33 33

30 39

Positive Positive

a Brain regions and Brodmann areas as described in Talairach and Tournoux’s atlas (Talairach & Tournoux, 1988). Talairach co-ordinates for the voxel of maximum activity in each cluster are given. The number of voxels activated in each cluster to a significance of P ≤ 0.05 is shown. Clusters of >20 voxels are reported.

1964

N.R. Horn et al. / Neuropsychologia 41 (2003) 1959–1966

Fig. 2. Negative correlations between activation during response inhibition and Eysenck’s Impulsivity score (left), and Barratt’s Impulsivity Scale (right) score. Transverse ‘glass brain’ views of statistical parametric maps of fMRI data showing regions significantly activated during response inhibition which correlate negatively with trait impulsivity scores. Significant correlations are shown at a threshold at P = 0.05, after inclusive masking with SPM of mean effect at P = 0.05. Correlation between activation and Eysenck’s Impulsivity score is shown on the left, and Barratt’s Impulsivity Scale on the right. Areas where P = 0.001 are circled, from left to right: right inferior parietal lobule; and superior frontal gyrus. Both are heteromodal association areas.

Table 4 Factor analysis of ventral frontal NoGo/Go activations, performance and questionnaire measures of impulsivity and NART IQ Factor

1

Left anterior OFC Left posterior OFC Right anterior OFC Right posterior OFC Eysenck impulsivity Barrat impulsivity NART IQ Error-rate Variance explained (%)

0.75 0.80 0.80 0.14 0.02 0.20 0.21 0.13 25

2 −0.08 0.27 −0.17 0.17 −0.14 0.07 −0.86 0.88 21

3

4 0.11 −0.05 0.12 0.88 0.78 0.00 0.07 0.14 18

0.11 0.13 −0.01 −0.21 0.51 0.95 −0.35 −0.20 18

Bold indicates loadings greater than 0.75 and italic >0.50 and <0.75.

in Garavan’s study2 (Garavan et al., 1999): our results support existing evidence for a specific role of this region in overriding a pre-potent response. This finding is compatible with the clinical literature suggesting that damage to the lateral orbitofrontal cortex is associated with behavioural disinhibition (Paradiso et al., 1999). However medial prefrontal cortical regions were also activated, and it may be argued that a lesion in one of a number of critical regions could result in poor impulse control. Our finding of activation in the dorsolateral prefrontal cortices replicates those reported in other studies (Casey et al., 1997; Garavan et al., 1999; Hager et al., 1998). This may reflect the executive cognitive components of the task such as vigilant attention for the letter V and keeping the response requirement in mind. Our findings of foci of activation in the posterior orbitofrontal cortex, the temporal poles, and the posterior cingulate indicate a widespread activation of the limbic network during response inhibition. We suggest the paralimbic activations may reflect processing of the motivational aspects of the task and their interaction with cognition. The more prominent medial prefrontal and paralimbic cortical activ2

Talairach and Tournoux co-ordinates x = 42, y = 40, z = −1.

ity in this study, compared to Garavan et al. (1999), may be attributed to the block design we used, which is likely to detect cognitive processing which takes place after initial response inhibition. We postulate that such delayed processing is likely to have an affective component related to success and failure. Transient happiness and sadness have been reported to activate limbic and paralimbic brain regions in PET studies (George et al., 1995; Paradiso et al., 1997, 1999). Other possible explanations for differences in these findings from earlier studies are, firstly, that the scanning protocol we used overcame artefacts which have been thought to lead to a loss of signal and reduced sensitivity to changes in functional signal in these areas Garavan et al. (1999), and secondly, that sample size and selection varies across studies: we studied a larger group-additionally, subjects’ age, health, motivation, and attention vary between studies. Activation of the left post-central gyrus in BA1 in the Go/NoGo analysis is consistent with increased sensory input in the region associated with the right hand, which had twice as much sensory input in the Go condition, compared with the NoGo condition. 4.3. Neural correlates of impulsivity measures This is the first study that directly examines the associations between brain regions activated during response inhibition and measures of trait impulsivity. We carried out two analyses. First, we examined which areas of activation in the NoGo/Go comparison covaried with personality, performance and IQ. Second, we used factor analysis to explore whether a unitary dimension of impulsivity could be identified in terms of orbitofrontal brain activation, personality and performance. High IVE-I was most positively associated with activation in paralimbic areas, including the right inferior frontal gyrus and right insula. This association was corroborated by factor 3 (Table 4) which loaded only on IVE-I and a region of right posterior OFC contiguous with the inferior frontal gyrus activation identified in the correlational

N.R. Horn et al. / Neuropsychologia 41 (2003) 1959–1966

1965

analysis. One possibility is that impulsive subjects recruit more neurones from these regions than less impulsive subjects, to achieve the same degree of behavioural inhibition. This together with a previous finding that IVE-I correlated better with serotonin function than BIS, suggests that IVE-I more closely relates to neurobiologically mediated aspects of impulsivity than BIS (Dolan, Anderson, & Deakin, 2001). Greater BIS impulsivity and error-rate were associated with activation in the left superior temporal gyrus. Since the associations were positive, they seem more likely to reflect cognitive processing of errors rather than a variation in brain function in impulsive subjects. Low BIS and IVE impulsivity scores were associated with greater activation in the anterior medial superior frontal gyrus and the temporoparietal association area—both heteromodal association areas. NART also correlated positively with heteromodal association areas. Both BIS and IVE-I correlated negatively with activation in the medial superior frontal gyrus but not at the threshold we set, which may have been too rigorous relative to the power of the experiment. This finding is consistent with evidence that medial prefrontal lesions result in poor impulse control. Cautious interpretation of these findings is necessary, as a unified measure of impulsivity has not been defined, and although trends were identified, a uniform pattern of association between anatomical brain regions and both impulsivity measures was not evident. The results are also not consistent with those of Garavan et al. (2002) who reported an association between absentmindedness (which correlates with BIS impulsivity) and cingulate activation during Go/NoGo. However, it should be noted that the task used in that study was a more cognitively demanding one, which would be expected to place greater demands on executive control processes mediated by the anterior cingulate. Notably, in our study there was no evidence from the factor analysis that the ventral frontal regions implicated in response inhibition related to a unitary dimension with loadings on all measures. From a functional perspective it may be considered that less impulsive subjects place greater demand on heteromodal association areas while more impulsive individuals place greater demand on paralimbic areas. This may be due to differences level of recruitment of neurones during response inhibition, and/or differences in post-stimulus cognitive processing: less impulsive subjects may preferentially update strategy, while more impulsive subjects may experience affective cognitions, as demonstrated with the finding that excitement engendered in making risky choices is associated with activation in the posterior part of the lateral orbitofrontal cortex (Elliott, Dolan, & Frith, 2000).

ences in activation patterns (Casey et al., 1997; George et al., 1995). Future research directed at children with ADHD may find personality trait measures helpful, but these have not been validated in children. We used the convention of describing our findings with reference to the maximum foci of activation that may have masked findings related to important areas which are adjacent, but functionally discrete. We did not systematically gather qualitative information about subjects’ cognitions or mood during scanning, although some spontaneously volunteered that they had, for example, been thinking that they had made no errors, or done very poorly, and one subject had been wondering about the randomisation pattern of target stimuli. Such information may have been helpful in explaining individual variations in results.

4.4. Limitations

Barratt, E. S. (1985). Impulsiveness defined within a systems model of personality. In E. P. Speilburger, & J. N. Butcher (Eds.), Advances in personality assessment (pp. 113–132). Hillsdale, NJ: Lawrence Erlbaum Associates. Barratt, E. S. (1994). Impulsiveness and aggression. In J. Monahan, & H. Steadman (Eds.), Violence and mental disorder: Developments in risk assessment (pp. 61–79). Chicago: University of Chicago Press.

In interpreting our results, we have been mindful that we have studied normal adult males, and results may differ and cannot be generalised to other groups, particularly as there is evidence that there may be gender and age-related differ-

5. Conclusions Our findings support the hypothesis that the anterior lateral orbitofrontal cortex is activated during response inhibition. In addition, a network of higher order association and paralimbic areas was activated. While these regions are known to be anatomically connected and have been activated in this study, the challenge of elucidating the precise sequences of neural activity will require more sophisticated techniques, such as the use of brain evoked-potential measurement and event-related fMRI paradigms. We detected differences between the functional type of brain region activated during response inhibition and personality trait measures. More impulsive individuals activated paralimbic areas, particularly the right inferior frontal gyrus extending to posterior lateral orbitofrontal cortex and anterior insula, while less impulsive individuals activated higher order association areas when inhibiting a pre-potent response.

Acknowledgements This study was supported with funding from the Mental Health Services of Salford and the Wellcome Trust. The authors thank the staff of the Neuroimaging Analysis Centre, University of Manchester for assistance, particularly: Dr. I. Holländer, Lecturer in functional MRI, Department of Gastroenterology and Ms. Y. Watson, Superintendent Radiographer, Department of Clinical Radiology. References

1966

N.R. Horn et al. / Neuropsychologia 41 (2003) 1959–1966

Barratt, E. S., & Patton, J. H. (1983). Impulsivity: Cognitive, behavioral, and psychophysiological correlates. In M. Zuckerman (Ed.), Biological bases of sensation seeking, impulsivity, and anxiety (pp. 77–116). Hillsdale, NJ: Lawrence Erlbaum Associates. Barratt, E. S., Stanford, M. S., Kent, T. A., & Felthous, A. R. (1997). Neuropsychological and cognitive psychophysiological substrates of impulsive aggression. Biological Psychiatry, 41, 1045–1061. Casey, B. J., Trainor, R. J., Orendi, J. L., Schubert, A. B., Nystrom, L. E., Giedd, J. N., Castellanos, X., Haxby, J. V., Noll, D. C., Cohen, J. D., Forman, S. D., Dahl, R. E., & Rapoport, J. L. (1997). A developmental functional MRI study of prefrontal activation during performance of a Go-No-Go task. Journal of Cognitive Neuroscience, 9, 835–847. Damasio, A. R. (2000). A neural basis for sociopathy. Archives of General Psychiatry, 57, 128–130. Damasio, A., Tranel, D., & Damasio, H. (1990). Individuals with psychopathic behaviour caused by frontal damage fail to respond autonomically to social stimuli. Brain and Behaviour Research, 41, 81–94. Dolan, M., Anderson, I. M., & Deakin, J. F. W. (2001). Relatinship between 5-HT function and impulsivity and aggression in male offenders with personality disorders. British Journal of Psychiatry, 178, 352–359. Elliott, R., Dolan, R. J., & Frith, C. D. (2000). Dissociable functions in the medial and lateral orbitofrontal cortex: Evidence from human neuroimaging studies. Cerebral Cortex, 10, 308–317. Eysenck, H. J., & Eysenck, S. B. G. (1991). Manual of the Eysenck Personality Scales (EPS Adult). Kent: Hodder and Stoughton. Friston, K. J., Holmes, A. P., Worsley, K. J., Poline, J. -B., Frith, C. D., & Frackowiak, R. S. J. (1995). Statistical parametric maps in functional imaging: A general approach. Human Brain Mapping, 2, 189–210. Garavan, H., Ross, T. J., & Stein, E. A. (1999). Right hemispheric dominance of inhibitory control: An event-related functional MRI study. Proceedings of the National Academy of Sciences USA, 96, 8301–8306. Garavan, H., Ross, T. J., Murphy, K., Roche, R. A. P., & Stein, E. A. (2002). Dissociable executive functions in the dynamic control of behaviour: Inhibition, error detection and correction. Neuroimage, 17, 1820–1829. George, M. S., Ketter, T. A., Parekh, P. I., Horwitz, B., Herscovitch, P., & Post, R. M. (1995). Brain activity during transient sadness and happiness in healthy women. American Journal of Psychiatry, 152, 341–351. Grafman, J., Schwab, K., Warden, D., Pridgen, A., Brown, H. R., & Salazar, A. M. (1996). Frontal lobe injuries, violence, and aggression: A report of the Vietnam Head Injury Study. Neurology, 46, 1231–1238. Hager, F., Volz, H. P., Gaser, C., Mentzel, H. J., Kaiser, W. A., & Sauer, H. (1998). Challenging the anterior attentional system with a continuous performance task: A functional magnetic resonance imaging approach. European Archives of Psychiatry and Clinical Neuroscience, 248, 161– 170. Holmes, A. P., Josephs, O., Buechel, C., & Friston, K. J. (1980). Statistical modelling of low-frequency compounds in fMRI. Neuroimage, 5, S480. Kawashima, R., Satoh, K., Itoh, H., Ono, S., Furumoto, S., Gotoh, R., Koyama, M., Yoshioka, S., Takahashi, T., Takahashi, K., Yanagisawa, T., & Fukuda, H. (1996). Functional anatomy of GO/NO-GO discrimination and response selection—A PET study in man. Brain Research, 728, 79–89. Kindlon, D., Mezzacappa, E., & Earls, F. (1995). Psychometric properties of impulsivity measures: Temporal stability, validity and factor structure. Journal of Child Psychology and Psychiatry, 36, 645– 661. Konishi, S., Nakajima, K., Uchida, I., Sekihara, K., & Miyashita, Y. (1998). No-go dominant brain activity in human inferior prefrontal

cortex revealed by functional magnetic resonance imaging. European Journal of Neuroscience, 10, 1209–1213. Krams, M., Rushworth, M. F., Deiber, M. P., Frackowiak, R. S., & Passingham, R. E. (1998). The preparation, execution and suppression of copied movements in the human brain. Experimental Brain Research, 120, 386–398. Monahan, J., Steadman, H. J., Applebaum, P. S., Robbins, P. C., Mulvey, E. P., Silver, E., Roth, L. H., & Grisso, T. (2000). Developing a clinically useful actuarial tool for assessing violence risk. British Journal of Psychiatry, 176, 312–319. Morgan, A. B., & Lilienfeld, S. O. (2000). A meta-analytic review of the relation between antisocial behaviour and neuropsychological measures of executive function. Clinical Psychology Review, 20, 113–156. Nelson, H., & Willison, J. (1991). The National Adult Reading Test (2nd ed.). Windsor: NFER-Nelson. Newman, J. P., Patterson, M. C., & Kosson, D. S. (1987). Response perseveration in psychopaths. Journal of Abnormal Psychology, 96, 145–148. Nobre, A., Coull, J., Frith, C., & Mesulam, M. (1999). Orbitofrontal cortex is activated during breaches of expectation in tasks of visual orientation. Nature Neuroscience, 2, 11–12. Paradiso, S., Robinson, R. G., Andreasen, N. C., Downhill, J. E., Davidson, R. J., Kirchner, P. T., Watkins, G. L., Boles Ponto, L. L., & Hichwa, R. D. (1997). Emotional activation of limbic circuitry in elderly normal subjects in a PET study. American Journal of Psychiatry, 154, 384– 389. Paradiso, S., Chemerinski, E., Yazici, K. M., Tartaro, A., & Robinson, R. G. (1999). Frontal lobe syndrome reassessed: Comparison of patients with lateral or medial frontal lobe damage. Journal of Neurology, Neurosurgery and Psychiatry, 67, 664–667. Paradiso, S., Johnson, D. L., Andreasen, N. C., O’Leary, D. S., Watkins, G. L., Boles Ponto, L. L., & Hichwa, R. D. (1999). Cerebral blood flow changes associated with attribution of emotional valence to pleasant, unpleasant, and neutral visual stimuli in a PET study of normal subjects. American Journal of Psychiatry, 156, 1618–1629. Parker, J. D. A., & Bagby, R. M. (1997). Impulsivity in adults: A critical review of measurement approaches. In Impulsivity: Theory, assessment, and treatment (pp. 142–157). London: The Guildford Press. Plutchik, R., & Van Praag, H. M. (1989). The measurement of suicidality. Progress in Neuropsychopharmacology and Biological Psychiatry, 13, S23–34. Plutchik, R., & Van Praag, H. M. (1995). The nature of impulsivity: Definitions, ontology, genetics, and relations to aggression. In E. Hollander, & D. Stein (Eds.). Impulsivity and Aggression (pp. 7–24). New York: Wiley. Raine, A., Buchsbaum, M. S., & LaCasse, L. (1997). Brain abnormalities in murderers indicated by positron emission tomography. Biological Psychiatry, 42, 495–508. Raine, A., Lencz, T., Bihrle, S., La Casse, L., & Colletti, P. (2000). Reduced prefrontal grey matter volume and reduced autonomic activity in antisocial personality disorder. Archives of General Psychiatry, 57, 119–127. Spitzer, R. L., Williams, J. B. W., Gibbon, M., & First, M. B. (1990). Structured clinical interview for DSM-III-R (Non-patient ed., SCID-NP, Version 1.0). Washington, DC: American Psychiatric Press. Stein, D. J., Hollander, E., & Liebowit, M. R. (1993). Neurobiology of impulsivity and the impulse control disorders. Journal of Neuropsychiatry, 5, 9–17. Stein, D. J., Towney, J., & Hollander, E. (1995). The neuropsychiatry of impulsive aggression. In E. Hollander, & D. Stein (Eds.), Impulsivity and aggression (pp. 91–105). New York: Wiley. Talairach, J., & Tournoux, P. (1988). Coplanar stereotactic atlas of the human brain. Stuttgart: George Thieme Verlag.