Neuroanatomical Correlates of Personality in the Elderly

Abstract

Extraversion and neuroticism are two important and frequently studied dimensions of human personality. They describe individual differences in emotional responding that are quite stable across the adult lifespan. Neuroimaging research has begun to provide evidence that neuroticism and extraversion have specific neuroanatomical correlates within the cerebral cortex and amygdala of young adults. However, these brain areas undergo alterations in size with aging, which may influence the nature of these personality factor-brain structure associations in the elderly. One study in the elderly demonstrated associations between perisylvian cortex structure and measures of self transcendence (Kaasinen et al., 2005), but the neuroanatomical correlates of extraversion and neuroticism, or other measures of the Five Factor Model of personality have not been explored. The purpose of the present study was to investigate the structural correlates of neuroticism and extraversion in healthy elderly subjects (n=29) using neuroanatomic measures of the cerebral cortex and amygdala. We observed that the thickness of specific lateral prefrontal cortex (PFC) regions, but not amygdala volume, correlates with measures of extraversion and neuroticism. The results suggest differences in the regional neuroanatomic correlates of specific personality traits with aging. We speculate that this relates to the influences of age-related structural changes in the PFC.

Introduction

Personality characteristics describe distinctive and recurrent patterns of thoughts, feelings and actions that occur in response to particular situational demands (Mischel, 2004). The Five Factor Model (FFM) is an influential model of human personality, reflecting several decades of psychological research (John, 1990; McCrae and Costa, 1990; McCrae and John, 1992; Costa and McCrae, 2000). This model provides a convenient method for summarizing a wide range of human behaviors and describes personality along the dimensions of extraversion, neuroticism, conscientiousness, agreeableness, and openness to experience. These factors are notably stable across the life span from early adulthood well into late-life, with no change or small decreases in neuroticism, extraversion, and openness; and variable small changes in agreeableness and conscientiousness primarily between college age and middle adulthood (Costa and McCrae, 1977, 1986, 1997; McCrae et al., 1999; Costa et al., 2000).

Two of the dimensions, extraversion and neuroticism, are of particular interest since their brain bases might contribute to mood and anxiety disorder predisposition throughout the lifespan (Rosenbaum et al., 1991; Watson and Clark, 1992; Costa and McCrae, 1996; Biederman et al., 2001; Widiger and Seidlitz, 2002; Duberstein et al., 2003; Khan et al., 2005). The introversion/extraversion factor captures the social dimension of personality. Extraverts have a preference for seeking and engaging in social interactions, whereas introverts prefer to avoid social situations and tend to be reserved or socially awkward. Extraversion may also be characterized as a sensitivity to positive or pleasure cues in the environment (Larsen and Ketelaar, 1991; McCrae and Costa, 1991; Costa and McCrae, 1992; Watson and Clark, 1992; Izard et al., 1993; Barrett and Pietromonaco, 1997; Lucas et al., 2000; Lucas and Diener, 2001; Pickering and Gray, 2001).

Neuroticism is characterized as a pervasive sensitivity to negative or punishment cues in the environment, like those present in social situations (Bolger and Schilling, 1991; Larsen and Ketelaar, 1991; McCrae and Costa, 1991; Costa and McCrae, 1992; Watson and Clark, 1992; Izard et al., 1993; Bolger and Zuckerman, 1995; Pickering and Gray, 2001). Individuals high in neuroticism also tend to automatically orient to novel situational cues (Wallace et al., 1991; Wallace and Newman, 1997, 1998) and more readily assess situations as threatening than those who are low in neuroticism (Schneider, 2004).

Studies using magnetic resonance imaging (MRI) combined with different quantitative techniques for structural analysis have begun to map the neuroanatomic relationships of extraversion and neuroticism in healthy young adults (Omura et al., 2005; Rauch et al., 2005; Wright et al., 2006b). These studies primarily point to associations between specific regions of the PFC - including orbitofrontal and dorsolateral regions - or amygdala, and extraversion and neuroticism. This is consistent with earlier functional imaging studies showing associations between extraversion, neuroticism or related measures and the activities of the PFC and amygdala during resting states (Johnson et al., 1999), or in response to specific activation procedures (Canli et al., 2001; Canli et al., 2002; Zald et al., 2002; Gusnard et al., 2003; Schwartz et al., 2003; Canli et al., 2004; Gray et al., 2005). However, the structure (and function) of the PFC and amygdala are altered with aging, and it remains uncertain how this effect may influence the relationships between these structures with extraversion and neuroticism. Specifically, lateral regions of the PFC diminish in size with age, while those on the medial and inferior surfaces are stable or increase across the adult life span (Salat et al., 2001; Salat et al., 2004; Grieve et al., 2005). These anatomic changes in the PFC may respectively relate to diminished executive functioning and working memory (Raz et al., 1999), but enhanced emotional control with aging (Carstensen et al., 2000). The amygdala is also known to exhibit mild atrophy with aging, and functional imaging studies suggest an altered profile of activation in this structure in young vs. elderly subjects (Iidaka et al., 2002; Gunning-Dixon et al., 2003; Mather et al., 2004; Fischer et al., 2005; Tessitore et al., 2005; Wedig et al., 2005; Wright et al., 2006a).

The main purpose of the present study is to examine the structural correlates of neuroticism and extraversion in healthy elderly subjects, given the known structural and functional changes in the PFC and amygdala with aging. As in earlier work in healthy young subjects (Wright et al., 2006b), we employed two quantitative techniques to analyze high-resolution MRI data for anatomic correlates of extraversion and neuroticism. PFC thickness was studied using the technique of cortical surface analysis (Fischl and Dale, 2000; Salat et al., 2004; Rauch et al., 2005; Wright et al., 2006b), and amygdala volume was measured using manual anatomic methods (Wedig et al., 2005; Wright et al., 2006a; Wright et al., 2006b).

Materials and Methods

Subjects

29 right-handed elderly subjects (17 female, 12 male) were studied (age in years: m=70.3, sd=6.6, range=61–84). All subjects underwent cortical thickness analyses and anatomic region of interest analysis of the amygdala (see below). Subjects were recruited from the local area via word of mouth and using advertisements. Written informed consent was obtained from each subject. The study was approved and conducted in accordance with guidelines established by the Partners Human Research Committee. Routine screening assessments were used to rule out the presence of neurological, medical, psychiatric conditions (including personality disorders), and contraindications to MRI. All subjects also underwent the Structured Clinical Interview for the Diagnostic and Statistical Manual 4^th edition (DSM-IV) (American Psychiatric Association, 1994) to confirm the absence of DSM-IV Axis I diagnoses (First et al., 1995), including the absence of dementia, and were required to be free of psychoactive medications. All subjects performed normally (m=29.4, sd=0.9, range 27–30) on the 30-point Mini-Mental State Examination (Folstein et al., 1975), which was used as an additional screen for cognitive disorders.

All subjects completed the NEO Five Factor Inventory (Costa and McCrae, 1992) (Table 1). Extraversion and neuroticism T-scores were calculated from gender specific normative data to control for effects of gender on the T-score. These T scores were used as the main variable of interest. As in our previous study we also calculated T-scores for openness, agreeableness and conscientiousness and used these in some analyses as described below. All analyses were confirmed using raw scores unless otherwise stated.

TABLE 1.

Personality Self-Report Group Characteristics

T-score	Extraversion	Neuroticism	Openness	Agreeableness	Conscientiousness
Mean	53.0	40.9	54.9	55.5	49.7
S.D.	10.9	7.3	9.5	7.8	10.2
Range	31–74	25–63	34–72	36–70	27–71

Imaging

All subjects underwent structural brain imaging using a Sonata 1.5 Tesla whole body imaging device (Siemens Medical Systems, Iselin NJ) with a 3-axis gradient head coil. Head movement was restricted using expandable foam cushions. After an automated scout image was acquired, two high-resolution 3D MPRAGE sequences (TR/TE/flip angle=7.25ms/3ms/7°) with an in-plane resolution of 1.3 mm, and 1 mm slice thickness were collected. The acquisition parameters were empirically optimized to increase gray/white and gray/cerebrospinal fluid contrast.

Cortical Thickness Measurement

These methods have been previously described in detail (Fischl and Dale, 2000; Rauch et al., 2005; Wright et al., 2006b), including in elderly subjects (Salat et al., 2004). Briefly, the two high-resolution structural scans for each participant were motion corrected and averaged to create a single average volume with a high signal-to-noise ratio. The resulting volume was used to segment cerebral white matter (Dale et al., 1999) and to estimate the gray/white interface. Topological defects in the gray/white estimate were corrected (Fischl et al., 2001), and this gray/white estimate was used as the starting point for a deformable surface algorithm designed to find the pial surface with sub-millimeter precision (Fischl and Dale, 2000). The entire cortical surface in each individual subject was then visually inspected and any inaccuracies in segmentation were manually corrected.

Cortical thickness measurements were obtained by reconstructing representations of the gray/white matter boundary and the pial surface, and then calculating the distance between those surfaces at each of approximately 320,000 points across the entire cortical mantle ((Dale et al., 1999; Fischl and Dale, 2000); software and complete documentation is available at http://www.nmr.mgh.harvard.edu/freesurfer). The accuracy of the thickness measures derived from this technique has been previously validated by direct comparisons with manual measures on post-mortem brains (Rosas et al., 2002), on MRI data (Kuperberg et al., 2003), and in elderly subjects (Salat et al., 2004). The surface representing the gray–white border was “inflated” (Fischl et al., 1999a), differences among individuals in the depth of gyri–sulci were normalized, and each subject’s reconstructed brain was then morphed and registered to an average spherical surface representation that optimally aligned sulcal and gyral features across subjects (Fischl et al., 1999a; Fischl et al., 1999b). This spherical morphing procedure was used to construct the cortical thickness correlation brain maps. Thickness measures were then mapped on the ‘inflated’ surface of each participant’s reconstructed brain (Fischl et al., 1999a). The data were smoothed on the surface tessellation using an iterative nearest-neighbor averaging procedure. Fifty iterations were applied, which is approximately equivalent to applying a two-dimensional Gaussian smoothing kernel along the cortical surface with a full-width/half-maximum of 13 mm. Data were then resampled for participants into a common spherical coordinate system (Fischl et al., 1999b). Data were also linearly transformed so that approximate Talairach coordinates (Talairach and Tournoux, 1988) could be derived at each point on the surface (Fischl et al., 2002) to allow comparisons with other studies.

Statistical analysis

Statistical surface maps were generated by computing two separate general linear models (GLMs) for the effects of neuroticism or extraversion T-scores on cortical thickness at each point. Similar maps were also generated for openness, agreeableness and conscientiousness T-scores. All analyses included the covariates of gender and age, which were entered in the model as external regressors to control for their effects. Parallel analyses were performed for raw scores. Based on our previous study in young adults (Wright et al., 2006b), our cortical a priori regions of interest included parts of the lateral and inferior prefrontal cortex. Based on the earlier findings, the degree of smoothing applied in the current study, and the combined surface areas of these regions of interest, we selected a statistical significance threshold of p < .001 (two-tailed), as a correction for multiple comparisons within a priori search territory. For all areas outside of the this territory cortex a threshold of p < .0001 (two-tailed).

Once sites with significant effects were identified as described above, a region of interest (ROI) based approach was used. For these analyses, thickness measures for each subject were obtained at the location corresponding to the coordinates of the peak p-value on the group maps (see Figs. 1–3). These thickness measures were then plotted against individual subject T-scores on a scattergram and fitted with a regression line. Partial correlation coefficients (r values) controlling for age and sex, and the associated p-values were then calculated. This ROI based approach was used primarily to display individual subject data points and obtain correlation coefficients. Because of concerns about collinearity influencing the partial correlation results, we also analyzed the data in a multiple linear regression model with age and sex entered as nuisance regressors. A separate model was utilized for each of the two main behavioral variables (extraversion and neuroticism), asking the question, which of the regions identified were the best predictors of these behavioral measures? To test for significant differences in the laterality of our findings, a similar ROI based approach was used whereby matching points in the left and right hemisphere were compared by calculating the r values at each site, using a Fisher r-to-z transformation and then examining for significant hemispheric differences using the test of differences (Cohen and Cohen, 1983).

Figure 1. Significant Correlations of Cortical Thickness and Extraversion in the Lateral Prefrontal Cortex.

(A) Colorized statistical map superimposed upon a partially inflated group average cortical surface. The lateral aspect the right hemisphere is shown. A significant positive correlation of cortical thickness with extraversion was found in the superior frontal cortex (SFC). (B) Colorized statistical map superimposed upon a partially inflated group average cortical surface. The lateral aspect the left hemisphere is shown with a significant positive correlation between cortical thickness and extraversion in the anterior parts of the middle frontal cortex (MFC). Dark gray regions are sulci, light gray are gyri. Colorized scale bars show the p-value for positive (red-yellow) and negative (blue) correlations. (C) Scatter plot and regression line demonstrating a significant positive correlation between right SFC thickness and extraversion. These values were extracted from the peak surface point of the SFC shown in (A). (D) Scatter plot and regression line from the same MFG site in (B).

Figure 3. Significant Correlations of Cortical Thickness and Openness in the Parietal Cortex.

(A) Colorized statistical map superimposed upon a partially inflated group average cortical surface. The lateral aspect the left hemisphere is shown. Significant inverse correlations of cortical thickness with openness were found in the inferior parietal cortex (IPC). Dark gray regions are sulci, light gray are gyri. Colorized scale bars show the p-value for positive (red-yellow) and negative (blue) correlations. (B) Scatter plot and regression line demonstrating a significant inverse correlation between IPC thickness and neuroticism. These values were extracted from the peak surface point of the IPC shown in (A).

Cortical Nomenclature

For consistency in comparison to our earlier study we use descriptive anatomic terminology to indicate the location of our significant effects. For example, in the lateral prefrontal cortex, we refer to the superior frontal cortex (e.g. the superior frontal gyrus and adjacent sulci), the middle frontal cortex (e.g. the middle frontal gryus and adjacent sulci) and the inferior frontal cortex (e.g. the inferior frontal gyrus and adjacent sulci). This terminology also allows clearer designation of effects in gryal and sulcal portions of cortex. Approximate Brodmann areas and Talairach coordinates are also included for each site of peak effect to allow for comparisons with other studies.

Amygdala Volumetric Analysis and Correlations

The amygdala was manually traced on each individual subjects’ motion corrected, averaged high-resolution 3D image (e.g. see Fig 4A,B). A technician who was blind to the results of the individual personality measures performed tracings. As amygdala volumes are known to vary with head size, we performed our analyses using an amygdala volume corrected for estimated intracranial volume. An automated atlas-based method that has been validated against manual intracranial volume measures (Buckner et al., 2004) was used to calculate estimated total intracranial volume (eTIV). For correction of the effects of head size, amygdala volumes were divided by eTIV then multiplied by 100%. The relationship of these corrected volumes to extraversion and neuroticism was then assessed using partial correlations with age and gender as covariates. Detailed descriptions about the tracing and eTIV procedures, and their reliability, have been presented in earlier studies (Wedig et al., 2005; Wright et al., 2006a; Wright et al., 2006b). For the amygdala-personality factor correlations, a statistical threshold of p < 0.025 was used to correct for examining the right and left amygdala.

Analyses of the other personality factors

Though not the primary purpose of this study we also examined the 3 other factors of the FFM (openness, agreeableness and conscientiousness) using both cortical surface based and amygdala volumetric analyses. We use the same methodology for these analyses as for neuroticisms and extraversion as described above. For cortical surface analyses we applied the non a priori threshold of p<.0001. For the amygdala, we used a threshold of p < .01 as an approximate correction for the number of comparisons that the analyses of these three factors entailed.

Results

Correlations with Extraversion and Cerebral Cortex Thickness

In the right hemisphere, surface-based analyses revealed a significant positive correlation between cortical thickness and extraversion T-scores in the superior frontal cortex (SFC, Fig 1, Table 2). Scatter plots based on ROI analysis at this peak confirmed the significant positive relationship between right superior frontal cortex thickness and extraversion (Fig. 1). Neuroticism scores were not significantly correlated with cortical thickness at this site (Table 2). In the left hemisphere, a significant positive correlation was found between extraversion and the thickness of the middle frontal cortex (MFC, Fig. 1, Table 2). Scatter plots confirmed the relationship between the MFC thickness and extraversion (Fig. 1). Neuroticism scores were not significantly correlated with cortical thickness at this site (Table 2).

TABLE 2.

Brain regions showing significant correlations with cortical thickness, extraversion, and neuroticism are listed and correspond to the sites shown in Figs. 1 and 2. The approximate Brodmann area designations are based on the Talairach coordinates of the peak point on the cortical surface maps. The Talairach X-coordinate indicates left-right, the Y-coordinate anterior-posterior, and the Z-coordinate superior-inferior. R values are for partial correlation coefficients controlling for age and sex.

Brain Region (Approx. Brodmann Area)	Hemisphere	Talairach Coordinates			Extraversion	Neuroticism
		X	Y	Z	R value	R value
Prefrontal Cortex
Extraversion
Superior Frontal Cortex s (A6)	Right	20	12	41	.643*	−.294
Middle Frontal Cortex (A10)	Left	−38	56	12	.638*	−.388
Neuroticism
Superior Frontal Cortex g (A6)	Right	17	15	44	.363	−.605*
Inferior Frontal Cortex (A44)	Right	37	21	24	.190	−.607*
Other Regions:
Neuroticism
Anterior Temporal Cortex (A38)	Right	48	15	−6	−.373	.678**

Statistical significance is as follows:

^*p < .001,

^**p < .0001. Note that for the sites where there were significant correlations with extraversion, there were weaker non-significant correlations in the opposite direction with neuroticism. For sites where neuroticism was significantly correlated with cortical thickness there were weaker non-significant correlations in the opposite direction with extraversion. The right superior frontal cortex finding in extraversion was located in the sulcus (s), but located in the gyrus (g) for neuroticism.

No region outside of the prefrontal cortex displayed statistically significant relationships between thickness and extraversion.

We compared the results from the right SFC with those from the corresponding point in the left SFC there was a statistically significant hemispheric difference between the correlations (right vs. left: r=.643 vs. r=.007, p=.007). Comparison of the results from the left MFC with those from the corresponding point in right MFC also indicated a statistically significant difference between the correlations (left vs. right: r=.638 vs. r=.079, p=.0033).

Correlations with Neuroticism and Cerebral Cortex Thickness

Surface bases analyses of the right hemisphere indicated significant inverse correlations between cortical thickness and neuroticism T-scores in the right superior and inferior frontal cortices (SFC, IFC; Fig. 2; Table 2). Scatter plots confirmed the significant negative correlation between neuroticism and cortical thickness in these regions (Fig. 2), but demonstrated no significant relationship with extraversion (Table 2). Of note, the right SFC site that was inversely correlated with neuroticism was located in gyral cortex and exhibited a modest but non-significant positive correlation with extraversion (Table 2). This site was in very close approximation to the right sulcal SFC region that was significantly positively associated with extraversion. When the correlation with neuroticism at identical points in the right and left SFC and IFC were compared, a statistically significant difference between hemispheres was present (right vs. left SFC: r=−.605 vs. r=−.061, p=.021; right vs. left IFC: r=−.607 vs. r=.051, p=.0014).

Figure 2. Significant Correlations of Cortical Thickness and Neuroticism in the Right Hemisphere.

(A) Colorized statistical map superimposed upon a partially inflated group average cortical surface. The lateral aspect the right hemisphere is shown. Significant inverse correlations of cortical thickness with neuroticism were found in the superior and inferior frontal cortex (SFC, IFC). A significant positive correlation between neuroticism and cortical thickness was found in the anterior temporal cortex (ATC). Dark gray regions are sulci, light gray are gyri. Colorized scale bars show the p-value for positive (red-yellow) and negative (blue) correlations. (B) Scatter plot and regression line demonstrating a significant inverse correlation between SFC thickness and neuroticism. These values were extracted from the peak surface point of the SFC locus shown in (A). (C) Scatter plot and regression line from the peak of the IFC locus in (A). (D) Scatter plot and regression line from peak of the ATC locus in (A).

One area in the right hemisphere outside of the PFC met the significance threshold for non a priori regions. This was in the mid-portion of the anterior temporal cortex (ATC), where thickness was positively correlated with neuroticism, but not extraversion, scores (Fig. 2; Table 2). When the correlations at identical points in the right and left were compared, a statistically significant difference was present (right vs. left ATC: r=−.678 vs. r=.062, p=.0014).

No regions of the left hemisphere demonstrated significant correlations between cortical thickness and neuroticism.

Multiple Linear Regression Models of Cortical Thickness, Extraversion and Neuroticism

For the first model, extraversion was entered as the dependent variable, with right SFC and left MFC thickness as the independent variables. The analysis was performed in a stepwise forward multiple linear regression model (p to enter model <.05). Age and sex were forced into the model a nuisance regressors (covariates). Result of this model indicated that both regions were significant predictors (overall model, F=11.6 p<.0005). For the second model, neuroticism was entered as the dependent variable, and analyses were performed similar fashion as for extraversion except that the independent variables were right SFC (gyral portion), right IFC, and right ATC. In this case only two of the regions entered in the model were significant predictors of neuroticism: right ATC and IFC (overall model, F=9.13 p<.0005). The right inferior region did not enter the model because it was strongly correlated with the right SFC (r=.611, p <.0005).

Absence of Correlations with Extraversion, Neuroticism and Amygdala Volume

No significant basic correlations between right or left amygdala volume and extraversion or neuroticism (T or raw scores) were found (all |r|<.2, p>.4). Similar results were obtained when partial correlations adjusting for the effects of gender and age were utilized (|r|<.3, p >.2) and when males and females were analyzed separately.

Anatomic Relationships of the Other NEO Factors: Openness, Agreeableness, and Conscientiousness

A significant negative correlation between the thickness of the right inferior parietal cortex (angular portion) and openness was found (r=−.706; p=.000038; Talairach x=43, y=−74, z=12) (Fig 3). This effect was selective for the right hemisphere (R vs L, r=−.706 vs. −.122; p=.0064), and none of the other personality factors were correlated with cortical thickness at this site (all |r| < .3; p>.1). No significant associations were observed between agreeableness or conscientiousness and cortical thickness. In the left amygdala, there was a trend positive association between agreeableness raw scores and left amygdala volume (r=.48, p=.012; partial correlation corrected for gender and age). Otherwise no significant correlations were found between openness, agreeableness, or conscientiousness (T or raw scores) and cortical thickness or amygdala volumes.

Discussion

The current results demonstrate that specific subregions of the right and left lateral PFC are selectively associated with extraversion and neuroticism in elderly individuals. Specifically, increased thickness of the right SFC and left MFC was selectively associated with greater levels of extraversion in the elderly, while decreased thickness in the right SFC and right IFC was selectively associated with greater levels of neuroticism. In one area outside of the PFC - the right ATC – neuroticism, but not extraversion, was significantly positively associated with thickness. Multiple linear regression modeling confirmed that thickness of the above noted regions were predictive of extraversion and neuroticism even when accounting for the effects of age and gender. Of note, although the regions were selected on the basis of extraversion or neuroticism associations with cortical thickness, there were often modest correlations in the opposite direction for the factor not utilized in the initial region identification process. This effect was particularly of note in the right SFC since in the sulcal region there was a significant positive correlation with extraversion with non-significant effects for neuroticism in the opposite direction. In the closely adjacent gyral portion of the right SFC the neuroticism effects demonstrated statistical significance while weaker effects in the opposite direction were present for extraversion. Although these two factors are considered independent according to the five-factor model of personality, they are often inversely associated. These results suggest that the cortical regions identified represent certain aspects of neuroticism and extraversion that are divergent but related. For example, the tendency to experience positive emotion found with extraversion may be paired with the tendency to experience low levels of negative emotion as found with emotional stability (the inverse of neuroticism).

Structural Correlates of Extraversion Across the Age Range

Two core aspects of extraversion are positive affect and a tendency to seek out and participate in social situations (Lucas and Diener, 2001; Lucas and Baird, 2004). Both these aspects of extraversion are related to hemispheric asymmetries in the PFC, as lesser right vs. left-sided anterior electrophysiological activity is associated with positive affect and approach behaviors (Tomarken et al., 1990; Tomarken et al., 1992; Davidson, 2004b, 2004a). A metabolic study in young adults supports this notion, demonstrating that greater extraversion is associated with lower resting activity in the right PFC (Johnson et al., 1999). Likewise, structural studies in young adults indicate that specific prefrontal cortical regions in extraverts are smaller in the right vs. the left hemisphere (Omura et al., 2005; Wright et al., 2006b). In contrast with those findings in young adults, the present study demonstrates that extraversion in elderly adults is associated with greater cortical thickness in specific regions of both the right and left PFC. Of note, the specific PFC subregions related to extraversion in the elderly were located more superiorly and anteriorly compared to those in the young. Interestingly, inferior and posterior regions of the bilateral PFC demonstrate significant thinning with increasing age in adults (Salat et al., 2004). Taken together, these findings suggest that the neural correlates of extraversion in the prefrontal cortex are altered with aging, perhaps related to atrophy in PFC regions that subserve extraversion in youth. We speculate that, as in young adults, the extraversion-associated PFC regions in the elderly may also be important for regulation of social contacts (i.e. disinhibiting social contacts extraverts, but inhibiting them in intraverts) (Bone and Montgomery, 1970; Paisey and Mangan, 1980; Izard et al., 1993; Wright et al., 2006b), but that age-related anatomic changes, life-time experiences or both alter the direction and laterality of the exact structure-function relationships.

Neuroticism, Cortical Thickness and Aging

An inverse relationship between neuroticism and left medial orbitofrontal cortex thickness was previously reported in healthy young subjects (Rauch et al., 2005; Wright et al., 2006b). Existing evidence indicates that there is extensive individual variation in OFC, particularly in more anterior areas where complex, abstract, or secondary reinforcers are represented (Kringelback & Rolls, 2004). Furthermore, the structural integrity of the OFC is important for the reappraisal of the affective or emotional significance of stimuli. Damage to the OFC in human and other primates impairs learning and reversal of stimulus-reinforcement associations, as well as impairing extinction to conditioned stimuli (Rolls et al., 1994; Rolls, 2004). However, our initial finding in young adults was not reproduced in the current cohort of elderly subjects. Instead we found that there were inverse relationships present between neuroticism and the thickness of specific regions of the right lateral PFC. Of note, the thickness of the medial OFC may increase with age (Salat et al., 2001; Salat et al., 2004; Grieve et al., 2005). Furthermore, research suggests that aging leads to decreased cognitive flexibility due to impaired reversal learning in the OFC (Schoenbaum et al., 2006). This is particularly interesting given that excessive sensitivity to negative cues in the environment is a core feature of neuroticism (Bolger and Schilling, 1991; Mroczek and Almeida, 2004). Taken together, these studies suggest that, as in the case for the lateral PFC and extraversion, age-related OFC structural changes may have led to a shift or change in the cortical thickness-personality relationships we observed. This could be due to age-related (as opposed to personality-related) functional and structural changes in these cortical regions interfering with the ability to detect these relationships. Alternatively, in the setting of the general preservation of personality traits that occurs with aging (Costa and McCrae, 1977, 1986, 1997; McCrae et al., 1999; Costa et al., 2000), age-related dysfunction in the OFC may lead to the compensatory involvement of other cortical areas. Another possibility is that the small decreases in neuroticism that may occur with aging (Costa et al., 1986; McCrae et al., 1999; Costa et al., 2000) result in a change in the underlying structural mechanisms supporting this specific personality factor, thus leading to a weaker relationship between neuroticism scores and OFC thickness. Conversely, the structural changes that occur with aging could lead to decreases in neuroticism levels.

Although OFC thickness was not significantly correlated with neuroticism, there were significant negative correlations with right SFC and IFC thickness, as well as significant positive correlations with right ATC thickness. There is very limited literature on relationships between neuroticism and the structure and function of the right lateral PFC or right ATC with aging. Work in young adults suggests that facets of neuroticism are associated with low levels of serotonin receptor binding in the dorsolateral PFC (Tauscher et al., 2001), and regions IFC have been implicated in self-regulatory control (Marsh et al., 2006) and decoding emotional intonation (Wildgruber et al., 2005). Thus, both neurochemical and functional alterations potentially relevant for neuroticism have been described in the lateral PFC, but additional work is needed to understand how these findings relate to the structural effects reported in the current study. With respect to the right ATC, a recent study suggests that damage to this region leads to decreased memory for unpleasant life events (Buchanan et al., 2006). On this basis we speculate that the positive correlations we observed between neuroticism and right ATC thickness could relate to an increased tendency to recall negative life events (under the assumption that increased thickness underlies functional enhancement). This notion, which also requires additional investigation, is consistent with the observation that those with high levels of neuroticism have a greater sensitivity for recalling events as negative (Schneider, 2004).

Openness, Self Transcendence and the Aging Brain

We are aware of only one earlier study reporting that brain structural differences in the elderly are associated with measures of personality. That study demonstrated grey matter voxel density changes in bilateral perisylvian temporal, parietal, and frontal cortices relate to measures of self transcendence (Kaasinen et al., 2005) using the Temperament and Character Inventory (TCI). However, the present study demonstrates selective structural correlates (i.e. cortical thickness) of personality using the Five Factor Model.

Self-transcendence is one of the three character traits of the TCI. It is characterized by individual openness to unusual and divergent feelings, thoughts, and behaviors, and as such may be most closely related to the openness measure of the FFM. The finding in the present study that openness is related to the thickness of the angular gyrus in the right inferior lateral parietal cortex potentially overlaps anatomically with the region identified in the self-transcendence study. However, although the anatomic localization of these findings was similar, the direction of the effects appears discordant at first glance. We observed an inverse relationship between openness and cortical thickness, but a positive relationship between self-transcendence and cortical gray matter density was described in the other study. It is uncertain how these two structural measures relate to each other, and these apparently discrepant findings may be complementary. For example, decreased cortical thickness could lead to local morphological or structural MRI signal intensity changes that increase the likelihood that voxel-based morphometry (VBM) will assign these voxels as gray matter. In this case, the openness results of this study and the self-transcendence results (Kaasinen et al., 2005) may indicate related underlying structural effects.

Amygdala Structural Metrics and Personality

The structure and function of the amygdala have been associated with measures of extraversion and neutroticism (Johnson et al., 1999; Canli et al., 2002; Omura et al., 2005) as well as with uninhibited and inhibited temperaments (Schwartz et al., 2003). However, we did not find any significant relationship between amygdala volume and these two personality factors in the current study of elderly individuals, or in an earlier study on young subjects (Wright et al., 2006b). There are several potential reasons for the apparent lack of a link between extraversion, neuroticism, and amygdala volumes in the current study. For example, whole amygdala volumes measurements may obscure variations at the level of individual subnuclei. Future studies examining the volume of specific amygdala subregions with very high resolution MRI will be of interest and should help to understand discrepancies between techniques utilizing MRI gray matter voxel density versus amygdala volume.

Potential Limitations, Future Directions, and Clinical Implications

Although the participants performed well on the MMSE and were well characterized from the psychiatric perspective using a structured clinical interview, more extensive cognitive evaluations or confirmatory informant interviews were not performed. This leaves open the possibility that some subjects may have had mild cognitive impairment and that this could have contributed to the findings. The main focus of this initial study was to determine if structural correlates of neuroticism and extraversion exist in the elderly, given the anatomic variability that accompanies aging. The current results support the existence of brain structure-personality correlates in the elderly, and suggest differences in these correlates in the young and elderly. However, the current elderly cohort was not prospectively matched for the personality measures of interest with the young subjects from our earlier study (Wright et al., 2006b), and as such we could not correctly perform direct between groups comparisons. Thus, our results may relate to cohort-specific differences in the distribution of personality measures between the two different cohorts. Future works in young, middle aged, and elderly adults with matching personality characteristics should examine structural (and functional) measures to directly explore the effects of aging on the neural correlates of personality.

Even matching personality measures between groups does not exclude the possibility of cohort effects in studies using cross-sectional design. As with most neuroimaging studies, this current study is cross-sectional in nature. Therefore, the results cannot be used to determine if the cortex structure-personality associations in the elderly reflect changes during a particular individual’s lifetime. Longitudinal studies are required to conclusively investigate aging effects on these structure-personality relations. Furthermore, ultra-high resolution structural MRI, postmortem studies, or both are needed to unravel the possible mechanisms of the anatomic effects we describe. These techniques may address important issues such as whether personality factor-brain structure associations relate to changes in specific cortical layers; in the cellular or subcellular composition of the cortex; or in the subnulcear or subregional structures of subcortical areas (e.g. amygdala).

One implication of having different cellular sources as the underlying cause of an anatomic size difference (e.g. in thickness or volume) is that the consequences for a size difference are not invariant. For example, specific subregions of the frontal lobe may be related to social approach or avoidance. In such a case enhancement or decrement of frontal lobe functions could lead to greater approach behaviors (via diminished inhibitory systems) in extraverts relative to introverts. As our approach at this point is purely anatomic, we cannot determine whether thinning (or thickening) might related to enhance or diminished cortical functioning. Furthermore, thinning due to loss of inhibitory interneurons, might enhance cortical function and activity (via disinhibition); alternatively, thinning due to decreased local cortical connectivity or neuron loss might lead to decreased cortical functioning. On this basis specific predications about the consequences of the direction of the thickness-personality measure relationship must be interpreted with caution. Additional studies assessing structure-function relationships and examining the microstructural contributors to cortical thickness variation are necessary to further address these functional meaning of the personality-structural metric associations we describe.

Finally, extraversion and neuroticism may play an important role in the development and expression of late-life psychiatric disorders in the elderly (Burvill et al., 1991; Henderson and Kay, 1997; Lyness et al., 1998; Seidlitz et al., 2001), and regions of the PFC, parietal and temporal cortices exhibit structural and functional changes in geriatric depression and psychosis (Kumar et al., 1997; Awata et al., 1998; Pantel et al., 1998; Ashtari et al., 1999; Kim et al., 1999; Thomas et al., 2003; Ballmaier et al., 2004; Aizenstein et al., 2005). Going forward it will be of clinical interest to investigate how the anatomical changes associated with specific personality factors in aging relate to the propensity for late-life neuropsychiatric syndromes. However, conceivably, it is only the abnormal representation of traits that might predispose to specific psychiatric disorders. In this regard it may be particularly important to focus on the extremes of personality for such studies.