Locus Coeruleus Is Associated with Higher Openness to Experience and IQ: Implications for the Noradrenergic System for Novelty Seeking in Daily Life Open Access

A person's ability to flexibly deploy their cognitive resources to offset the effects of natural age-related decline or neurodegenerative disease is described as cognitive reserve (CR; Stern et al., 2020; Cabeza et al., 2018; Stern, 2009). There are important interindividual differences in cognitive ability emerging in younger adults that are shaped by lifetime experiences, some of which may offer protection from neuropathology later in life (Livingston et al., 2020). The main proxies of CR are sociobehavioral variables including educational status, occupational attainment, IQ, and cognitive efficiency measured using standardized neuropsychological tests (Stern et al., 2020). However, several findings (Karsazi et al., 2021; Ihle et al., 2019; Franchow, Suchy, Thorgusen, & Williams, 2013; Cuttler & Graf, 2007) suggest that a particular personality trait—openness to experience (OE)—may offer protection from the detrimental effect of neuropathology later in life. A possible explanation is that because high openness promotes a receptivity to novel experiences, which are pursued with high interest, this necessitates flexible cognitive engagement with the world. Conversely, a person who scores lower on the openness trait is more inflexible in their daily routine and more fixed in their beliefs and attitudes (Coleman, Furnham, & Treglown, 2023; Rammstedt, Lechner, & Danner, 2018; Schretlen, van der Hulst, Pearlson, & Gordon, 2010; Moutafi, Furnham, & Crump, 2006; Ackerman & Heggestad, 1997; Costa & McCrae, 1992).

Scoring higher on the OE trait is also associated with greater IQ levels (fluid and crystallized/verbal) in both younger and older adults (Rammstedt et al., 2018; Schretlen et al., 2010; Ackerman & Heggestad, 1997) and is inversely related to Alzheimer's disease (AD) biomarkers (Tautvydaitė et al., 2017), suggesting that this personal disposition might be related to better cognitive and biological health outcomes. Similarly, a study of 845 older adults found that, over a period of 20 years, OE was found to be protective against cognitive slowdown (Sharp, Reynolds, Pedersen, & Gatz, 2010). Openness is also not entirely fixed in its expression over time, and it has been demonstrated that 30 weeks of a cognitive training intervention can increase scores on the OE scale (Jackson, Hill, Payne, Roberts, & Stine-Morrow, 2012). Taken together, these findings suggest that openness is an important dispositional factor that is related to key CR proxies and has a role in shaping the resilience trajectory in the course of healthy and pathological ageing.

The extent to which the personal disposition of openness and the construct of reserve are related through a common neurobiological mechanism is not currently known. However, Robertson (Robertson, 2013, 2014) has postulated that more frequent activation (and the related integrity) of the locus coeruleus (LC)—noradrenergic system over the lifespan (the main source of noradrenaline [NA] in the brain) could be a key candidate affecting CR and resilience capabilities. Specifically, it is hypothesized that the frequent upregulation of the noradrenergic system through cognitive stimulation and exposure to novelty might be one of the key neurobiological components for building CR and, therefore, resilience to neurodegenerative pathologies across the lifespan. In the last decades, several studies reported that the LC-NA system integrity was related to greater brain and cognitive health (Plini et al., 2021, 2023, 2024; Dahl et al., 2022a, 2022b; Elman et al., 2021; Jacobs et al., 2021; Wilson et al., 2013), particularly to better attentive and mnemonic functions both in healthy and clinical populations (Plini et al., 2021, 2023, 2024; Dutt, Li, Mather, Nation, & Alzheimer's Disease Neuroimaging Initiative, 2021; Dahl et al., 2019; Clewett et al., 2016).

There are multiple explanations for the association between OE and specific activation patterns of the LC-NA system. First, openness promotes exploration of our environment in pursuit of novelty (https://dictionary.apa.org/openness-to-experience). The LC appears to underlie novelty signaling patterns (Krebs, 2013, 2018; Robertson, 2013, 2014; Devilbiss & Waterhouse, 2011), which project to new network configurations within the hippocampus when encoding new versus familiar experiential episodes (specifically resetting of hippocampal maps using cellular compartmental analysis in rats: Grella et al., 2019). It has also been demonstrated in animal studies that exposure to repeated novel stimuli is a key factor underpinning the benefits of environmental enrichment upon memory function (Grilli et al., 2009; Naka, Shiga, Yaguchi, & Okado, 2002; Sara, Vankov, & Hervé, 1994). Blocking noradrenergic transmission pharmacologically disrupts the benefits of this novelty recognition (Veyrac et al., 2009; Sara et al., 1994). In the longer term, repeated phasic NA responding causes greater LC innervation and metabolic health, increasing autophagic factors, which are linked to reduced AD pathology (Omoluabi et al., 2021). Second, a receptive and open mind depends on cognitive flexibility (Chen, He, & Fan, 2022; Robertson, 2013, 2014). The LC-NA system is hypothesized to underlie an antagonistic relationship between exploitative versus explorative states. The adaptive gain theory (Devilbiss & Waterhouse, 2011; Aston-Jones & Cohen, 2005) proposes that a phasic LC activity is driven by an exploit mode for prioritizing goal- or task-related information. When the utility of the current goal wanes, a tonic pattern of LC activity emerges that is associated with an exploratory mode promoting search for other alternative behaviors. Flexibly shifting between explore and exploit modes is a neurobehavioral mechanism that could plausibly support the trait of openness, which also varies along an exploration–exploitation dimension (these cognitive operations are typically also involved in problem-solving capability and IQ expression—Birney & Beckmann, 2022; Robertson, 2013, 2014; Colom, Karama, Jung, & Haier, 2010). Narrowing of openness is associated with exploiting familiar routines and behaviors, whereas an expansion of openness is an exploratory mode promoting curiosity-led behaviors (Herz, Baror, & Bar, 2020; Costa & McCrae, 1985, 1992).

The aim of the current study is to test the noradrenergic theory of CR by investigating whether the signal intensity (a parameter of tissue density—integrity) of the LC in humans (measured by voxel-based morphometry; Ashburner & Friston, 2000) is related to the expression of the OE trait measured by using the Neuroticism Extraversion Openness (NEO)-Five Factor Personality Inventory (FFI; Costa & McCrae, 1985, 1992). Specifically, it is hypothesized that people with higher expression of the OE trait will have greater LC signal intensity (proxy of tissue density—namely, interpreted as parameter of structural integrity/neuronal health) because they are more likely to engage with a set of noradrenergically mediated behaviors that promotes neuroplasticity within the LC-NA system (Mather, 2021; Robertson, 2013, 2014). Personality is a relatively stable but malleable construct that is associated with daily behaviors (Sutin, Ferrucci, Zonderman, & Terracciano, 2011; Fleeson & Gallagher, 2009; Martin, Friedman, & Schwartz, 2007; Rhodes & Smith, 2006) and is well positioned to interact with the LC-NA system to build reserve and neuroprotective effects against disease from a young age (Matchett, Grinberg, Theofilas, & Murray, 2021; Sutin et al., 2018; Mather & Harley, 2016; Robertson, 2013, 2014). Indeed, it is known that AD biomarkers are found in healthy brains even from the second and third decades of life (Braak, Thal, Ghebremedhin, & Del Tredici, 2011). A secondary aim of this study was to investigate whether LC signal intensity and IQ are related, and whether the mediation of the LC-NA system may help explain the relationship between the OE trait and IQ, which is already reported in the literature (Anglim et al., 2022; DeYoung, 2020; Rammstedt et al., 2018; DeYoung, Quilty, Peterson, & Gray, 2014; Bartels et al., 2012; Schretlen et al., 2010; Moutafi et al., 2006; Ackerman & Heggestad, 1997).

The current study will apply voxel based morphometry (VBM) analyses utilizing 3-Tesla T1-weighted MRI scans from 135 healthy young participants (age range = 20–35; 96 males, 39 females) derived from the LEMON (Leipzig Study for Mind–Body–Emotion Interactions) data set (Babayan et al., 2019). The relationship between the structural signal intensity of the LC and the OE trait was examined, and as a control procedure, the relationships between the LC and the other four dimensions of the Big Five personality inventory were also tested. Finally, to assess the differential contribution of neuromodulators to the OE trait, the specificity of the noradrenergic hypothesis was tested using VBM to measure the relative involvement of the other main neuromodulatory subcortical nuclei projecting to the cortex (dorsal raphe [DR] and median raphe [MR] 5-HT, ventral tegmental area [VTA] DA, and nucleus basalis of Meynert [NBM] Ach) to the expression of the OE trait. The same models were repeated while investigating the relationship between LC and IQ.

The data were provided by Max Plank Institute (https://www.mpg.de/en) from the LEMON data set by Babayan and colleagues (2019). The behavioral data were downloaded from nitrc.org (https://fcon_1000.projects.nitrc.org/indi/retro/MPI_LEMON.html accessed in July 2020), and the raw MRI data were downloaded from the following source between July and September 2020 https://fcon_1000.projects.nitrc.org/indi/retro/MPI_LEMON/downloads/download_MRI.html. Because the LEMON data are organized in different Excel files, the behavioral data including neuropsychological, psychological, and medical variables were all manually merged as to build a comprehensive sheet for all the participants comprised in the LEMON initiative. The analyses were carried out only on the participants in the age range between 20 and 35 years (39 female, 96 male). Participants with current psychological and psychiatric disorders, such as anorexia and depression and substance abuse, were excluded. The LEMON data set also includes a smaller sample of older adults (aged over 60 years). However, because of the limited number and imprecise age information, we opted to concentrate solely on younger participants. This approach allows for a more focused investigation into the nature of our hypothesis, free from potential interference from age-related brain effects and statistical concerns associated with smaller sample sizes and an unbalanced design with younger participants.

Personality traits were assessed using the NEO-FFI questionnaire by Costa and McCrae (1992). The NEO-FFI is a 44-item self-report measure on a Likert scale (1–5) investigating the five main components of personality: OE, conscientiousness (C), extraversion (E), agreeableness (A), and neuroticism (N). These five domains are derived from a set of questions that assess emotional, interpersonal, experiential, attitudinal, and motivation styles, and the interaction between these five dimensions defines personality profiles. Greater values reported reflect greater adherence to the items of the five facets. For example, individuals who score high in conscientiousness tend to be well organized and less impulsive, whereas those who score high in extraversion are inclined to be assertive and highly sociable. On the other hand, individuals with high scores in N tend to be irritable, shy, and moody whereas low scores in A indicate antagonism, lack of compassion, and stubbornness. As described earlier, high scores in OE are informative about the disposition to be open minded, prone to creativity, and willing to engage new experiences. Conversely, people who score low in OE are generally disposed to routine and to conservative thinking and tend to avoid changes and challenges.

IQ was measured using the WST (Wortschatztest) vocabulary test (Schmidt & Metzler, 1992), which is a multiple-choice vocabulary test capable to estimate verbal “crystalized” intelligence on the base of vocabulary knowledge (similar to the English National Adult Reading Test and Italian Test di Intelligenza Breve). The test consists of discriminating nonexisting words among real words. The decision to use crystallized IQ arises from the fact that a previous study found it associated with OE (Rammstedt et al., 2018; Schretlen et al., 2010; Ackerman & Heggestad, 1997) and because it is conceived as a proxy of CR and several studies on CR have used vocabulary-based IQ. (Livingston et al., 2020; Stern et al., 2020; Cabeza et al., 2018; Stern, 2009). Therefore, because the background of our study falls within the concept of reserve, we have decided to not focus on fluid, which is more closely associated with the actual neuropsychological functioning. In Table 1 the average personality trait scale levels and IQ values together with total intracranial volume (TIV) are reported.

3-Tesla high-resolution T1-weighted images in NIfTI format underwent a preliminary manual quality control to detect major motion or reconstruction artifacts. This procedure was carried out referring to the rating scale guidelines (1 = poor, 2 = fair, 3 = good, 4 = excellent) of the Human Connectome Project (https://www.humanconnectome.org/). Participants with low definition (excessive blurriness) and/or marked ringing, inhomogeneities, and motion artifacts were removed from the data set. To perform volumetric analyses, the images were processed using CAT12 (Computational Anatomy Toolbox: https://www.neuro.uni-jena.de/cat/) implemented in SPM12. The segmentation was run following the default CAT12 settings, except for the voxel size that was settled at 1-mm isotropic voxel size, because LEMON scans were gathered with 1-mm³ voxel size. On completion of this process, 135 young participants (96 male, 39 female) out of 153 were considered suitable for the analyses. The processed and modulated whole-brain images (Montreal Neurological Institute [MNI] gray matter + white matter) were smoothed using SPM12 interface with a 2-mm³ FWHM kernel according to previous studies investigating the LC-NA system with MPRAGE T1 scans (Plini et al., 2021, 2023, 2024; Tang et al., 2023; Dutt et al., 2021; Dutt, Li, Mather, Nation, & Alzheimer's Disease Neuroimaging Initiative, 2020). Lastly, to better account for individual volumetric variability in the VBM analyses, the TIV was calculated for each subject using CAT12 interface (Statistical Analyses – Estimate TIV).

As described in detail in our previous works (Plini et al., 2021, 2023, 2024), the ROIs were developed on the base of previously published atlases. The ROIs were symmetrically corrected from the original atlases to avoid overlapping borders with other structures and to avoid possible biases of “induced lateralization”. Indeed, since these neuromodulators are symmetrical nuclei (Mai & Paxinos, 2012), we stressed the need of using symmetrical ROIs to better reflect the histological properties of these nuclei. All the ROIs were 1 mm³ isotropic voxel size and oriented in the MNI space as the processed images of CAT12. See Figure 1 for ROI's volumetric info. For the current study we used the LC “omini-comprehensive” probabilistic mask we developed earlier (Plini et al., 2021). The “omni-comprehensive” LC mask solves the inconsistent LC spatial localization reported by previous works (Dahl et al., 2019, 2022b; Rong et al., 2020; Liu et al., 2019; Betts, Cardenas-Blanco, Kanowski, Jessen, & Düzel, 2017; Tona et al., 2017; Keren et al., 2015; Keren, Lozar, Harris, Morgan, & Eckert, 2009), while enabling to comprise the whole anatomical LC regions defined across lifespan, without encroaching other pontine and cerebellar regions, and without crossing the walls of the fourth ventricles (further details can be found in supplementary materials of Plini et al., 2021 and at this link: https://www.youtube.com/watch?v=90bsA6Jqxs4). The MR and DR ROIs were provided by Beliveau and colleagues (2015), the VTA mask was obtained by downloading the VTA MNI probabilistic map from the atlas made by Pauli, Nili, and Tyszka (2018) from the NeuroVault website (https://neurovault.org/ accessed on 15 December 2018). The NMB was developed on the base of the probabilistic MNI maps of the acetylcholine cells of the Forebrain are provided by SPM Anatomy Toolbox 2.2c (https://www.fzjuelich.de/inm/inm1/EN/Forschung/_docs/SPMAnatomyToolbox/SPMAnatomyToolbox_node.html accessed on 15 December 2018) by Koulousakis, Andrade, Visser-Vandewalle, and Sesia (2019); Schulz, Pagano, Bonfante, Wilson, and Politis (2018); Liu, Chang, Pearce, and Gentleman (2015); Kilimann and colleagues (2014); George and colleagues (2011); Zaborszky and colleagues (2008).

As earlier studies investigating the neuromodulators with magnetization prepared rapid gradient echo T1 images (Plini et al., 2021, 2023, 2024; Tang et al., 2023; Dutt et al., 2020, 2021; Jethwa, Dhillon, Meng, Auer, & Alzheimer's Disease Neuroimaging Initiative, 2019; De Marco & Venneri, 2018; Schulz et al., 2018), the processed and smoothed whole-brain images were used for the VBM analyses in CAT12 (basic models -> implementing multiple regression models). The analyses were divided into two main branches. The first one investigated the LC-NA system and the five personality traits of NEO-FFI in five different statistical models (OE, conscientiousness, extraversion, agreeableness, and neuroticism). The second one investigated the relationship between the LC-NA system and IQ (WST – vocabulary). All the analyses followed the same procedures and had the same set of continuous covariates (age, TIV, sex). After the estimation of the statistical models (“estimating the statistical model”) and after checking for the design orthogonality (“checking for design orthogonality”), the “results” were checked using the following contrast in the SPM12 “contrast interface”: 0 0 0 0 1 (for positive relationship) and 0 0 0 0 −1 (for negative relationship). Both relationships were always tested as to control the reliability of the findings. A further step in the analyses pipeline was to indicate the LC binary mask as an inclusive mask to isolate the LC involvement in the models. Eventually, the statistical threshold was settled at p < .01 and later increased progressively until the results disappeared (namely, p < .001, p < .05 FWE). Each set of voxel-wise analyses, after having examined the LC, was systematically repeated considering the other ROIs' binary masks in this order: DR, MR, VTA, and NBM as controlling assurance for the LC findings.

To perform Bayesian analyses (ROI-based analyses), the average signal intensity of the ROIs was extracted in FMRIB Software Library by using the binary masks on the smoothed (2-mm³ FWHM kernel) whole-brain images. In FMRIB Software Library terminal, the flags of “fslstats,” “-k,” (mask), and “-m” (output mean) were used to gather the average voxel intensities for each ROI subject by subject. This procedure was also carried out to extract the average signal intensity from clusters of significant voxels outcoming from the VBM analyses to calculate more accurate Bayes Factors (BF) and to perform mediation analyses.

In JASP (https://jasp-stats.org/), a Bayesian model was performed to compare within the same model the differential strengths of the neuromodulators predicting OE (linear regression model, OE entered as dependent variable, and the five ROIs as covariates). TIV, age, and sex were also included and added to the “null model.” The differential relationship of the five ROIs were compared with the null model, and the analyses were run with default parameters with the only exception of Bayesian information criteria selected in the advanced option of the JASP interface. The same models were run for the other personality traits.

Both Bayesian and Pearson's correlation matrices were built considering the following continuous variables: OE, conscientiousness, extraversion, agreeableness, neuroticism, WST-IQ, and the five personality traits with the STAI-G-X2 anxiety scale and Perceived Stress Questionnaire (PSQ). Similarly, STAI-G-X2 and PSQ were entered in a correlation matrix with LC, MR, DR, MR, VTA, and NBM to explore a possible relationship with the neuromodulatory subcortical system. BFs were generated in the JASP interface (Bayesian correlation model) on the base of the average signal intensity of the significant clusters outcoming from the VBM analyses (when the ROIs were surviving the statistical thresholds settled). The average signal intensity of the whole ROI was considered in the Bayesian correlation matrices only for the ROIs, which showed no significant cluster of voxels in the VBM analyses. Finally, partial correlations were performed in SPSS 25 (https://www.ibm.com/products/spss-statistics). The relationships between LC and IQ were controlled for OE in two different batches: First, from the LC–IQ relationship, OE was regressed out, and then from the LC–OE relationship, the effect of IQ was regressed out. These analyses had the aim to better differentiate the LC variance across these different domains.

Last, to better investigate the association between OE and LC signal intensity, in an ANCOVA model, the OE trait was treated as independent factor divided in three levels (low, mid, and high OE expression) and the LC was treated as a dependent continuous variable while controlling for age, sex, and TIV. Post hoc analyses both applied Tukey and Bonferroni corrections (these results are provided in the Appendix Tables A1 and B1).

Mediation analyses with parallel multiple mediators (Model 4) was carried out in SPSS using the PROCESS toolbox by Andrew Hayes. OE was entered as predictor (X) and IQ (Y) as outcome in accordance with our framework conceiving that greater OE may induce greater IQ levels because of novelty exposure and cognitive stimulation via the noradrenergic system. This model is also consistent with certain evidence considering IQ as a subcomponent of OE and while being related to OE across various studies (Anglim et al., 2022; DeYoung, 2020; Bartels et al., 2012; Moutafi et al., 2006).

As multiple mediators, the average signal intensity of LC, MR, DR, VTA, and NBM were entered. The models were covaried for age, TIV, and sex. A standard model was set with 95% confidence intervals and 10,000 bootstrap samples.

As can be observed in Table 2, across the 135 healthy participants, LC signal intensity was the strongest predictor of the OE personality trait compared with the other neuromodulators, while controlling for age, sex, and TIV. Two large bilateral clusters, for 153 voxels within the LC region, positively related to greater OE values for p < .01 threshold. As shown in Figure 2, the spatial localizations of the LC results overlapped the LC region defined by the previously published LC atlases and masks (García-Gomar et al., 2022; Dahl et al., 2019, 2022b; Rong et al., 2020; Liu et al., 2019; Betts et al., 2017; Tona et al., 2017; Keren et al., 2009, 2015). By increasing the statistical threshold up to p < .001, only a cluster of 22 voxels survived in the LC core. However, no LC voxels remained when FWE correction was applied.

Regarding other personality traits, 67 voxels within the LC region were associated with conscientiousness. Only five voxels of the LC clusters survived when more conservative thresholds were set (p < .001); however, the cluster did not survive FWE correction. In Table 2, the minor and negligible associations between the five ROIs and the other personality traits are reported in detail.

As control procedure, when the opposite relationships were tested, no associations were found between the LC and five personality traits, neither for the other ROIs. The only exception (not reported in tables) was the negative relationship between the VTA and conscientiousness. Lower VTA signal intensity (68 voxels) related to greater conscientiousness values. However, this result did not survive the p < .001 threshold.

As reported in Table 2, 23 LC voxels (for p < .01) mostly left lateralized rostrally were related to greater IQ. The LC spatial localization overlaps the LC core defined in the previously published LC atlases. When the statistical threshold was increased to p < .001, the cluster disappeared. No associations between IQ and the other ROIs were found, demonstrating specificity for the relationship between IQ and LC. No significant clusters across the five ROIs were found when the opposite relationships were tested. However, ROI-based analyses survived Bonferroni corrections (for more details, see Appendix Tables A1 and B1).

The abovementioned VBM analysis found that the LC was a key predictor of the OE trait. However, a further Bayesian approach (ROI-based analyses) provides a means to compare the strength of evidence for different neuromodulator seeds in predicting OE trait scores. Here, we employed a model in which each neuromodulator seed is examined as a standalone predictor and also in a combined way, which compares the predictive strength when different neuromodulator seeds also examined in combination if they explain more variance. The Bayesian linear regression model showed that as a standalone variable, the LC signal intensity has the strongest relationship with OE against the null model (BF₁₀ 338.401–BF_M 9.544). The combined effect for LC and VTA signal intensity has the strongest relationship among single and combined models (BF₁₀ 993.661–BF_M 33.699), suggesting a disproportionate catecholaminergic involvement in OE trait accounting for 14% (LC alone) to 18% (LC + VTA) of its variance. Moreover, the combined effect of LC and serotoninergic DR was the third strongest model related to OE, accounting for 17% of variance. A summary of these analyses is reported in detail in Table 3.

Correlation matrices revealed that among the five personality traits, only the OE scale related significantly with IQ. Specifically, greater OE values related to greater IQ scores (Pearson's r = .391, BF₁₀ = 3305.75, for p < .001). For greater detail, refer to the correlation matrices in Appendix Tables C1 and D1.

Partial correlations revealed that the relationship between LC and IQ disappeared when the effect of OE trait was regressed out in the model. On the other hand, the relationship between OE trait and LC remained significant when controlling for IQ. These two partial correlations together are suggestive that the relationship between OE and LC might be more stable than the relationship between LC and IQ. For greater detail refer to the partial correlation table of Appendix Table D1.

Mediation analyses with multiple parallel mediators were carried out to clarify possible mediation effects of the neuromodulators. Although controlling for age, TIV, and sex, bootstrap (10,000) confidence intervals were used to examine the role of the five subcortical ROIs in mediating the relationship between the OE trait and IQ. Only LC alone significantly mediates the relationship between OE (X) and IQ (Y), indicating that greater OE was predictive of greater IQ score and that this relationship was not mediated by the other neuromodulators. Indeed, the total indirect effect of the parallel mediators was not significant. These results suggest that the strong relationship found between OE and IQ is disproportionally influenced by the noradrenergic system over the other neuromodulators, as previously observed in the VBM analyses (see Table 4 for details).

The present study investigated the relationship between MRI signal intensity of different neuromodulators' seeds and the FFI score in 135 healthy young participants. As hypothesized, greater structural signal intensity of the LC was found to be related to higher scores in the OE trait and with a greater IQ score. When neuromodulator seeds were considered together in a Bayesian model, catecholamines (LC + VTA) were the strongest predictor of OE, followed by the LC as a stand-alone predictor of OE, and, third, the LC combined with the serotoninergic DR. These results are independent of the effects of sex, age, and TIV. These findings are the first evidence linking neuromodulatory MRI signal intensity to personality traits.

Given that the OE trait is commonly associated with intelligence (both fluid and crystalized/verbal) and is frequently named as the “intellect” trait in the literature (Anglim et al., 2022; DeYoung, 2013, 2020; Rammstedt et al., 2018; Bartels et al., 2012; Schretlen et al., 2010; Moutafi et al., 2006; DeYoung, Peterson, & Higgins, 2005; Ackerman & Heggestad, 1997), we examined the extent to which LC-NA signal intensity might account for this relationship. Mediation analyses revealed that, in comparison with the other neuromodulatory nuclei, only the LC accounted for the relationship between OE and IQ, supporting the assertion that the LC-NA system may be a key neurobiological substrate mediating the relationship between OE and intellectual expansion. We also established that the relationship between the LC and OE remained when the effect of IQ was regressed out, demonstrating that unique variance in the openness trait is associated with LC-NA system signal intensity. Finally, it was apparent that the LC was not exclusively associated with the OE trait. We also observed a weaker relationship between LC and the conscientiousness (C) scale.

We interpret the key findings of the current study in the context of the noradrenergic theory of CR (Krebs et al., 2013, 2018; Robertson, 2013, 2014), proposing that more protracted responses to novelty and chronic exposure to enriched environments in high-OE trait individuals are underpinned by the activation of the LC-NA system. Why might novelty be an important factor for increasing LC signal intensity? One potential underlying mechanism could be increased frequency of LC phasic firing in response to novel events. Ghosh and Omoluabi (Ghosh et al., 2021; Omoluabi et al., 2021) examined single unit recordings from the LC during novelty exposure in rats. After 6 weeks of LC stimulation (LC phasic activation for 20 min 5 days per week), rats showed increased response to novel stimuli and greater LC health (increased macrophage activation and greater LC fiber density), along with increased cognitive functions and reduced AD biomarkers (LC pretangle) in comparison with rats that did not undergo the LC phasic activation. Rats not exposed to novelty showed greater LC neurodegeneration with increased AD biomarker (LC pretangle) and poorer cognitive outcomes. Omoluabi and colleagues (2021) concluded that LC-phasic firing in response to novelty was protective against the detrimental effect of the AD biomarker (pretangle tau production in the LC), preventing LC fiber loss resulting in greater LC axonal integrity and more preserved cognition.

Consistent with the aforementioned animal study, a recent fMRI study in humans carried on 128 healthy individuals from the Harvard aging study (Prokopiou et al., 2022) found LC activity significantly increased while novel stimuli were presented. They also reported that a lower novelty-related LC activity was associated with greater cognitive decline related to AD biomarkers' levels (beta-amyloid PET). Previously, Krebs, Park, Bombeke, and Boehler (2018) also similarly demonstrated that phasic LC activity occurs when contextually unexpected stimulus are presented, namely, when a novel stimulus that is incongruent to context arises.

Collectively, these studies are suggestive that LC phasic firing and novelty exposure are associated with greater LC white matter integrity, reduced AD biomarkers, and better cognitive health. Because greater MRI signal intensity reflect greater tissue density (namely, greater density of gray matter cells and white matter fiber innervations), the current study provides evidence of how chronic novelty exposure in human may affect LC-NA structural variations and cognition, possibly resulting in greater brain and cognitive health consistently as reported by Omoluabi and colleagues (Omoluabi et al., 2021). See Figure 3 for a summary of the proposed neurobiological mechanism underlying the association between LC MRI signal intensity and OE.

Although the VBM analyses revealed that the LC was the key predictor of OE, a Bayesian multiple regression model explored the combined effects of LC with the other subcortical nuclei. This analysis revealed that LC + VTA was a stronger predictor of OE than the LC alone. An additive involvement of the dopaminergic system in shaping personality may explain this relationship: Exploratory behavior that characterizes openness will be further reinforced by the reward value of novel experiences via the mesolimbic dopamine system. Alternatively, the role of DA in supporting the OE trait may have a genetic basis as catalyst for the conversion of DA to NA. (Barnes, Dean, Nandam, O'Connell, & Bellgrove, 2011). Furthermore, the role of the DR nucleus, although only a negligible contributor to the model on its own, also combined with the LC to predict OE trait variation. Serotoninergic involvement in OE trait expression may involve increased sensitivity to stress during exploration of novel environments. A previous work has found that cerebral 5-HTT levels are associated with OE (Ren et al., 2021; Kalbitzer et al., 2009). However, 30% of DR is composed of catecholaminergic neurons (Kirby, Pernar, Valentino, & Beck, 2003; Ordway, Stockmeier, Cason, & Klimek, 1997; Baker et al., 1991; Farley & Hornykiewicz, 1977); therefore, this association might reflect common dopaminergic and noradrenergic involvement in OE expression. Overall, the synergic involvement of these three main neuromodulatory nuclei may contribute to shaping the nature of openness via different pathways—through VTA-DA mediated reward sensitivity, DR-5-HT potentiation of stress-related responses during exposure to novelty, as well as overarching explore–exploit behavioral patterns promoted by the LC-NA system (Ren et al., 2021; Zmorzyński, Styk, Klinkosz, Iskra, & Filip, 2021; DeYoung, 2013; DeYoung et al., 2011; DeYoung, Peterson, Séguin, & Tremblay, 2008; Tochigi et al., 2006). We observed positive relationships between the signal intensity of such nuclei and OE; therefore, these interpretations are based on the assumption that greater structural signal intensity (tissue density – integrity) of these nuclei would ensure an adequate neuromodulatory functioning building the ground for a more developed OE trait. Conversely, lower signal intensity (reduced tissue density – integrity) of these nuclei would undermine the normal neuromodulator biosynthesis underlying OE trait expression. This interpretation is supported by the literature reporting how variations of neuromodulators' concentrations can bidirectionally affect personality traits expression (Kanen et al., 2021; Käckenmester, Bott, & Wacker, 2019; Fischer, Lee, & Verzijden, 2018; Ward, Sreenivas, Read, Saunders, & Rogers, 2017; Narita et al., 2015; Tochigi et al., 2006). By contrast, reduced signal intensity of the neuromodulatory subcortical system can reflect poorer bioavailability of such neuromodulators (Matchett et al., 2021; Mather & Harley, 2016; Mai & Paxinos, 2012).

A unique contribution of the LC-NA system in this study was its mediatory role in accounting for a relationship between OE and IQ. This intercorrelation highlights LC-NA system centrality in higher order cognition. Greater IQ and higher OE scores are both conceived as reserve proxies that reduce the risk of cognitive decline and dementia (Karsazi et al., 2021; Ihle et al., 2019; Rammstedt et al., 2018; Tautvydaitė et al., 2017; Franchow et al., 2013; Schretlen et al., 2010; Sharp et al., 2010; Cuttler & Graf, 2007; Ackerman & Heggestad, 1997). The integrity of the LC-NA system, given its role in neurodegeneration, might be a common factor contributing to the expression of these two constructs in terms of brain and cognitive health (Robertson, 2013, 2014; DeYoung, 2013). These findings are therefore suggestive that the LC signal intensity, even at a young age, can significantly affect cognition and personality expression with the potential to build resilience to neurodegeneration in later life. Although the current study is the first to link LC-NA signal intensity to OE, there is previous evidence that the noradrenergic system is associated with greater cognitive performance and intelligence (Plini et al., 2021, 2023, 2024; Dahl et al., 2019, 2022a, 2022b; Dutt et al., 2021; Elman et al., 2021; Jacobs et al., 2021; Tsukahara & Engle, 2021; Liu et al., 2020; Clewett et al., 2016; Zhao, Kong, & Qu, 2014; Wilson et al., 2013).

Although OE showed the strongest relationship to the LC, a secondary VBM association was also observed between the C trait, conscientiousness, and the LC. The C trait, which reflects perseverance and focus, is associated with increased attention performance and reduced speed of processing (Yoneda et al., 2023; Sutin, Stephan, Damian, et al., 2019; Chapman et al., 2017) and improved memory (Sutin, Stephan, Luchetti, & Terracciano, 2019; Allen, Laborde, & Walter, 2019; Luchetti, Terracciano, Stephan, & Sutin, 2016). These cognitive domains also implicate LC-NA system integrity and functioning across several studies (Plini et al., 2021, 2023, 2024; Dahl et al., 2022a, 2022b; Grueschow, Kleim, & Ruff, 2022; Bari et al., 2020; Unsworth & Robison, 2017; Aston-Jones & Cohen, 2005; Aston-Jones, Rajkowski, & Cohen, 2000). Moreover, further work has found that greater C scores are associated with both greater resistance to dementia (Yoneda et al., 2023; Kaup, Harmell, & Yaffe, 2019) and lower risk of mild cognitive impairment and AD (Aschenbrenner et al., 2020; Terracciano, Stephan, Luchetti, Albanese, & Sutin, 2017; Wilson, Schneider, Arnold, Bienias, & Bennett, 2007). It should be noted that the role of additional variables needs further investigation. For example, it is documented that individuals who are high in C are more prone to engage regular physical activity (Sutin et al., 2016), which is also known to be related to better brain and cognitive health and to the LC-NA system (Plini et al., 2024). Clearer dissociation of the role of these different protective factors is warranted.

The current findings demonstrates that the C trait combined with greater OE expression in healthy individuals is associated with greater signal intensity in the LC. It remains to be seen whether these personality traits, through interaction with LC-NA system over the lifespan, provide greater resilience to neurodegeneration. However, these significant clinical implications warrant longitudinal investigation.

Cross-sectional designs are limited in terms of causal inference. A longitudinal study would provide more accurate understanding of these relationships and would enable mapping of possible trajectories between the LC-NA system and personality development. However, personality traits measured by NEO-FFI tend to be stable throughout lifetime (Gnambs, 2014; McCrae, Kurtz, Yamagata, & Terracciano, 2011). This gives us more confidence in the mediation analyses and in the partial correlations, which may help us understand the actual nature of these relationships.

Another potential limitation is selection bias—people who were recruited via the LEMON study might be, by nature of their voluntary participation, individuals who are “highly open” to engage in new experiences and this might limit the representativeness of the sample to the general population. This possible selection bias may also have been exacerbated by excluding individuals with ongoing psychological or psychiatric symptoms. The current sample reported low levels of anxiety and had a low level of neuroticism compared with other scales such as OE or C. Therefore, the possible detrimental effect of “high tonic” LC firing commonly observed in response to stress and anxiety could affect or reduce the LC–OE relationship and their possible neuroprotective outcomes (Ghosh et al., 2021; Omoluabi et al., 2021; Matchett et al., 2021; Rorabaugh et al., 2017; Rector, Bagby, Huta, & Ayearst, 2012). Nonetheless, it is worth mentioning we found no relationships between LC signal intensity and anxiety scales, which could affect the present set of findings.

Other minor limitations are the relatively small size of the sample, which can explain why the results did not survive the most conservative multiple comparison corrections on a voxel-wise approach, whereas it remained significant after Bonferroni correction on an ROI analysis (see Tables 2B and 2C in Appendix Tables A1 and B1). In the same vein, a more balanced sample of male and female participants may have led to differentiation between sex because the effect of sex in our analyses was close to significance.

In addition, the limitations of in vivo MRI studies also apply to the current work; ex vivo histological investigations would provide greater reliability on the matter.

Lastly, the constraints of the self-report personality questionnaires with forced choices, along with the IQ, were measured on the base of vocabulary only. However, within these limitations, we made the attempt to account for the confounding variables, controlling for age, sex, and TIV. In the same vein, we also systematically tested with numerous control analyses different relationships, and we also tested the antithetic hypothesis to consider the alternative direction of effects. Sixty VBM control analyses contrasted to our main analyses yielding greater confidence in the validity of our key findings.

Together, these findings are suggestive that personal attitudes to life, captured by the personality of openness (and conscientiousness), might be influenced, in part, by activation of the noradrenergic system. These findings may provide explanatory ground concerning the neuropsychological dynamics underlining the association between OE and resilience to neurodegenerative diseases while supporting Robertson's theory (Karsazi et al., 2021; Ihle et al., 2019; Cuttler & Graf, 2007; Tautvydaitė et al., 2017; Robertson, 2013, 2014; Franchow et al., 2013; Sharp et al., 2010). Indeed, as outlined by Robertson (Robertson, 2013, 2014), LC-NA system plays an important role underpinning all the attentional processes and, in particular, the exposure to novelty along with managing and developing problem-solving skills. This interacting with greater conscientiousness may ensure consistent cognitive control eliciting optimal noradrenergic tone (Robertson, 2013, 2014; Aston-Jones & Cohen, 2005; Aston-Jones et al., 2000). In keeping with this, a dispositional “openness” trait might drive a greater noradrenergic tone throughout lifetime, leading to more frequent phasic activation of LC-NA system yielding to greater LC integrity and greater brain and cognitive health (Plini et al., 2021, 2023, 2024; Prokopiou et al., 2022; Dutt et al., 2021; Mather, 2021; Omoluabi et al., 2021; Matchett et al., 2021; Mather & Harley, 2016; Clewett et al., 2016; Robertson, 2013, 2014). This openness trait throughout the lifespan might interact with other reserve variables, such as IQ via the mediation of the LC-NA system. Indeed, several studies reported greater OE is linked to greater crystallized intelligence (Rammstedt et al., 2018; Schretlen et al., 2010; Ackerman & Heggestad, 1997). Alternatively, it is possible that LC-NA characteristics could contribute to the openness personality characteristic or that the relationship is explained by other variables, and because of this, future studies are needed.

However, the newly established link between personality traits, LC-NA system, and cognition outlines the intercorrelation between neuropsychological variables and opens to potential psychological interventions targeting the LC-NA system. Cognitive and behavioral approaches (cognitive behavioral therapy), which focus on OE, in people who score low in OE and C, might be considered among the preventing strategies for neurodegenerative diseases (Forgeard et al., 2019). This might be beneficial in improving people's well-being both in enhancing the executive/attentional domain (Forgeard et al., 2019; Jackson et al., 2012) and in building the resilience ground to neurodegenerative diseases. Personality traits (OE especially) might be considered as critical components to focus on for brain health potentially influencing the neuroprotective role of LC-NA system, particularly in the face of neurodegeneration. Future studies should replicate these findings and design longitudinal investigations considering biological biomarkers, and personological and cognitive variables using high-resolution MRI accompanied by ex vivo histological studies, to better understand the neuropsychological dynamics that underpin our novel findings.

Thanks are extended to Francesca Fabbricatore for proofreading the article and for the thoughtful comments provided throughout the course.

Corresponding author: Emanuele R. G. Plini, School of Psychology & Trinity College Institute of Neuroscience, Trinity College Dublin, 42A Pearse Street, Dublin, Ireland, e-mail: [email protected].

Emanuele R. G. Plini: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Visualization; Writing—Original draft; Writing—Review & editing. Ian H. Robertson: Conceptualization; Supervision; Writing—Review & editing. Méadhbh B. Brosnan: Formal analysis; Methodology; Software; Writing—Review & editing. Paul M. Dockree: Conceptualization; Funding acquisition; Methodology; Resources; Supervision; Writing—Review & editing.

The project was founded by the Irish Research Council—Irish Research Council Laureate Consolidator Award (2018–23) IRCLA/2017/306 to P. M. D.

Retrospective analysis of the citations in every article published in this journal from 2010 to 2021 reveals a persistent pattern of gender imbalance: Although the proportions of authorship teams (categorized by estimated gender identification of first author/last author) publishing in the Journal of Cognitive Neuroscience (JoCN) during this period were M(an)/M = .407, W(oman)/M = .32, M/W = .115, and W/W = .159, the comparable proportions for the articles that these authorship teams cited were M/M = .549, W/M = .257, M/W = .109, and W/W = .085 (Postle and Fulvio, JoCN, 34:1, pp. 1–3). Consequently, JoCN encourages all authors to consider gender balance explicitly when selecting which articles to cite and gives them the opportunity to report their article's gender citation balance.