- Email: [email protected]
The Relationship Between General Intelligence and Cortical Structure in Healthy Individuals
NEUROSCIENCE RESEARCH ARTICLE S. Bajaj et al. / Neuroscience 388 (2018) 36–44
The Relationship Between General Intelligence and Cortical Structure in Healthy Individuals Sahil Bajaj, a* Adam Raikes, a Ryan Smith, a Natalie S. Dailey, a Anna Alkozei, a John R. Vanuk a and William D. S. Killgore a,b a
Social, Cognitive and Affective Neuroscience Laboratory (SCAN Lab), Department of Psychiatry, College of Medicine, University of Arizona, Tucson, AZ 85724, USA b
McLean Hospital, Department of Psychiatry, Harvard Medical School, Belmont, MA, USA
Abstract—Considerable work in recent years has examined the relationship between cortical thickness (CT) and general intelligence (IQ) in healthy individuals. It is not known whether specific IQ variables (i.e., perceptual reasoning [PIQ], verbal comprehension IQ [VIQ], and full-scale IQ [FSIQ]) are associated with multiple cortical measures (i.e., CT, cortical volume (CV), cortical surface area (CSA) and cortical gyrification (CG)) within the same individuals. Here we examined the association between these neuroimaging metrics and IQ in 56 healthy adults. At a cluster-forming threshold (CFT) of p < 0.05, we observed significant positive relationships between CT and all three IQ variables in regions within the posterior frontal and superior parietal lobes. Regions within the temporal and posterior frontal lobes exhibited positive relationships between CV and two IQ variables (PIQ and FSIQ) and regions within the inferior parietal lobe exhibited positive relationships between CV and PIQ. Additionally, CV was positively associated with VIQ in the left insula and with FSIQ within the inferior frontal gyrus. At a more stringent CFT (p < 0.01), the CT–PIQ, CT–VIQ, CT–FSIQ, and CV–PIQ relationships remained significant within the posterior frontal lobe, as did the CV–PIQ relationship within the temporal and inferior parietal lobes. We did not observe statistically significant relationships between IQ and either CSA or CG. Our findings suggest that the neural basis of IQ extends beyond previously observed relationships with fronto-parietal regions. We also conclude that CT and CV may be more useful metrics than CSA or CG in the study of intellectual abilities. Ó 2018 IBRO. Published by Elsevier Ltd. All rights reserved.
Key words: cortical measures, brain structure–function, intelligence, verbal comprehension, perceptual reasoning, full-scale IQ.
function (Burgaleta et al., 2014; Gregory et al., 2016). Mathematically, many of these structural measures are interrelated. For example, CV is the product of two, genetically distinct properties: CT and CSA (Winkler et al., 2010). CG positively correlates with both total brain volume and CSA but negatively with CT (Gautam et al., 2015; Hogstrom et al., 2013). Given the complex relationships between these heterogeneous morphometric properties, it is crucial to study the simultaneous relevance of each of these measures to cognition (e.g., their individual and joint relationships with measures of human intellectual ability [IQ)]). For instance, if two structural measures, such as CT and CSA, both have a positive association with cognitive abilities, then CV should also exhibit a positive, and possibly stronger, relationship than either CT or CSA alone. Fundamentally, despite the fact that some cortical measures (CT, CV, CSA and CG) may covary, these measures reflect different facets of brain structure and could each contribute uniquely to cognitive function. Specifically, CT and CSA are considered highly heritable and genetically independent (Eyler et al., 2011; Kremen et al., 2010; Panizzon et al., 2009). Due to their
INTRODUCTION Surface-based structural brain analysis techniques, involving an estimation of morphometric measures such as cortical thickness (CT), cortical volume (CV), cortical surface area (CSA), and cortical gyrification (CG) have expanded our understanding of the impact of various neurodegenerative diseases on brain structure. These measures provide specific indices of several unique aspects of brain structure. These measures also provide insight into the dynamics of brain structure during neurodevelopment (Burgaleta et al., 2014; White et al., 2010), the effects of aging on brain structure (Bajaj et al., 2017; Burgaleta et al., 2014; Hogstrom et al., 2013), and the relationships between brain structure and
*Corresponding author. Address: 1501 N Campbell Avenue, Department of Psychiatry, Room # 7304B, University of Arizona, Tucson, AZ 85724, USA. E-mail address: [email protected] (S. Bajaj). Abbreviations: CFT, cluster-forming threshold; CG, cortical gyrification; CSA, cortical surface area; CT, cortical thickness; CV, cortical volume; FSIQ, full-scale IQ; IQ, intelligence; PIQ, perceptual reasoning; VIQ, verbal comprehension. https://doi.org/10.1016/j.neuroscience.2018.07.008 0306-4522/Ó 2018 IBRO. Published by Elsevier Ltd. All rights reserved. 36
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
unique regional variation with the timing of prenatal perturbations, these measures are often understood as separate morphometric variables during neurodevelopment, aging, and disease (Eyler et al., 2011; Hogstrom et al., 2013; Panizzon et al., 2009). Moreover, CV is considered distinct from, though mathematically related to, CT and CSA (i.e., due to the non-uniform, integrated influence of CT and CSA) (Gerrits et al., 2016). Further, CG represents the convolution of brain structure and is influenced by both neuronal density and intra-cranial volume (Toro and Burnod, 2005; Welker, 1990). This folding increases the potential for a greater number of neurons within the same space. When coupled with the fact that CSA may facilitate better neuronal signal processing, increased CG and CSA may provide more efficient structural organization to facilitate brain connectivity and functional development (Luders et al., 2008; Murre and Sturdy, 1995; Ruppin et al., 1993). This supports the idea that the organization and gyral structure of the cortex contributes to its function. With respect to structural measurements and cognition, the relationship between general intelligence (IQ) and structural measures, particularly CT and CSA, is likely age-dependent. Notably, Schnack and colleagues reported that among children (10 years of age), higher standardized IQ scores were associated with thinner cortex within the left hemisphere (Schnack et al., 2015). This pattern differed in young adults, where thicker left hemisphere cortex was associated with higher IQ scores, suggesting that maturation-related structural changes may play an important role in the expression of IQ. Consistent with these findings, Shaw and colleagues reported a shift from a negative correlation between CT and IQ in early childhood (age range 3.8–8.4 years) to a positive correlation in late childhood (age range 8.6–11.7 years) and into early adulthood (age range 11.8–29 years) (Shaw et al., 2006). In other work with healthy children and adolescents, changes in CT, but not CSA, were associated with changes in performance IQ, verbal IQ, and full-scale IQ (Burgaleta et al., 2014). Furthermore, higher CG in middle-aged participants (44–48 years) was associated with better mental flexibility, larger brain volume, but also thinner cortex (Gautam et al., 2015). Therefore, relationships between IQ and the structural characteristics of the cortex appear to exhibit dynamic, age- and developmental stage-dependent changes. A recent meta-analysis suggested that much of the previous literature may over-estimate/over-simplify the association between brain volume and IQ (Pietschnig et al., 2015). The association instead appears to be more complex than suggested by simple volumetric measures. Surface-based cortical measures, rather than purely volumetric measures, may therefore provide more useful information about the complex and widespread relationships between cortical structure and general IQ. Prior work has generally focused on individual structural measures in isolation and their association with IQ. Given that neural structure–function relationships are likely more complex than can be accounted for by simple volumetric measures, it is crucial to examine the parallel associations between
37
multiple brain structure measures (CT, CV, CSA, and CG) and multiple aspects of general IQ (perceptual reasoning (PIQ), verbal comprehension and full-scale IQ) in the same sample of healthy young adults. The primary aim of the current study is therefore to investigate the vertex-wise associations between these structural metrics and these different IQ variables. Based on prior research, we hypothesized that both CT and CV would be positively associated with all domains of IQ. However, due to inconsistent and limited evidence regarding the relationship between CSA or CG and cognitive abilities, we had no specific hypothesis regarding the relationships between these specific metrics and IQ. Thus, these latter analyses were primarily for exploratory purposes.
EXPERIMENTAL PROCEDURES Participants Fifty-six healthy participants between 18 and 45 years of age (mean age = 30.8 ± 8.1 years, 27 females; 29 males) participated in this study. Participants were recruited from the New England area and screened via a comprehensive telephone interview. Individuals with any history of psychiatric, neurological, or significant medical problems, current use of psychotropic medications, or current use of illicit substances were excluded. Participants were all primary native English speakers, with 14.95 years of formal education (SD = 2. 18 years). 69.6% of participants were Caucasian, 14.2% were African American, 8.9% were Asian, 3.6% reported polyethnic backgrounds and 3.6% reported other ethnic backgrounds. All participants provided written informed consent prior to enrollment. The study protocol was approved by the Institutional Review Boards of McLean Hospital and Partners Healthcare, and the U.S. Army Human Research Protections Office. Other behavioral data and structural estimates from this sample have been reported elsewhere (Killgore et al., 2013), but the vertex-wise cortical measures and their associations with IQ reported in this study are novel and have not been previously reported. Data acquisition Neuroanatomical data. We recorded high-resolution T1-weighted magnetic resonance imaging data using a 3-Tesla Siemens TIM Trio whole-brain MR scanner located at the McLean Hospital Imaging Center. Before the scan, each participant was instructed to rest, relax and try his/her best to minimize movement during the entire scan. Head movement was further minimized with foam padding placed comfortably about the head. T1weighted data for each participant were acquired using a 3D magnetization-prepared rapid acquisition gradient echo (MPRAGE) sequence, which consisted of 128 sagittal slices (slice thickness = 1.33 mm, voxel resolu tion = 1.33 1 1 mm, field of view (FOV) = 256 mm with TR/TE/FA/inversion time of 2100 ms/2.25 m s/12°/1100 ms).
38
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
General intelligence (IQ) assessment. A trained research technician, supervised by a licensed neuropsychologist, administered the IQ assessment to each participant. All participants completed the Wechsler Abbreviated Scale of Intelligence II (WASI-II; (Wechsler, 1999). The WASI-II provides estimates of IQ in the domains of perceptual reasoning (Block Design and Matrix Reasoning; PIQ) and verbal comprehension (Vocabulary and Similarities; VIQ). The combination of PIQ and VIQ provides an estimate of Full-Scale IQ (FSIQ). PIQ reflects overall visuospatial intellectual abilities (e.g., spatial processing, visual-motor integration, and attentiveness to non-verbal tasks). VIQ reflects overall language capacity (e.g., measures of verbal expressive, verbal comprehension, and verbal reasoning abilities). FSIQ indicates overall cognitive and intellectual ability. The four subtests of the WASI-II (Block Design, Matrix Reasoning, Vocabulary, and Similarities) have strong associations with more extensive measures of general cognitive abilities (Hays et al., 2002; Wechsler, 1999). FSIQ scores have high reliability (0.98), and also have a correlation (r = 0.92) with scores on the full Wechsler Adult Intelligence Scale (WAIS)-III (Wechsler, 1997, 1999). Data analysis
Image processing and brain structure measures. We used the ‘‘recon-all” pipeline in FreeSurfer (version 6.0) (https://surfer.nmr.mgh.harvard.edu) to process the anatomical brain images for all the participants (Dale et al., 1999; Fischl et al., 1999) and for estimating the structural measures. Processing involved basic imagepreprocessing steps, including motion-correction, brain extraction (i.e., removal of non-brain tissue), automated transformation to the standard MNI co-ordinate system, volumetric segmentation into cortical and sub-cortical matter, intensity correction, and parcellation of the cerebral cortex into gyral and sulcal matter (Desikan et al., 2006). The technical details of these steps are documented in previous publications (Dale et al., 1999; Fischl et al., 1999; Fischl et al., 2004). To inspect FreeSurfer’s preprocessing accuracy, standard quality control steps were performed, which involved a careful visual inspection of raw T1-weighted images, skull-stripped brain volumes, and pial surfaces. The structural measures (CT, CV, CSA, and CG) were estimated separately for the left and the right hemisphere for each participant. CT was estimated as the shortest distance between the white matter (WM) surface (white–gray matter interface) and the pial surface (grey matter–CSF interface) (Fischl and Dale, 2000). CV was estimated as the amount of grey matter that lies between white–grey matter interface and pial matter (Winkler et al., 2010). CSA was estimated as the sum of the areas of the triangles making up the surface model and is defined as the extent of 2dimensional surface enclosed by the outer layer of the cerebral cortex (Fischl et al., 1999). Finally, CG was estimated as the ratio of the area on the outer surface to the area on the pial surface (Schaer et al., 2008).
Statistical analysis. We fit separate generalized linear models (GLMs) to the left and right hemispheres within FreeSurfer’s statistical engine to examine structure–IQ relationships. Standardized measures of each IQ subdomain (PIQ, VIQ, and FSIQ – independent variables) were regressed on each structural measure (dependent variables). For this analysis, we selected two threshold levels. First, to minimize Type II error and inform future hypothesis testing, we used a liberal cluster-forming threshold (CFT) of p < 0.05. To minimize Type I error, we also report results with a stricter CFT of p < 0.01. CFTs are used to detect significant brain clusters based on the number of clustered voxels, whose voxel-wise statistic values are above the pre-defined threshold (Woo et al., 2014). Previously, liberal CFTs have been used to determine clusters after correcting for multiple comparisons across the surfaces (Fuglset et al., 2016). However, liberal cluster-forming height thresholds increase cluster size and may highlight more than one region, resulting in reduced anatomical specificity. Since cluster-wise false positive rates of cortical measures may depend on smoothing levels, our analysis involved liberal, as well as, stringent CFTs at two different smoothing levels to better understand the associations between cortical structure and general IQ. As the sample size in our study is small and the clusters are not expected to be localized with a high degree of specificity, we selected a smaller Gaussian smoothing kernel size of 10 mm with a CFT of p < 0.05 and a moderately larger smoothing kernel size of 15 mm with a CFT of p < 0.01. Moreover, unlike volume-based analysis, a larger smoothing kernel size in surface-based analysis never extends into bone/ air/white matter. For all analyses, we used a clusterwise threshold (CWP) of p < 0.05 (corrected for multiple comparisons using Monte-Carlo simulations) to identify significant clusters. All models were controlled for age and sex.
RESULTS Associations between general IQ and each cortical measure CT and IQ. We identified clusters within the posterior frontal and superior parietal lobes that showed significant relationships between thicker cortex and higher IQ at a liberal CFT of p < 0.05 as well as at a strict CFT of p < 0.01. Fig. 1 displays the clusters showing these associations between CT and PIQ (A–D), VIQ (E–G), and FSIQ (H–K) at CFT of p < 0.05 at FWHM of 10 mm. Fig. 2 displays the clusters showing these associations between CT and PIQ (A, B), VIQ (C–F), and FSIQ (G–I) at CFT of p < 0.01 at FWHM of 15 mm. No clusters showed negative association between CT and any IQ measures at CFT of p < 0.05. Identified clusters are summarized in Table 1. CV and IQ. We identified clusters within the temporal, posterior frontal, and parietal lobes as well as inferior frontal gyrus that showed significant relationships between higher CV and higher IQ, especially at CFT of
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
39
Fig. 1. Relationships between cortical thickness (CT) and IQ. We found positive associations between CT and all three subscales of IQ – Perceptual Reasoning IQ (PIQ) (A–D), Verbal Comprehension IQ (VIQ) (E–G), and Full-Scale IQ (FSIQ) (H–K) (CFT: p 0.05, CWP: p 0.05, corrected for multiple comparisons).
p < 0.05. However, at CFT of p < 0.01, higher CV was only positively associated with PIQ. Fig. 3 displays the clusters showing these associations between CV and PIQ (A–C), VIQ (D, E), and FSIQ (F, G) at CFT of p < 0.05 at FWHM of 10 mm. Fig. 4 displays the clusters showing these associations between CV and PIQ (A–C) at CFT of p < 0.01 at FWHM of 15 mm. No clusters showed negative association between CV and any IQ measures at CFT of p < 0.05. Identified clusters are summarized in Table 2.
idea that IQ is associated with quantifiable differences in brain structure within regions that include, but are not limited to, the fronto-parietal regions most emphasized in previous work on general IQ (Jung and Haier, 2007). Our results further suggest that CT and CV may be more useful than CG and CSA in studying the contribution of cortical structure to IQ.
CSA and IQ. There were no significant relationships found between CSA and IQ at CFT of p < 0.05.
Our findings further advance the results reported from several previous studies. We observed significant relationships between multiple aspects of general IQ and morphometric characteristics in areas associated with working memory, motor-skills, and physical exercise (e.g., the precentral gyrus [PreCG]) (Burgaleta et al., 2014; Killgore and Schwab, 2012; Miller and Cohen, 2001). Previous work indicates that brain areas, such as the basal ganglia, frontal cortex, and related dopaminergic pathways, play important roles in both motor performance and cognitive performance abilities (Diamond, 2000; Nieoullon, 2002; Wassenberg et al., 2005). As expected, we also found significant relationships between IQ and cortical areas responsible for (a) integrating and processing information, including the parahippocampal gyrus and precuneus/cuneus cortex (PreC/CunC) (Cavanna and Trimble, 2006; Hulshoff Pol et al., 2006; Westlye et al., 2009), (b) creative thinking
CG and IQ. There were no significant relationships found between CG and IQ at CFT of p < 0.05.
DISCUSSION In this study we explored the association between general IQ and multiple measures of brain structure. We identified significant positive associations between all three IQ scales (PIQ, VIQ, and FSIQ) and two structural measures (CT and CV) for several regions distributed across the brain. On the other hand, IQ was not significantly correlated with CSA or CG. Thus, greater thickness and volume of the cortex were associated with better intellectual functioning, whereas surface area and gyrification were not. These findings support the
Associations between general IQ measures and structural measures
40
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
Fig. 2. Relationships between cortical thickness (CT) and IQ. We found positive associations between CT and all three subscales of IQ – Perceptual Reasoning IQ (PIQ) (A, B), Verbal Comprehension IQ (VIQ) (C–F), and Full-Scale IQ (FSIQ) (G–I) (CFT: p 0.01, CWP: p 0.05, corrected for multiple comparisons).
and memory retrieval, including the PreC (Chen et al., 2015), (c) visual identification and recognition, such as regions within the ventral temporal cortex e.g. fusiform gyrus (Bar et al., 2001; McCandliss et al., 2003), and (d) integration and retrieval of semantic knowledge, including the medial temporal lobes (McClelland and Rogers, 2003). These results are also consistent with previous work highlighting the relationships between IQ and CT within the temporal lobe (Choi et al., 2008). In our sample the left insula was the only brain region to show significant positive association between higher CV and VIQ. The insular cortex is strongly interconnected with the temporal cortex (Mesulam and Mufson, 1982) and limbic system, and has previously been implicated in learning and memory functions (Balderas et al., 2015; Bermudez-Rattoni, 2014; Wu et al., 2017). In the left hemisphere of most people, this region plays a crucial role in processing and integrating cognitive information associated with auditory perception and receptive language (Mirman et al., 2015). A large body of previous work has highlighted the contribution of fronto-parietal regions to general IQ, and has led to the Parieto-Frontal Integration Theory of Intelligence (P-FIT) (Jung and Haier, 2007). This theory suggests the structure and function of brain are intricately connected. For instance, inferior frontal regions (such as
the pars opercularis) play a direct role in goal-directed cognitive performance (Demonet et al., 2005). Additionally, the parietal regions further contribute through integrating sensory information (i.e., first processed in occipital and temporal cortices), allocating attention, and facilitating effective problem solving (Gevins and Smith, 2000; Graham et al., 2010; Karama et al., 2009; Smith et al., 2004). Our findings generally support this theory but add to a growing body of work suggesting that brain regions outside of this frontoparietal system also contribute to general IQ (Colom et al., 2009; Hearne et al., 2016). In other words, our findings support the position that structural correlates of general IQ are widespread and not strictly limited to areas identified in the P-FIT model, which nevertheless remains provisional and experimental. Our findings are also consistent with those of a meta-analysis of 12 structural and 16 functional imaging studies that presented an updated neurocognitive model of general IQ. This model includes not only structures within the frontal, parietal, temporal, and occipital lobes but also the insula, posterior cingulate cortex, and several subcortical regions (Basten et al., 2015). In contrast to CV and CT, we found no significant associations between IQ and either CSA or CG. Previous studies have inconsistently identified relationships between CSA and measures of general IQ
41
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
Table 1. Relationships between cortical thickness (CT) and IQ scales – Perceptual Reasoning IQ (PIQ), Verbal Comprehension IQ (VIQ) and Full-Scale IQ (FSIQ) Clusters showing significant relationships Between CT and PIQ scores Cluster number
Maxima
Peak co-ordinates (MNI: X, Y, Z)
CWP
Number of vertices within the cluster
FreeSurfer label
Association
1 2 3 4 5 6 7
6.04 3.01 4.88 3.56 3.21 5.67 5.02
42.6, 8.1, 54.7 28.8, 43.8, 17.3 59.1, 3.1, 26.3 28.6, 28.5, 20.5 22.5, 58.3, 21.1 43.0, 7.9, 53.5 58.4, 1.3, 28.1
0.00 0.02 0.01 0.00 0.03 0.00 0.01
4822 1812 2245 2400 1557 4074 1861
LPreCG* LFG* RPreCG* RPHG* RPreC* LPreCG** RPreCG**
(+) (+) (+) (+) (+) (+) (+)
Between CT and VIQ scores 1 6.23 2 4.08 3 4.54 4 4.31 5 4.00 6 2.86 7 6.72 8 4.28 9 3.09
28.3, 20.5, 68.6 54.0, 17.9, 3.6 17.7, 56.0, 21.3 12.6, 43.5, 71.4 58.6, 6.4, 26.7 9.6, 95.9, 10.2 27.8, 20.9, 68.3 12.3, 45.2, 69.8 26.1, 77.0, 6.6
0.00 0.00 0.02 0.00 0.00 0.00 0.00 0.01 0.00
5595 2776 1951 3218 3420 2421 5099 2306 1505
LPreCG* LTTG* RPreC* RPreC* RPreCG* RCunC* LPreCG** RPreC** RFG**
(+) (+) (+) (+) (+) (+) (+) (+) (+)
Between CT and FSIQ scores 1 6.66 2 3.37 3 4.89 4 4.76 5 3.02 6 6.58 7 4.55 8 3.26
42.7, 8.6, 54.9 28.8, 43.8, 17.3 11.2, 47.4, 65.9 59.0, 5.3, 26.7 28.7, 29.4, 19.6 43.1, 8.2, 53.8 58.6, 1.8, 27.3 33.8, 83.0, 13.3
0.00 0.00 0.03 0.00 0.00 0.00 0.02 0.01
6865 2184 2469 3673 2779 6118 1801 1308
LPreCG* LFG* RPreC* RPreCG* RPHG* LPreCG** RPreCG** RLOC**
(+) (+) (+) (+) (+) (+) (+) (+)
Abbreviations: Left/Right Precentral Gyrus (L/R: PreCG); Left/Right Fusiform Gyrus (L/R: FG); Right Parahippocampal Gyrus (RPHG); Right Precuneus Cortex (RPreC); Left Transverse Temporal Gyrus (LTTG); Right Cuneus Cortex (RCunC); Right Lateral Occipital Cortex (RLOC). * Significant at cluster-forming threshold (CFT) of p < 0.05 (FWHM 10 mm) and cluster-wise threshold of p (CWP) < 0.05 (corrected for multiple comparisons using Monte Carlo simulation). ** Significant at cluster-forming threshold (CFT) of p < 0.01 (FWHM 15 mm) and cluster-wise threshold of p (CWP) < 0.05 (corrected for multiple comparisons using Monte Carlo simulation).
(Burgaleta et al., 2014; Schnack et al., 2015). Additionally, developmental changes in CSA are generally less notable than those for CV (Ostby et al., 2009). However, Schnack and colleagues found that by age 10, CSA is often larger among children with greater IQ, developing to an apex during adolescence (Schnack et al., 2015). Given that all of our participants were within the age range of 18–45, one potential explanation for the non-significant relationship between CSA and general IQ is that by early adulthood (i.e., age 18) cortical expansion, which leads to increased CSA, is complete in most individuals and instead begins contracting, potentially resulting in higher CT and CV (Giedd et al., 1999). Moreover, CSA contraction within several brain areas, including the bilateral PreCG and CunC, was previously reported to be more prominent among adults with higher IQ (Schnack et al., 2015). CG, however, is a conceptually distinct morphological measure from CT or CV. Prior work has postulated that stronger cognitive abilities may be related to greater CG due to a larger number of neurons within the convoluted cortex as well as larger surface area, which facilitates the processing of information (Gautam et al.,
2015; Gregory et al., 2016; Luders et al., 2008). However, we found no association between CG and measured IQ. Notably, CG is a relatively smooth structural measure throughout the cortex, and often requires a larger surface area to show significant associations with cognitive measures (Razlighi et al., 2016). Thus, CG and CSA are not entirely independent metrics, and neither was independently associated with IQ in the present study. Limitations The present findings should be interpreted in light of several limitations. First, we had a widespread age range over a relatively small sample size, which could account for some of the vertex-wise structural differences across the three IQ domains. Second, the absence of significant relationships between both CSA and CG and IQ was unexpected and suggests the necessity of continuing to explore these relationships for different age groups, given typical developmental and structural changes that may continue into the third decade of life (Lenroot and Giedd, 2006). Additionally, we controlled for sex in our models.
42
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
Table 2. Relationships between cortical volume (CV) and IQ scales – Perceptual Reasoning IQ (PIQ), Verbal Comprehension IQ (VIQ) and Full-Scale IQ (FSIQ) Clusters showing significant relationships Between CV and PIQ scores Cluster number
Maxima
Peak co-ordinates (MNI: X, Y, Z)
CWP
Number of vertices within the cluster
FreeSurfer label
Association
1 2 3 4 5 6 7 8 9 10 11
4.20 3.99 3.83 4.22 3.68 3.53 4.74 4.09 3.87 3.61 3.50
41.7, 9.1, 37.9 39.6, 31.9, 7.8 49.0, 8.6, 39.8 49.2, 2.2, 25.5 46.7, 7.0, 27.6 38.6, 76.1, 12.7 42.0, 9.8, 36.7 48.1, 9.0, 39.5 43.7, 28.8, 4.0 47.1, 6.5, 28.0 39.0, 76.3, 13.3
0.00 0.00 0.01 0.01 0.00 0.00 0.00 0.04 0.01 0.03 0.04
2464 3482 2917 1957 3202 2003 2246 1765 2479 2047 1320
LITG* LTTG* LPreCG* RSTG* RPreCG* RIPC* LITG** LPreCG** LTTG** RPreCG** RIPC**
(+) (+) (+) (+) (+) (+) (+) (+) (+) (+) (+)
31.0, 17.5, 13
0.01
2442
LIns*
(+)
41.7, 30.6, 5.7 31.5, 13.5, 13.2 49.6, 5.1, 27.4
0.01 0.05 0.05
2743 2247 2135
LTTG* LPOC* RPreCG*
(+) (+) (+)
Between CV and VIQ scores 1 3.25 Between CV and FSIQ scores 1 3.92 2 2.73 3 3.20
Abbreviations: Left Inferior Temporal Gyrus (LITG); Left Transverse Temporal Gyrus (LTTG); Left/Right Precentral Gyrus (L/R: PreCG); Right Superior Temporal Gyrus (RSTG); Right Inferior Parietal Cortex (RIPC); Left Insula (LIns); Left Pars Opercularis (LPOC). * Significant at cluster-forming threshold (CFT) of p < 0.05 (FWHM 10 mm) and cluster-wise threshold of p (CWP) < 0.05 (corrected for multiple comparisons using Monte Carlo simulation). ** Significant at cluster-forming threshold (CFT) of p < 0.01 (FWHM 15 mm) and cluster-wise threshold of p (CWP) < 0.05 (corrected for multiple comparisons using Monte Carlo simulation).
However, men and women show different developmental trajectories of cognitive abilities and brain morphology measurements (YurgelunTodd et al., 2002) and sex-specific relationships between brain morphology and cognition merit further attention. Future studies incorporating larger sample sizes with consistent representations across age ranges are necessary to corroborate the associations between the structural and IQ measures reported in our study.
CONCLUDING REMARKS
Fig. 3. Relationships between cortical volume (CV) and IQ. We found positive associations between CV and all three subscales of IQ – Perceptual Reasoning IQ (PIQ) (A–C), Verbal Comprehension IQ (VIQ) (D, E), and Full-Scale IQ (FSIQ) (F, G) (CFT: p 0.05, CWP: p 0.05, corrected for multiple comparisons).
The present findings advance our understanding of how brain structure is associated with IQ by including, for the first time, multiple metrics of brain structure and IQ in the same study. We provide evidence that greater general IQ may be functionally related to CT and volume extending beyond previously reported fronto-parietal regions typically associated with human IQ. These findings therefore add to a growing literature on the neural basis of individual differences in general cognitive ability.
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
43
Chen QL et al (2015) Individual differences in verbal creative thinking are reflected in the precuneus. Neuropsychologia 75:441–449. https://doi.org/10.1016/j. neuropsychologia.2015.07.001. Choi YY et al (2008) Multiple bases of human intelligence revealed by cortical thickness and neural activation. J Neurosci 28:10323–10329. https://doi.org/10.1523/ JNEUROSCI.3259-08.2008. Colom R, Haier RJ, Head K, A´lvarez-Linera J, Quiroga MA´, Shih PC, Jung RE (2009) Fig. 4. Relationships between cortical volume (CV) and IQ. We found positive associations Gray matter correlates of fluid, between CV and Perceptual Reasoning IQ (PIQ) (A–C) (CFT: p 0.01, CWP: p 0.05, corrected crystallized, and spatial intelligence: for multiple comparisons). testing the P-FIT model. Intelligence 37:124–135. https://doi.org/10.1016/j. intell.2008.07.007. CONFLICTS OF INTEREST STATEMENT Dale AM, Fischl B, Sereno MI (1999) Cortical surface-based analysis. I. Segmentation and surface reconstruction. Neuroimage All the authors declared no conflicts of interest. 9:179–194. https://doi.org/10.1006/nimg.1998.0395. Demonet JF, Thierry G, Cardebat D (2005) Renewal AUTHOR CONTRIBUTIONS of the neurophysiology of language: functional neuroimaging. Physiol Rev 85:49–95. https://doi.org/10.1152/physrev.00049. SB conducted the neuroimaging analyses and wrote the 2003. initial draft of the manuscript and organized the Desikan RS et al (2006) An automated labeling system for revisions. AR and RS provided critical comments and subdividing the human cerebral cortex on MRI scans into gyral contributed to the writing of the manuscript. NSD, AA based regions of interest. Neuroimage 31:968–980. https://doi. org/10.1016/j.neuroimage.2006.01.021. and JRV contributed to the writing of revisions of the Diamond A (2000) Close interrelation of motor development and manuscript. WDSK obtained the funding, designed the cognitive development and of the cerebellum and prefrontal study, supervised staff training, data collection, and cortex. Child Dev 71:44–56. analysis, and contributed to writing revisions of the Eyler LT et al (2011) Genetic and environmental contributions to manuscript. regional cortical surface area in humans: a magnetic resonance imaging twin study. Cereb Cortex 21:2313–2321. https://doi.org/ 10.1093/cercor/bhr013. FUNDING Fischl B, Dale AM (2000) Measuring the thickness of the human cerebral cortex from magnetic resonance images. Proc Natl Acad This research was supported by a grant from the U.S. Sci U S A 97:11050–11055. https://doi.org/10.1073/ Army Medical Research and Materiel Command to pnas.200033797. WDSK (W81XWH-09-1-0730). The opinions, Fischl B, Sereno MI, Dale AM (1999) Cortical surface-based analysis. interpretations, conclusions and recommendations in II: Inflation, flattening, and a surface-based coordinate system. this paper are solely those of the authors and are not Neuroimage 9:195–207. https://doi.org/10.1006/nimg.1998.0396. necessarily endorsed by the Department of Defense or Fischl B et al (2004) Automatically parcellating the human cerebral the U.S. Army Medical Research and Materiel Command. cortex. Cereb Cortex 14:11–22. Fuglset TS, Endestad T, Hilland E, Bang L, Tamnes CK, Landro NI, Ro O (2016) Brain volumes and regional cortical thickness in REFERENCES young females with anorexia nervosa. BMC Psychiatry 16:404. https://doi.org/10.1186/s12888-016-1126-9. Bajaj S, Alkozei A, Dailey NS, Killgore WDS (2017) Brain aging: Gautam P, Anstey KJ, Wen W, Sachdev PS, Cherbuin N (2015) uncovering cortical characteristics of healthy aging in young Cortical gyrification and its relationships with cortical volume, adults. Front Aging Neurosci 9:412. https://doi.org/10.3389/ cortical thickness, and cognitive performance in healthy mid-life fnagi.2017.00412. adults. Behav Brain Res 287:331–339. https://doi.org/10.1016/j. Balderas I, Rodriguez-Ortiz CJ, Bermudez-Rattoni F (2015) bbr.2015.03.018. Consolidation and reconsolidation of object recognition memory. Gerrits NJ et al (2016) Cortical thickness, surface area and Behav Brain Res 285:213–222. https://doi.org/10.1016/j. subcortical volume differentially contribute to cognitive bbr.2014.08.049. heterogeneity in Parkinson’s disease. PLoS One 11. https://doi. Bar M et al (2001) Cortical mechanisms specific to explicit visual org/10.1371/journal.pone.0148852 e0148852. object recognition. Neuron 29:529–535. Gevins A, Smith ME (2000) Neurophysiological measures of working Basten U, Hilger K, Fiebach CJ (2015) Where smart brains are memory and individual differences in cognitive ability and different: a quantitative meta-analysis of functional and structural cognitive style. Cereb Cortex 10:829–839. brain imaging studies on intelligence. Intelligence 51:10–27. Giedd JN et al (1999) Brain development during childhood and https://doi.org/10.1016/j.intell.2015.04.009. adolescence: a longitudinal MRI study. Nat Neurosci 2:861–863. Bermudez-Rattoni F (2014) The forgotten insular cortex: its role on https://doi.org/10.1038/13158. recognition memory formation. Neurobiol Learn Mem Graham S et al (2010) IQ-related fMRI differences during cognitive 109:207–216. https://doi.org/10.1016/j.nlm.2014.01.001. set shifting. Cereb Cortex 20:641–649. https://doi.org/10.1093/ Burgaleta M, Johnson W, Waber DP, Colom R, Karama S (2014) cercor/bhp130. Cognitive ability changes and dynamics of cortical thickness Gregory MD, Kippenhan JS, Dickinson D, Carrasco J, Mattay VS, development in healthy children and adolescents. Neuroimage Weinberger DR, Berman KF (2016) Regional variations in brain 84:810–819. https://doi.org/10.1016/j.neuroimage.2013.09.038. gyrification are associated with general cognitive ability in Cavanna AE, Trimble MR (2006) The precuneus: a review of its humans. Curr Biol 26:1301–1305. https://doi.org/10.1016/j. functional anatomy and behavioural correlates. Brain cub.2016.03.021. 129:564–583. https://doi.org/10.1093/brain/awl004.
44
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
Hays JR, Reas DL, Shaw JB (2002) Concurrent validity of the Wechsler abbreviated scale of intelligence and the Kaufman brief intelligence test among psychiatric inpatients. Psychol Rep 90:355–359. https://doi.org/10.2466/pr0.2002.90.2.355. Hearne LJ, Mattingley JB, Cocchi L (2016) Functional brain networks related to individual differences in human intelligence at rest. Sci Rep 6:32328. https://doi.org/10.1038/srep32328. Hogstrom LJ, Westlye LT, Walhovd KB, Fjell AM (2013) The structure of the cerebral cortex across adult life: age-related patterns of surface area, thickness, and gyrification. Cereb Cortex 23:2521–2530. https://doi.org/10.1093/cercor/bhs231. Hulshoff Pol HE et al (2006) Genetic contributions to human brain morphology and intelligence. J Neurosci 26:10235–10242. https:// doi.org/10.1523/JNEUROSCI.1312-06.2006. Jung RE, Haier RJ (2007) The Parieto-Frontal Integration Theory (PFIT) of intelligence: converging neuroimaging evidence. Behav Brain Sci 30:135–154. https://doi.org/10.1017/ S0140525X07001185. discussion 154–187. Karama S et al (2009) Positive association between cognitive ability and cortical thickness in a representative US sample of healthy 6 to 18 year-olds. Intelligence 37:145–155. https://doi.org/10.1016/j. intell.2008.09.006. Killgore WD, Olson EA, Weber M (2013) Physical exercise habits correlate with gray matter volume of the hippocampus in healthy adult humans. Sci Rep 3:3457. https://doi.org/10.1038/ srep03457. Killgore WD, Schwab ZJ (2012) Sex differences in the association between physical exercise and IQ. Percept Mot Skills 115:605–617. https://doi.org/10.2466/06.10.50.PMS.115.5.605617. Kremen WS et al (2010) Genetic and environmental influences on the size of specific brain regions in midlife: the VETSA MRI study. Neuroimage 49:1213–1223. https://doi.org/10.1016/j. neuroimage.2009.09.043. Lenroot RK, Giedd JN (2006) Brain development in children and adolescents: insights from anatomical magnetic resonance imaging. Neurosci Biobehav Rev 30:718–729. https://doi.org/ 10.1016/j.neubiorev.2006.06.001. Luders E et al (2008) Mapping the relationship between cortical convolution and intelligence: effects of gender. Cereb Cortex 18:2019–2026. https://doi.org/10.1093/cercor/bhm227. McCandliss BD, Cohen L, Dehaene S (2003) The visual word form area: expertise for reading in the fusiform gyrus. Trends Cogn Sci 7:293–299. McClelland JL, Rogers TT (2003) The parallel distributed processing approach to semantic cognition. Nat Rev Neurosci 4:310–322. https://doi.org/10.1038/nrn1076. Mesulam MM, Mufson EJ (1982) Insula of the old world monkey. I. Architectonics in the insulo-orbito-temporal component of the paralimbic brain. J Comp Neurol 212:1–22. https://doi.org/ 10.1002/cne.902120102. Miller EK, Cohen JD (2001) An integrative theory of prefrontal cortex function. Annu Rev Neurosci 24:167–202. https://doi.org/10.1146/ annurev.neuro.24.1.167. Mirman D, Chen Q, Zhang Y, Wang Z, Faseyitan OK, Coslett HB, Schwartz MF (2015) Neural organization of spoken language revealed by lesion-symptom mapping. Nat Commun 6:6762. https://doi.org/10.1038/ncomms7762. Murre JM, Sturdy DP (1995) The connectivity of the brain: multi-level quantitative analysis. Biol Cybern 73:529–545. Nieoullon A (2002) Dopamine and the regulation of cognition and attention. Prog Neurobiol 67:53–83. Ostby Y, Tamnes CK, Fjell AM, Westlye LT, Due-Tonnessen P, Walhovd KB (2009) Heterogeneity in subcortical brain development: a structural magnetic resonance imaging study of brain maturation from 8 to 30 years. J Neurosci 29:11772–11782. https://doi.org/10.1523/JNEUROSCI.1242-09.2009.
Panizzon MS et al (2009) Distinct genetic influences on cortical surface area and cortical thickness. Cereb Cortex 19:2728–2735. https://doi.org/10.1093/cercor/bhp026. Pietschnig J, Penke L, Wicherts JM, Zeiler M, Voracek M (2015) Meta-analysis of associations between human brain volume and intelligence differences: How strong are they and what do they mean? Neurosci Biobehav Rev 57:411–432. https://doi.org/ 10.1016/j.neubiorev.2015.09.017. Razlighi QR et al (2016) Dynamic patterns of brain structure-behavior correlation across the lifespan. Cereb Cortex. https://doi.org/ 10.1093/cercor/bhw179. Ruppin E, Schwartz EL, Yeshurun Y (1993) Examining the volume efficiency of the cortical architecture in a multi-processor network model. Biol Cybern 70:89–94. Schaer M, Cuadra MB, Tamarit L, Lazeyras F, Eliez S, Thiran JP (2008) A surface-based approach to quantify local cortical gyrification. IEEE Trans Med Imaging 27:161–170. https://doi. org/10.1109/TMI.2007.903576. Schnack HG et al (2015) Changes in thickness and surface area of the human cortex and their relationship with intelligence. Cereb Cortex 25:1608–1617. https://doi.org/10.1093/cercor/bht357. Shaw P et al (2006) Intellectual ability and cortical development in children and adolescents. Nature 440:676–679. https://doi.org/ 10.1038/nature04513. Smith AB, Taylor E, Brammer M, Rubia K (2004) Neural correlates of switching set as measured in fast, event-related functional magnetic resonance imaging. Hum Brain Mapp 21:247–256. https://doi.org/10.1002/hbm.20007. Toro R, Burnod Y (2005) A morphogenetic model for the development of cortical convolutions. Cereb Cortex 15:1900–1913. https://doi. org/10.1093/cercor/bhi068. Wassenberg R et al (2005) Relation between cognitive and motor performance in 5- to 6-year-old children: results from a large-scale cross-sectional study. Child Dev 76:1092–1103. https://doi.org/ 10.1111/j.1467-8624.2005.00899.x. Wechsler D (1997) Wechsler adult intelligence scaleÒ – third edition (WAISÒ-III). San Antonio, TX: Pearson Assessment Inc. Wechsler D (1999) Manual for the wechsler abbreviated scale of intelligence. San Antonio, TX: The Psychological Corporation. Welker W (1990) Why does cerebral cortex fissure and fold? A review of determinants of gyri and sulci. Cereb Cortex 8:3–136. doi: citeulike-article-id:1263649. Westlye LT, Walhovd KB, Bjornerud A, Due-Tonnessen P, Fjell AM (2009) Error-related negativity is mediated by fractional anisotropy in the posterior cingulate gyrus–a study combining diffusion tensor imaging and electrophysiology in healthy adults. Cereb Cortex 19:293–304. https://doi.org/10.1093/cercor/bhn084. White T, Su S, Schmidt M, Kao CY, Sapiro G (2010) The development of gyrification in childhood and adolescence. Brain Cogn 72:36–45. https://doi.org/10.1016/j.bandc.2009.10.009. Winkler AM et al (2010) Cortical thickness or grey matter volume? The importance of selecting the phenotype for imaging genetics studies. Neuroimage 53:1135–1146. https://doi.org/10.1016/j. neuroimage.2009.12.028. Woo CW, Krishnan A, Wager TD (2014) Cluster-extent based thresholding in fMRI analyses: pitfalls and recommendations. Neuroimage 91:412–419. https://doi.org/10.1016/j. neuroimage.2013.12.058. Wu N, Wang F, Jin Z, Zhang Z, Wang LK, Zhang C, Sun T (2017) Effects of GABAB receptors in the insula on recognition memory observed with intellicage. Behav Brain Funct 13:7. https://doi.org/ 10.1186/s12993-017-0125-4. Yurgelun-Todd DA, Killgore WD, Young AD (2002) Sex differences in cerebral tissue volume and cognitive performance during adolescence. Psychol Rep 91:743–757. https://doi.org/10.2466/ pr0.2002.91.3.743.
(Received 27 February 2018, Accepted 5 July 2018) (Available online 19 July 2018)
The Relationship Between General Intelligence and Cortical Structure in Healthy Individuals Sahil Bajaj, a* Adam Raikes, a Ryan Smith, a Natalie S. Dailey, a Anna Alkozei, a John R. Vanuk a and William D. S. Killgore a,b a
Social, Cognitive and Affective Neuroscience Laboratory (SCAN Lab), Department of Psychiatry, College of Medicine, University of Arizona, Tucson, AZ 85724, USA b
McLean Hospital, Department of Psychiatry, Harvard Medical School, Belmont, MA, USA
Abstract—Considerable work in recent years has examined the relationship between cortical thickness (CT) and general intelligence (IQ) in healthy individuals. It is not known whether specific IQ variables (i.e., perceptual reasoning [PIQ], verbal comprehension IQ [VIQ], and full-scale IQ [FSIQ]) are associated with multiple cortical measures (i.e., CT, cortical volume (CV), cortical surface area (CSA) and cortical gyrification (CG)) within the same individuals. Here we examined the association between these neuroimaging metrics and IQ in 56 healthy adults. At a cluster-forming threshold (CFT) of p < 0.05, we observed significant positive relationships between CT and all three IQ variables in regions within the posterior frontal and superior parietal lobes. Regions within the temporal and posterior frontal lobes exhibited positive relationships between CV and two IQ variables (PIQ and FSIQ) and regions within the inferior parietal lobe exhibited positive relationships between CV and PIQ. Additionally, CV was positively associated with VIQ in the left insula and with FSIQ within the inferior frontal gyrus. At a more stringent CFT (p < 0.01), the CT–PIQ, CT–VIQ, CT–FSIQ, and CV–PIQ relationships remained significant within the posterior frontal lobe, as did the CV–PIQ relationship within the temporal and inferior parietal lobes. We did not observe statistically significant relationships between IQ and either CSA or CG. Our findings suggest that the neural basis of IQ extends beyond previously observed relationships with fronto-parietal regions. We also conclude that CT and CV may be more useful metrics than CSA or CG in the study of intellectual abilities. Ó 2018 IBRO. Published by Elsevier Ltd. All rights reserved.
Key words: cortical measures, brain structure–function, intelligence, verbal comprehension, perceptual reasoning, full-scale IQ.
function (Burgaleta et al., 2014; Gregory et al., 2016). Mathematically, many of these structural measures are interrelated. For example, CV is the product of two, genetically distinct properties: CT and CSA (Winkler et al., 2010). CG positively correlates with both total brain volume and CSA but negatively with CT (Gautam et al., 2015; Hogstrom et al., 2013). Given the complex relationships between these heterogeneous morphometric properties, it is crucial to study the simultaneous relevance of each of these measures to cognition (e.g., their individual and joint relationships with measures of human intellectual ability [IQ)]). For instance, if two structural measures, such as CT and CSA, both have a positive association with cognitive abilities, then CV should also exhibit a positive, and possibly stronger, relationship than either CT or CSA alone. Fundamentally, despite the fact that some cortical measures (CT, CV, CSA and CG) may covary, these measures reflect different facets of brain structure and could each contribute uniquely to cognitive function. Specifically, CT and CSA are considered highly heritable and genetically independent (Eyler et al., 2011; Kremen et al., 2010; Panizzon et al., 2009). Due to their
INTRODUCTION Surface-based structural brain analysis techniques, involving an estimation of morphometric measures such as cortical thickness (CT), cortical volume (CV), cortical surface area (CSA), and cortical gyrification (CG) have expanded our understanding of the impact of various neurodegenerative diseases on brain structure. These measures provide specific indices of several unique aspects of brain structure. These measures also provide insight into the dynamics of brain structure during neurodevelopment (Burgaleta et al., 2014; White et al., 2010), the effects of aging on brain structure (Bajaj et al., 2017; Burgaleta et al., 2014; Hogstrom et al., 2013), and the relationships between brain structure and
*Corresponding author. Address: 1501 N Campbell Avenue, Department of Psychiatry, Room # 7304B, University of Arizona, Tucson, AZ 85724, USA. E-mail address: [email protected] (S. Bajaj). Abbreviations: CFT, cluster-forming threshold; CG, cortical gyrification; CSA, cortical surface area; CT, cortical thickness; CV, cortical volume; FSIQ, full-scale IQ; IQ, intelligence; PIQ, perceptual reasoning; VIQ, verbal comprehension. https://doi.org/10.1016/j.neuroscience.2018.07.008 0306-4522/Ó 2018 IBRO. Published by Elsevier Ltd. All rights reserved. 36
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
unique regional variation with the timing of prenatal perturbations, these measures are often understood as separate morphometric variables during neurodevelopment, aging, and disease (Eyler et al., 2011; Hogstrom et al., 2013; Panizzon et al., 2009). Moreover, CV is considered distinct from, though mathematically related to, CT and CSA (i.e., due to the non-uniform, integrated influence of CT and CSA) (Gerrits et al., 2016). Further, CG represents the convolution of brain structure and is influenced by both neuronal density and intra-cranial volume (Toro and Burnod, 2005; Welker, 1990). This folding increases the potential for a greater number of neurons within the same space. When coupled with the fact that CSA may facilitate better neuronal signal processing, increased CG and CSA may provide more efficient structural organization to facilitate brain connectivity and functional development (Luders et al., 2008; Murre and Sturdy, 1995; Ruppin et al., 1993). This supports the idea that the organization and gyral structure of the cortex contributes to its function. With respect to structural measurements and cognition, the relationship between general intelligence (IQ) and structural measures, particularly CT and CSA, is likely age-dependent. Notably, Schnack and colleagues reported that among children (10 years of age), higher standardized IQ scores were associated with thinner cortex within the left hemisphere (Schnack et al., 2015). This pattern differed in young adults, where thicker left hemisphere cortex was associated with higher IQ scores, suggesting that maturation-related structural changes may play an important role in the expression of IQ. Consistent with these findings, Shaw and colleagues reported a shift from a negative correlation between CT and IQ in early childhood (age range 3.8–8.4 years) to a positive correlation in late childhood (age range 8.6–11.7 years) and into early adulthood (age range 11.8–29 years) (Shaw et al., 2006). In other work with healthy children and adolescents, changes in CT, but not CSA, were associated with changes in performance IQ, verbal IQ, and full-scale IQ (Burgaleta et al., 2014). Furthermore, higher CG in middle-aged participants (44–48 years) was associated with better mental flexibility, larger brain volume, but also thinner cortex (Gautam et al., 2015). Therefore, relationships between IQ and the structural characteristics of the cortex appear to exhibit dynamic, age- and developmental stage-dependent changes. A recent meta-analysis suggested that much of the previous literature may over-estimate/over-simplify the association between brain volume and IQ (Pietschnig et al., 2015). The association instead appears to be more complex than suggested by simple volumetric measures. Surface-based cortical measures, rather than purely volumetric measures, may therefore provide more useful information about the complex and widespread relationships between cortical structure and general IQ. Prior work has generally focused on individual structural measures in isolation and their association with IQ. Given that neural structure–function relationships are likely more complex than can be accounted for by simple volumetric measures, it is crucial to examine the parallel associations between
37
multiple brain structure measures (CT, CV, CSA, and CG) and multiple aspects of general IQ (perceptual reasoning (PIQ), verbal comprehension and full-scale IQ) in the same sample of healthy young adults. The primary aim of the current study is therefore to investigate the vertex-wise associations between these structural metrics and these different IQ variables. Based on prior research, we hypothesized that both CT and CV would be positively associated with all domains of IQ. However, due to inconsistent and limited evidence regarding the relationship between CSA or CG and cognitive abilities, we had no specific hypothesis regarding the relationships between these specific metrics and IQ. Thus, these latter analyses were primarily for exploratory purposes.
EXPERIMENTAL PROCEDURES Participants Fifty-six healthy participants between 18 and 45 years of age (mean age = 30.8 ± 8.1 years, 27 females; 29 males) participated in this study. Participants were recruited from the New England area and screened via a comprehensive telephone interview. Individuals with any history of psychiatric, neurological, or significant medical problems, current use of psychotropic medications, or current use of illicit substances were excluded. Participants were all primary native English speakers, with 14.95 years of formal education (SD = 2. 18 years). 69.6% of participants were Caucasian, 14.2% were African American, 8.9% were Asian, 3.6% reported polyethnic backgrounds and 3.6% reported other ethnic backgrounds. All participants provided written informed consent prior to enrollment. The study protocol was approved by the Institutional Review Boards of McLean Hospital and Partners Healthcare, and the U.S. Army Human Research Protections Office. Other behavioral data and structural estimates from this sample have been reported elsewhere (Killgore et al., 2013), but the vertex-wise cortical measures and their associations with IQ reported in this study are novel and have not been previously reported. Data acquisition Neuroanatomical data. We recorded high-resolution T1-weighted magnetic resonance imaging data using a 3-Tesla Siemens TIM Trio whole-brain MR scanner located at the McLean Hospital Imaging Center. Before the scan, each participant was instructed to rest, relax and try his/her best to minimize movement during the entire scan. Head movement was further minimized with foam padding placed comfortably about the head. T1weighted data for each participant were acquired using a 3D magnetization-prepared rapid acquisition gradient echo (MPRAGE) sequence, which consisted of 128 sagittal slices (slice thickness = 1.33 mm, voxel resolu tion = 1.33 1 1 mm, field of view (FOV) = 256 mm with TR/TE/FA/inversion time of 2100 ms/2.25 m s/12°/1100 ms).
38
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
General intelligence (IQ) assessment. A trained research technician, supervised by a licensed neuropsychologist, administered the IQ assessment to each participant. All participants completed the Wechsler Abbreviated Scale of Intelligence II (WASI-II; (Wechsler, 1999). The WASI-II provides estimates of IQ in the domains of perceptual reasoning (Block Design and Matrix Reasoning; PIQ) and verbal comprehension (Vocabulary and Similarities; VIQ). The combination of PIQ and VIQ provides an estimate of Full-Scale IQ (FSIQ). PIQ reflects overall visuospatial intellectual abilities (e.g., spatial processing, visual-motor integration, and attentiveness to non-verbal tasks). VIQ reflects overall language capacity (e.g., measures of verbal expressive, verbal comprehension, and verbal reasoning abilities). FSIQ indicates overall cognitive and intellectual ability. The four subtests of the WASI-II (Block Design, Matrix Reasoning, Vocabulary, and Similarities) have strong associations with more extensive measures of general cognitive abilities (Hays et al., 2002; Wechsler, 1999). FSIQ scores have high reliability (0.98), and also have a correlation (r = 0.92) with scores on the full Wechsler Adult Intelligence Scale (WAIS)-III (Wechsler, 1997, 1999). Data analysis
Image processing and brain structure measures. We used the ‘‘recon-all” pipeline in FreeSurfer (version 6.0) (https://surfer.nmr.mgh.harvard.edu) to process the anatomical brain images for all the participants (Dale et al., 1999; Fischl et al., 1999) and for estimating the structural measures. Processing involved basic imagepreprocessing steps, including motion-correction, brain extraction (i.e., removal of non-brain tissue), automated transformation to the standard MNI co-ordinate system, volumetric segmentation into cortical and sub-cortical matter, intensity correction, and parcellation of the cerebral cortex into gyral and sulcal matter (Desikan et al., 2006). The technical details of these steps are documented in previous publications (Dale et al., 1999; Fischl et al., 1999; Fischl et al., 2004). To inspect FreeSurfer’s preprocessing accuracy, standard quality control steps were performed, which involved a careful visual inspection of raw T1-weighted images, skull-stripped brain volumes, and pial surfaces. The structural measures (CT, CV, CSA, and CG) were estimated separately for the left and the right hemisphere for each participant. CT was estimated as the shortest distance between the white matter (WM) surface (white–gray matter interface) and the pial surface (grey matter–CSF interface) (Fischl and Dale, 2000). CV was estimated as the amount of grey matter that lies between white–grey matter interface and pial matter (Winkler et al., 2010). CSA was estimated as the sum of the areas of the triangles making up the surface model and is defined as the extent of 2dimensional surface enclosed by the outer layer of the cerebral cortex (Fischl et al., 1999). Finally, CG was estimated as the ratio of the area on the outer surface to the area on the pial surface (Schaer et al., 2008).
Statistical analysis. We fit separate generalized linear models (GLMs) to the left and right hemispheres within FreeSurfer’s statistical engine to examine structure–IQ relationships. Standardized measures of each IQ subdomain (PIQ, VIQ, and FSIQ – independent variables) were regressed on each structural measure (dependent variables). For this analysis, we selected two threshold levels. First, to minimize Type II error and inform future hypothesis testing, we used a liberal cluster-forming threshold (CFT) of p < 0.05. To minimize Type I error, we also report results with a stricter CFT of p < 0.01. CFTs are used to detect significant brain clusters based on the number of clustered voxels, whose voxel-wise statistic values are above the pre-defined threshold (Woo et al., 2014). Previously, liberal CFTs have been used to determine clusters after correcting for multiple comparisons across the surfaces (Fuglset et al., 2016). However, liberal cluster-forming height thresholds increase cluster size and may highlight more than one region, resulting in reduced anatomical specificity. Since cluster-wise false positive rates of cortical measures may depend on smoothing levels, our analysis involved liberal, as well as, stringent CFTs at two different smoothing levels to better understand the associations between cortical structure and general IQ. As the sample size in our study is small and the clusters are not expected to be localized with a high degree of specificity, we selected a smaller Gaussian smoothing kernel size of 10 mm with a CFT of p < 0.05 and a moderately larger smoothing kernel size of 15 mm with a CFT of p < 0.01. Moreover, unlike volume-based analysis, a larger smoothing kernel size in surface-based analysis never extends into bone/ air/white matter. For all analyses, we used a clusterwise threshold (CWP) of p < 0.05 (corrected for multiple comparisons using Monte-Carlo simulations) to identify significant clusters. All models were controlled for age and sex.
RESULTS Associations between general IQ and each cortical measure CT and IQ. We identified clusters within the posterior frontal and superior parietal lobes that showed significant relationships between thicker cortex and higher IQ at a liberal CFT of p < 0.05 as well as at a strict CFT of p < 0.01. Fig. 1 displays the clusters showing these associations between CT and PIQ (A–D), VIQ (E–G), and FSIQ (H–K) at CFT of p < 0.05 at FWHM of 10 mm. Fig. 2 displays the clusters showing these associations between CT and PIQ (A, B), VIQ (C–F), and FSIQ (G–I) at CFT of p < 0.01 at FWHM of 15 mm. No clusters showed negative association between CT and any IQ measures at CFT of p < 0.05. Identified clusters are summarized in Table 1. CV and IQ. We identified clusters within the temporal, posterior frontal, and parietal lobes as well as inferior frontal gyrus that showed significant relationships between higher CV and higher IQ, especially at CFT of
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
39
Fig. 1. Relationships between cortical thickness (CT) and IQ. We found positive associations between CT and all three subscales of IQ – Perceptual Reasoning IQ (PIQ) (A–D), Verbal Comprehension IQ (VIQ) (E–G), and Full-Scale IQ (FSIQ) (H–K) (CFT: p 0.05, CWP: p 0.05, corrected for multiple comparisons).
p < 0.05. However, at CFT of p < 0.01, higher CV was only positively associated with PIQ. Fig. 3 displays the clusters showing these associations between CV and PIQ (A–C), VIQ (D, E), and FSIQ (F, G) at CFT of p < 0.05 at FWHM of 10 mm. Fig. 4 displays the clusters showing these associations between CV and PIQ (A–C) at CFT of p < 0.01 at FWHM of 15 mm. No clusters showed negative association between CV and any IQ measures at CFT of p < 0.05. Identified clusters are summarized in Table 2.
idea that IQ is associated with quantifiable differences in brain structure within regions that include, but are not limited to, the fronto-parietal regions most emphasized in previous work on general IQ (Jung and Haier, 2007). Our results further suggest that CT and CV may be more useful than CG and CSA in studying the contribution of cortical structure to IQ.
CSA and IQ. There were no significant relationships found between CSA and IQ at CFT of p < 0.05.
Our findings further advance the results reported from several previous studies. We observed significant relationships between multiple aspects of general IQ and morphometric characteristics in areas associated with working memory, motor-skills, and physical exercise (e.g., the precentral gyrus [PreCG]) (Burgaleta et al., 2014; Killgore and Schwab, 2012; Miller and Cohen, 2001). Previous work indicates that brain areas, such as the basal ganglia, frontal cortex, and related dopaminergic pathways, play important roles in both motor performance and cognitive performance abilities (Diamond, 2000; Nieoullon, 2002; Wassenberg et al., 2005). As expected, we also found significant relationships between IQ and cortical areas responsible for (a) integrating and processing information, including the parahippocampal gyrus and precuneus/cuneus cortex (PreC/CunC) (Cavanna and Trimble, 2006; Hulshoff Pol et al., 2006; Westlye et al., 2009), (b) creative thinking
CG and IQ. There were no significant relationships found between CG and IQ at CFT of p < 0.05.
DISCUSSION In this study we explored the association between general IQ and multiple measures of brain structure. We identified significant positive associations between all three IQ scales (PIQ, VIQ, and FSIQ) and two structural measures (CT and CV) for several regions distributed across the brain. On the other hand, IQ was not significantly correlated with CSA or CG. Thus, greater thickness and volume of the cortex were associated with better intellectual functioning, whereas surface area and gyrification were not. These findings support the
Associations between general IQ measures and structural measures
40
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
Fig. 2. Relationships between cortical thickness (CT) and IQ. We found positive associations between CT and all three subscales of IQ – Perceptual Reasoning IQ (PIQ) (A, B), Verbal Comprehension IQ (VIQ) (C–F), and Full-Scale IQ (FSIQ) (G–I) (CFT: p 0.01, CWP: p 0.05, corrected for multiple comparisons).
and memory retrieval, including the PreC (Chen et al., 2015), (c) visual identification and recognition, such as regions within the ventral temporal cortex e.g. fusiform gyrus (Bar et al., 2001; McCandliss et al., 2003), and (d) integration and retrieval of semantic knowledge, including the medial temporal lobes (McClelland and Rogers, 2003). These results are also consistent with previous work highlighting the relationships between IQ and CT within the temporal lobe (Choi et al., 2008). In our sample the left insula was the only brain region to show significant positive association between higher CV and VIQ. The insular cortex is strongly interconnected with the temporal cortex (Mesulam and Mufson, 1982) and limbic system, and has previously been implicated in learning and memory functions (Balderas et al., 2015; Bermudez-Rattoni, 2014; Wu et al., 2017). In the left hemisphere of most people, this region plays a crucial role in processing and integrating cognitive information associated with auditory perception and receptive language (Mirman et al., 2015). A large body of previous work has highlighted the contribution of fronto-parietal regions to general IQ, and has led to the Parieto-Frontal Integration Theory of Intelligence (P-FIT) (Jung and Haier, 2007). This theory suggests the structure and function of brain are intricately connected. For instance, inferior frontal regions (such as
the pars opercularis) play a direct role in goal-directed cognitive performance (Demonet et al., 2005). Additionally, the parietal regions further contribute through integrating sensory information (i.e., first processed in occipital and temporal cortices), allocating attention, and facilitating effective problem solving (Gevins and Smith, 2000; Graham et al., 2010; Karama et al., 2009; Smith et al., 2004). Our findings generally support this theory but add to a growing body of work suggesting that brain regions outside of this frontoparietal system also contribute to general IQ (Colom et al., 2009; Hearne et al., 2016). In other words, our findings support the position that structural correlates of general IQ are widespread and not strictly limited to areas identified in the P-FIT model, which nevertheless remains provisional and experimental. Our findings are also consistent with those of a meta-analysis of 12 structural and 16 functional imaging studies that presented an updated neurocognitive model of general IQ. This model includes not only structures within the frontal, parietal, temporal, and occipital lobes but also the insula, posterior cingulate cortex, and several subcortical regions (Basten et al., 2015). In contrast to CV and CT, we found no significant associations between IQ and either CSA or CG. Previous studies have inconsistently identified relationships between CSA and measures of general IQ
41
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
Table 1. Relationships between cortical thickness (CT) and IQ scales – Perceptual Reasoning IQ (PIQ), Verbal Comprehension IQ (VIQ) and Full-Scale IQ (FSIQ) Clusters showing significant relationships Between CT and PIQ scores Cluster number
Maxima
Peak co-ordinates (MNI: X, Y, Z)
CWP
Number of vertices within the cluster
FreeSurfer label
Association
1 2 3 4 5 6 7
6.04 3.01 4.88 3.56 3.21 5.67 5.02
42.6, 8.1, 54.7 28.8, 43.8, 17.3 59.1, 3.1, 26.3 28.6, 28.5, 20.5 22.5, 58.3, 21.1 43.0, 7.9, 53.5 58.4, 1.3, 28.1
0.00 0.02 0.01 0.00 0.03 0.00 0.01
4822 1812 2245 2400 1557 4074 1861
LPreCG* LFG* RPreCG* RPHG* RPreC* LPreCG** RPreCG**
(+) (+) (+) (+) (+) (+) (+)
Between CT and VIQ scores 1 6.23 2 4.08 3 4.54 4 4.31 5 4.00 6 2.86 7 6.72 8 4.28 9 3.09
28.3, 20.5, 68.6 54.0, 17.9, 3.6 17.7, 56.0, 21.3 12.6, 43.5, 71.4 58.6, 6.4, 26.7 9.6, 95.9, 10.2 27.8, 20.9, 68.3 12.3, 45.2, 69.8 26.1, 77.0, 6.6
0.00 0.00 0.02 0.00 0.00 0.00 0.00 0.01 0.00
5595 2776 1951 3218 3420 2421 5099 2306 1505
LPreCG* LTTG* RPreC* RPreC* RPreCG* RCunC* LPreCG** RPreC** RFG**
(+) (+) (+) (+) (+) (+) (+) (+) (+)
Between CT and FSIQ scores 1 6.66 2 3.37 3 4.89 4 4.76 5 3.02 6 6.58 7 4.55 8 3.26
42.7, 8.6, 54.9 28.8, 43.8, 17.3 11.2, 47.4, 65.9 59.0, 5.3, 26.7 28.7, 29.4, 19.6 43.1, 8.2, 53.8 58.6, 1.8, 27.3 33.8, 83.0, 13.3
0.00 0.00 0.03 0.00 0.00 0.00 0.02 0.01
6865 2184 2469 3673 2779 6118 1801 1308
LPreCG* LFG* RPreC* RPreCG* RPHG* LPreCG** RPreCG** RLOC**
(+) (+) (+) (+) (+) (+) (+) (+)
Abbreviations: Left/Right Precentral Gyrus (L/R: PreCG); Left/Right Fusiform Gyrus (L/R: FG); Right Parahippocampal Gyrus (RPHG); Right Precuneus Cortex (RPreC); Left Transverse Temporal Gyrus (LTTG); Right Cuneus Cortex (RCunC); Right Lateral Occipital Cortex (RLOC). * Significant at cluster-forming threshold (CFT) of p < 0.05 (FWHM 10 mm) and cluster-wise threshold of p (CWP) < 0.05 (corrected for multiple comparisons using Monte Carlo simulation). ** Significant at cluster-forming threshold (CFT) of p < 0.01 (FWHM 15 mm) and cluster-wise threshold of p (CWP) < 0.05 (corrected for multiple comparisons using Monte Carlo simulation).
(Burgaleta et al., 2014; Schnack et al., 2015). Additionally, developmental changes in CSA are generally less notable than those for CV (Ostby et al., 2009). However, Schnack and colleagues found that by age 10, CSA is often larger among children with greater IQ, developing to an apex during adolescence (Schnack et al., 2015). Given that all of our participants were within the age range of 18–45, one potential explanation for the non-significant relationship between CSA and general IQ is that by early adulthood (i.e., age 18) cortical expansion, which leads to increased CSA, is complete in most individuals and instead begins contracting, potentially resulting in higher CT and CV (Giedd et al., 1999). Moreover, CSA contraction within several brain areas, including the bilateral PreCG and CunC, was previously reported to be more prominent among adults with higher IQ (Schnack et al., 2015). CG, however, is a conceptually distinct morphological measure from CT or CV. Prior work has postulated that stronger cognitive abilities may be related to greater CG due to a larger number of neurons within the convoluted cortex as well as larger surface area, which facilitates the processing of information (Gautam et al.,
2015; Gregory et al., 2016; Luders et al., 2008). However, we found no association between CG and measured IQ. Notably, CG is a relatively smooth structural measure throughout the cortex, and often requires a larger surface area to show significant associations with cognitive measures (Razlighi et al., 2016). Thus, CG and CSA are not entirely independent metrics, and neither was independently associated with IQ in the present study. Limitations The present findings should be interpreted in light of several limitations. First, we had a widespread age range over a relatively small sample size, which could account for some of the vertex-wise structural differences across the three IQ domains. Second, the absence of significant relationships between both CSA and CG and IQ was unexpected and suggests the necessity of continuing to explore these relationships for different age groups, given typical developmental and structural changes that may continue into the third decade of life (Lenroot and Giedd, 2006). Additionally, we controlled for sex in our models.
42
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
Table 2. Relationships between cortical volume (CV) and IQ scales – Perceptual Reasoning IQ (PIQ), Verbal Comprehension IQ (VIQ) and Full-Scale IQ (FSIQ) Clusters showing significant relationships Between CV and PIQ scores Cluster number
Maxima
Peak co-ordinates (MNI: X, Y, Z)
CWP
Number of vertices within the cluster
FreeSurfer label
Association
1 2 3 4 5 6 7 8 9 10 11
4.20 3.99 3.83 4.22 3.68 3.53 4.74 4.09 3.87 3.61 3.50
41.7, 9.1, 37.9 39.6, 31.9, 7.8 49.0, 8.6, 39.8 49.2, 2.2, 25.5 46.7, 7.0, 27.6 38.6, 76.1, 12.7 42.0, 9.8, 36.7 48.1, 9.0, 39.5 43.7, 28.8, 4.0 47.1, 6.5, 28.0 39.0, 76.3, 13.3
0.00 0.00 0.01 0.01 0.00 0.00 0.00 0.04 0.01 0.03 0.04
2464 3482 2917 1957 3202 2003 2246 1765 2479 2047 1320
LITG* LTTG* LPreCG* RSTG* RPreCG* RIPC* LITG** LPreCG** LTTG** RPreCG** RIPC**
(+) (+) (+) (+) (+) (+) (+) (+) (+) (+) (+)
31.0, 17.5, 13
0.01
2442
LIns*
(+)
41.7, 30.6, 5.7 31.5, 13.5, 13.2 49.6, 5.1, 27.4
0.01 0.05 0.05
2743 2247 2135
LTTG* LPOC* RPreCG*
(+) (+) (+)
Between CV and VIQ scores 1 3.25 Between CV and FSIQ scores 1 3.92 2 2.73 3 3.20
Abbreviations: Left Inferior Temporal Gyrus (LITG); Left Transverse Temporal Gyrus (LTTG); Left/Right Precentral Gyrus (L/R: PreCG); Right Superior Temporal Gyrus (RSTG); Right Inferior Parietal Cortex (RIPC); Left Insula (LIns); Left Pars Opercularis (LPOC). * Significant at cluster-forming threshold (CFT) of p < 0.05 (FWHM 10 mm) and cluster-wise threshold of p (CWP) < 0.05 (corrected for multiple comparisons using Monte Carlo simulation). ** Significant at cluster-forming threshold (CFT) of p < 0.01 (FWHM 15 mm) and cluster-wise threshold of p (CWP) < 0.05 (corrected for multiple comparisons using Monte Carlo simulation).
However, men and women show different developmental trajectories of cognitive abilities and brain morphology measurements (YurgelunTodd et al., 2002) and sex-specific relationships between brain morphology and cognition merit further attention. Future studies incorporating larger sample sizes with consistent representations across age ranges are necessary to corroborate the associations between the structural and IQ measures reported in our study.
CONCLUDING REMARKS
Fig. 3. Relationships between cortical volume (CV) and IQ. We found positive associations between CV and all three subscales of IQ – Perceptual Reasoning IQ (PIQ) (A–C), Verbal Comprehension IQ (VIQ) (D, E), and Full-Scale IQ (FSIQ) (F, G) (CFT: p 0.05, CWP: p 0.05, corrected for multiple comparisons).
The present findings advance our understanding of how brain structure is associated with IQ by including, for the first time, multiple metrics of brain structure and IQ in the same study. We provide evidence that greater general IQ may be functionally related to CT and volume extending beyond previously reported fronto-parietal regions typically associated with human IQ. These findings therefore add to a growing literature on the neural basis of individual differences in general cognitive ability.
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
43
Chen QL et al (2015) Individual differences in verbal creative thinking are reflected in the precuneus. Neuropsychologia 75:441–449. https://doi.org/10.1016/j. neuropsychologia.2015.07.001. Choi YY et al (2008) Multiple bases of human intelligence revealed by cortical thickness and neural activation. J Neurosci 28:10323–10329. https://doi.org/10.1523/ JNEUROSCI.3259-08.2008. Colom R, Haier RJ, Head K, A´lvarez-Linera J, Quiroga MA´, Shih PC, Jung RE (2009) Fig. 4. Relationships between cortical volume (CV) and IQ. We found positive associations Gray matter correlates of fluid, between CV and Perceptual Reasoning IQ (PIQ) (A–C) (CFT: p 0.01, CWP: p 0.05, corrected crystallized, and spatial intelligence: for multiple comparisons). testing the P-FIT model. Intelligence 37:124–135. https://doi.org/10.1016/j. intell.2008.07.007. CONFLICTS OF INTEREST STATEMENT Dale AM, Fischl B, Sereno MI (1999) Cortical surface-based analysis. I. Segmentation and surface reconstruction. Neuroimage All the authors declared no conflicts of interest. 9:179–194. https://doi.org/10.1006/nimg.1998.0395. Demonet JF, Thierry G, Cardebat D (2005) Renewal AUTHOR CONTRIBUTIONS of the neurophysiology of language: functional neuroimaging. Physiol Rev 85:49–95. https://doi.org/10.1152/physrev.00049. SB conducted the neuroimaging analyses and wrote the 2003. initial draft of the manuscript and organized the Desikan RS et al (2006) An automated labeling system for revisions. AR and RS provided critical comments and subdividing the human cerebral cortex on MRI scans into gyral contributed to the writing of the manuscript. NSD, AA based regions of interest. Neuroimage 31:968–980. https://doi. org/10.1016/j.neuroimage.2006.01.021. and JRV contributed to the writing of revisions of the Diamond A (2000) Close interrelation of motor development and manuscript. WDSK obtained the funding, designed the cognitive development and of the cerebellum and prefrontal study, supervised staff training, data collection, and cortex. Child Dev 71:44–56. analysis, and contributed to writing revisions of the Eyler LT et al (2011) Genetic and environmental contributions to manuscript. regional cortical surface area in humans: a magnetic resonance imaging twin study. Cereb Cortex 21:2313–2321. https://doi.org/ 10.1093/cercor/bhr013. FUNDING Fischl B, Dale AM (2000) Measuring the thickness of the human cerebral cortex from magnetic resonance images. Proc Natl Acad This research was supported by a grant from the U.S. Sci U S A 97:11050–11055. https://doi.org/10.1073/ Army Medical Research and Materiel Command to pnas.200033797. WDSK (W81XWH-09-1-0730). The opinions, Fischl B, Sereno MI, Dale AM (1999) Cortical surface-based analysis. interpretations, conclusions and recommendations in II: Inflation, flattening, and a surface-based coordinate system. this paper are solely those of the authors and are not Neuroimage 9:195–207. https://doi.org/10.1006/nimg.1998.0396. necessarily endorsed by the Department of Defense or Fischl B et al (2004) Automatically parcellating the human cerebral the U.S. Army Medical Research and Materiel Command. cortex. Cereb Cortex 14:11–22. Fuglset TS, Endestad T, Hilland E, Bang L, Tamnes CK, Landro NI, Ro O (2016) Brain volumes and regional cortical thickness in REFERENCES young females with anorexia nervosa. BMC Psychiatry 16:404. https://doi.org/10.1186/s12888-016-1126-9. Bajaj S, Alkozei A, Dailey NS, Killgore WDS (2017) Brain aging: Gautam P, Anstey KJ, Wen W, Sachdev PS, Cherbuin N (2015) uncovering cortical characteristics of healthy aging in young Cortical gyrification and its relationships with cortical volume, adults. Front Aging Neurosci 9:412. https://doi.org/10.3389/ cortical thickness, and cognitive performance in healthy mid-life fnagi.2017.00412. adults. Behav Brain Res 287:331–339. https://doi.org/10.1016/j. Balderas I, Rodriguez-Ortiz CJ, Bermudez-Rattoni F (2015) bbr.2015.03.018. Consolidation and reconsolidation of object recognition memory. Gerrits NJ et al (2016) Cortical thickness, surface area and Behav Brain Res 285:213–222. https://doi.org/10.1016/j. subcortical volume differentially contribute to cognitive bbr.2014.08.049. heterogeneity in Parkinson’s disease. PLoS One 11. https://doi. Bar M et al (2001) Cortical mechanisms specific to explicit visual org/10.1371/journal.pone.0148852 e0148852. object recognition. Neuron 29:529–535. Gevins A, Smith ME (2000) Neurophysiological measures of working Basten U, Hilger K, Fiebach CJ (2015) Where smart brains are memory and individual differences in cognitive ability and different: a quantitative meta-analysis of functional and structural cognitive style. Cereb Cortex 10:829–839. brain imaging studies on intelligence. Intelligence 51:10–27. Giedd JN et al (1999) Brain development during childhood and https://doi.org/10.1016/j.intell.2015.04.009. adolescence: a longitudinal MRI study. Nat Neurosci 2:861–863. Bermudez-Rattoni F (2014) The forgotten insular cortex: its role on https://doi.org/10.1038/13158. recognition memory formation. Neurobiol Learn Mem Graham S et al (2010) IQ-related fMRI differences during cognitive 109:207–216. https://doi.org/10.1016/j.nlm.2014.01.001. set shifting. Cereb Cortex 20:641–649. https://doi.org/10.1093/ Burgaleta M, Johnson W, Waber DP, Colom R, Karama S (2014) cercor/bhp130. Cognitive ability changes and dynamics of cortical thickness Gregory MD, Kippenhan JS, Dickinson D, Carrasco J, Mattay VS, development in healthy children and adolescents. Neuroimage Weinberger DR, Berman KF (2016) Regional variations in brain 84:810–819. https://doi.org/10.1016/j.neuroimage.2013.09.038. gyrification are associated with general cognitive ability in Cavanna AE, Trimble MR (2006) The precuneus: a review of its humans. Curr Biol 26:1301–1305. https://doi.org/10.1016/j. functional anatomy and behavioural correlates. Brain cub.2016.03.021. 129:564–583. https://doi.org/10.1093/brain/awl004.
44
S. Bajaj et al. / Neuroscience 388 (2018) 36–44
Hays JR, Reas DL, Shaw JB (2002) Concurrent validity of the Wechsler abbreviated scale of intelligence and the Kaufman brief intelligence test among psychiatric inpatients. Psychol Rep 90:355–359. https://doi.org/10.2466/pr0.2002.90.2.355. Hearne LJ, Mattingley JB, Cocchi L (2016) Functional brain networks related to individual differences in human intelligence at rest. Sci Rep 6:32328. https://doi.org/10.1038/srep32328. Hogstrom LJ, Westlye LT, Walhovd KB, Fjell AM (2013) The structure of the cerebral cortex across adult life: age-related patterns of surface area, thickness, and gyrification. Cereb Cortex 23:2521–2530. https://doi.org/10.1093/cercor/bhs231. Hulshoff Pol HE et al (2006) Genetic contributions to human brain morphology and intelligence. J Neurosci 26:10235–10242. https:// doi.org/10.1523/JNEUROSCI.1312-06.2006. Jung RE, Haier RJ (2007) The Parieto-Frontal Integration Theory (PFIT) of intelligence: converging neuroimaging evidence. Behav Brain Sci 30:135–154. https://doi.org/10.1017/ S0140525X07001185. discussion 154–187. Karama S et al (2009) Positive association between cognitive ability and cortical thickness in a representative US sample of healthy 6 to 18 year-olds. Intelligence 37:145–155. https://doi.org/10.1016/j. intell.2008.09.006. Killgore WD, Olson EA, Weber M (2013) Physical exercise habits correlate with gray matter volume of the hippocampus in healthy adult humans. Sci Rep 3:3457. https://doi.org/10.1038/ srep03457. Killgore WD, Schwab ZJ (2012) Sex differences in the association between physical exercise and IQ. Percept Mot Skills 115:605–617. https://doi.org/10.2466/06.10.50.PMS.115.5.605617. Kremen WS et al (2010) Genetic and environmental influences on the size of specific brain regions in midlife: the VETSA MRI study. Neuroimage 49:1213–1223. https://doi.org/10.1016/j. neuroimage.2009.09.043. Lenroot RK, Giedd JN (2006) Brain development in children and adolescents: insights from anatomical magnetic resonance imaging. Neurosci Biobehav Rev 30:718–729. https://doi.org/ 10.1016/j.neubiorev.2006.06.001. Luders E et al (2008) Mapping the relationship between cortical convolution and intelligence: effects of gender. Cereb Cortex 18:2019–2026. https://doi.org/10.1093/cercor/bhm227. McCandliss BD, Cohen L, Dehaene S (2003) The visual word form area: expertise for reading in the fusiform gyrus. Trends Cogn Sci 7:293–299. McClelland JL, Rogers TT (2003) The parallel distributed processing approach to semantic cognition. Nat Rev Neurosci 4:310–322. https://doi.org/10.1038/nrn1076. Mesulam MM, Mufson EJ (1982) Insula of the old world monkey. I. Architectonics in the insulo-orbito-temporal component of the paralimbic brain. J Comp Neurol 212:1–22. https://doi.org/ 10.1002/cne.902120102. Miller EK, Cohen JD (2001) An integrative theory of prefrontal cortex function. Annu Rev Neurosci 24:167–202. https://doi.org/10.1146/ annurev.neuro.24.1.167. Mirman D, Chen Q, Zhang Y, Wang Z, Faseyitan OK, Coslett HB, Schwartz MF (2015) Neural organization of spoken language revealed by lesion-symptom mapping. Nat Commun 6:6762. https://doi.org/10.1038/ncomms7762. Murre JM, Sturdy DP (1995) The connectivity of the brain: multi-level quantitative analysis. Biol Cybern 73:529–545. Nieoullon A (2002) Dopamine and the regulation of cognition and attention. Prog Neurobiol 67:53–83. Ostby Y, Tamnes CK, Fjell AM, Westlye LT, Due-Tonnessen P, Walhovd KB (2009) Heterogeneity in subcortical brain development: a structural magnetic resonance imaging study of brain maturation from 8 to 30 years. J Neurosci 29:11772–11782. https://doi.org/10.1523/JNEUROSCI.1242-09.2009.
Panizzon MS et al (2009) Distinct genetic influences on cortical surface area and cortical thickness. Cereb Cortex 19:2728–2735. https://doi.org/10.1093/cercor/bhp026. Pietschnig J, Penke L, Wicherts JM, Zeiler M, Voracek M (2015) Meta-analysis of associations between human brain volume and intelligence differences: How strong are they and what do they mean? Neurosci Biobehav Rev 57:411–432. https://doi.org/ 10.1016/j.neubiorev.2015.09.017. Razlighi QR et al (2016) Dynamic patterns of brain structure-behavior correlation across the lifespan. Cereb Cortex. https://doi.org/ 10.1093/cercor/bhw179. Ruppin E, Schwartz EL, Yeshurun Y (1993) Examining the volume efficiency of the cortical architecture in a multi-processor network model. Biol Cybern 70:89–94. Schaer M, Cuadra MB, Tamarit L, Lazeyras F, Eliez S, Thiran JP (2008) A surface-based approach to quantify local cortical gyrification. IEEE Trans Med Imaging 27:161–170. https://doi. org/10.1109/TMI.2007.903576. Schnack HG et al (2015) Changes in thickness and surface area of the human cortex and their relationship with intelligence. Cereb Cortex 25:1608–1617. https://doi.org/10.1093/cercor/bht357. Shaw P et al (2006) Intellectual ability and cortical development in children and adolescents. Nature 440:676–679. https://doi.org/ 10.1038/nature04513. Smith AB, Taylor E, Brammer M, Rubia K (2004) Neural correlates of switching set as measured in fast, event-related functional magnetic resonance imaging. Hum Brain Mapp 21:247–256. https://doi.org/10.1002/hbm.20007. Toro R, Burnod Y (2005) A morphogenetic model for the development of cortical convolutions. Cereb Cortex 15:1900–1913. https://doi. org/10.1093/cercor/bhi068. Wassenberg R et al (2005) Relation between cognitive and motor performance in 5- to 6-year-old children: results from a large-scale cross-sectional study. Child Dev 76:1092–1103. https://doi.org/ 10.1111/j.1467-8624.2005.00899.x. Wechsler D (1997) Wechsler adult intelligence scaleÒ – third edition (WAISÒ-III). San Antonio, TX: Pearson Assessment Inc. Wechsler D (1999) Manual for the wechsler abbreviated scale of intelligence. San Antonio, TX: The Psychological Corporation. Welker W (1990) Why does cerebral cortex fissure and fold? A review of determinants of gyri and sulci. Cereb Cortex 8:3–136. doi: citeulike-article-id:1263649. Westlye LT, Walhovd KB, Bjornerud A, Due-Tonnessen P, Fjell AM (2009) Error-related negativity is mediated by fractional anisotropy in the posterior cingulate gyrus–a study combining diffusion tensor imaging and electrophysiology in healthy adults. Cereb Cortex 19:293–304. https://doi.org/10.1093/cercor/bhn084. White T, Su S, Schmidt M, Kao CY, Sapiro G (2010) The development of gyrification in childhood and adolescence. Brain Cogn 72:36–45. https://doi.org/10.1016/j.bandc.2009.10.009. Winkler AM et al (2010) Cortical thickness or grey matter volume? The importance of selecting the phenotype for imaging genetics studies. Neuroimage 53:1135–1146. https://doi.org/10.1016/j. neuroimage.2009.12.028. Woo CW, Krishnan A, Wager TD (2014) Cluster-extent based thresholding in fMRI analyses: pitfalls and recommendations. Neuroimage 91:412–419. https://doi.org/10.1016/j. neuroimage.2013.12.058. Wu N, Wang F, Jin Z, Zhang Z, Wang LK, Zhang C, Sun T (2017) Effects of GABAB receptors in the insula on recognition memory observed with intellicage. Behav Brain Funct 13:7. https://doi.org/ 10.1186/s12993-017-0125-4. Yurgelun-Todd DA, Killgore WD, Young AD (2002) Sex differences in cerebral tissue volume and cognitive performance during adolescence. Psychol Rep 91:743–757. https://doi.org/10.2466/ pr0.2002.91.3.743.
(Received 27 February 2018, Accepted 5 July 2018) (Available online 19 July 2018)