Content uploaded by Donald L Rowe
Author content
All content in this area was uploaded by Donald L Rowe on Mar 12, 2014
Content may be subject to copyright.
May 8, 2007 12:11 WSPC/179-JIN 00143
Journal of Integrative Neuroscience, Vol. 6, No. 1 (2007) 35–74
c
Imperial College Press
Research Report
BRAIN STRUCTURE AND FUNCTION CORRELATES OF
GENERAL AND SOCIAL COGNITION
DONALD L. ROWE
The Br ain Dynamics Center, Westmead Millennium Institute
Westmead Hospital and Western Clinical School, University of Sydney
NSW 2145, Australia
Psychological Medicine, Western Clinical School, University of Sydney
NSW 2145, Australia
donrowe@med.usyd.edu.au
www.brain-dynamics.net
www.brainprofiling.com
NICHOLAS J. COOPER
The Brain Resource International Database and the Brain Resource Company
NSW 2007, Australia
BELINDA J. LIDDELL
The Brain Resource International Database and the Brain Resource Company
NSW 2007, Australia
The Br ain Dynamics Center, Westmead Millennium Institute
Westmead Hospital and Western Clinical School, University of Sydney
NSW 2145, Australia
C. RICHARD CLARK
School of Psychology, Flinders University
The Brain Resource International Database and the Brain Resource Company
NSW 2007, Australia
EVIAN GORDON
The Brain Resource International Database and the Brain Resource Company
Sydney, NSW 2007, Australia
The Br ain Dynamics Center, Westmead Millennium Institute
Westmead Hospital and Western Clinical School, University of Sydney
Sydney, NSW 2145, Australia
Psychological Medicine, Western Clinical School, University of Sydney
NSW 2145, Australia
35
May 8, 2007 12:11 WSPC/179-JIN 00143
36 Rowe et al.
LEANNE M. WILLIAMS
The Br ain Dynamics Center, Westmead Millennium Institute
Westmead Hospital and Western Clinical School, University of Sydney
NSW 2145, Australia
Psychological Medicine, Western Clinical School, University of Sydney
NSW 2145, Australia
Received 29 November 2006
Accepted 28 February 2007
Aims: To examine how general (e.g., memory, attention) and social (emotional and inter-
personal processes) cognition relate to measures of brain function and structure.
Methods: PCA was used to identify general and social cognitive factors from the Brain
Resource International Database in 1,316 subjects. The identified factors were correlated
with each subject’s corresponding brain structure (MRI) and function (EEG/ERP) data.
Results: Seven core cognitive factors were identified for general and three for social. General
cognition was correlated with global grey matter, while social cognition was negatively
correlated with grey matter in fronto-temporal-somatosensory regions. Executive function,
information processing speed and verbal memory performance were correlated with delta-
theta qEEG, while most general cognitive factors negatively correlated with beta qEEG.
Faster information processing speed was correlated with alpha qEEG. Executive function
and information processing speed was correlated with negative-going ERP amplitude and
slower ERP latency at frontal sites, but at posterior sites negative correlations were found.
Discussion: In contrast to general cognition, social cognition is identified by different func-
tional (automated) activity and more localized neural structures. Only general cognition,
requiring more effortful, controlled processing is related to brain function measures, par-
ticularly in frontal cortices.
Integrative Significance: Recording measures from multiple modalities including MRI,
EEG/ERP, social and general cognition within the same subject provides a method of
brain profiling for use in cognitive-neurotherapy and pharmacological studies.
Keywords: Integrative neuroscience; neuropsychology; personalized medicine; neurophysi-
ology; peak performance.
1. Introduction
There is growing consensus that cognitive functions may be considered in terms of
general and social cognitive abilities. General cognitive abilities include functions
such as memory, executive planning and processing speed, whereas social cogni-
tive functions define individuals’ perceptions, emotion and social behavior [1, 2].
Much of the focus on social cognition arose from work in the area of emotional
intelligence [49]. This considers social cognition as a set of abilities distinct from
general intellectual abilities. Yet, it remains unknown whether brain structure and
brain function correlates distinguish general and social cognitive functions within
the same subjects. Moreover, the links between general-social cognition and brain
structure-function will add to the construct validity of both domains of cognition.
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 37
Understanding the neurobiological basis of cognitive and social abilities in terms
of these correlates will be important in establishing objective markers of both norma-
tive and abnormal performance [34, 92, 101]. Within the normative context, knowl-
edge of the correlates of general versus social cognition will provide a framework
for interpreting individual profiles of strengths and weaknesses. Breakdowns in both
general and social cognitive function are characteristic of many psychiatric disor-
ders [40, 92], and information on how these deficits link to underlying brain biology
will help in the development of objective markers for both diagnosis and treat-
ment evaluation [42, 84]. The availability of biological correlates of general and
social cognitive performance will also aid in the development of new pharmacological
compounds, by providing an interpretative framework linking biology to cognitive
endpoints.
Core domains of general cognitive ability include information processing speed,
memory (verbal memory and working memory), attention, and sensory-motor, ver-
bal (such as fluency) and executive planning functions [16, 73]. The concept of a
global general ability (or “g” factor) is also commonly used [50, 61]. Social cogni-
tive functions, on the other hand, include aspects of emotional intelligence (such as
emotional regulation and control and the capacity for effective interpersonal interac-
tion) and emotion processing biases (such as the “negativity bias” towards expecting
and perceiving negative events) [1]. To date, research has focused on identifying the
brain structure and function correlates of the “g” factor, and to a lesser extent
the domains of general cognition [19, 95]. In terms of brain structure, both grey
and white matter volume has been linked to general ability [19, 36]. Similarly, elec-
troencephalographic (EEG) measures of brain function have been correlated with
several measures of general cognitive ability [75, 89, 95]. Given the focus on broad
measures of cognitive ability, the precise relationships between specific domains of
general-social cognitive function and brain structure-function remain undetermined.
Fewer studies have examined relationships between social cognitive performance
and brain function/structure correlates. One such study from our group to date sug-
gests that social cognitive performance (in terms of emotion processing measures)
may be comparatively independent of grey matter volume [102]. Moreover, social
cognitive measures show relationships with brain function over age that suggest dif-
ferent biological mechanisms to general cognition [102]. Other findings have empha-
sized the importance of frontal lobe activity in various aspects of social cognition
such as “theory of mind” [8, 93]. Other recent work has identified the importance
of specific neuronal types such as recently discovered mirror neurons. These are
thought to play a critical role in social cognitive domains including theory of mind
and empathy [11, 29, 68]. These studies provide significant motivation for further
exploration of the relationships between both general and social cognitive perfor-
mance and brain structure and function.
Most previous studies focus on one aspect of general cognition or an underlying
brain structure or function, and do not integrate other important behavioral and
biological measures as discussed above. In addition, no study to our knowledge has
May 8, 2007 12:11 WSPC/179-JIN 00143
38 Rowe et al.
brought together general and social cognitive measures in the same subjects, with
measures of brain structure and function. Yet, such an integrated approach is impor-
tant to determining the neurobiological correlates of general and social cognition.
In this study, we acquired data on general and social cognitive function, together
with measures of brain structure and function within the same subjects using the
standardized methodology of the Brain Resource International Database [31]−[34].
We sought to identify core domains of general and social cognitive function. These
domains were then correlated with brain structure and brain function (psychophys-
iological) measures. We then examined the results to determine whether domains of
social and general cognitive function are distinguished in terms of these measures,
and the different underlying neurobiological mechanisms.
2. Method
2.1. Participants
A total of 1316 healthy subjects (50.5% females; age range: 16 to 60 years; mean age =
32.77, SD =12.95) were selected from the Brain Resource International Database
(BRID; www.brainresource.com). Participants were screened and tested using the
standardized protocols in the BRID [33, 34]. Exclusion criteria included head injury,
history of psychiatric illness [screened using the SPHERE; 44], neurological disor-
ders or other serious medical conditions including a history of psychiatric diagno-
sis, and/or a personal history of drug or alcohol addictions assessed using the World
Health Organization’s AUDIT. All participants were asked to refrain from drinking
caffeine and smoking cigarettes 2 hours before the study session. Informed written
consent was provided in accordance with human research ethical requirements.
2.2. Data acquisition and analysis
2.2.1. Overview
All participants completed a standardized, computerized battery of tests; a bat-
tery of neuropsychological tests to assess general cognitive function (IntegNeuro
TM
),
a battery of psychological tests to assess social cognition, and a suite of psychophys-
iological measures (NeuroMarker
TM
) including brain function in resting conditions
(electroencephalogram, EEG) and task-activation conditions (event-related poten-
tials, ERPs). Structural MRI data was acquired for a subset of 223 subjects. The
protocols for data acquisition in each of these modalities are addressed in turn.
2.2.2. General cognition tests
The tests of general cognitive functions tested by the IntegNeuro
TM
battery are as
follows:
(1) Motor tapping: Tapping with finger of dominant and non-dominant hand to
assess basic sensori-motor function and coordination.
(2) Spot the real word: Assessment of premorbid intelligence and verbal functions.
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 39
(3) Timing — Ability to accurately estimate temporal events up to 12 seconds in
duration.
(4) Span of visual memory: To assess “online” or working memory capacity
(assesses similar constructs to the Corsi Blocks test).
(5) Digit span (forwards and backwards): Assesses working memory capacity for
numerical information.
(6) Verbal interference: Assesses similar constructs to the Stroop, with interference
from color in Part 1, and interference from word reading in Part 2.
(7) Memory recall and learning: To assess similar constructs to the Rey Auditory
Verbal Learning Test (RAVLT) and California Verbal Learning Test (CAVLT).
It includes immediate recall trials (1–4), a recognition memory trial, a recall with
distractor trial, and short-delay (Trial 6) and long-delay (Trial 7) recall trials.
(8) Switching of attention, Parts 1 and 2: To assess similar constructs to the
Trails A and B.
(9) Word generation: Assesses verbal fluency (F, A, S) and animal category fluency.
(10) Sustained attention/working memory: An N-Back test of sustained attention
and low-load working memory updating.
(11) Executive function maze: A maze completion test of planning, error monitoring
and decision-making. Assesses similar constructs to the Austin Maze.
These tests have been shown to have excellent validity in terms of relationships
with conventional paper-and-pencil measures of similar constructs and in terms of
assessment via a different (web-based) platform [73, 91]. They also have sound test-
retest reliability and cross-site consistency [71, 100], and age and sex norms have
been established [16].
2.2.3. Social cognition tests
The psychological tests of social cognition include the following:
a
(1) Brain Resource Inventory of Emotional intelligence Factors (BRIEF): A short
12-item test of emotional intelligence which includes empathy/intuition, social
relationships and self-esteem. The BRIEF has been found to have good validity
against published EQ scales [49].
(2) Depression Anxiety and Stress Scale, 21-item version [50, 59, 60], which has
also been validated in terms of related constructs assessed using Brain Resource
methodology [50].
(3) NEO-FFI, a shortened version of the NEO-Personality Inventory-Revised, to
assess trait aspects of temperament, including openness, conscientiousness,
extraversion, agreeableness and neuroticism [21, 64]. It has been shown that
a
Subsequent versions of the standardized Brain Resource test batteries include tests of emotion recognition
and emotional memory, which have also been considered key aspects of social cognition.
May 8, 2007 12:11 WSPC/179-JIN 00143
40 Rowe et al.
these aspects of temperament relate to but do not account for the variation in
emotional intelligence assessed by the BRIEF [49].
2.2.4. Brain function data
Electroencephalographic (EEG) data for resting quantified EEGs (qEEGs) and
event-related potentials (ERPs) was acquired according to the standardized
methodology used for the Brain Resource International Database. This methodology
has been frequently published in articles in is this current volume and the following
references [3, 17, 40, 48, 100]. A summary of the measures is provided below.
Quantified EEG
Resting conditions consisted of 2 minutes resting eyes closed and 2 minutes resting
eyes open, in which EEG data were acquired. EEG recordings were subsequently
processed via Fast Fourier transform to compute absolute power (qEEG) within the
delta (1.5–3.5 Hz), theta (4–7.5 Hz), alpha (8–13 Hz), and beta (14.5–30 Hz) bands,
and the ratio of theta to beta1 (14.5–20 Hz) power [41, 100]. The data were then log
transformed. Alpha peak amplitude and frequency were also quantified.
Auditory oddball task
The oddball task is designed to elicit brain activation during a fundamental aspect
of information processing: selective attention to infrequent task-relevant “signals”
(target stimuli), with suppression of responses to more frequent task-irrelevant
“noise” (“background” stimuli). A pseudorandom sequence of infrequent target tones
(1000 Hz) is presented among frequent background tones (500 Hz), each for 50 ms
(5 ms rise and fall) at a volume of 75dB, with an inter-stimulus interval (ISI) of
1 second. Subjects are instructed to respond via button press only to target tones
and to ignore background tones. From this task, we quantified:
(1) Task-related qEEG: delta, theta, alpha, beta for target and background stimulus
epochs.
(2) Task-related ERPs: N100, P200, N200 and P300 to target stimuli and N100 and
P200 to background stimuli, scored in terms of peak amplitude and peak latency
using a previously published baseline to peak method [40, 100].
Sustained attention/Working memory task
This task is an N-Back continuous performance task, corresponding to the sustained
attention test used to assess General Cognitive functions. Subjects are presented
with a pseudo random sequence of letters (B, C, D or G) on the center of a computer
screen, with 200 ms duration and ISI of 2.5 s. They are instructed to respond (via
button press) only when the same letter appears twice in a row [20]. The process
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 41
of focal interest is the ongoing updating of working memory that occurs for non-
consecutive letters [15]. Key components elicited by working memory updating are
the N100, P150 ms, N300 and P450, peaking around 100 ms, 150 ms, 300 ms and
400 ms, respectively, and scored for both peak magnitude and peak latency [40, 48].
These ERP components are illustrated in Fig. 1.
Go-NoGo ERP data
The Go-NoGo task is designed to assess inhibition of automated responses, and has
also been considered a test of conflict in pre-potent responses [51, 69]. Subjects are
instructed to respond (via tapping the response bar) as quickly as possible when
the word “PRESS” appears in green (“GO” trials), and to stop immediately it
appears in red (“No-Go” trials). These stimuli are presented for 500 ms, with an
ISI of 1 second. There are three times as many “Go” to “No-Go” trials to elicit
automated “Go” responses and to maximize the need for inhibition on “No-Go”
trials. The “No-Go” trials are of focal interest, and assess inhibitory capacity. ERP
components of interest are the fronto-central N300, related P200 and subsequent
P350. We also scored the N100. The ERP components scored in this task are shown
in Fig. 2.
2.2.5. MRI data
Magnetic resonance images were acquired using a 1.5 Tesla Siemens (Erlangen, Ger-
many) Vision Plus system at Westmead Hospital (n = 117) and a 1.5 Tesla Siemens
(Erlangen, Germany) Sonata at Perrett Imaging, Flinders University, Australia
(n = 106). 3D T1-weighted partitions were acquired in the sagittal plane using a
3D MPRAGE sequence (TR = 9.7 ms; TE = 4 ms; Echo train: 7; Flip Angle = 12
◦
;
TI = 200 ms; NEX = 1). A total of 180 contiguous 1 mm slices were acquired with
Fig. 1. Group average ERP waveforms elicited by the sustained attention/working memory task
showing the respective component labels: N100, P150, N300, and P450.
May 8, 2007 12:11 WSPC/179-JIN 00143
42 Rowe et al.
Fig. 2. Group average ERP waveforms for the Go-NoGo task showing the respective component
labels: N100, P200, N300, and P350.
a 256 × 256 matrix with an in plane resolution of 1 mm × 1 mm resulting in 1mm3
isotropic voxels. The cross-site consistency of images from these scanners has been
established [35].
Images were processed using SPM2 (Welcome Department of Cognitive Neurol-
ogy, London, UK), running on MATLAB 6.5 (MathWorks, Natick, USA), using our
previously published protocol [35]. Images were spatially normalized by transform-
ing each brain into a standardized stereotactic space which approximates Talairach
space. This process was performed, using a custom T1-image template based on our
223 brain images, which matches the ICBM 152 template (Montreal Neurological
Institute). Segmentation into grey matter versus white matter and CSF was based
on a cluster analysis method that accounted for voxel signal intensity, together with
an aprioriexpectation of the anatomical location of the different tissue types. A
correction was made to preserve quantitative tissue volumes following the normal-
ization procedure [4].
We quantified grey matter volume according to specific regions of interest (ROI)
defined using the anatomical boundaries of the neuroanatomical atlas of Tzourio-
Mazoyer and colleagues [97]. ROIs include frontal lobe (encompassing the lateral
portion, defined by the inferior frontal gyrus including opercular and triangular
parts, and lateral sections of superior and middle frontal gyri, the medial portion,
defined by the medial frontal gyrus and medial sections of the superior and middle
frontal gyri as well as the anterior cingulate, the orbital frontal, and supplemen-
tary motor areas), temporal lobe (encompassing the fusiform, superior temporal
middle temporal and, inferior temporal gyri), parietal and somatosensory-related
cortices (encompassing superior parietal, inferior parietal, supramarginal, postcen-
tral and precentral regions and insula) regions, occipital lobe (encompassing cuneus,
lingual gyrus, superior occipital gyrus, and middle occipital gyrus), and subcortical
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 43
limbic structures (hippocampus, amygdala) and basal ganglia structures (thalamus,
putamen and caudate) [97].
We also quantified white matter according to ROIs of interest, including for the
frontal cortex, corpus callosum, occipital cortex, temporal cortex (extracapsule) and
parietal cortex.
2.3. Data analysis
2.3.1. Identification of general and social cognition factors:
Principal component analysis (PCA)
Principal components analysis (PCA) was chosen over exploratory factor analysis
(EFA) to permit derivation of factor coefficients and the reduction of individual test
scores into a core set of composite general and social cognitive factors [12]. Given
that cognitive test scores have previously been found to correlate [46], we used an
obliminal rotation (direct obliminal, with delta = 0). Outliers greater than three
standard deviations from the mean were removed. Outliers and missing values were
then replaced with age-appropriate scores derived using regression onto the age-
predicted score, and accounted for less than 5% of the total data. We use the term
“factor” to refer to the results of PCA to avoid confusion with ERP terminology
(“components”).
Separate coefficient matrices for both social cognitive, general cognitive and “g”
factors were derived from the PCA and used to calculate the factor scores for each
individual. This was computed by applying a regression of the coefficients from
each of the factor matrices to the standardized scores (z-scores) for each measure to
calculate the final 11 factor scores for each individual.
Further details of this PCA analysis (beyond the scope of this journal) are
provided in supplementary materials. http://www.brainnet.org.au/publications/
pubs
abstract.jsp?PubID=939.
2.3.2. Identification of brain structure and brain function correlates of
general and social cognition factors: Correlation analyses
The factor scores generated by the PCA for both general cognition and social cog-
nition were correlated with brain function (EEG, ERPs) and brain structure (MRI)
measures using Pearson’s bivariate correlation analyses. For correlations involv-
ing MRI data, height and age was included as a proxy covariate for head size to
ensure that the results were not simply due to this factor. Given the large number
of possible correlations we used a stringently corrected alpha level of p<0.005,
with a minimum correlation coefficient magnitude of 0.15. In addition, we used
an extent threshold for correlations involving brain function measures, defined
by the presence of significant correlations for at least three contiguous recording
sites.
May 8, 2007 12:11 WSPC/179-JIN 00143
44 Rowe et al.
3. Results
Results are presented in relation to the identification of general and social cognitive
factors and in relation to the correlations between these factors and brain struc-
ture/brain function measures.
3.1. PCA identification of general and social cognition factors
3.1.1. General cognitive factors
PCA was first undertaken with a subgroup of 410 subjects, who had a complete set
of scores an all general cognitive tests. A larger set of subjects (n = 891) were used
as a second subgroup to perform a confirmatory PCA. In this confirmatory PCA, we
focused on confirmation of factor structure for those factors which were not defined
by loadings from the verbal memory and learning test. The PCA for both groups
revealed the same underlying structure (see Table 1).
The following factors were obtained from the analysis with the associated pattern
matrix illustrated in Table 1. PCA for the subgroup of 410 subjects with complete
data revealed seven factors with rounded eigenvalues of 1.0 or greater, and which
together accounted for 60.1% of the variance in test scores. These factors are listed
below and in Table 1;
(1) Information Processing Speed, defined by test scores which rely on speed of
responding for optimal performance, including verbal interference (Parts 1
and 2), switching of attention (Parts 1 and 2) and choice reaction time;
(2) Verbal Memory with loadings from immediate recall, short and long delay recall
and recognition memory scores from the verbal memory and learning test.
(3) Working Memory Capacity, defined by loadings from the forwards and reverse
digit span test scores, reflecting aspects of short term and working memory
capacity;
(4) Vigilance and Sustained Attention, defined by loadings primarily from the sus-
tained attention and working memory N-back test;
(5) Sensorimotor Function, defined by loadings from the motor tapping test, specif-
ically pauses between taps for both dominant and non-dominant hand;
(6) Verbal Processing, defined primarily by performance on word generation scores,
including both verbal fluency (for letters F, A, S) and animal category fluency;
(7) Executive Function (Visuo-Spatial), loaded primarily on tests of complex rea-
soning including the maze executive function test, visual interference (Stroop)
test, and switching of attention (part 2, Trails B) test.
In addition, the first unrotated component extracted from a PCA was included
a general cognitive factor “g” as a measure of overall cognitive performance. This
overall solution accounted for 20.35% of the variance.
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 45
Table 1. PCA Pattern Matrix for Neuropsychological (Cognitive) measures with n = 410 showing 7 factors; information
processing speed, verbal memory, working memory capacity, vigilance and sustained attention, sensorimotor function, verbal
processing and executive function (visual spatial). Bolded loadings (> 0.7) followed by italicized loadings (> 0.5) are the highest
within a column and were used in the interpretation of the factor identify.
Information Working Vig. & Sensori- Executive
Processing Verbal Memory Sustained Motor Verbal Function
Measure Speed Memory Capacity Attention Function Processing (vis. spat.)
VI Word Naming Score 0.68 0.15 0.04 −0.04 0.00 −0.03 −0.05
VI Color Naming Score 0.65 −0.03 −0.10 0.01 −0.05 −0.08 −0.27
SOA: Part 1 Completion Time −0 .65 −0.01 −0.16 0.02 0.00 −0.05 −0.03
SOA: Part 2 Completion Time −0 .59 −0.12 −0.31 −0.16 −0.04 −0.03 0.11
Choice Reaction Time −0.49 0.06 0.08 0.41 −0.04 −0.22 −0.01
VML Recognition Accuracy −0.01 0.76 0.07 0.08 −0.02 −0.34 0.03
VML Rejection Accuracy −0.04 0.44 −0.21 −0.11 −0.04 0.30 −0.03
VML Long Delay Recall 0.12 0.79 −0.06 −0.04 0.01 0.12 0.02
VML Total Recall Trials 1-4 0.06 0.74 0.05 0.00 0.00 0.16 −0.03
Reverse Digits Trials Correct 0.16 0.05 0.67 0.10 −0.10 −0.02 −0.06
Forward Digits Trials Correct 0.10 −0.09 0.71 −0.04 −0.08 0.13 0.03
Sustained Attention WM RT −0.15 0.03 0.19 0.78 0.11 0.13 − 0.03
Sustained Attention WM Errors 0.26 −0.12 −0.38 0.64 −0.02 −0.01 0.11
Dominant Hand Finger Tapping −0.09 −0.01 0.02 0.04 0.77 0.05 −0.02
Nondominant Hand Finger Tapping 0.13 −0.01 −0.15 0.03 0.73 −0.05 0.03
Letter Fluency (FAS) Average −0.01 0.02 0.39 −0.08 0.10 0.68 0.04
Animal Category Fluency 0.08 0.10 −0.04 0.18 −0.11 0.71 −0.11
SVM Trials Correct 0.02 0.09 0.30 −0.19 0.15 −0.15 −0.47
Maze Completion Time −0.22 0.06 0.13 −0.02 0.08 −0.01 0.82
Maze Overruns 0.07 0.02 0.03 −0.04 0.01 −0.10 0.91
∗
Abbreviations include Switching of Attention (SOA), Verbal Memory and Learning (VML), Working Memory (WM), Span
of Visual Memory (SVM), Vigilance (Vig.) and Visual Spatial (vis. spat.).
May 8, 2007 12:11 WSPC/179-JIN 00143
46 Rowe et al.
3.1.2. Social cognitive factors
PCA of the social cognitive tests listed in Table 2 together accounted for 50.6% of
the variance in test scores. The pattern matrix is shown in Table 3. On the basis of
the structure of loadings, we proposed the following factor labels and interpretations:
(1) Negativity. This factor was defined by high loadings from the DASS depression,
anxiety, and stress scores, as well as high NEO-FFI, neuroticism abut low agree-
ableness, indicating a bias towards negative affect and a temperament associated
with the expectation of negative outcomes;
(2) Sociability. This factor was defined by high loadings from the NEO-FFI mea-
sures of extraversion, openness and agreeableness, and the BRIEF measures of
empathy/intuition and social/relationships;
Table 2. Social Cognition tests included in the PCA.
From Test Test Score Description
NEO-FFI Neuroticism scale score Tendency to experience negative emotions
NEO-FFI Extraversion scale score Energy, urgency, sensation seeking
NEO-FFI Openness scale score Openness to new experiences
NEO-FFI Agreeableness scale score Tendency to be compassionate, cooperative
NEO-FFI Conscientiousness scale score Tendency for self-discipline, striving for achievement
DASS Stress score Difficulty relaxing, agitation, irritability, impatience
DASS Anxiety score Anxious affect, autonomic arousal, skeletal muscle
effects
DASS Depression score Anhedonia, dysphoria, hopelessness, inertia
BRIEF Empathy/Intuition score Empathetic and intuitive ability
BRIEF Social/Relationships score Sociability and relationship skills
BRIEF Self Esteem score Confidence and positive attitude to self
Table 3. PCA pattern matrix for the social cognitive measures showing 3 factors; Negativity,
Sociability and Emotional Control. Bolded loadings (> 0.7) followed by italicized loadings (> 0.5)
are the highest within a column and were used in the interpretation of the factor identity.
Measure Negativity Sociability Emotional Control
Neuroticism 0.68 −0.09 −0.29
Extraversion −0.19 0.58 0.27
Openness −0.03 0.60 −0 .50
Agreeableness −0.32 0.44 −0.11
Conscientiousness −0.17 0.04 0.60
Stress 0.85 0.14 0.03
Anxiety 0.78 0.09 0.10
Depression 0.78 −0.03 −0.14
Empathy/Intuition 0.23 0.71 0.07
Social/Relationships 0.18 0.68 0.17
Self Esteem 0.04 0.13 0.73
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 47
Table 4. Correlations between social cognitive factors and frontal grey matter for left (L) and
right (R) hemispheres. Correlations significant between p<0.005 and p<0.0005 are indicated by
yellow shading.
Measure Negativity Sociability Emotional Control
Information Processing Speed 0.02 0.09 −0 .22
Verbal Memory −0.01 0.16 −0 .17
Working Memory Capacity 0.11 0.03 −0 .15
Vig. & Sustained Attention 0.04 −0.02 −0.02
Sensori-motor Function 0.10 −0.00 −0.06
Verbal Processing 0.08 0.11 −0 .19
Executive Function (vis. spat.) 0.00 0.03 0.12
“g” 0.06 0.06 −0 .24
(3) Emotional Control. This factor was defined by lower levels of NEO-FFI open-
ness, but higher levels of NEO-FFI conscientiousness and BRIEF self-esteem
suggesting higher self-control and emotion regulation.
3.2. General cognition — social cognition
The correlations between the factors of general and social cognition are listed in
Table 4. Significant negative correlations were found between emotional control and
information processing, verbal memory, working memory capacity, verbal process-
ing and “g”. Sociability was also positively correlated with verbal memory. Cor-
relations were significant at r =0.15 (t[1, 1178] = 5.21, p<0.0001) to r =0.24
(t[1, 1178] = 8.49, p<0.0001). The remaining correlations were generally very
small or non-significant.
3.3. Social cognition — brain structure (MRI)
The correlations between social cognition and cortical and subcortical grey matter
are presented in Tables 5–8. The significant correlations (p<0.005), as highlighted,
range from −0.15 (t[1, 221] = −2.26, p<0.005) to −0.2(t[1, 221] = −3.03, p<
0.002).
3.4. General cognition — brain structure (MRI)
The correlations (r) between neuropsychological/cognitive factors and various struc-
tures in the cortex are illustrated in Tables 9–12. The significant correlations
(p<0.005), as highlighted, range from 0.15 (t[1, 221] = 2.26, p<0.005) to 0.41
(t[1, 221] = 6.68, p<0.0001).
3.5. Social cognition — brain function (EEG, ERPs)
There were no significant correlates between the measured psychophysiological mea-
sures and the social cognitive factors.
May 8, 2007 12:11 WSPC/179-JIN 00143
48 Rowe et al.
Table 5. Correlations between social cognitive factors and frontal grey matter for left (L)
and right (R) hemispheres. Correlations significant between p<0.005 and p<0.0005 are
indicated by grey shading.
Frontal cortex Hemisphere Negativity Sociability
Emotional
Control
Superior
L
-0.00 -0.10 -0.13
R
0.02 -0.12 -0.14
Superior orbital
L
0.01 -0.11 -0.14
R
0.05 -0.09 -0.17
Middle
L
-0.01 -0.09 -0.15
R
0.02 -0.11 -0.16
Middle orbital
L
0.00 -0.12 -0.12
R
0.04 -0.09 -0.16
Inferior opercular
L
-0.02 -0.11 -0.12
R
0.01 -0.08 -0.15
Inferior triangular
L
0.01 -0.13 -0.15
R
0.02 -0.09 -0.13
Inferior orbital
L
0.01 -0.14 -0.12
R
0.04 -0.11 -0.14
Supplementary motor area
L
0.02 -0.10 -0.18
R
0.02 -0.13 -0.18
Superior medial
L
-0.00 0.07 -0.14
R
0.05 -0.10 -0.14
Anterior cingulate
L
-0.00 -0.11 -0.19
R
0.03 -0.11 -0.20
Table 6. Correlations between social cognitive factors and the hippocampus, amygdala,
occipital cortex and temporal cortex for left (L) and right (R) hemispheres. Correlations
significant between p<0.005 and p<0.0005 are indicated by grey shading.
Cortical or Subcortical
region
Hemisphere Negativity Sociability
Emotional
Control
Hippocampus
L
0.02 -0.10 -0.07
R
0.04 -0.10 -0.08
Amygdala
R
-0.00 -0.12 -0.09
L
0.01 -0.15 -0.08
Occipital cortex
R
Superior occipital
L
0.02 -0.15 -0.09
R
0.02 -0.15 -0.09
Middle occipital
L
0.04 -0.14 -0.19
R
-0.00 -0.13 -0.17
Inferior occipital
L
0.03 -0.10 -0.16
R
0.01 -0.04 -0.17
Temporal cortex
Fusiform gyrus
L
0.05 -0.12 -0.11
R
0.03 -0.12 -0.13
Superior temporal
L
0.02 -0.05 -0.13
R
0.01 -0.10 -0.14
Middle temporal
L
0.01 -0.10 -0.15
R
0.03 -0.11 -0.18
Inferior temporal
L
0.02 -0.12 -0.12
R 0.03 -0.11 -0.15
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 49
Table 7. Correlations between social cognitive factors and the somatosensory cortices and
the basal ganglia and thalamus for left (L) and right (R) hemispheres. Correlations signifi-
cant between p<0.005 and p<0.0005 are indicated by grey shading.
Somatosensory cortices Hemisphere Negativity Sociability
Emotional
Control
Insula
L
-0.00 -0.11 -0.09
R
0.02 -0.10 -0.12
Precentral gyrus
L
-0.00 -0.11 -0.17
R
0.00 -0.12 -0.17
Postcentral gyrus
L
0.02 -0.11 -0.16
R
0.03 -0.11 -0.17
Superior parietal gyri
L
0.01 -0.12 -0.15
R
0.02 -0.12 -0.13
Inferior parietal gyri
L
0.00 -0.09 -0.17
R
0.00 -0.10 -0.17
Supramarginal gyri
L
0.05 -0.10 -0.15
R
0.03 -0.12 -0.16
Basal ganglia
Caudate
L
-0.01 -0.09 -0.13
R
0.03 -0.08 -0.13
Putamen
L
0.05 -0.09 -0.07
R
0.04 -0.11 -0.08
Thalamus
L
-0.02 -0.12 -0.12
R
0.00 -0.12 -0.13
Table 8. Correlations between social cognitive factors and white matter by left (L) and
right (R) hemispheres. Correlations significant between p<0.005 and p<0.0005 are
indicated by grey shading.
White matter volume Hemisphere Negativity Sociability
Emotional
Control
Frontal cortex
L
-0.04 -0.14 -0.08
R
-0.06 -0.15 -0.07
Corpus collosum
L
-0.04 -0.15 -0.04
R
-0.08 -0.14 -0.02
Occipital cortex
L
-0.06
-0.19 -0.01
R
-0.05 -0.18 -0.01
Temporal cortex
(extracapsule)
L
-0.04 -0.12 -0.03
R -0.03 -0.13 -0.04
Parietal cortex
L
-0.02 -0.16 -0.03
R
-0.04 -0.17 -0.03
3.6. General cognition — brain function (EEG, ERPs)
The correlates between neuropsychological/cognitive factors and psychophysiolog-
ical measures are illustrated in Tables 13–16. The significant (p<0.005) corre-
lations considered were in the range of 0.15 (t[1, 410] = 3.06, p<0.005) to 0.34
(t[1, 410] = 7.3, p<0.0001). The resting brain function qEEG correlates of the
general cognitive factors are presented in Table 13, and the activation-task related
qEEG correlations in Table 14. Correlations between activation-task ERPs and gen-
eral cognition factors are presented in Tables 15–17. Note that P300 latency for odd-
May 8, 2007 12:11 WSPC/179-JIN 00143
50 Rowe et al.
Table 9. Correlations between general cognitive factors and frontal grey matter for left (L) and
right (R) hemispheres. Correlations significant between p<0.005 and p<0.0001 are indicated by
grey shading, and p<0.0001 are indicated by dark grey shading.
Frontal Cortex
Hemi-
sphere
"g"
Info. Proc.
Speed
Verbal
Memory
WM
Capacity
Vig. &
Sustained
Attention
Sensori-
motor
Function
Verbal
Proc.
Executive
Function
Superior
L
0.37 0.33 0.21 0.10 0.11 -0.09 0.11 0.38
R
0.36 0.31 0.21 0.11 0.12 -0.06 0.11 0.39
Superior orbital
L
0.33 0.28 0.18 0.09 0.09 -0.06 0.11 0.36
R
0.32 0.27 0.18 0.09 0.11 -0.07 0.07 0.36
Middle
L
0.35 0.30 0.20 0.08 0.08 -0.09 0.11 0.37
R
0.34 0.29 0.21 0.08 0.08 -0.09 0.10 0.36
Middle orbital
L
0.32 0.27 0.19 0.06 0.09 -0.05 0.11 0.35
R
0.33 0.27 0.21 0.07 0.11 -0.07 0.10 0.37
Inferior opercular
L
0.37 0.31 0.22 0.11 0.08 -0.08 0.12 0.37
R
0.36 0.30 0.26 0.10 0.08 -0.12 0.12 0.37
Inferior triangular
L
0.33 0.28 0.22 0.10 0.13 0.01 0.07 0.34
R
0.37 0.32 0.22 0.08 0.08 -0.12 0.11 0.39
Inferior orbital
L
0.31 0.26 0.16 0.08 0.14 -0.05 0.10 0.34
R
0.30 0.26 0.17 0.04 0.12 -0.10 0.08 0.35
Rolandic opercular
L
0.37 0.31 0.17 0.11 0.10 -0.11 0.14 0.38
R
0.37 0.32 0.19 0.13 0.14 -0.11 0.13 0.37
Supplementary motor
area
L
0.34 0.27 0.22 0.10 0.11 -0.06 0.12 0.36
R
0.32 0.25 0.20 0.10 0.13 -0.06 0.14 0.35
Superior medial
L
0.06 0.05 0.07 -0.01 0.00 0.05 0.00 0.10
R
0.41 0.38 0.23 0.09 0.13 -0.10 0.11 0.40
Anterior cingulate L 0.28 0.22 0.12 0.06 0.04 -0.10 0.12 0.33
R
0.31 0.25 0.15 0.08 0.05 -0.08 0.12 0.34
Table 10. Correlations between general cognitive factors and the hippocampus, amygdala, occipital
cortex and temporal cortex for left (L) and right (R) hemispheres. Correlations significant between
p<0.005 and p<0.0001 are indicated by grey shading, and p<0.0001 are indicated by dark grey
shading.
Cortical or
Subcortical region
Hemi-
sphere
"g"
Info.
Proc.
Speed
Verbal
Memory
WM
Capacity
Vig. &
Sustained
Attention
Sensori-
motor
Function
Verbal
Proc.
Executive
Function
Hippocampus
L
0.23 0.18 0.09 0.09 0.15 -0.09 0.10 0.27
R
0.22 0.17 0.10 0.08 0.16 -0.05 0.07 0.27
Amygdala
L
0.21 0.16 0.08 0.09 0.17 -0.10 0.10 0.25
R
0.19 0.13 0.06 0.06 0.14 -0.09 0.06 0.26
Occipital cortex
Superior occipital
L
0.34 0.33 0.15 0.08 0.16 -0.13 0.10 0.33
R
0.37 0.34 0.17 0.13 0.15 -0.13 0.10 0.34
Middle occipital
L
0.30 0.23 0.18 0.09 0.13 -0.08 0.12 0.34
R
0.27 0.22 0.19 0.13 0.13 -0.08 0.09 0.28
Inferior occipital
L
0.28 0.21 0.17 0.08 0.13 -0.06 0.12 0.33
R
0.30 0.24 0.21 0.12 0.16 -0.07 0.14 0.30
Temporal cortex
Fusiform gyrus
L
0.24 0.18 0.11 0.07 0.14 -0.07 0.08 0.30
R
0.25 0.19 0.13 0.08 0.14 -0.09 0.09 0.29
Superior temporal
L
0.36 0.31 0.18 0.09 0.12 -0.10 0.11 0.37
R
0.33 0.27 0.15 0.11 0.13 -0.11 0.12 0.36
Middle temporal
L
0.32 0.27 0.18 0.09 0.12 -0.07 0.09 0.35
R
0.31 0.25 0.16 0.09 0.13 -0.08 0.10 0.34
Inferior temporal L 0.26 0.19 0.14 0.10 0.16 -0.04 0.08 0.31
R
0.24 0.17 0.15 0.10 0.15 -0.04 0.09 0.30
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 51
Table 11. Correlations between general cognitive factors and the somatosensory cortices and the
basal ganglia and thalamus for left (L) and right (R) hemispheres. Correlations significant between
p<0.005 and p<0.0001 are indicated by grey shading, and p<0.0001 are indicated by dark grey
shading.
Somatosensory
Cortices
Hemi-
sphere
"g"
Info.
Proc.
Speed
Verbal
Memory
WM
Capacity
Vig. &
Sustaine
d
Attention
Sensori-
motor
Function
Verbal
Proc.
Executive
Function
Insula
L
0.35 0.29 0.17 0.13 0.15 -0.09 0.10 0.36
R
0.34 0.29 0.19 0.10 0.14 -0.09 0.09 0.36
Precentral gyrus
L
0.33 0.26 0.19 0.10 0.09 -0.07 0.13 0.36
R
0.31 0.25 0.21 0.09 0.10 -0.08 0.10 0.34
Postcentral gyrus
L
0.36 0.29 0.19 0.11 0.10 -0.05 0.11 0.38
R
0.32 0.26 0.21 0.10 0.10 -0.07 0.08 0.33
Superior parietal gyri
L
0.34 0.33 0.21 0.09 0.14 -0.05 0.06 0.33
R
0.29 0.28 0.18 0.06 0.15 -0.08 0.05 0.30
Inferior parietal gyri
L
0.36 0.31 0.22 0.10 0.11 -0.05 0.09 0.36
R
0.33 0.30 0.23 0.07 0.09 -0.04 0.05 0.35
Supramarginal gyri
L
0.35 0.30 0.19 0.12 0.14 -0.09 0.08 0.36
R
0.36 0.31 0.23 0.12 0.12 -0.09 0.09 0.37
Basal ganglia
Caudate
L
0.24 0.18 0.09 0.11 0.10 -0.08 0.07 0.29
R
0.15 0.09 0.05 0.07 0.10 -0.09 0.05 0.22
Putamen
L
0.31 0.26 0.15 0.13 0.15 -0.09 0.08 0.32
R
0.24 0.19 0.11 0.07 0.13 -0.09 0.05 0.28
Thalamus
L
0.18 0.13 0.05 0.04 0.09 -0.16 0.09 0.24
R
0.15 0.11 0.06 0.02 0.09 -0.14 0.08 0.20
Table 12. Correlations between general cognitive factors and white matter for left (L) and right (R)
hemispheres. Correlations significant between p<0.005 and p<0.0001 are indicated by grey
shading.
White Matter
Volume
Hemi-
sphere
"g"
Info.
Proc.
Speed
Verbal
Memory
WM
Capacity
Vig. &
Sustaine
d
Attention
Sensori-
motor
Function
Verbal
Proc.
Executive
Function
Frontal cortex
L
0.03 0.00 -0.08 0.04 0.03 0.07 0.04 0.08
R
0.06 0.03 -0.08 0.05 0.03 0.06 0.06 0.10
Corpus collosum
L
-0.09 -0.10 -0.30 -0.08 -0.03 -0.13 0.01 0.00
R
-0.04 -0.07 -0.27 -0.04 -0.02 -0.17 0.04 0.04
Occipital cortex
L
0.08 0.01 -0.21 0.05 0.00 -0.16 0.11 0.14
R
0.01 -0.05 -0.17 -0.01 -0.02 -0.11 0.07 0.10
Internal capsule
L
0.02 -0.03 -0.20 0.00 -0.09 -0.10 0.05 0.11
R
0.10 0.02 -0.17 0.09 -0.06 -0.13 0.10 0.17
Temporal cortex
(extracapsule)
L
0.03 -0.02 -0.18 -0.02 0.01 -0.11 0.10 0.11
R
0.07 0.00 -0.19 0.02 0.03 -0.10 0.13 0.12
Parietal cortex
L
0.04 0.02 -0.18 -0.03 0.00 -0.13 0.05 0.09
R
0.04 0.01 -0.18 -0.01 -0.02 -0.14 0.04 0.09
ball targets was also significantly correlated with the information processing speed
factor along midline sites (−0.15 to −0.17), but is not illustrated in figures.
4. Discussion
To our knowledge, this is the first study to obtain measurements from multi-
ple modalities that include measures of brain structure (MRI), brain function
(EEG/ERP), and general (e.g., memory, attention) and social (emotional and inter-
personal processes) cognition, from within the same subject. Through integrative
May 8, 2007 12:11 WSPC/179-JIN 00143
52 Rowe et al.
Table 13. The most significant correlations between general cognitive factors and resting eyes
closed (EC: left within each cell) and eyes open (EO: right) qEEG. Correlations significant at
p<0.0001 are indicated by red and blue shading. Range extends from r =< −0.2 (blue) to
r => 0.2 (red). Note that orange to red coloring means larger amplitude is correlated with better
performance, while blue coloring means smaller amplitude is correlated with better performance.
Delta Theta Alpha Beta Theta/Beta1
Measure
EC - EO EC - EO EC - EO EC - EO EC - EO
Information
Processing
Speed
Verbal
Memory
Executive
Function
(Visual
Spatial)
Working
Memory
Capacity
Sensori-motor
Function
Table 14. The most significant correlations between general cognitive factors and phasic auditory
oddball target and background (Bg) qEEG. Correlations significant at p<0.0001 are indicated by
red and blue shading. Range extends from r =< −0.2 (blue) to r => 0.2 (red). Note that orange to
red coloring means larger amplitude or faster peak frequency is correlated with better performance,
while blue coloring means smaller amplitude is correlated with better performance.
Measure
Delta
Oddball
Target
Delta
Oddball
Bg
Theta
Oddball
Target
Theta
Oddball
Bg
Beta
Oddball
Target
Beta
Oddball
Bg
Alpha (pk)
Oddball Bg
Alpha (pk
freq)
Oddball Bg
Information
Processing
Speed
Verbal
Memory
Executive
Function
(Visual
Spatial)
Working
Memory
Capacity
Sensori-motor
Function
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 53
Table 15. The most significant correlations between general cognitive factors and auditory oddball
target and background ERP components for amplitude and latency. Correlations significant at
p<0.0001 are indicated by red and blue shading. Range extends from r =< −0.2 (blue) to r => 0.2
(red). Note that orange to red coloring means larger ERP amplitude for either negative-going or
positive-going components is correlated with better performance, while blue coloring means smaller
amplitude is correlated with better performance. Similarly, orange to red coloring for ERP latency
means earlier or faster ERP component is correlated with better cognitive performance, whereas
blue coloring means later or slower ERP component is correlated with better performance.
0ddball Target stimuli Oddball Background stimuli
Measure
Oddball
P200
amplitude
Oddball
N200
amplitude
Oddball
P300
amplitude
Oddball
N200
latency
Oddball P200
amplitude
Oddball
N100
latency
Information
Processing
Speed
Verbal Memory
Executive
Function (Visual
Spatial)
Working Memory
Capacity
Table 16. The most significant correlates between General Cognitive factors and the N-Back
(background) component of sustained attention and working Memory (WM) ERP test, represent-
ing the updating of the letter sequence in working memory. Correlations significant at p<0.0001
are indicated by red and blue shading. Range extends from r =< −0.2 (blue) to r => 0.2(red).
Note that orange to red coloring means larger ERP amplitude for either negative-going or posi-
tive-going components is correlated with better performance, while blue coloring means smaller
amplitude is correlated with better performance. Similarly, orange to red coloring for ERP latency
means earlier or faster ERP component is correlated with better cognitive performance, whereas
blue coloring means later or slower ERP component is correlated with better performance.
Measure WM P150
amplitude
WM N300
amplitude
WM P450
amplitude
WM N100
latency
WM P450
latency
Information
Processing
Speed
Verbal Memory
Executive
Function (Visual
Spatial)
Vigilance &
Sustained
Attention
a
a
This factor has defining loadings from the N-back task used to elicit sustained attention/working memory ERPs.
May 8, 2007 12:11 WSPC/179-JIN 00143
54 Rowe et al.
Table 17. The most significant correlates between cognitive factors and Go-NoGo background
(NoGo) ERP components. Correlations significant at p<0.0001 are indicated by red and blue
shading. Range extends from r =< −0.2 (blue) to r => 0.2 (red). Note that orange to red coloring
means larger ERP amplitude for either negative-going or positive-going components is correlated
with better performance, while blue coloring means smaller amplitude is correlated with better per-
formance. Similarly, orange to red coloring for ERP latency means earlier or faster ERP component
is correlated with better cognitive performance, whereas blue coloring means later or slower ERP
component is correlated with better performance.
Measure
NoGo N100
amplitude
NoGo N300
amplitude
NoGo P200
amplitude
NoGo P350
amplitude
NoGo N100
latency
NoGo P350
latency
Information
Processing Speed
Verbal Memory
Executive Function
(Visual Spatial)
Vigilance &
Sustained
Attention
analysis of this large dataset, we first identified core factors of general and social
cognitive function, and then examined whether these two aspects of cognition are
distinguished by different brain structure and brain function correlates. It was shown
that general cognition comprises seven core factors and social cognition, three core
factors and that these general and social cognitive factors are largely independent
of each other. In terms of correlates, general and social cognitive factors were dis-
tinguished by opposing relationships with cortical and subcortical grey matter. In
addition, only general cognitive factors were related to task-related brain function.
Taken together, these findings suggest that social cognitive functions may rely on
largely automatic processes which do not improve with grey matter volume, and
are independent of effortful task-related brain activity, while general cognitive func-
tions may involve controlled processes which depend on grey matter and effortful
brain activity. These results will provide a valuable platform for identifying profiles
of strengths and weaknesses across general and social cognitive abilities within the
normative population, and for identifying more comprehensive cognitive markers of
psychiatric disorder.
In summary, the following key findings were identified:
(1) The principal component analysis revealed seven core general cognitive factors:
information processing speed, verbal memory, working memory capacity, vigi-
lance/sustained attention, sensori-motor function, verbal processing, and exec-
utive function (visual spatial).
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 55
(2) In parallel, three independent social cognitive factors were identified: negativity,
sociability and emotional control.
(3) Social cognitive factors showed small or no correlation with general cognitive fac-
tors with the exception of emotional control and sociability with verbal memory.
(4) In terms of brain structure, general and social cognitive factors showed opposing
patterns of relationship with grey matter. While general cognitive factors showed
moderate and positive (up to 0.41) correlations with distributed cortical and
subcortical grey matter, social cognitive factors showed more subtle and negative
(up to −0.20) correlations with more focal regions of grey matter.
(5) In terms of brain structure, only general cognitive factors correlated with mea-
sures of EEG and ERPs. There were no correlations with these measures for
social cognitive factors.
(6) Better general cognitive performance was associated with increases in resting EEG
slow-wave activity (delta, theta), but reductions in faster wave (beta) activity.
(7) Better performance on these general cognitive factors was also generally
associated with greater ERP amplitude and slower ERP latency during acti-
vation tasks. These effects tended to follow the topography of the ERP compo-
nents (predominantly frontal effects for negative-going components, and parietal
effects for positive-going components).
(8) These brain function correlates of general cognition were most apparent for
information processing speed, verbal memory and executive function factors
which feasibly reflect more effortful processing.
These findings are considered in more detail in the following sections. Rather than
provide an exhaustive account of these numerous relationships, we have highlighted
those correlates which reflect distinctive profiles for general and social cognition and
which point the way for future investigations.
4.1. General cognition and social cognition
Correlations between general and social cognitive factors were predominantly non-
significant or very small. The exception was small negative correlations between
emotional control and information processing speed, verbal memory, working mem-
ory capacity, verbal processing, executive function and “g”, and between sociability
and verbal memory (see Table 4). The findings parallel the social cognition and
brain structure correlates listed in Table 5 and discussed in Sec. 4.4. This finding is
very logical given the general cognition factors are significantly correlated with brain
structure (see Sec. 4.2), and therefore the two measures should show similar correla-
tions with social cognition, despite being modality independent. As discussed further
below, the findings suggest that increased general cognitive performance occurs at
the expense of emotional control. In contrast, it is sociability which shows a positive
relationship with verbal memory performance. This suggests that individuals with
May 8, 2007 12:11 WSPC/179-JIN 00143
56 Rowe et al.
better verbal memory and learning ability also show a social cognitive profile that
suggests these individuals are better skilled at social interaction.
The negativity factor being a composite of depression, stress, anxiety and neu-
roticism variables, is a measure that is highly prevalent in a range of psychiatric
disorders. The fact that it shows little or no correlates with the general cognitive
factors suggests that this factor of social cognition is underpin by neurobiological
mechanisms that are functionally very independent to those mechanisms underlying
general cognition. Clinically, this finding can be very helpful given clinicians are often
required to differentiate the organic or psychological source of impairment in general
cognition. These results suggest that depression, other than in the most severe cases,
is unlikely to be the source of significant impairment in general cognitive function.
Our research in social cognitive performance across the life span shows interesting
age-related changes, such that while areas of social cognitive performance improve
with age, other areas, such as information processing speed decline [102]. This work is
consistent with the current finding of a negative relationship between social cognitive
and general cognitive performance on some factors. These findings suggest that there
may be compensatory changes occurring between independent domains of cognition.
As certain aspects of general cognition decline, there is a relative improvement in
specific areas of social cognition that leads to more successful adaptive behavior in
the face of aging. These hypotheses warrant further investigation in future studies
and may have implications for the successful treatment of degenerative conditions,
or identification of factors such as negativity that may reflect poor adaptation.
4.2. Brain structure and general cognition
The correlates between structural measures and the general cognitive factors
revealed that much of the associated variance could be accounted for by the exec-
utive function factor followed by the information processing speed, verbal memory
and to a lesser extent the sustained attention factor. Working memory capacity,
sensorimotor function, and verbal processing factors were found to show little or no
association with structural measures (see Tables 9–12).
The composite factor of general cognitive performance was significantly corre-
lated with a distributed network of grey matter, but not white matter. This is
consistent with prior studies that show strong correlations between “g” and a dis-
tributed range of grey matter structures in the cortex [19, 36]. The findings in this
study indicate that cognitive performance in normals is predominantly associated
with grey matter volume rather than the underlying white matter connectivity.
Although, the executive function and sensorimotor factors did show small positive
correlations with white matter regions (see Table 12), and white matter hyperin-
tensities has been associated with reduced cognitive performance in a subset of the
elderly [72]. In contrast, the verbal memory factor displayed significant negative
correlations with white matter volume suggesting that while specific increases cor-
tical white matter volume is associated with increased performance in the area of
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 57
executive function and sensorimotor activity, it is associated with decreased verbal
memory and learning performance. This is consistent with our prior findings that
show an age-related decline in white matter is associated with reduced memory
performance [9]. The dissociation between executive function and verbal memory
in this study may reflect the auditory nature of the verbal memory and learning
task, and the concomitant absence of active visual or sensorimotor processing, as
in the executive function factor. It may also reflect the fact that we are hard wired
for learning language through imitation such that dedicated and modular structures
exist for such functions [6, 14, 26, 77]. In such nuclei, the high level of localized func-
tional specificity may occur at the expense of long range corticocortical connectivity
thereby reducing the associations with more distant nuclei [86].
For the verbal memory factor, there were significant positive correlations with
the frontal, occipital, temporal and parietal cortices (see Tables 9–11). Interestingly,
there was no significant correlation observed between verbal memory and the hip-
pocampus (see Table 10). In contrast, the executive function, information processing
and sustained attention factors were significantly correlated with the hippocampus.
The sustained attention task, like the executive function and information process-
ing factors, requires an active updating of working memory over a sustained time
frame. These findings would suggest that hippocampal volume is more associated
with working memory ability during sustained mental effort or information process-
ing, rather than during more verbal memory and learning processes when basic short
term memory is accessed.
4.2.1. Brain structure — executive function
The executive function (visual spatial) factor, showed a similar pattern of positive
correlations with volume of brain structure, as observed for the general cognitive
factor. However, the structural correlates with executive function were of greater
magnitude in many areas suggesting executive function may be a more sensitive
measure of neural reserve. In this study, “g” is a composite of all cognitive factors,
some of which have little or no correlation, which may partially explain the relatively
smaller but still significant correlates in comparison with the executive function
factor.
The most significant correlates with executive function included most areas of the
frontal and temporal cortices, sensorimotor strip, and parietal and occipital cortices.
This is not surprising given the strong executive function, visual and sensorimotor
processes required in the tasks that comprise the executive function factor, and
the traditional relationships between executive function and the prefrontal cortex,
cerebellum and motor control, and visual function and the occipital and parietal
cortices.
Of interest, was the finding that executive function did not strongly correlate
with white matter volume with the exception of a small correlation with the right
internal capsule (see Table 12). This correlate may be related to the visuo-spatial
May 8, 2007 12:11 WSPC/179-JIN 00143
58 Rowe et al.
nature of the tasks and the traditional association of this domain of cognitive pro-
cessing with the right cerebral hemisphere [47]. This could signify that right hemi-
spheric tasks involved in visuo-spatial processing rely on the temporal synchrony
of important long range connections between spatially distant nuclei specialized for
visual processing versus spatial mapping [37, 62, 70].
4.2.2. Brain structure — information processing
The information processing speed component displayed similar correlates with mea-
sures of brain structure, as observed for the executive function factor, with the excep-
tion of no significant correlates with white matter and further, overall smaller and
less spatially diffuse correlates. The most distinctive differences appear to reflect less
involvement of the thalamus and cingulate. Otherwise, the information processing
speed factor appears to be associated with very similar areas of grey matter volume
as the executive function factor. This may be expected, given that this factor also
comprises tasks that tap heavily into visual spatial processing and sensori-motor
abilities.
4.3. Brain structure — social cognition
Of particular interest were the negative correlations between brain structure and
social cognition, but positive correlations with general cognition. The social cognitive
factors were also found to correlate which a much more reduced set of cortical
regions and there were no significant correlations observed between the general and
social cognitive factors. These findings suggest that while increased grey matter is
associated with increased general cognitive performance this occurs at the expense of
social cognitive performance.
b
In addition, it appears that increased performance on
social cognitive factors is underpinned by very different functional behavior in similar
but more localized neural structures. This is consistent with our recent findings that
suggest that social cognition compared to general cognition is underpin by opposing
biological mechanisms [102].
The distinct feature of the social cognitive factors is that it reflects more auto-
mated an involuntary processes of emotion and personality [13, 76, 99], whereas the
general cognitive factors reflect more effortful, controlled and intentional processing
[90, 96]. Therefore, it is not surprising that the social cognitive factors displayed
such disparate correlations to the general cognitive factors. Of relevance here is the
psychophysiological (brain function) findings, which did not show any correlations
with the social cognitive factors. This suggests that social cognition is not signif-
icantly association with resting EEG conditions or the phasic event-related condi-
tions that reflect more effortful and intentional processing. It appears that social
b
Note that this effect remains significant when controlling for age and sex, although the magnitude of the
correlations do decrease (up to 0.23) for the cognitive factors, but nevertheless, remain significant.
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 59
cognitive function would be better evaluated via functional brain measures, particu-
larly those which use activation tasks related to social cognition. Such findings have
been observed in functional event-related and structural paradigms that involve the
unconscious and automated processing of facial emotions [56, 57, 99, 101].
4.4. Brain structure — sociability and emotional control
4.4.1. Brain structure — sociability
The sociability factor was found to be negatively correlated with a distinct set
of neural structures compared with the emotional control factor, suggesting these
two subjective social measures tap into independent neuronal structures (see
Tables 5–8). Significant negative correlates were found between sociability scores,
and parietal/occipital white matter and grey matter of the occipital cortex. The
sociability factor loads positively on empathy/intuition, social/relationships, open-
ness, extraversion and agreeableness measures. This suggests that sociability is
affected by the occipital cortex, and parietal and occipital white matter, such that
larger volumes are associated with decreased sociability. This relationship is consis-
tent with a prior study where authors found a negative correlation between extraver-
sion and the cuneus in the occipital cortex [53]. These authors hypothesized that
personality predicts the brains response to cognitive demand. This would suggest
that more introverted tendencies or less sociability is generally associated with better
general cognition. Presumably, those with low sociability may have greater cognitive
reserve and perform better overall than those with high sociability. This was found
to be the case in this study with significant positive correlations between regions of
the visual system and the general cognitive factors (see Sec. 3.4).
4.4.2. Brain structure — emotional control
The emotional control factor was correlated with a more diverse range of cortical
structures compared to the sociability factor. These included areas of the parietal
and frontal lobes including the pre- and postcentral gyrus, bilateral middle frontal
and right superior frontal gyrus, supplementary motor area, anterior cingulate, mid-
dle and inferior occipital gyri, inferior parietal gyri, supramarginal gyri, and middle
temporal gyri. Recall, that the emotional control factor is associated with self dis-
cipline and achievement, self esteem and confidence and closedness to new ideas.
This is consistent with suggestions that the middle frontal and superior frontal gyri
extending medially to the anterior cingulate, and the inferior parietal cortex, are
associated with self-awareness, and findings that show these structures are acti-
vated by introspective tasks [30]. Several studies investigating different aspects of
emotional processing have also observed activation in similar regions of the pre-
frontal and parietal lobes [25, 39, 74]. For example, it has been suggested that the
medial prefrontal cortex has a general role in emotional processing, and the anterior
May 8, 2007 12:11 WSPC/179-JIN 00143
60 Rowe et al.
cingulate in emotional recall and imagery [74]. These regions may provide the crit-
ical resources in the top-down control processes necessary for conscious emotional
control, in response to incoming emotional information [103]. Certainly, increased
grey matter, as shown in this study, does correlate with increased cognitive capacity.
Presumably, while this may relate to greater introspection and openness, it appears
to be at the expense of emotional stability.
4.5. Brain function (psychophysiology) — general cognition
Compared to the structural findings, the psychophysiological measures displayed
more significant and specific correlates with a range of general cognitive factors,
but not the social cognitive factors. In general, the correlations were smaller than
observed with the MRI measures, but were still significant (up to 0.34; see Sec. 3.6).
Note that the sample size was larger for the psychophysiological data suggest-
ing that despite the smaller correlations, the findings represent genuine associa-
tions between neurophysiology and general cognitive function. With the exception
of the verbal processing factor, all general cognitive factors displayed significant
correlates. As with the imaging findings, the information processing speed, exec-
utive function (visual-spatial) and verbal memory factors displayed the most sig-
nificant correlates with the widest spatial distribution, in addition to being asso-
ciated with a broad range of frequencies and ERP components. The remaining
factors of vigilance and sustained attention, verbal processing, working memory
capacity and sensori-motor function displayed more specific correlates. This is in
contrast to the structural measures which displayed little or no correlates with
these factors, suggesting the psychophysiological measures show greater sensitivity
to these cognitive factors and are a better functional indicator of general cognitive
capacity.
For the ERP measures, there was greater specificity and complexity across the
correlations with cognitive factors. Consistent with the MRI findings, the informa-
tion processing and executive function factors were correlated with the most ERP
measures, with the other cognitive factors showing more specific ERP correlates.
These relationships are described in more detail below.
4.5.1. Resting qEEG — general cognition
The pattern of correlations between qEEG measures and cognitive factors is sum-
marized in Tables 13–17. As illustrated, the information processing speed and exec-
utive function factors show the most significant and positive correlations, both with
delta and theta activity (resting and phasic states) and theta/beta1 activity (resting
state only was examined; see Tables 13 and 14). In contrast, increased resting and
phasic (event-related) beta activity was associated with decreased cognitive per-
formance, especially for working memory capacity, information processing speed,
sensori-motor function and executive function factors, particularly at fronto-central
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 61
sites (see Tables 13 and 14). These results suggest that increased delta and theta
EEG activity and decreased beta in healthy subjects is associated with improved
general cognitive performance. This is contrary to what may be expected given
this pattern is typically associated with low states of cortical arousal and cognitive
inactivity [7, 66, 67], poorer cognitive performance [43], and pathological conditions
such as attention deficit disorder (ADD) and dementia [18, 27, 43, 54, 55]. However,
increases in event-related theta EEG activity are known to be associated with height-
ened cognitive processes [52], as was found in this study for theta qEEG (increased)
during the oddball paradigm (see Table 14). Similar patterns were also observed
for other factors, including verbal memory and information processing speed, sug-
gesting increased theta during brain activity is observed as a normal state of active
information processing.
Our results and these studies suggest there is an optimal resting EEG state
(reflected in resting delta, theta qEEG and theta/beta1 ratio) that maintains suffi-
cient cortical arousal while also allowing for sufficient cortical flexibility. If the ratio
is too low, the brain may not be able to easily switch into the increased theta states
that are associated with effortful activity, which leads to poorer cognitive perfor-
mance. Alternatively, with a higher theta/beta1 ratio, the brain can more efficiently
switch into the increased theta states that are associated with effortful cognitive
activity, resulting in improved cognitive performance. However, abnormally large
ratios are also detrimental where poor cognitive performance may result, in addition
to indicating a potential pathology such as dementia and ADD [18, 27, 43, 54, 55].
These findings impact on treatment interventions such as neurofeedback that are
aimed at normalizing the theta/beta1 ratio in patients. Most clinicians are also
accustomed to applying the same methodology to peak performance when it comes
to cognitive performance training (inhibition of slow activity and reward of beta
activity or decreasing theta/beta ratio). However, this study provides evidence and
a theoretical foundation for a different methodology of peak performance training
in healthy subjects.
The model of an optimal resting EEG state that maintains sufficient cortical
flexibility appears logical if we examine the EEG model of Robinson and colleagues
[79, 80, 82, 83, 85, 88]. In this model, delta-theta EEG activity arises due to a dif-
ferential excitatory and inhibitory activity in cortical and subcortical networks that
alters the characteristic 1/f spectral shape between 0 and 7 Hz that is observed in
the eyes closed resting EEGs. This 1/f spectral property is referred to as the Power-
law scaling, as indicated by the decrease in log power with increasing log frequency
(1/f) that is seen in both temporal and spatial spectra [28]. If we consider neocortical
excitatory pyramidal and the inhibitory stellate cells, increased inhibitory activity is
associated with a decrease in delta-theta EEG and a flattening of the characteristic
1/f EEG spectral shape. In contrast, reduced inhibitory and increased excitatory
activity leads to an increase in delta-theta activity and convergence of the spec-
tral shape to 1/f [88]. The model suggests there is an optimal state of delta-theta
May 8, 2007 12:11 WSPC/179-JIN 00143
62 Rowe et al.
activity that is generated by a balance between inhibitory and excitatory processes.
Too much inhibitory activity leads to an over inhibited cortex, while too much exci-
tatory activity can lead to instability and seizure [81]. This type of activity and the
1/f spectra has been considered to indicate aspects of self-organized criticality that
can permit a system to rapidly react to external perturbations, thereby inducing
new and flexible dynamic states [5, 28]. Presumably, this leads to an increase in cor-
tical flexibility that permits the cortex to readily switch into more effortful cognitive
states.
4.5.2. ERP amplitude — general cognition
For ERP measures, a consistent pattern of significant negative correlations over
prefrontal and fronto central sites were observed between the executive function
factor and positive-going ERP amplitudes including the NoGo P200 and P350 com-
ponent, working memory background P150 and P450 components, oddball target
P200 and P300 components, and the oddball background P200 component. Sim-
ilar correlates were observed for the information processing speed factor, and to
a lesser extent the verbal memory factor. These results would suggest that an
increase in the amplitude of positive-going ERP components at frontal sites is
associated with a decrease in cognitive performance. The opposite pattern was
found for negative-going ERP components with significant positive correlations
between amplitude and executive function at prefrontal and fronto central sites.
Sites included the NoGo N300, working memory background N100, N300, and odd-
ball target N200 components (see Tables 15–17). This pattern was also apparent,
but to a lesser extent for the information processing speed and verbal memory fac-
tors, although correlations with the NoGo N100 amplitude were more significant (see
Table 17).
These above results suggest that an increase in the amplitude of negative ERP
components at prefrontal and fronto central sites, is associated with increased gen-
eral cognitive ability on more effortful tasks of executive function, information pro-
cessing and verbal memory and learning. In contrast, an increase in the amplitude
of positive-going components is associated with a decrease in general cognitive per-
formance on such tasks. According to the ERP model of Rennie and Robinson
et al. [79], negative-going components are initially generated by positive thalam-
ocortical volleys that represent broad excitatory activation of their target cortical
areas [87]. By contrast, positive-going components reflect more inhibitory and focal
actions such that there is a narrowing or focus of the thalamocortical activation in
the neocortex. This process is exemplified by the searchlight hypothesis [23] that
proposes that cortico-TC feedback may serve as a mechanism which provides focal
enhancement and suppression of specific cortical networks [22–24, 65]. These studies
would suggest that broader activation of cortical regions in the frontal cortex, rather
than inhibition, may be associated with increased cognitive capacity for more and
complex effortful processing.
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 63
In contrast to the above findings, the negative-going components were also found
to show negative correlations with cognitive performance at posterior (parietal-
occipital) sites, as opposed to the positive-going components where there were
positive correlations evident at these sites. This would suggest the converse mecha-
nism may be operating within posterior regions of the cortex: broad thalamocortical
activation is reduced whereas inhibitory or focal activation is increased. This pat-
tern of activity in the posterior cortex may reflect the local neural connectivity of
the visual cortex, which is structurally dissimilar to the frontal cortex. There is
direct retinotopic mapping from visual sensory receptors to vertical columns in the
visual cortex [22, 65]. The extent of the structural differentiation is such that at
the local level, the visual cortex contains less diffuse and less combinatorial con-
nections between neighboring neurons compared to the frontal cortex. In addition,
the processing in the visual cortex can be considered relatively simple compared to
the complex processing that occurs in the frontal cortex, and which is believed to
be responsible for human intelligence, complex reasoning, creativity and personality
[87]. The frontal cortex is built with a different neural structure, with diffuse and
more intricate connectivity between neighboring cells. Therefore, activation in the
frontal cortex during more effortful and complex general cognitive processing could
occur across a greater spatial extent than in the visual cortex. Hence, the larger
negative-going components in the frontal cortex are associated with more diffuse
activation whereas reduced ERPs negative-going, but larger positive-going compo-
nents recorded over posterior regions are associated with more localized activation.
The findings from this study suggest that both patterns are associated with improved
general cognitive function in healthy individuals, but this effect shows spatial
specificity.
4.5.3. ERP latency — general cognition
For ERP latency measures, significant positive correlations were identified between
executive function and the ERP latency of several early components, including NoGo
N100 at frontal to central sites, oddball background N100 at prefrontal and fronto
central sites, and working memory background N100 at central to prefrontal sites
(see Tables 15–17). This pattern was also present for the information processing
speed factor but to a lesser extent.
The N100 components discussed above reflect early attentional processing of the
paradigm stimulus, and suggest that slightly delayed attentional processing is asso-
ciated with increased cognitive performance. This is somewhat counterintuitive to
findings that often associate cognitive abnormalities and poorer cognitive perfor-
mance with delayed ERP components [63, 78, 98]. Like the theta/beta1 measure-
ment, perhaps there is a more optimal latency of event-related processing in the
frontal cortex that facilitates more efficient processing [87].
As discussed in Sec. 4.5.2, assuming the increase in the amplitude of negative-
going components is associated with a spatially more diffuse cortical activation at
May 8, 2007 12:11 WSPC/179-JIN 00143
64 Rowe et al.
frontal sites, the delay in latency could be a further consequence of this, or may be
associated with such activity. More complex neural activities occurring in the frontal
cortex in response to higher loading cognitive tasks, reflected in the executive func-
tion and information processing factors, could be associated with certain delays in
neural processing leading to a delay in the latency of frontal ERP components.
Scientific evidence is emerging to suggest optimal delays in the transfer of neural
impulses via thalamocortical circuitry through to the neocortex is crucial for increas-
ing the probability of neocortical activation in response to a stimulus [45, 58, 94]. It
is hypothesized that these spatial and temporal relationships between neural activ-
ity in corticothalamic and thalamocortical pathways are responsible for the higher
intelligence and complex information processing capacity observed in humans [87].
4.5.4. Early ERP latency — information processing
The information processing factor displayed a similar pattern of correlates to the
executive function factor, with the exception of significant negative correlations with
some ERP latencies. These negative correlations were observed over fronto-central
and parietal regions for the NoGo P350 component, and over centroparietal regions
for the oddball target N200 and P300 components. These findings suggest that for
the information processing speed factor, faster ERP components at the middle and
later stages of processing are more advantageous (see Tables 15–17). Recall, that
the information processing speed factor consists of tasks that are dependant on
completion and reaction times, rather than more careful and structured manipu-
lation of the stimuli as in the executive function factor. Presumably, this style of
more rapid propagation does benefit from faster neural processing, as reflected by
the faster latency ERP components. However, as the processing load increases, as
observed predominantly with the executive function factor, we suggest that ERP
latencies will increase as the cortex shifts into a more optimal range of processing
that leads to sufficient temporal delays that permit processing of complicated stimuli
and relationships [87].
4.5.5. Specific qEEG findings — general cognition
Compared to the executive function factor, the delta and theta correlates for the
information processing factor were more significant and more diffuse. Similarly, the
resting delta qEEG activity was more significantly correlated with the verbal mem-
ory factor, but was not correlated with beta. These differences in resting delta,
theta and beta correlates between the various cognitive factors suggests it is pos-
sible to localize resting psychophysiological patterns that are associated with dis-
tinct domains of cognitive performance. Such data can be invaluable to scientists
or clinicians who are seeking to improve an identifiable deficit in a specific cogni-
tive domain via psychophysiological means. Alternative treatments could be directly
tested, either via a pharmacological compound that is designed to target a specific
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 65
EEG frequency or a neurotherapy technique that trains a patient to normalize a
specific EEG frequency. Once again, these findings suggest there are optimal levels
of resting delta, theta and beta EEG activity that is associated with increased per-
formance in specific domains of general cognition. For example, from this data it
appears that delta and theta frequencies are marginally more important for general
information processing than the delta frequency is related to verbal memory.
Another distinct feature of the information processing speed factor was that it
correlated positively with alpha power during the resting state (Table 13) and alpha
peak frequency during event-related qEEG (Table 14). This suggests greater alpha
power and faster peak frequency is associated with increased performance within this
cognitive domain. The significance of the alpha rhythm is that it is a measure of the
strength and delay of signal transfer in cortico-thalamocortical loops [88]. Increased
alpha power, particularly in the eyes open state, suggests signals are being boosted
through to specific areas of the cortex, while increased frequency suggests the prop-
agation of neural activity through cortico-thalamocortical loops is faster. Collec-
tively, this may increase the efficiency and accuracy of thalamocortical transfer [87].
Consistent with this, alpha measured during the oddball task demonstrated pos-
itive correlations with information processing speed. Specifically, positive correla-
tions were observed for alpha peak amplitude (backgrounds, over occipital regions)
and alpha peak frequency (backgrounds, over fronto-central regions). Such measures
are commonly associated with cognitive processes, in particular working memory
[17, 52]. Indeed, alpha peak frequency also showed positive correlations with execu-
tive function factor (fronto-centrally) and with the working memory capacity factor
(particularly centrally).
4.5.6. ERPs — vigilance and sustained attention
The vigilance and sustained attention factor deserves separate comment since it is
the only psychophysiological ERP measure that has a corresponding general cog-
nitive factor that was derived from the psychometric performance during the ERP
task. This factor displayed significant correlations with the ERP background P450
amplitude (positive) and latency (negative) generated in the working memory ERP
paradigm (see Table 16). In addition, the NoGo P350 amplitude across fronto central
and parietal sites showed significant positive correlations with the vigilance factor.
These results suggest that sustained mental effort and vigilance processes involv-
ing the discrimination of relevant from irrelevant information, are reflected in
increased positive-going amplitudes and decreased latencies, and therefore are asso-
ciated with better cognitive performance.
5. Summary and Conclusion
We have identified important relationships between neuroanatomical (MRI) and
cognitive measures, and psychophysiological and cognitive measures. Increased grey
May 8, 2007 12:11 WSPC/179-JIN 00143
66 Rowe et al.
matter volume is associated with increased cognitive capacity (general positive cor-
relations) but this appears to be at the expense of sociability and emotional control
(general negative correlations). This is consistent with view that social cognitive
function is underpin by distinct structures and neural function, and often repre-
sents more automatic and unintentional processing, in contrast to effortful and
intentional cognitive processing involved in general cognition as measured in this
study. This may partly explain why the social cognitive factors show predominantly
small or non-significant correlations with psychophysiological measures, given such
measures are more functional measures and therefore correlate significantly with
effortful general cognitive factors. Therefore it suggests that social cognitive per-
formance is better assessed by more functional ERP paradigms that directly assess
the brains early automatic processing of emotion. In addition, the sociability factor
compared with the emotional control factor was associated with an independent set
of neuroanatomical structures suggesting distinct cortical regions subserve specific
functions dedicated to social interaction. Level of sociability was negatively associ-
ated with parietal and occipital structures including both grey and white matter.
Emotional control was negatively associated with frontal and sensorimotor struc-
tures but was not associated with white matter.
The executive function factor was the most significant correlate with measures of
neuroanatomical structure and psychophysiological activity, even more so than the
general “g” factor. Other factors such as verbal memory, and vigilance and sustained
attention, were found to have more localized structural, as well as more specific
psychophysiological correlates. Consistent with the imaging findings, the executive
function factor also displayed the most significant and robust range of psychophys-
iological correlates. In summary, executive function was negatively associated with
beta activity, but positively associated with delta, theta EEG and theta/beta1 ratio,
across both resting and phasic brain states, contrary to findings in pathological con-
ditions. These findings could suggest that an optimal level of resting delta-theta
(high) and beta (low) EEG activity provides increased neural flexibility that per-
mits the brain to switch into more active delta-theta states, which are conducive for
improved cognitive performance.
An unexpected finding was the positive correlates between executive function
and the latency of early ERP components at frontal sites. This suggests that a
delay in the formation of ERP components is associated with increased cognitive
performance in the executive function and information processing speed domains.
This was interpreted to suggest that an optimal delay in ERP components in frontal
neural circuits is associated with more elaborate and complex processing of intrinsic
and extrinsic stimuli that reflects more complicated tasks. In addition, the informa-
tion processing speed factor correlated negatively with ERP latency of some later
components, consistent with its strong loading on measures of processing speed and
presumably the correlation with faster neural processing.
This study suggests that general and social cognitive performance are largely
independent and are associated with the activity of discrete and distributed neural
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 67
structures and specific psychophysiological activity. This information may help sci-
entists to develop agents which target psychophysiological endophenotypes or neural
structures to promote a change in a targeted area of general cognitive or social cogni-
tive performance. However, the cognitive or social correlate or the direction of change
may sometimes be counterintuitive and reference to a large normative database of
correlates between structure (MRI), psychophysiology and function appears useful
in determining this. Inversely, scientists may examine changes in cognitive function
due to a pharmacological agent or disease, and need to infer the possible under-
lying structural or neurophysiological (functional) changes. The large database of
structural and psychophysiological correlates makes this step possible.
The above suggestion of optimal qEEG and ERP activity is of interest to both
therapists and scientists examining the psychophysiological and social and general
cognitive effects of pharmacological agents. The results suggest that therapists must
not use qEEGs as a simple bidirectional marker. For example, increased theta/beta
ratio is typically interpreted as a pathological condition which must be inhibited
by treatment (e.g., neurotherapy). Instead, the findings here suggest that increased
theta/beta is not necessarily a pathological sign and therapists must be careful to
determine whether or not the functional cognitive correlate is abnormal, and if not,
still maintain an optimal theta/beta ratio and prevent over-inhibition. This is where
reference to the correlates of cognitive measures or factors can aid to identify what is
the optimal level in the individual. Scientists in the area of drug design and pharma-
cology can also take a similar approach. Scientists wanting to increase performance
in a specific cognitive domain may search for a compound that attenuates or raises
the particular qEEG or ERP measure that is observed to correlate with that cog-
nitive domain. However once again, he or she will need to be careful to titrate the
dosage until an optimal change in the psychophysiological measure is observed.
Research has also identified structural, neurophysiological and neuropsychologi-
cal markers that are associated with conditions such as schizophrenia, mild cognitive
impairment, Alzheimer’s disease and Attention Deficit Disorder [3, 10, 38, 43, 104].
These markers alone show a low level of specificity but when combined using an inte-
grative approach and multiple measures from a large test battery tapping into struc-
tural, neurophysiological and neuropsychological domains, the specificity between
disorders can be significantly increased [34]. This can provide a more specialized
understanding of neuropsychological function and specific disorders.
Overall the correlates found in this study were significant and within the small to
medium range, but the large sample size suggests the relationships observed are gen-
uine findings that occur for a significant number of individuals. However, the power
and clarity of the study could be improved in future work by analyzing the corre-
lates between tests that tap into discrete cognitive domains and by using functional
imaging methods and more event-related qEEG and ERP measures, rather than the
resting EEG and subset of ERP measures which were used.
May 8, 2007 12:11 WSPC/179-JIN 00143
68 Rowe et al.
Acknowledgments
We acknowledge the support of the Brain Resource International Database (under
the auspices of The Brain Resource Company — www.brainresource.com) for use
of the data reported in this study. We also thank the individuals who gave their
time to take part in the study. All scientific decisions are made independent of
any BRC commercial decisions via the independently operated scientific division,
BRAINnet (www.brainnet.org.au), which is overseen by the independently funded
Brain Dynamics Center and scientist members. L. M. Williams also holds a Senior
Research Fellowship funded Pfizer Pharmaceutical Company.
References
[1] Adolphs R, The neurobiology of social cognition, Curr Opin Neurobiol 11:231–239,
2001.
[2] Adolphs R, How do we know the minds of others? Domain-specificity, simulation, and
enactive social cognition, Brain Res 1079:25–35, 2006.
[3] Alexander DM, Arns MW, Paul RH, Rowe DL, Cooper N, Esser AH, Fallahpour
K, Stephan BC, Heesen E, Breteler R, Williams LM, Gordon E, EEG markers for
cognitive decline in elderly subjects with subjective memory complaints, JIntegr
Neurosci 5:49–74, 2006.
[4] Ashburner J, Friston KJ, Voxel-based morphometry — The methods, NeuroImage
11:805–821, 2000.
[5] Bak P, Tang C, Wiesenfeld K, Self-organized criticality: An explanation of 1/f noise,
Phys Rev Lett 59:381–384, 1987.
[6] Bates E, Thal D, Finlay B, Clancy B, Early language development and its neural
correlates, Handbook Neuropsychol, Vol 8, Part II, Elsevier, Amsterdam, 2002.
[7] Belyavin A, Wright NA, Changes in electrical activity of the brain with vigilance,
Electroence Clin Neuro 66:137–144, 1987.
[8] Bird CM, Castelli F, Malik O, Frith U, Husain M, The impact of extensive medial
frontal lobe damage on “Theory of Mind” and cognition, Brain 127:914–928, 2004.
[9] Brickman AM, Zimmerman ME, Paul RH, Grieve SM, Tate DF, Cohen RA, Williams
LM, Clark CR, Gordon E, Regional white matter and neuropsychological functioning
across the adult lifespan, Biol Psych 60:444–453, 2006.
[10] Brown KJ, Gonsalvez CJ, Harris AW, Williams LM, Gordon E, Target and non-
target ERP disturbances in first episode vs. chronic schizophrenia, Clin Neurophysiol
113:1754–1763, 2002.
[11] Carr L, Iacoboni M, Dubeau MC, Mazziotta JC, Lenzi GL, Neural mechanisms of
empathy in humans: A relay from neural systems for imitation to limbic areas, PNAS
100:5497–5502, 2003.
[12] Cattell RB, The Scientific Use of Factor Analysis in Behavioral and Life Sciences,
Plenum, New York, 1978.
[13] Chartrand TL, van Baaren RB, Bargh JA, Linking automatic evaluation to mood and
information processing style: Consequences for experienced affect, impression forma-
tion, and stereotyping, JExpPsychGen135:70–77, 2006.
[14] Chomsky N, Language and Mind, Harcourt Brace Jovanovich, New York, 1969.
May 8, 2007 12:11 WSPC/179-JIN 00143
Brain Structure and Function Corre lates of General and Social Cognition 69
[15] Clark CR, Orr RS, Wright EK, Weber DL, Working memory updating to visual verbal
stimuli: A high resolution ERP study, in Koga K, Nagata K, Hirata K (eds.), Brain
Topogr Today, Elsevier Science, Japan, Tokyo, 1997.
[16] Clark CR, Paul RH, Williams LM, Arns M, Fallahpour K, Handmer C, Gordon E, Stan-
dardized assessment of cognitive functioning during development and aging using an
automated touchscreen battery, Arch Clin Neuropsychol 21:449–467, 2006.
[17] Clark CR, Veltmeyer MD, Hamilton RJ, Simms E, Paul R, Hermens D, Gordon E,
Spontaneous alpha peak frequency predicts working memory performance across the
age span, Int J Psychophysiol 53:1–9, 2004.
[18] Clarke AR, Barry RJ, McCarthy R, Selikowitz M, Brown CR, EEG evidence for a
new conceptualization of attention deficit hyperactivity disorder, Clin Neurophysiol
113:1036–1044, 2002.
[19] Colom R, Jung RE, Haier RJ, Distributed brain sites for the g-factor of intelligence,
NeuroImage 31:1359–1365, 2006.
[20] Cooper NJ, Keage H, Hermens D, Williams LM, Debrota D, Clark CR, Gordon
E, The dose-dependent effect of methylphenidate on performance, cognition and
psychophysiology, JIntegrNeurosci4:123–144, 2005.
[21] Costa PT, McCrae RR, Four ways five factors are basic, Pers Indiv Differ 13:653–665,
1992.
[22] Crabtree JW, Organization in the somatosensory sector of the cat’s thalamic reticular
nucleus, JCompNeurol