Selective increase of cortical thickness in high-performing
elderly—structural indices of optimal cognitive aging
Anders M. Fjell,a,*Kristine B. Walhovd,aIvar Reinvang,a,bArvid Lundervold,cDavid Salat,d
Brian T. Quinn,dBruce Fischl,d,eand Anders M. Daled,f,g
aUniversity of Oslo, Department of Psychology, Norway
bRikshospitalet University Hospital, Department of Psychosomatic Medicine, Oslo, Norway
cDepartment of Physiology and Locus on Neuroscience, University of Bergen, Norway
dMGH-NMR Center, Harvard University, MA 02129, USA
eMIT Computer Science and Artificial Intelligence Laboratory, MA 02129, USA
fMR Center, Norwegian University of Science and Technology (NTNU), Norway
gDepartments of Neurosciences and Radiology, University of California, San Diego, CA 92093, USA
Received 21 May 2005; revised 22 July 2005; accepted 3 August 2005
Available online 19 September 2005
The aim of this study was to identify cortical areas important for
optimal cognitive aging. 74 participants (20–88 years) went through
neuropsychological tests and two MR sessions. The sample was split
into two age groups. In each, every participant was classified as ‘‘high’’
or ‘‘average’’ on fluid ability tests and on neuropsychological tests
related to executive function. The groups were compared with regard
to thickness on a point-by-point basis across the entire cortical mantle.
The old high fluid performers had thicker cortex than the average
performers in large areas of cortex, while there was minimal difference
between the groups of high vs. average executive function. Further-
more, the old group with high fluid function had thicker cortex than
the young participants in the posterior cingulate and adjacent areas.
Further analyses showed that the latter was a result of a complex aging
pattern, differing between the two performance groups, with decades of
cortical thickening and subsequent thinning.
D 2005 Elsevier Inc. All rights reserved.
Keywords: Aging; Cingulate gyrus; Cognition; Fluid function; Executive
function; Cortical thickness
How is optimal cognitive aging characterized neuroanatomi-
cally? While most people experience some decline of cognitive
functions with increasing age, others are able to maintain a very
high level of intellectual function throughout the entire life span.
The mechanisms causing this wide array of individual cognitive
differences in aging are largely unknown, but one would expect
neural, perhaps volumetrically detectable, correlates. High function
may be upheld into old age by preservation of the original neuronal
circuits. In this case, optimal aging is a matter of being as similar as
possible to younger persons. Alternatively, recruitment of alter-
native neuroanatomical structures during cognitive processing may
help keep up a high level of function, even when the primary areas
responsible for that function have begun to deteriorate. A potential
neuroanatomical correlate of the latter may be thickening of
specific cortical areas.
This article focuses on the importance of cortical thickness for
optimal cognitive aging.
The rationale for relating volumetric measures to cognitive
function is that (1) there is considerable individual variation in the
volume of brain structures and (2) larger volume may be associated
with higher cognitive capacity. The exact reason for such a
relationship is unknown, but candidate explanations involve
number of neurons or synaptic connections, degree of complex
circuitry, dendritic expansion, myelin thickness, metabolic effi-
ciency, efficient neurotransmitter production, release and reuptake,
and brain reserve capacity (Deary and Caryl, 1997). However, even
though a relationship between general intellectual ability and total
brain volume is established (for a review, see Vernon et al., 2000),
more specific relationships have been harder to demonstrate. Van
Petten et al. (2004) gave an overview of 11 studies of healthy
participants (including Baare et al., 1999; Gunning-Dixon and Raz,
2003; Gur et al., 1998; Hanninen et al., 1997; MacLullich et al.,
2002; Raz et al., 1993, 1998; Salat et al., 2002; Sanfilipo et al.,
2002; Schretlen et al., 2000; Sowell et al., 2001) and concluded
that it has been difficult to establish robust relationships between
neuropsychological functioning and gross morphometric character-
istics. All these studies included measures of frontal volume, but
with the exception of Salat et al. (2002) and Van Petten et al.
(2004), either rather gross or restricted measures have been used
(e.g. the entire frontal lobe or dorsolateral prefrontal gray volume).
1053-8119/$ - see front matter D 2005 Elsevier Inc. All rights reserved.
* Corresponding author. University of Oslo, Department of Psychology,
POB 1094 Blindern, 0317 Oslo, Norway. Fax: +47 22 84 50 01.
E-mail address: email@example.com (A.M. Fjell).
Available online on ScienceDirect (www.sciencedirect.com).
NeuroImage 29 (2006) 984 – 994
Still, some relationships have been identified. Schretlen et al.
(2000) showed that perceptual comparison speed, executive
ability, and frontal lobe volume each made significant contribu-
tions to fluid intelligence, indicating that the processing speed and
frontal-executive theories of cognitive aging complement one
another. However, diverse results have been reported. Salat et al.
(2002) found that greater orbital prefrontal volume selectively
predicted worse working memory performance in older adults.
One reason for the discrepant findings may be the use of
predefined regions of interests (ROIs). This may pose two
opposite problems. Less refined volumetric measures can reduce
the possibility of identifying robust relationships since the func-
tional organization of the frontal lobe is extremely complex and
specialized. On the other hand, the use of predefined restricted
ROIs may also lead to omission of important information since we
do not know the exact localization of complex cognitive functions.
An approach less constrained by predefined neuroanatomical
borders may be more suitable for exploring the relationship
between morphometric properties and cognitive function. Thus, a
point-by-point strategy may be more sensitive and less biased.
This was employed in the present study, where thickness was
compared for groups of interest at each vertex across the entire
The present paper seeks to relate two broad domains of
cognitive function to cortical thickness. The first domain is Ffluid_
abilities, a type of intelligence that is applied to novel problems and
is relatively independent of educational and cultural influences. In
this paper, fluid abilities refer to scores on psychometric tests
related to visuo-constructive abilities (block design) and visual
reasoning (matrices). These two tests have the highest g-loading in
the WAIS-R battery (Deary, 2001). The other domain is termed
Fexecutive_, or somewhat inaccurately, Ffrontal_ abilities. Normal
aging is known to lead to a reduction of the cortical volume or
thickness of the frontal lobes (Salat et al., 1999a,b; Tisserand et al.,
2002; Raz et al., 2004). This selective loss is taken to support the
view that Fexecutive_ or Ffrontal_ abilities decline relatively more
with increasing age than other cognitive functions (e.g. West, 1996;
West and Bell, 1997; Chao and Knight, 1997; Lowe and Rabbitt,
1997). However, the distinction between Fexecutive_ vs. Ffluid_
tests is not uncomplicated (Rabbitt, 1997). Still, there is evidence
in favor of a conceptual distinction. For instance, Lowe and Rabbitt
(1997) found that age-related variance on executive tasks remained
after controlling for speed and intelligence. Such evidence,
combined with scientific interest in the role of the frontal cortex
in cognitive aging, justifies a study of the relationship between the
different functions and brain characteristics.
The following questions were investigated:
(1) Can structural cortical characteristics explain why some
elderly show superior cognitive function? A general linear model
(GLM) approach was employed contrasting the cortical thickness
at each vertex of groups of old participants with high versus
average functioning with regard to both fluid and executive
abilities, respectively (see Materials and methods).
(2) If cortical thickness can contribute to explain the difference
between high and average functioning old groups, the following
needs to be put to the test: does optimal aging only require
preservation of cortical volume or does it call for changes in brain
morphometry, such as thickening in certain parts of the cortex? If
elderly uphold cognitive function mainly by preserving cortical
thickness, we will expect to see differences between high and
average performers in areas where cortical thickness is known to
decrease with age. If, however, optimal cognitive aging is
characterized by actual change, we may see thickening of selective
cortical areas with age. Cortical areas less prone to age reductions
will be neuroanatomical candidates for possible thickening. Based
on previous research using a point-by-point approach (Salat et al.,
2004), areas where thickening may occur include the temporal lobe
and medial structures such as the cingulate cortex. If the old high
performers have thicker cortex than the young (superior or average
performers) in specific areas, this implies actual changes, and
similar thickening in the old average performers is not to be
expected. However, if the only difference between the old and the
young participants is that the old have thinner cortex but to varying
degrees according to level of function, then limitation of cortical
decrement seems to be the more important mechanism in optimal
Materials and methods
Table 1 summarizes the sample, consisting of 74 volunteers (41
F/33 M) between 20 and 88 years. The participants were
community dwellers recruited among employees from a local
hospital, or through charity organizations, activity centers for the
elderly, and newspaper ads. They were screened by interview for
diseases and traumas known to affect CNS functioning, and criteria
for exclusion were neurological conditions or use of medication
known to influence central nervous system functioning (for more
specific exclusion criteria, see Walhovd and Fjell, 2002). All
participants were required to not use a hearing aid and have normal
or corrected to normal vision. They were given a moderate sum of
money to refund possible costs related to their participation. All
were examined with the Norwegian version of the Wechsler
Abbreviated Scale of Intelligence (WASI; Wechsler, 1999).
Participants with an IQ score of below 85 were excluded. Table
1 summarizes the characteristics of the total sample and the sample
divided into two age groups by the median age.
The frontal lobes, or circuits involving parts of the frontal
cortex, support a number of human higher-order cognitive
Young (n = 35)Old (n = 39) Total (n = 74)
For the Beck Depression Inventory (Beck DI), we have data from only 66
of the 74 participants. t test showed that no significant differences between
the groups existed for education (t = 1.862, df = 67.28, n.s.) and IQ (t =
0.338, df = 72, n.s.), while significant differences in mean score of MMS
(t = 2.473, df = 2.47, P < 0.05) and Beck DI (t = ?5.13, df = 62.55, P <
0.05) existed in the two age groups. In addition, IQ did not correlate with
age (r = ?0.03, n.s.).
A.M. Fjell et al. / NeuroImage 29 (2006) 984–994
functions, including regulation of behavior and cognition, response
monitoring and inhibition, and working memory and attention (e.g.
Balota et al., 2000; Damasio, 1994; Luria, 1980; West and Baylis,
1998). These functions are distinguished from more general fluid
abilities by their neuroanatomical and behavioral correlates.
Evidence for the frontal basis of these abilities comes from
behavioral patient studies (Bryan and Luszcz, 2000; Stuss et al.,
2001) and brain imaging studies (Bush et al., 2000; Chao and
Knight, 1997). However, involvement of other cortical and
subcortical structures are also critical in performing such tasks
(see, for example, Alexander et al., 1986; Rubin, 1999). In the
present study, a battery of five clinically validated tests related to
executive function was employed, in addition to two tests of fluid
cognitive function (tests with high loadings on fluid intelligence).
The tests used in the present study are well-validated neuro-
psychological instruments, described in depth elsewhere, e.g.
Lezak (1995) and Spreen and Strauss (1991). Here, they will only
be described briefly.
Tests of neuropsychological fluid function
Block design and matrix reasoning from the WASI battery are
fluid measures known to be less influenced by cultural experiences
but dramatically affected by age (Woodruff-Pak, 1997). Block
design involves copying small geometric designs with four or nine
plastic cubes while viewing a constructed model or a picture within
a specified time limit. Matrix reasoning requires the participant to
complete logical arrangements of designs with missing parts,
Tests of executive function
Five tests assumed related to executive functions were used.
Stroop test (MacLeod, 1991). The version of the test employed
in this study consists of three conditions. The simple tasks require
the participant to name the ink color of rows of colored circles and
to read color words (Stroop 1 and 2, respectively). The complex
task (Stroop 3) requires the participant to name the ink of words
that are color-incongruent (that is, the word meanings and ink
colors mismatch, e.g. the word blue printed in yellow ink). Much
literature links performance on the Stroop test to the function of the
prefrontal cortex in both aging and neurological conditions (West
and Bell, 1997; Brown and Marsden, 1998; Rafal and Henik,
Trail Making Test part A and B. TMT-A consists of consec-
utively numbered circles arranged randomly on a sheet of paper,
and the participant is required to draw a line between the circles in
ascending order as quickly as possible. In TMT-B, half the
numbers are replaced with letters, and the task is to connect each
number with a letter and each letter with a number (1–A–2–B–
3–C etc.). TMT-B is a commonly used test of prefrontal function
(Rasmusson et al., 1998; Lezak, 1995) and is considered a measure
of the ability to flexibly shift the course of an ongoing activity.
Digit span Backward (Wechsler, 1981) requires the participant
to mentally reverse an orally presented string of digits. This
involves double tracking in that both the memory and the reversing
operations must proceed simultaneously. Performance depends
upon working memory and cognitive regulation and manipulation
to a stronger degree than in the forward span task. It is thus
assumed that the test depends more on frontal structures than its
Corsi Block Tapping Test (CBTT) consists of nine black 1 2 in.
cubes fastened in a random order to a black surface, and the
participant is required to repeat a tapping pattern tapped by the
examiner. This test is one of the most commonly used tests of
non-verbal short-term memory in clinical neuropsychology.
Controlled Word Association test is a measure of a person’s
ability to make verbal associations to specified letters (here: F, A,
and S) within a time limit of 1 min per letter. COWAT is regarded a
measure of frontal function since frontal lesions generally result in
reduced scores, and the task tends to involve bilateral frontal and
temporal lobe activation. Studies have indicated a relationship with
tests like oral spelling, digit span, Stroop, and mental calculations
(Boone et al., 1998; Lezak, 1995).
For Stroop 3 and TMT-B, the influence from the simple
conditions (Stroop 1, TMT-A) was regressed out, and all
analyses were performed on the residuals, removing the effect
of lower-order cognitive functions like motor speed, visual
search, and color naming. Each participant’s score on the five
tests was then converted to t scores (mean of 50, standard
deviation (SD) of 10) based on the mean and the SD of the
sample (i.e. not age-corrected), and a composite score was
calculated as the mean of the t scores. Inter-item reliability
analysis of the 5 measures yielded a Cronbach’s alpha of 0.61,
indicating substantial overlap, and also non-shared variance
between the different variables. Since exclusion of any of the
variables would not have lead to an increase in the alpha, all
were included. This composite score will be referred to as
Fexecutive function_. Following the same procedure, a composite
score based on the two WASI performance subtests (block
design, matrix reasoning) was calculated, and this will be
referred to as Ffluid function_.
MRI scanning and volumetric analyses
A Siemens Symphony Quantum 1.5 T MR scanner with a
conventional head coil was used. The pulse sequences used for
morphometric analysis were: two 3D magnetization prepared
gradient echo (MP-RAGE), T1-weighted sequences in succession
(TR/TE/TI/FA = 2730 ms/4 ms/1000 ms/7-, matrix = 192 ? 256,
FOV = 256 mm), with a scan time of 8.5 min per volume. Each
volume consisted of 128 sagittal slices with slice thickness = 1.33
mm and in-plane pixel size of 1 mm ? 1 mm. The image files in
DICOM format were transferred to a Linux workstation for
The automated procedures for volumetric measurement of the
entire cortical mantle are described by Salat et al. (2004). Cortical
thickness measurements were obtained by reconstructing repre-
sentations of the gray/white matter boundary (Dale and Sereno,
1993; Dale et al., 1999) and the cortical surface and then
calculating the distance between those surfaces at each point
across the cortical mantle. This method uses both intensity and
continuity information from the entire 3D MR volume in
segmentation and deformation procedures to construct represen-
tations of cortical thickness. The maps are created using spatial
intensity gradients across tissue classes and are therefore not
simply reliant on absolute signal intensity. The maps produced
are not restricted to the voxel resolution of the original data and
thus are capable of detecting submillimeter differences between
groups (Fischl and Dale, 2000). Thickness measures may be
mapped on the Finflated_ surface of each participant’s recon-
structed brain (Dale and Sereno, 1993; Fischl et al., 1999),
allowing visualization of data across the entire cortical surface
without interference from cortical folding. Maps were smoothed
A.M. Fjell et al. / NeuroImage 29 (2006) 984–994
using a circularly symmetric Gaussian kernel across the surface
with a standard deviation of 12.6 mm and averaged across
participants using a non-rigid high-dimensional spherical averag-
ing method to align cortical folding patterns (Fischl et al., 1999).
This procedure provides accurate matching of morphologically
homologous cortical locations among participants on the basis of
each individual’s anatomy while minimizing metric distortion,
resulting in a mean measure of cortical thickness for each group
at each point on the reconstructed surface. The gray/white matter
boundaries are illustrated in Fig. 1. Statistical comparisons of
global data and surface maps were generated by computing a
general linear model of the effects of each variable on thickness
at each vertex. Instead of using a corrected P value threshold, a
scale with the actual P values is displayed in the figures. Since
group comparisons are done, the number of participants in each
group is limited, and a harsh criterion for multiple comparisons
may be too conservative.
High vs. average fluid function old groups
The different function groups were distinguished by dividing
the sample at the median score (Fig. 2). The high and the average
fluid groups had a mean age of 67.4 years 71.9 years, respectively
(t = 1.85, df = 37, n.s.), a statistically equal gender distribution
(t = ?1.785, df = 37, n.s.), and the respective fluid t scores were
50.3 (SD = 2.8) and 38.1 (SD = 4.9) (t = ?9.1, df = 37, P <
0.0001). Even though the age difference between the groups was
statistically insignificant, age was included in the GLM as a
regressor to ensure that differences between groups were not due
to age. The IQ means of the high and average functioning groups
were 121.8 (SD = 8.6) and 104.6 (SD = 9.6), respectively. The
GLMs comparing the high functioning with the average function-
ing old group showed differences in a variety of cortical areas,
especially in the right hemisphere. The results are displayed in
Fig. 3. All significant differences indicated thicker cortex for the
high relative to the average functioning group. The largest
difference between the groups was found in posterior parts of
the cingulate gyrus in right hemisphere. Furthermore, some frontal
and prefrontal areas differed between the groups in both hemi-
spheres, as well as the medial structure and the gyrus of the
High vs. average executive function old groups
The group with high versus average executive function had a
mean age of 69.1 years and 70.1 years respectively (t = 0.04,
df = 37, n.s.) and an equal gender distribution (t = ?0.791, df =
37, n.s.). The respective executive t scores were 51.5 (SD = 2.8)
and 42.2 (SD = 4.9) (t = ?7.3, df = 37, P < 0.0001). Of the 39
old participants, 25 got the same classification on both domains,
while 14 changed. As evident from Fig. 4, apart from a tendency
towards thicker cortex in a small part of the right frontal middle
gyrus in the high performers, there is essentially no difference
between the high executive performers and those with average
Old high and average performers compared to young
Since the cortical differences between the fluid groups were
much larger than the differences between the executive groups,
further analyses were restricted to fluid function only. By GLM,
the high functioning old group was compared to young
participants (mean age 33.7 years), who on average had higher
fluid function (old high mean = 50.3, SD = 3.8, young mean =
56.3; SD = 5.9, t = 4.058, df = 53, P < 0.0001). The young
group had thicker cortex in most areas. However, the old high
performers showed thicker cortex in certain restricted areas,
mainly in the posterior half of the right cingulate gyrus and in a
small area of the left subcallosal gyrus. To rule out the possibility
that this distinguishes high from average performers at any age,
the high and the average young group were compared. No
significant differences in these areas were identified. The results
are illustrated in Figs. 5 and 6.
Fig. 1. Gray matter segmentation. White and gray matter boundaries in the
T1 volume of a young woman. Gray matter is represented by the area
between the red and the green line.
Fig. 2. Distribution of high and average old cognitive performers. The
scatterplots show the t scores of the high and the average old fluid and
frontal performers. A median split was used to divide the participants into
high and average functioning groups. The t scores are based on the mean
and standard deviation of the test scores of the sample itself. Thus, the
scores are not corrected for age and cannot be related to the population
mean. However, full-scale IQ (WASI) was above 100 (the population mean)
for both groups.
A.M. Fjell et al. / NeuroImage 29 (2006) 984–994
Next, the average functioning old group was compared to the
young, and the results showed differences in the form of thicker
cortex of the young group throughout most of the cortical mantle.
Generally, the differences were larger in the right than in the left
hemisphere. The thickening of the right cingulate cortex was not
seen for the average functioning group of elderly. The results are
illustrated in Fig. 7. Scatterplots illustrating the individual data
points for four selected vertexes on the surface of the cortex are
presented in Fig. 8.
Based on the results from the GLMs, posterior cingulate was
chosen as an area deserving further exploration. Regression
analyses were performed for the high and the average
functioning parts of the sample separately (young and old
pooled together) to investigate whether the age slope of the two
function groups differed. Regression analyses with cortical
thickness at a point of the posterior cingulate as the dependent
variable, and age and age squareasindependentvariables,showed
that, for the high functioning group, there was indeed a non-linear
relationship(allbeta’sarestandardized:y =0.32?age(P <0.01)?
0.28 ? age2(P < 0.01),
indicating a prolonged thickening prior to a later decline, while the
same prolonged thickening was not observed for the average
functioning group (y = 0.15 ? age (n.s.) ? 0.17 ? age2(n.s.), F =
1.618 (n.s.), R2= 0.09) (Fig. 9). For the other three points picked for
further investigation, within the insula, inferior temporal gyrus, and
occipital gyrus, no non-linear effects were observed (for all age2
values, P > 0.30).
F = 8.178 (P < 0.001), R2= 0.32),
Cortical correlates of optimal aging
The present data indicate that cortical thickness is related to
superior fluid cognitive function in higher age. Since no
corresponding group differences were identified in young groups
of high vs. average fluid ability, the observed effects are likely not
only due to characteristics that are observable early in life. Similar
Fig. 3. Fluid function and cortical thickness. The cortical thickness of old participants with high and normal function on the two performance (fluid) tests of
WASI was compared by general linear modeling (different onset, same slope, assumed) across the entire cortical mantle. Red and yellow indicate areas of
thicker cortex for the high functioning group. In the right hemisphere, the main differences were found in temporal middle gyrus and temporal inferior gyrus,
gyrus cuneus, the gyrus and sulcus of the insula, gyrus rectus, the gyrus of the cingulate isthmus, and the posterior cingulate gyrus. In the left hemisphere, no
effects were identified in gyrus cuneus, insula, posterior cingulate gyrus, and temporal inferior and middle gyrus. Effects found in left but not in right
hemisphere included fronto-marginal gyrus, orbital gyrus, and subcallosal gyrus.
Fig. 4. Executive function and cortical thickness. The cortical thickness of old participants with high and normal function on a battery of neuropsychological
tests of executive functioning was compared by general linear modeling (different onset, same slope, assumed) across the entire cortical mantle. Red and yellow
indicate areas of thicker cortex for the high functioning group.
A.M. Fjell et al. / NeuroImage 29 (2006) 984–994
differences in cortical thickness were not found for groups of high
vs. average executive abilities. Therefore, we will focus on fluid
abilities here. Furthermore, there was a tendency for stronger
relationships in the right hemisphere, which was not surprising
since the fluid tests consist of visuospatial material (Prabhakaran et
al., 1997). Thus, in the following, we will concentrate on
discussing these right hemisphere effects. The superior and average
functioning old groups differed in the right middle and inferior
temporal gyrus (Broadman’s areas [BA] 21 and 20, respectively).
These brain structures are involved in several complex cognitive
tasks (Cabeza and Nyberg, 2000). For instance, functional studies
indicate a role for the inferior temporal gyrus in visual perception
(Ishai et al., 1999; Herath et al., 2001) and syllogistic reasoning
(Goel et al., 2004). It is unreasonable to expect a strict anatomical
correspondence between the results from structural and functional
imaging studies, but the abovementioned functions have some
resemblance with the visuo-constructive and visual reasoning task
used in the present study.
Furthermore, and somewhat more surprising, significant differ-
ences were found in large sections of the posterior cingulate gyrus
(BA 29, 30, 23, 31), the isthmus of the cingulate gyrus (BA 26),
and the adjacent gyrus cuneus (BA 18, 19). Gyrus cuneus is part of
the visual association cortex, and functional brain imaging has
demonstrated involvement of the cingulate gyrus in many higher
order cognitive functions. Both the anterior and posterior cingulate
have been implicated in tasks involving integration of complex
relations among stimuli (Kroger et al., 2002). Posner and Petersen
(1990) suggested that the cingulate is part of an executive attention
system needed when a supervisory mechanism is required to
resolve a cognitive conflict. Such situations may generally involve
high task demands, including planning and decision making, error
monitoring and correction, and novel and not well-learned
responses. These functions partly overlap with the requirements
involved in the fluid tests in the present study. Furthermore, the
posterior cingulate has been found to be a part of a neural network
involved in relational reasoning, also including the secondary
Fig. 5. High-performing elderly and young participants. The cortical thickness of old participants with high function on the two performance (fluid) tests of
WASI was compared by general linear modeling (different onset, same slope, assumed) across the entire cortical mantle to the group of younger participants.
Red and yellow indicate areas of thicker cortex for the high functioning old group, blue indicates thicker cortex for the young group. Differences were
identified in several areas, mainly the parietal superior gyrus, the inferior angular part of parietal gyrus, frontal superior gyrus, the temporal medial lingual part
of the occipital gyrus, the occipital pole, and gyrus precuneus, mainly anterior sections, in both hemispheres, in addition to the precentral and postcentral gyri
and the orbital gyrus in the left hemisphere. However, in the right hemisphere, the high functioning old participants exhibited thickening of the posterior half of
the cingulate gyrus, and in the left, the same phenomenon was demonstrated in a small area of the subcallosal gyrus.
Fig. 6. High- versus average-performing young participants. The cortical thickness of young participants with high function on the two performance (fluid) tests
of WASI was compared by general linear modeling (different onset, same slope, assumed) across the entire cortical mantle to a group of young average-
performing participants. Blue indicates areas of thicker cortex for the high functioning group. Differences were found in left hemisphere postcentral gyrus and
the occipital pole.
A.M. Fjell et al. / NeuroImage 29 (2006) 984–994
visual cortex and the medial anterior frontal cortex (Ruff et al.,
The main focus of cognitive research has been on the anterior
part of the cingulate cortex (BA 25, 24), which has been shown
to be involved in both higher order cognition and regulation of
emotion (for a review, see Bush et al., 2000). Recently, posterior
cingulate cortex as a part of larger neural networks has gained
more interest (Buckner, 2004). Since the neuroanatomy and
cytoarchitecture of the cingulate cortex is very complex (Elston
et al., 2005), different parts of the structure are obviously
involved in different aspects of human cognition. Studies of
monkeys have shown that the posterior cingulate is close to or
encompass major cortical connections to medial temporal
structures important for memory function and that it may be
highly associated with visuospatial functions (Kobayashi and
Amaral, 2003). In a recent review, Buckner (2004) suggested that
dysfunction of a neural network including the medial temporal
lobe and the precuneus, extending into the posterior cingulate
Fig. 7. Average-performing elderly and young participants. The cortical thickness of old participants with average function on the two performance (fluid) tests
of WASI was compared by general linear modeling (different onset, same slope, assumed) across the entire cortical mantle to the group of younger participants.
Red and yellow indicate areas of thicker cortex for the average-functioning old group, blue indicates thicker cortex for the young group. Differences in the form
of thicker cortex of the young group were found throughout most of the cortical mantle, both laterally and medially, in the frontal (including the middle and
superior frontal gyri, rectus gyrus, fronto-marginal, frontal inferior opercular, frontal inferior triangular, transverse frontopolar, and orbital gyri), temporal
(including the temporal middle gyrus, the temporal pole gyrus, inferior posterior parts, and the superior lateral aspect of the temporal gyrus and temporal
superior planum polar gyrus), and occipital (including occipital pole gryus, the middle occipital gyrus, and the occipital inferior gyrus) lobes, in addition to
areas in the parietal lobes, as lateral sections of the precentral gyrus and lateral and medial sections of postcentral gyrus, paracentral gyrus, gyri cuneus and
precuneus, parietal superior gyrus and the inferior supramarginal part of the parietal gyrus, the cingulate isthmus gyrus and posterior parts of the cingulate
gyrus, and ventral part of gyrus cuneus.
Fig. 8. Scatterplots for selected vertexes on the cortical surface. Scatterplots showing the cortical thickness for each participant at four different points (PCG:
posterior cingulate gyrus; insula; ITG: inferior temporal gyrus; the occipital gyrus) on the cortex. The participants are grouped based on age (old versus young)
and fluid performance (high versus average). The colored lines indicate the mean cortical thickness for each group at each point on the brain surface.
A.M. Fjell et al. / NeuroImage 29 (2006) 984–994
and retrosplenial cortex, may be central in explaining the
memory deficits associated with Alzheimer’s disease. The
cortical thickness in posterior cingulate differed between the
function groups in the present study. As seen in the non-linear
analyses, these differences are results of complex interactions
between the groups and the rate of early thickening and later
thinning of the cerebral cortex in this area.
Fluid ability fundamentally depends on visuospatial orient-
ing. One may reason that a ceiling effect exists for these
capabilities in young age, but more variance may be found in
aging. Preservation, as well as prolonged thickening of relevant
areas may help uphold such basal and essential capacities. The
latter may, as evident from Fig. 9, account for a relationship
between increased thickness of the posterior cingulate cortex
and fluid ability in old age. Based on the present cross-sectional
data, it appears that, even though the cortical thickness of the
posterior cingulate eventually starts decreasing also in elderly
with high fluid function, this decrease may not result in a
thinning relative to young adulthood since this area appears to
become thicker throughout middle adulthood in persons with
high fluid ability.
Finally, some distinct but scattered frontal lobe areas were
identified as significantly different between the high and average
old groups. Parts of the superior and middle frontal gyrus (BA 6, 8,
9, 10, 11) and gyrus rectus (part of BA 11), which may be
considered an extension of cingulate gyrus onto the frontal lobes,
were related to superior fluid function in high age. As parts of the
frontal and prefrontal cortex, these structures are close to areas
involved in working memory and executive and control functions
necessary in performance on fluid neuropsychological tests. Even
though no previous studies have pointed to a role for the
morphometric characteristics of this cluster of gyri in successful
aging, brain imaging studies have indicated that some cortical
regions are more critical than others in fluid intelligence tasks, e.g.
the frontal lobes (Duncan et al., 2000; Gray et al., 2003). Thus, the
present finding may be understood in relation to previous research
on the general function of the frontal lobe in human higher-order
Interhemispheric comparison is not the topic of this paper, but it
is noted that medial structures, including posterior sections of the
cingulate gyrus, the isthmus of the cingulate gyrus and gyrus
rectus, differed between high and average scorers in both hemi-
spheres. Furthermore, frontal and prefrontal areas differed between
the groups in both hemispheres, but not invariably the same areas
Evidently, the present results point to a possible role of cortical
thickness of specific brain areas in achieving or maintaining
superior fluid cognitive performance in high age. This result stands
in contrast to the findings related to executive abilities. Thickness
measures could only to a very modest extent explain the difference
between elderly of high and average performance on neuro-
psychological tests related to executive functions. The small effects
indicate that cortical thickness does not distinguish well between
superior vs. average executive function in high age. It must be
noted, however, that since analyses were performed on the
residuals when the effects of performance on simpler test
conditions were regressed out, this is a very strict procedure. In
the fluid domain, no such correction was performed, and, as such,
the results are not directly comparable.
Still, one reason for the lack of differences between the
executive groups may be that neuropsychological tests of
executive function traditionally are validated by use of patient
populations or functional brain imaging studies. However, the
necessity of a cortical area for maintenance of function does not
entail that normal variation in morphometric characteristics of
these areas is related to cognitive abilities. Based on the present
data, it seems that established neuropsychological tests of frontal
pathology are not related to differences in cortical thickness in an
average to superior functioning population. This result is in
accordance with the study by Van Petten et al. (2004), where
volumes of neither superior, middle, nor inferior frontal gyrus
correlated significantly with executive tests. The authors noted
that the lack of significant relationships could reflect limitations
in the regions measured, which did not cover the orbital gyrus,
gyrus rectus, or cingulate gyrus. In the present study, the whole
cortical surface is measured on a point-by-point basis, thus
making the latter explanation less probable. At least two
additional explanations for the lack of correspondence between
the test scores and cortical thickness exist. First, structural
changes in the frontal lobe do not necessarily entail changes in
the thickness of cortex. Second, other parts of circuits involving
frontal cortex, e.g. within the striatum (Rubin, 1999), may be
more powerful predictors of age decline in the functions referred
to as executive.
High and average fluid functioning old versus young
As expected, both the superior and the average functioning old
had thinner cortex than the young, and the thinning was much
more profound in the average than in the high function group.
This confirms the effect of age on cortical thickness. Of more
interest here, however, was that the high functioning group
displayed thicker cortex than the young group in mainly two
distinct areas: a small area of the subcallosal gyrus and posterior
parts of cingulate cortex. The differences were not found between
high and average functioning young participants, indicating that
older high performers actually have thicker cortex than young
high performers as a result of aging. Especially interesting is the
thickening of posterior parts of cingulate gyrus since this area may
support evaluative functions such as monitoring sensory events
and the organism’s own behavior in the service of spatial
Fig. 9. Non-linear regression lines for high and average performers in PCG.
Scatterplots showing the non-linear relationships between age and cortical
thickness in a point in posterior cingulated gyrus (see Fig. 8) for the high
and the average fluid performers (old and young polled together).
A.M. Fjell et al. / NeuroImage 29 (2006) 984–994
orientation and memory (Vogt et al., 1992). As argued, this
structure may also serve as an important link between neocortex
and several brain areas, including hippocampus. Evident from the
non-linear analyses, high performers will decline in thickness in
posterior cingulated after a certain age, but the thinning follows
decades of thickening and is not as profound as the thinning seen
in average performers.
Decline in the efficiency of more specialized cognitive
subprocesses may be compensated by increase of monitoring and
control. It can be argued that cognitive activity itself can alter the
brain morphometry in relevant structures (e.g. Draganski et al.,
2004) and may lead to actual increase of cortical thickness in
certain critical areas. As a consequence, one may speculate that
thicker cortex in specific areas in the high-functioning group may
be a structural effect of successful compensatory processing.
However, this result needs to be replicated and further investigated,
ideally with a longitudinal design, before it can be readily
accepted. Furthermore, it is important to reconcile this compensa-
tion view with findings of increased heritability estimates for
cognitive function with age (McClearn et al., 1997). The latter
could be interpreted as evidence against an environmentalist view
of brain and cognitive aging. However, a link between these
finding may be established within a framework of cognitive
reserve, where genetic factors may influence the thickness of
certain cortical areas and this may either directly or indirectly be
related to changes in cognitive subprocesses.
Limitations and conclusion
The conclusions from the present study should be drawn with
caution, and limitations should be noted: first, the subject groups
studied here are not large. Second, while predictions can be made
regarding cortical thickness based on ability group status, the
cortical thickness measures indeed also show scatter. For instance,
as seen from Fig. 8, there are a few persons with very thin cortex in
the average fluid ability older group. Furthermore, it should be
noted that predictions cannot easily be made from cortical
thickness in young adulthood since there are several with thin
cortices especially in the posterior cingulate in the young high fluid
ability group. It seems that mechanisms acting to enhance cortical
thickness of the posterior cingulate along with fluid ability work
throughout the adult life span, and the group effects are not
observable until late adulthood.
In conclusion, the present data imply a relationship between
cortical thickness and cognitive function in higher age. While
previous reports have been inconsistent, the point-by-point
strategy employed here may be an appropriate and sensitive
approach to study the relationship between cortical structure and
cognitive function. Furthermore, while a close relationship
between fluid cognitive function and cortical thickness was
found, only very small effects were observed for executive
functions. Finally, the present results indicate that both cortical
preservation and, to some extent, thickening may be necessary to
ensure optimal cognitive aging. Both processes are detectable by
morphometric methods. Of course, longitudinal studies are
necessary to ensure that the effects observed really are functions
of age. Still, the differences between the old performance groups
versus the differences between the young performance groups
give some indications that the observed effects are due to age,
and not only a reflection of characteristics that are also observed
at an earlier age.
The first two authors have contributed equally to the present
paper, and their names are presented in random order. Support for
this research was provided by the Norwegian Research Council,
the Institute of Psychology at the University of Oslo, the National
Institutes of Health (R01-NS39581, R01-RR16594, P41-RR14075,
and R01-RR13609), the Mental Illness and Neuroscience Discov-
ery (MIND) Institute, and in part by the Biomedical Informatics
Research Network Project (BIRN, http://www.nbirn.net), which is
funded by the National Center for Research Resources at the
National Institutes of Health (NCRR BIRN Morphometric Project
Alexander, G.E., Delong, M.R., Strick, P.L., 1986. Parallel organization of
functionally segregated circuits linking basal ganglia and cortex. Annu.
Rev. Neurosci. 9, 357–381.
Baare, W.M.C., Pol, H.E.H., Hijman, R., Mali, W.P.T., Viergever, M.A.,
Kahn, R.S., 1999. Volumetric analysis of frontal lobe regions in
schizophrenia: relation to cognitive functions and symptomatology.
Biol. Psychiatry 45, 1597–1605.
Balota, D.A., Dolan, P.O., Duchek, J.M., 2000. Memory changes in healthy
young and older adults. In: Tulving, E., Craik, F.I.M. (Eds.), Handbook
of Memory. Oxford Univ. Press, Oxford, pp. 395–410.
Boone, K.B., Ponto ´n, M.O., Gorsuch, R.L., Gonza ´lez, J.J., Miller, B.L.,
1998. Factor analysis of four measures of prefrontal lobe functioning.
Arch. Clin. Neuropsychol. 13, 585–595.
Brown, R.G., Marsden, C.D., 1998. Internal versus external cues and the
control of attention in Parkinson’s disease. Brain 111, 323–345.
Bryan, J., Luszcz, M.A., 2000. Measurement of executive function:
considerations for detecting adult age differences. J. Clin. Exp.
Neuropsychol. 22, 40–55.
Buckner, R.L., 2004. Memory and executive function in aging and AD:
multiple factors that cause decline and reserve factors that compensate.
Neuron 44, 195–208.
Bush, G., Luu, P., Posner, M.I., 2000. Cognitive and emotional influences
in anterior cingulate cortex. Trends Cogn. Sci. 4, 215–222.
Cabeza, R., Nyberg, L., 2000. Imaging cognition: II. An empirical review
of 275 PET and fMRI studies. J. Cogn. Neurosci. 12, 1–47.
Chao, L.L., Knight, R.T., 1997. Prefrontal deficits in attention and
inhibitory control with aging. Cereb. Cortex 7, 63–69.
Dale, A.M., Sereno, M.I., 1993. Improved localization of cortical activity
by combining EEG and MEG with MRI cortical surface reconstruction:
a linear approach. J. Cogn. Neurosci. 5, 162–176.
Dale, A.M., Fischl, B., Sereno, M.I., 1999. Cortical surface-based analysis
I: segmentation and surface reconstruction. NeuroImage 9, 179–194.
Damasio, A.R., 1994. Descartes’ error. Emotion, Reason, and the Human
BrainAvon Books, New York.
Deary, I.J., 2001. Human intelligence differences: a recent history. Trends
Cogn. Sci. 5, 127–130.
Deary, I.J., Caryl, P.G., 1997. Neuroscience and human intelligence
differences. Trends Neurosci. 20, 365–371.
Draganski, B., Gaser, C., Busch, V., Schuierer, G., Bogdahn, U., May, A.,
2004. Neuroplasticity: changes in grey mater induced by training.
Nature 427, 311–312.
Duncan, J., Seitz, R.J., Kolodny, J., Bor, D., Herzog, H., Ahmed, A.,
Newell, F.N., Emslie, H., 2000. A neural basis for general intelligence.
Science 289, 457–460.
Elston, G.N., Benavides-Piccione, R., DeFelipe, J., 2005. A study of
pyramidal cell structure in the cingulate cortex of the macaque monkey
with comparative notes on inferotemporal and primary visual cortex.
Cereb. Cortex 15, 64–73.
A.M. Fjell et al. / NeuroImage 29 (2006) 984–994
Fischl, B., Dale, A.M., 2000. Measuring the thickness of the human
cerebral cortex from magnetic resonance images. Proc. Natl. Acad. Sci.
Fischl, B., Sereno, M.I., Dale, A.M., 1999. Cortical surface-based analysis:
II. Inflation, flattening, and a surface-based coordinate system. Neuro-
Image 9, 195–207.
Goel, V., Makale, M., Grafman, J., 2004. The hippocampal system mediates
logical reasoning about familiar spatial environments. J. Cogn. Neuro-
sci. 16, 654–664.
Gray, J.R., Chabris, C.F., Braver, T.S., 2003. Neural mechanisms of general
fluid intelligence. Nat. Neurosci. 6, 316–322.
Gunning-Dixon, F.M., Raz, N., 2003. Neuroanatomical correlates of
selected executive functions in middle-aged and older adults: a
prospective MRI study. Neuropsychologia 41, 1929–1941.
Gur, R.E., Cowell, P., Turetsky, B.I., Gallacher, F., Cannon, T., Bilker, W.,
Gur, R.C., 1998. A follow-up magnetic resonance imaging study of
schizophrenia. Arch. Gen. Psychiatry 55, 145–152.
Hanninen, T., Hallikainen, M., Koivisto, K., Partanen, K., Laakso, M.P.,
Riekkinen, P.J., Soininen, H., 1997. Decline of frontal lobe volumes
in subjects with age-associated memory impairment. Neurology 48,
Herath, P., Kinomura, S., Roland, P.E., 2001. Visual recognition: evidence
for two distinctive mechanisms from a PET study. Hum. Brain Mapp.
Ishai, A., Ungerleider, L.G., Martin, A., Haxby, J.V., 1999. Distributed
representation of objects in the human ventral visual pathway. Proc.
Natl. Acad. Sci. 96, 9379–9384.
Kobayashi, Y., Amaral, D.G., 2003. Macaque monkey retrosplenial cortex:
II. Cortical afferents. J. Comp. Neurol. 466, 48–79.
Kroger, J.K., Sabb, F.W., Fales, C.L., Bookheimer, S.Y., Cohen, M.S.,
Holyoak, K.J., 2002. Recruitment of anterior dorsolateral prefrontal
cortex in human reasoning: a parametric study of relational complexity.
Cereb. Cortex 112, 477–485.
Lezak, M.D., 1995. Neuropsychological Assessment, 3rd edR Oxford Univ.
Lowe, C., Rabbitt, P., 1997. Cognitive models of aging and frontal lobe
deficits. In: Rabbitt, P. (Ed.), Methodology of Frontal and Executive
Function. Psychology Press, East Sussex, UK, pp. 39–59.
Luria, A.R., 1980. Higher Cortical Functions in Man. Basic Books, New
MacLeod, C., 1991. Half a century of research on the Stroop effect: an
integrative review. Psychol. Rev. 109, 163–203.
MacLullich, A.M.J., Ferguson, K.J., Deary, I.J., Seckl, J.R., Starr, J.M.,
Wardlaw, J.M., 2002. Intracranial capacity and brain volumes are
associated with cognition in healthy elderly men. Neurology 59,
McClearn, G.E., Johansson, B., Berg, S., Pedersen, N.L., Ahern, F., Petrill,
S.A., Plomin, R., 1997. Substantial genetic influence on cognitive
abilities in twins 80 or more years old. Science 276, 1560–1563.
Posner, M.I., Petersen, S.E., 1990. The attention system of the human brain.
Annu. Rev. Neurosci. 13, 25–42.
Prabhakaran, V., Smith, J.A., Desmond, J.E., Glover, G.H., Gabrieli, J.D.,
1997. Neural substrates of fluid reasoning: an fMRI study of neocortical
activation during performance of the Raven’s Progressive Matrices Test.
Cogn. Psychol. 33, 43–63.
Rabbitt, P., 1997. Introduction: methodologies and models in the study of
executive function. In: Rabbitt, P. (Ed.), Methodology of Frontal and
Executive Function. Psychology Press, UK, pp. 1–38.
Rafal, R., Henik, A., 1994. The neurology of inhibition: integrating
controlled and automatic processes. In: Dagenbach, D., Carr, T.H.
(Eds.), Inhibitory Processes in Attention, Memory, and Language.
Academic Press, San Diego, CA, pp. 1–51.
Rasmusson, D.X., Zonderman, A.B., Kawas, C., Resnick, S., 1998. Effects
of age and dementia on the trail making test. Clin. Neuropsychol. 12,
Raz, N., Torres, I.J., Spencer, W.D., Millman, D., Baertschi, J.C., Sarpel,
G., 1993. Neuroanatomical correlates of age-sensitive and age-
invariant cognitive abilities. An in vivo MRI investigation. Intelli-
gence 17, 407–422.
Raz, N., Gunning-Dixon, F.M., Head, D., Dupuis, J.H., Acker, J.D., 1998.
Neuroanatomical correlates of cognitive aging: evidence from structural
magnetic resonance imaging. Neuropsychology 12, 95–114.
Raz, N., Gunning-Dixon, F., Denise, H., Rodrigue, K.M., Williamson, A.,
Acker, J.D., 2004. Agning, sexual dimorphism, and hemispheric
asymmetry of the cereb cortex: replicability of regional differences in
volume. Neurobiol. Aging 25, 377–396.
Rubin, D.C., 1999. Frontal–striatal circuits in cognitive aging: evidence for
caudate involvement. Aging Neuropsychol. Cogn. 6, 241–259.
Ruff, C.C., Knauff, M., Fangmeier, T., Spreer, J., 2003. Reasoning and
working memory: common and distinct neuronal processes. Neuro-
psychologia 41, 1241–1253.
Salat, D.H., Kaye, J.A., Janowsky, J.S., 1999a. Prefrontal gray matter and
white matter volumes in healthy aging and Alzheimer disease. Arch.
Neurol. 56, 338–344.
Salat, D.H., Kaye, J.A., Janowsky, J.S., 1999b. Selective preservation and
degeneration within the prefrontal cortex in aging and Alzheimer
disease. Arch. Neurol. 58, 1403–1408.
Salat, D.H., Kaye, J.A., Janowsky, J.S., 2002. Greater orbital prefrontal
volume selectively predicts worse working memory performance in
older adults. Cereb. Cortex 12, 494–505.
Salat, D.H., Buckner, R.L., Snyder, A.Z., Greve, D.N., Desikan, R.S.R.,
Busa, E., Morris, J.C., Dale, A.M., Fischl, B., 2004. Thinning of the
cerebral cortex in aging. Cereb. Cortex 14, 721–730.
Sanfilipo, M., Lafargue, T., Rusinek, H., Arena, L., Loneragan, C., Lautin,
A., Rotrosen, J., Wolkin, A., 2002. Cognitive performance in
schizophrenia: relationship to regional brain volumes and psychiatric
symptoms. Psychiatry Res. 116, 1–23.
Schretlen, D., Pearlson, G.D., Anthony, J.C., Aylward, E.H., Augustine,
A.M., Davis, A., Barta, P., 2000. Elucidating the contributions of
processing speed, executive ability, and frontal lobe volume to normal
age-related differences in fluid intelligence. J. Int. Neuropsychol. Soc.
Sowell, E.R., Delis, D., Stiles, J., Jernigan, T.L., 2001. Improved
memory functioning and frontal lobe maturation between childhood
and adolescence: a structural MRI study. J. Int. Neuropsychol. Soc. 7,
Spreen, O., Strauss, E., 1991. A Compendium of Neuropsychological Tests:
Administration, Norms, and Commentary. Oxford Univ. Press, New
Stuss, D.T., Floden, D., Alexander, M.P., Levine, B., Katz, D., 2001. Stroop
performance in focal lesion patients: dissociation of processes and
frontal lobe lesion location. Neuropsychologia 39, 771–786.
Tisserand, D.J., Pruessner, J.C., Argita, E.J.S., van Boxtel, M.P.J., Evans,
A.C., Jolles, J., Uylings, H.B., 2002. Regional frontal cortical volumes
decrease differentially in aging: an MRI study to compare volumetric
approaches and voxel-based morphometry. NeuroImage 17, 657–669.
Van Petten, C., Plante, E., Davidson, P.S.R., Kuo, T.Y., Bajuscak, L.,
Glisky, E.L., 2004. Memory and executive function in older adults:
relationships with temporal and prefrontal gray matter volumes and
white matter hyperintensities. Neuropsychologia 42, 1313–1335.
Vernon, P.A., Wicket, J.C., Bazana, P.G., Stelmack, R.M., 2000. The
neuropsychology and neurophysiology of human intelligence. In:
Sternberg, R.J. (Ed.), Handbook of Intelligence. Cambridge Univ.
Press, New York, pp. 245–264.
Vogt, B.A., Finch, D.M., Olson, C.R., 1992. Functional heterogeneity in
cingulate cortex: the anterior executive and posterior evaluative regions.
Cereb. Cortex 2, 435–443.
Walhovd, K.B., Fjell, A.M., 2002. The relationship between P3 and
neuropsychological function in an adult life span sample. Biol. Psychol.
Wechsler, D., 1981. Wechsler Adult Intelligence Scale-Revised. The
Psychological Corporation, San Antonio, TX.
Wechsler, D., 1999. Wechsler Abbreviated Scale of Intelligence. The
Psychological Corporation, San Antonio, TX.
A.M. Fjell et al. / NeuroImage 29 (2006) 984–994
West, R.K., 1996. An application of prefrontal cortex function theory to Download full-text
cognitive aging. Psychol. Bull. 120, 272–292.
West, R., Baylis, G.C., 1998. Effects of increased response domination and
contextual disintegration on the Stroop interference effect in older
adults. Psychol. Aging 13, 206–217.
West, R., Bell, M.A., 1997. Stroop color–word interference and electro-
encephalogram activation: evidence for age-related decline in prefrontal
functioning. Neuropsychology 11, 421–427.
Woodruff-Pak, D.S., 1997. The Neuropsychology of Aging. Blackwell
Publishers, Malden, USA.
A.M. Fjell et al. / NeuroImage 29 (2006) 984–994