Download Structural divisions and functional fields in the human cerebral cortex 1

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Clinical neurochemistry wikipedia , lookup

Limbic system wikipedia , lookup

Binding problem wikipedia , lookup

Premovement neuronal activity wikipedia , lookup

Neuromarketing wikipedia , lookup

Evolution of human intelligence wikipedia , lookup

Nervous system network models wikipedia , lookup

Neuroscience and intelligence wikipedia , lookup

Embodied language processing wikipedia , lookup

Affective neuroscience wikipedia , lookup

Selfish brain theory wikipedia , lookup

Neurogenomics wikipedia , lookup

Synaptic gating wikipedia , lookup

Brain wikipedia , lookup

Artificial general intelligence wikipedia , lookup

Emotional lateralization wikipedia , lookup

Functional magnetic resonance imaging wikipedia , lookup

Neuroinformatics wikipedia , lookup

Human multitasking wikipedia , lookup

Activity-dependent plasticity wikipedia , lookup

Environmental enrichment wikipedia , lookup

Haemodynamic response wikipedia , lookup

Brain morphometry wikipedia , lookup

Neuroanatomy wikipedia , lookup

Magnetoencephalography wikipedia , lookup

Time perception wikipedia , lookup

Brain Rules wikipedia , lookup

Neurolinguistics wikipedia , lookup

Cortical cooling wikipedia , lookup

Feature detection (nervous system) wikipedia , lookup

Cognitive neuroscience wikipedia , lookup

Holonomic brain theory wikipedia , lookup

Neurophilosophy wikipedia , lookup

Neuropsychology wikipedia , lookup

Neuropsychopharmacology wikipedia , lookup

Cognitive neuroscience of music wikipedia , lookup

Neural correlates of consciousness wikipedia , lookup

Metastability in the brain wikipedia , lookup

Human brain wikipedia , lookup

Aging brain wikipedia , lookup

Neuroeconomics wikipedia , lookup

History of neuroimaging wikipedia , lookup

Neuroesthetics wikipedia , lookup

Inferior temporal gyrus wikipedia , lookup

Connectome wikipedia , lookup

Neuroplasticity wikipedia , lookup

Cerebral cortex wikipedia , lookup

Transcript
Brain Research Reviews 26 Ž1998. 87–105
Structural divisions and functional fields in the human cerebral cortex
Per E. Roland
a
a,)
, Karl Zilles
1
b
DiÕision of Human Brain Research, Department of Neuroscience, Doktorsringen 12, The Karolinska Institute, S-171 77 Stockholm Sweden
b
C and O Vogt, Brain Research Institute, Heinrich Heine UniÕersity, D-4000, Dusseldorf
1 Germany
¨
Abstract
The question of what is a cortical area needs a thorough definition of borders both in the microstructural and the functional domains.
Microstructural parcellation of the human cerebral cortex should be made on multiple criteria based on quantitative measurements of
microstructural variables, such as neuron densities, neurotransmitter receptor densities, enzyme densities, etc. Because of the inter-individual variations of extent and topography of microstructurally defined areas, the final microstructurally defined areas appear as
population maps. In the functional domain, columns, patches and blobs signifying synaptically active parts of the cortex appear as cortical
functional fields. These fields are the largest functional entities of the cerebral cortex according to the cortical field hypothesis. In its
strong version, the cortical field hypothesis postulates that all neurons and synapses within the fields perform a co-operative computation.
A number of such fields together provide the functional contribution of the cerebral cortex. The functional parcellation of the human
cerebral cortex must be based on field population maps, which after intersection analysis appear as functional domains. The major
structural–functional hypothesis to be examined is whether these functional domains are equi-territorial to the microstructurally defined
meta-maps. The cortical hypothesis predicts that, if two brain tasks make use of one or several identical or largely overlapping fields, they
cannot be performed simultaneously without errors or increases in latency. Evidence for such interference is presented. This evidence
represents a restriction in the parallel processing of the human brain. In the posterior part of the brain not only visual cortical areas may
qualify for parallel processing, but also the somatosensory cortices appear to have separate functional streams for the detection of
microgeometry and macrogeometry. q 1998 Elsevier Science B.V. All rights reserved.
Keywords: Human cerebral cortex; Computer; Field; Parallel
Contents
1. Introduction .
.......................................................................
88
2. Microstructural organisation of the cerebral cortex
...................................................
88
3. Population maps . . . . . . . . . . . . . . . . . .
3.1. Visual area V2 described as a population map
...................................................
...................................................
90
91
4. Functional parcellation of the cortex, the cortical field activation hypothesis .
4.1. Population maps of functional fields . . . . . . . . . . . . . . . . . . .
4.2. Multiple fields account for all types of brain tasks, examples . . . . . .
.....................................
.....................................
.....................................
91
94
94
...........................................................
...........................................................
95
96
............................
97
.......................................
99
................................................................
99
5. Intersection analysis . . . . . . . .
5.1. Intersection analysis, examples
6. Another example of intersection analysis of short term and long term memory representations
7. Formulation of the structural-functional problem in cortical parcellation .
8. The interference principle .
9. Parallel processing in the human brain . . . . . . . . . . . . . .
9.1. A case for parallel processing in the somatosensory cortex
)
1
............................................
............................................
Corresponding author. Fax: q46 Ž8. 309045; E-mail: [email protected]
Published on the World Wide Web on 20 February 1998.
0165-0173r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved.
PII S 0 1 6 5 - 0 1 7 3 Ž 9 7 . 0 0 0 5 8 - 1
100
100
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
88
10. Conclusions
.......................................................................
Acknowledgements .
References
101
.....................................................................
102
..........................................................................
102
1. Introduction
This paper presents some ideas on how the cerebral
cortex of man could be parcelled based on structural and
functional criteria. Any parcellation is based on an assumption of what is a cortical area. Since cortical areas are
thought to reflect the principle of organization of the
cerebral cortex, the issue of parcelling the cortex is also
fundamental for the organization of the cerebral cortex.
The question of what is a cortical area needs several
qualifications. Is it possible to define an area by one single
criterion, for example cytoarchitectonics? What is a border
of a cortical area? If cortical areas have borders how are
these borders defined? Here we suggest the approach that
quantitative criteria, that is criteria based on measurements
of microstructure and functional activations of the human
brain should be used to define and delimit areas.
The microstructure of the human cerebral cortex is
defined by more than a single structural modality. It
consists of Ž1. the cytoarchitecture, the packing density
and laminar distribution of pyramidal and non-pyramidal
neurons; Ž2. the myeloarchitecture, i.e. the packing density
and laminar distribution of myelinated fibers; Ž3. the receptor architecture, the density of neurotransmitter receptors in fmolrmg protein and laminar distribution of different receptor types; Ž4. the density of reuptake sites on
nerve terminals for different neurotransmitters; Ž5. distribution of other neurochemical variables such as the densities of enzymes, specific structural proteins, etc.; Ž6. the
architecture of synapses. Since especially the densities of
receptors, reuptake sites and enzymes is modifiable to
some extent even under non-pathological conditions, this
illustrates that the distinction between structure and function is fuzzy. The measurements of microstructural variables should be taken in vivo as well as in postmortem
brains. By functional activations is meant immediate biochemical changes and changes in ion channels which
accompany synaptic activity and are locally resulting in
changes in the regional cerebral blood flow ŽrCBF., regional cerebral metabolism, optical signals, extended
dipoles, etc.
Although it is possible to delimit areas based on microstructural measurements and delimit functional activations in single subjects, the concept of a cortical area
should be more general. We therefore show that this
concept should be realized as population Žprobability.
maps. The quantitative nature of data often implies the use
of statistics or mathematics to unambiguously delimit the
borders of microstructurally defined areas or functional
fields.
We present the cortical field hypothesis ŽCFH., i.e. the
functional organization of the human cerebral cortex is
based on functional fields each occupying a certain, relatively large territory of the cortex. Such fields are the
largest functional entities. Some major postulates of the
CFH are presented with examples to illustrate these. It is
not the purpose of this review to present all experimental
evidence for the CFH. This has been done elsewhere in
more detail w1,2x. Rather this article is an attempt to derive
some principles of how the relation between structure and
function could be studied; to present some quantitative
methods which will be necessary tools for such studies,
and to point out some consequences of the CFH for the
organization of the cerebral cortex in parallel and nonparallel processing. These aspects are illustrated with typical examples. The literature cited also should be regarded
as examples, since this communication by no means is
exhaustive in this respect.
2. Microstructural organisation of the cerebral cortex
It is generally assumed that the microstructural parcellation of the human cerebral cortex is a more or less settled
issue when it comes to cytoarchitectonics and myeloarchitectonics Žfor example Ref. w3x.. Although Brodmann’s
map is known to neurobiologists as the accepted parcellation of the cerebral cortex based on cytoarchitecture, it
reflects a parcellation based on subjective criteria of where
the borders should be. This map and others w4–8x are
usually derived from observations on a single or very few
brains, and they neglect the large variations in extent and
topography of microstructure from brain to brain w9–11x
ŽFigs. 1 and 2.. Moreover, the presented maps are useless
since they cannot be translated with any faithfulness into
the true three dimensional format of the brain, nor into a
flat-mounted 2D format. New attempts to investigate the
microstructural parcellation of the cerebral cortex in man
must be based on quantitatiÕe, obserÕer-independent cytoarchitectonic and myeloarchitectonic studies, take the
intersubject variability into account and be truly 3D volumes transferable to a standard anatomical format. Further,
no cortical area should be parcelled by a single criterion,
such as that of cytoarchitecture. Instead, additional inde-
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
Fig. 1. Brodmann’s areas based on cytoarchitectural studies of non-quantitative nature on mainly a single brain w139x.
89
90
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
Fig. 2. Individual variations of the extent of cytoarchitectural areas 4a and 1 in three subjects. ŽA. The borders of 4a Žred. and area 1 Žyellow. are shown in
the standard brain format w33x. ŽB. Three dimensional reconstruction and surface rendering of areas 4 and 1 based on quantitative cytoarchitectonics ŽSee
text.. The surface rendering of the areas is done on the standard brain format of the Human Brain Atlas w33x. Note that there is no relation between the
gross morphology and the extent and localization, neither did such correlations appear in the original brains ŽFrom Roland and Zilles w11x by permission..
pendent criteria like myeloarchitecture and the receptor
architecture should be used to, by multiple criteria, delimit
a microstructural area whenever possible w11,12x. Published examples of a combined cyto- and receptorarchitectural approach in humans demonstrate the effectiveness of
this concept w12–19x.
Quantitative cytoarchitectonics and delimitation of cortical cytoarchitectural areas by statistical measures is a
relatively recent development. The methodology is described in w12,13,20,21x. Software to correct for the distortions due to fixation, sectioning, and embedding of the
histological sections is necessary as well as magnetic
resonance imaging of the post mortem brains prior to
sectioning ŽFor details, see Refs. w22,23x..
Even if it is possible to define objectively the borders
based on quantitative measurements of the microstructure
and receptor structure, the extent and localization of such
delimited areas vary from brain to brain w9–11,24–26x.
The localization and extent of a cortical area defined on
quantitative measurements, therefore, can only be in the
form of a three-dimensional population Žprobability. map.
3. Population maps
A population map of a microstructurally defined area is
a three dimensional representation in standard anatomical
format of the delimitation of the area in a population of
subjects. Let the microstructural quantitative variable be
m. Further let the area in question be defined as m
exceeding a certain value m c . It is then possible in an
individual brain of subject Si to define a volume within
which m ) m c . In the 3D pictorial standard anatomical
representation of Si ’s postmortem brain all voxels within
this volume thus will have values larger than m c . Let there
be n brains examined, and let i be an indicator function
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
with the value of 1 when Si g in the volume in which m
) m c and 0 otherwise; then one can for each voxel in
standard anatomical format determine the fraction of subjects Ž h. for which m ) m c , i.e.
hm) mc s
1
n
h
Ý Si
i
m) mc
Ž 1.
The 3D standard anatomical representation of h thus
constitutes a population map of that area. In other words
the population map of an area, voxel by voxel shows the
fraction of brains belonging to that area. Others have
suggested that morphometric variability can be described
as probability maps, a concept closely related to population
maps w27,28x.
The variation of microstructurally quantitatively defined
areas depends on the standard anatomical format chosen.
This issue is examined by Roland et al. w29x. As an
example on how a population map can be constructed,
consider visual area V2.
91
constructed brain was transformed into the format of the
Human Brain Atlas ŽHBA. w33x.
In this way V2 was defined by borders of densities of
neurons ŽAmunts, Malikovic et al., unpublished observations. and laminar myelin densities in individual brains
w12x. These densities and GABA A receptor densities
changed at the V1rV2 and V2rV3, V2rVP borders w12x.
Each brain then contained a volume designated as V2.
When transformed into standard format, the five images
basically had identical gross morphological format. The
volumes of V2, however, due to individual differences,
covered slightly different parts of the standard brain format
volume. Fig. 3 shows the population map of V2, based on
five brains. The border of V2 in Fig. 3 is not absolute, but
defined by the 50 percentile i.e. voxels for which one half
of the population have V2, and the other half another area.
The delimitation of functional areas is also by population
maps Žsee later..
3.1. Visual area V2 described as a population map
4. Functional parcellation of the cortex, the cortical
field activation hypothesis
The volume density of neurons in V2 in five postmortem brains was estimated with the grey level index
ŽGLI. procedure w30x. Each pixel represents neuronal density in a field measuring 27 mm per side. Neuronal density
profiles, oriented orthogonally to the laminae and extending from the boundary between layers I and II to that
between layer VI and white matter were extracted from the
images. These profiles indicated the neuronal density
throughout all six lamina. Since the length of the profile
curves depends on the distance from the pial surface to the
white matter border, the profile curves were standardized
to a cortical thickness of 100% w12,21x. A mean feature
vector X 1 was calculated from a block of 10 adjacent
profile curves, and another mean feature vector X 2 from a
neighbouring block of 10 profile
curves. The Mahalanobis
X
distance w31x D 2 s Ž X 1 yX 2 . Cy1 Ž X 1 y X 2 . was calculated from the vectors and the inverse of the pooled
covariance matrix Cy1 . Plotting the D 2 values as a function of the position of the blocks of profiles revealed
maxima at locations at which major differences in laminar
patterns occurred between area V2 and the neighbouring
areas. Statistical significance was established by computing Hotelling’s statistic T 2 s D 2rŽ1rn1 q 1rn 2 . in which
n1 s n 2 s 10, the number of profiles from which each of
the mean vectors had been calculated w13,21x. In this way,
borders between cortical areas in general can be determined observer-independently based on laminar neuronal
densities.
By using the magnetic resonance images of the postmortem brains acquired before histological processing as
reference, compensation for shrinkage and distortion was
done by algorithms according to Schormann et al.
w22,23,32x. This software was also used for the 3D-reconstruction based on the sections. Subsequently the 3D-re-
It is often quoted that the elements of neuronal activity
are segregated into functional columns w3,34,35x. Whereas
this arrangement is obvious in konio-cortices or perhaps in
parakonio-cortices, the increased synaptic and neuronal
activity elsewhere seems to occur more like in fields, blobs
or patches, questioning the columnar organisation as a
general principle of organisation w36–39x. At a more coarse
level, studies in which tracer methods have been used to
reveal the activation of the cortex indicate that the information processing in isocortex is always dependent on
activation of cortical fields with a size ranging from some
600 to 2000 mm3 w1,40–47x, Žsee Ref. w2x for a review..
This is the cortical field activation hypothesis which in its
weak formulation states that neurons in the cerebral cortex
change their biochemical activity in large distinct ensembles w2x
The field is a theoretical construct w1,2x based on the
actual observed macroscopically extended activations in
man and primates, and on electro-physiological and
anatomical observations, plus theoretical considerations all
summarized in Ref. w2x. The field is assumed to exist as the
domain of active ion-channels, active synapses, and active
neurons engaged locally in functionally contributing to the
task being performed by the brain. The cortical tissue
surrounding the field is apparently not activated. Thus the
density of active ion-channels, active synapses and active
neurons is higher in the field than in the surrounding
cortical tissue. In essence the Cortical Field Hypothesis
ŽCFH. in its weak formulation does not address the issue
on how in detail the synapses and neurons within the field
are activated. It only postulates that the field is the largest
functional element.
The active ion-channels can be measured in experimental animals with fluorophores w48–51x, or with radioactive
92
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
ligands w2x. The activity of the synapses increases the
energy metabolism in the synapses and active synapses
therefore can be visualized in autoradiograms w52,53x.
When the synaptic activity is visualised in autoradiograms
of primates, the field is seen to be composed of patches,
bands, stripes — and in koniocortex — columns of increased glucose phosphorylation w36,37,54–56x. Similar
fields of patchy nature can also be observed with optical
methods in the homotypical isocortex of non-human primates and man w57,58x. The bandlike, patchy and stripelike
fine organization does not present any contradiction to the
observation that in perpendicular cortical penetrations the
neurons may respond to the same stimulus properties w3x.
Based on other observations, mainly in non-human
primates one may formulate a stronger Õersion of the
CFH. The stronger version relies on assumptions. The
main issue is that the patches, bands, stripes and columns
of activity constituting a field are postulated to perform
some form of cooperative computation. This is exemplified
in the single unit electrode recordings w59–63x, although a
direct correspondence between patches, stripes and bands
mainly reflecting synaptic activity and action potentials of
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
local neurons is not to be expected. Multiunit recordings in
awake behaving animals also seem to support the idea that
synchronisation of neuronal populations is field like and
distributed to multiple fields w64x. The fine organization of
the field is inhomogeneous in as much as synaptic excitation and synaptic inhibition vary within the field Žand
autoradiographically give rise to the columns, patches,
bands and stripes., but the parts in which excitation dominates and the parts in which inhibition dominates are all
hypothesized to form some form of cooperative computation. In the agranular and the konio-parakoniocortices, the
fields are topologically organised Žsomatotopical, topotopical, retinotopical.. Outside these zones, the organisation is
non-topical.
The anatomical reason for the internal organization of
fields has become more evident in recent years. In motor
cortex, somatosensory cortex, visual association area V4
and in prefrontal cortex, the intrinsic cortical connectivity
is organised in a patch-like fashion w65,66x. A small part of
the cortex is connected with surrounding cortex in a patchor band-like fashion, the patches are interrupted by gaps in
which no neurons are connected with the small part. The
widths of both patches and gaps seem dependent on the
dendritic spread of pyramidal neurons in layers 2 and 3
w65,66x. This pattern can be matched with the terminal
arborizations of axon collaterals and horizontal intrinsic
connections w67,68x. Also cortico-cortical fiber terminals
seem to produce a similar pattern of patches and gaps in
the target area w66,69–73x. Axon terminals of cortico-cortical connexions and callosal terminals can span several mm
in monkey cortex w74–76x. These factors and in addition
the dispersed connectivity pattern of cortico-cortical groups
of axons from a small cortical locus Žw77,78x for example.
are major structural determinants in the formation of functional fields.
Under the strong version of the CFH, each field is
supposed to make some functional contribution to the task
being performed by the brain. We refer to this as the
corporate computation of the neurons in the field. At a
given moment, the synapses in the patches and bands in
which excitation prevails are activated due to depolariza-
93
tion of the afferent terminals and of the horizontal intracortical excitatory terminals. The afferent terminals being
depolarized may originate from the thalamus and from
several other cortical fields. For a short period of time, one
may perhaps somewhat simplified, regard the afferent
sources as fixed, meaning that the fast computation in the
field is determined by the particular combination of thalamic and cortical afferents Žleaving out the modulatory
amine afferents.. Each patch within the field is assumed to
receive the same combination of depolarized afferents
from the fixed sources. However the horizontal excitatory
intracortical connexions within and between patches must
have some role in making the computation corporate. Thus
the functional contribution of the field is the corporate
computation being performed by the neurons constituting
the field. At any moment the combination of afferents
depolarizing the dendrites within the field may change and
thereby the functional contribution changes. Presumably
can parts of the network constituting the field be lesioned,
but still be functional w79x. The CFH in its strong version
fails if the field at any given moment contains patches
performing computations of two different kinds or if the
horizontal cortical connexions are unimportant for the
computation.
If it is so that the field is surrounded by less active
synapses and neurons it must have borders and a shape.
The borders of functional fields can be defined in anatomically coherent images of the change in any physiological–
biochemical variable, for example DrCBF, the change in
regional cerebral metabolic rates DrCMR, change in
neurotransmitter binding DBmax, change in optical signal,
as fields of EPSP’s and IPSP’s, as a BOLD signal in
f-MRI, opening of several types of ion-channels etc. In
such images the problem of determining where an increase
stops and noise or decreases take over can be treated in a
strict mathematical way w80x. The general idea is to embed
the quantitatively measured physiological–biochemical
signal or a derived statistical signal into a one parameter
family of successively smoothed maps by linear operations. Under scale-space theory, the field is represented as
a scale-space blob. For a precise definition of a scale-space
Fig. 3. Population map of visual area V2. The color scale extends from dark blue Ž0.0. to white Ž1.0. for the probability of V2 localization. The probability
space in which half of the population has their V2 Žthe 50% overlap border. is seen as a transition between yellow and green. The individual
cytoarchitectural V2 maps have been filtered with a 4 mm Gaussian isotropic filter ŽFrom Roland et al. w29x by permission..
Fig. 4. Population map of functional fields in the somatosensory motor region of the left hemisphere. Six subjects discriminated the shapes of rectangular
parallelepipeda with their right hand w96x. Red color: six out of six subjects having their activation, yellow: five out of six subjects having their activation,
in orange was four out of six, green: three out of six, light blue: two out of six, and dark blue: one out of six. This functional population map is
superimposed on the average MRI of the anatomy of the six subjects. The MRI is by 8 mm3 voxels, hence the blurring ŽCourtesy of Dr. Anders Ledberg,
Division of Human Brain Research, Karolinska Institute, 1997..
Fig. 5. Maps of the average extent of a group of ten subjects doing a somatosensory reaction time task with the right hand. Sagittal section of the left
hemisphere through the lateral parietal operculum. The approximate location of the areas 1 and 2 is shown. Note the field activations in areas 3b, 2 and
cortex lining the interparietal sulcus as well as the lateral parietal operculum and the scattered fields in the premotor cortex. Fields are shown superimposed
on the standard Human Brain Atlas w33x.
94
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
blob, see Ref. w80x. As a scale-space blob the intensity is
linked over different scales to define a five-dimensional
object with extent in three spatial dimensions one intensity
dimension and one scale-space dimension. Given the intensity and the spatial extent, the borders of the field or
scale-space blob can be uniquely mathematically represented at an optimal scale w80–82x.
That the field is the largest functional element implies
that information processing is not done by the synapses
and neurons in a single independent column, a single
independent patch or a single independent band. If this was
so, it would be impossible to see changes in activity of the
cerebral cortex by imaging methods which integrate over
larger tissue blocks Žof the approximate size of 40 mm3 or
larger.. Thus many columns, patches, stripes of increased
synaptic activity must be present within the tissue sampled
in order to be detected by PET or f-MRI. Still one might
claim that the patches, stripes and columns of increased
activity belong to independent population of neurons, for
example excited by independent Žparallel. afferent axons,
and that no intracortical communication exists between the
activated neuron populations. This objection does not affect the CFH in its weak formulation, but — if it is true
— falsifies the CFH in its strong formulation. The field
activation concept transcends different scales. A field can
be defined as a cortical domain of active Žopen. ion-channels or as an active large neuronal population at the
macroscopical scale. This should not be confused with the
methods used to identify fields. At the macroscopical scale
f-MRI and PET may not have the spatial resolution to
segment areas which are functionally compartmentalized,
for example V1 and V2 where the CFH does not apply.
Since most neuroimaging studies of man are studies in
which the statistically significant voxels of a group of
individuals are summarized as a collection of the centers
of gravity of fields or points of maximal activity, the field
character is not evident from such studies. However, recently the extent or number of statistically significant
voxels have become a statistical parameter w83–86x. The
significant parts of the cerebral volume appear then as
clusters of voxels and these statistical images are often
called cluster images and the results from experiments
analyzed this way have supported the field activation
hypothesis Žw87–91x and many others.. Thus, when the
cerebral cortex is activated during a particular task, multiple activated volumes appear, but since the cortex is thin
Ž1.5–3 mm. a more appropriate term may be fields w2x.
The exact extent of fields is difficult to ascertain by
conventional approaches, because the extent of significant
clusters of voxels by the current methods appear by thresholding statistical images of the type of Student’s t and
similar images w84,85,92x. Furthermore, these statistical
images have so far mainly been based on results of a
whole group of subjects, thus not reflecting the individual
variances in increases and decreases of the rCBF. With the
advent of 3D acquisition PET, it is now possible to study
the extent and the variation of extent in activated cortical
fields among individuals w93x. Functional magnetic resonance imaging ŽfMRI. in part has the same problems. The
exact extent of the activated volumes is not clear from the
fMRI images, because nearby located veins give a stronger
signal than does the cortical tissue w94,95x. A reduction in
voxel volume with the purpose of increasing the spatial
resolution of fMRI unfortunately reduces the available
signal, but the noise level remains constant. This in practice reduces the signal to noise ratio so that the spatial
resolution of PET and fMRI images are still comparable.
4.1. Population maps of functional fields
With the new method w82x, fields can be delimited in a
statistical image based on analysis of several individuals or
from statistical images from single individuals. Only the
latter will be true population maps, that is 3D-maps which
show the variations in location, extent and amplitude of the
change in the quantitatively measured physiological variable.
Fig. 4 shows the variations in extent of activation of the
motor cortex and somatosensory areas in six subjects
discriminating the shapes of rectangular parallelepipeda
w93,96x. Thus as expected, the extent of functional fields
also vary among subjects, even in a standard anatomical
format. This implies that functional fields also should be
represented as population maps. It is possible then to
delimit the volume for which a certain percentage of a
population is represented by a particular functional field.
4.2. Multiple fields account for all types of brain tasks,
examples
The cortical field hypothesis was presented as a series
of axioms of which one is:
Multiple fields working together account for all types of
brain activity, in the awake state Žsee 16.6.2 in Ref. w2x..
This implies that no function can be subserved by the
brain, by the activation of one field only.
This means for example also that simple point-like
stimuli in a behavioral context should activate multiple
fields. A point-like stimulus on a receptor sheet may be
expected to activate only a very restricted part of the
primary sensory cortex. But according to the cortical field
hypothesis the effect of stimulating a small part of, say the
skin surface should give rise to a field with a somatotopically corresponding localization, reasonably large to be
detected with f-MRI or PET and additional fields of
similar or larger size elsewhere. In a recent experiment,
nine subjects were engaged in a simple somatosensory
reaction-time task w97x. The signal to react to was a 2 mm
in diameter stylus which suddenly, at random intervals,
indented the pulp of the right index-finger by 2.8 mm for
1000 ms. The subjects reacted by pressing a response key
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
with the right thumb. The control was holding the key with
the thumb on, but otherwise resting w98x. Although the
peripheral stimulus was restricted to some 15 mm2 of the
skin surface the primary somatosensory cortex was activated corresponding to area 3b in a rather large field
extending more than 7 mm medio-laterally on the postcentral gyrus ŽFig. 5.. In addition larger fields appeared in the
cortex lining the postcentral sulcus, and the somatosensory
association areas in the anterior part of the superior and
inferior parietal lobule including the parietal operculum
ŽFig. 5.. Other types of brain works, in which larger parts
of the receptor sheet have been stimulated are associated
with similar size fields in the homotypical isocortex
w44,45,47,99–103x. Thus even point-like stimuli are associated with larger fields in homotypical isocortex where field
size does not seem to be dependent on the size of the
stimulus on the peripheral receptor sheet.
There have been no challenges to the CFH in its weak
form that the cortical field is the largest functional element. However there have been challenges to the axiom
that multiple fields account for all types of brain activity in
the awake state. One apparent challenge is the claim that
there is only one area specifically devoted to color perception w104x. The discrepancy is more apparent than real. It is
possible that by subtracting the measurement of activity of
one condition from that of a slightly different condition,
that only a single field will appear. Cognitive subtraction
paradigms w105x assume that brain functions can be separated into components, of which some can be
eliminatedradded by manipulating the task conditions.
Thus by using a control condition which contains one
cognitive component less than the task condition, one can
isolate the cortical fields and subcortical neural populations subserving that component. By repetitively subtracting the measured activity Žregional cerebral blood flow
rCBF, regional cerebral metabolism rCMR and BOLD
signal, etc.. of control from the measured activity during
the task across subjects, one arrives at a statistical image
from which the fields changing ŽrCBF, rCMR. significantly will appear. The conclusion usually drawn from
such experiments is that the fields isolated this way represent the neuronal populations responsible for that cognitive
component. In the example of w104,106x, only one site
appeared as being specific for color perception compared
to the control. This of course does not imply that this
region alone subserves perception of color. The limitations
of interpretation of results from subtraction paradigms
have been pointed out earlier w2,107,108x. The permissible
conclusion of the above color experiment is that the synaptic activity, as reflected in the field increase of rCBF, is
higher during color perception compared to the control
Žcontaining no color components.. Of course color perception would not be possible if not the blob regions of V1,
the stripe region of V2 and the V4 were active as a
minimum. Indeed, Gulyas and Roland w45x in a subtraction
paradigm, in which controls were detection of similar
95
non-colored dots, found several fields appearing as participating in color perception, some of which were located in
V1 and V2 w109x. That the rCBF cancel out by subtraction
in a certain region just signifies that the task and its control
activated that region to the same extent or that no significant changes occur. This does not permit any conclusion as
to the functional contribution of such areas which cancel
out and hence no conclusion about whether the fieldŽs.
found significant represent or do not represent a particular
perceptual or cognitive ‘component’.
It is usually assumed that the cerebral cortex is functionally specialized and that each field and corresponding
neuron population make a specific functional contribution
reflected in the computation taking place in the field. On
the other hand, the CFH postulates that multiple fields
working together account for all types of brain activity in
the awake state Ž16.6.2 in Ref. w2x.. The simultaneous
activation of multiple fields is the function being performed by the brain Ž16.6.3 in Ref. w2x.. No object, scene
or pattern is represented in a single field Ž16.6.9 in Ref.
w2x.. Is it still possible to envisage that a certain brain task
could be performed by a single cortical field? First it is
difficult to imagine the field working in isolation, because
each part of the cortex is connected to by 5 to 20 other
regions w110,111x. Second, although neurons locally may
engage in synchronous oscillatory activity, such activity
will tend to spread and distribute over many cortical
regions w112x. Rather the field activation reflects the intrinsic synaptic activity plus the afferent synaptic activity from
5–20 other regions of the brain sending afferents to the
field territory. The composition of active afferents may
change, giving the possibilities of alternative functional
contributions of the field or a field covering mainly identical territory. This high connectivity between cortical areas
will tend to always give activation patterns of multiple
fields associated with the performance of a certain brain
task.
Besides the limitations of subtraction paradigms, cognitive subtractive categorical designs have been criticised for
neglecting the interaction between task components w108x.
A brain function is logically defined against no task conditions. The problem Ždiscussed elsewhere in Refs. w2,113x.
is that in no task conditions the brain may well be active
with all sorts of intrinsic activities. A clear reference state
is difficult to establish Žsee, though, Refs. w98,114,115x..
That is why the subtraction technique has become so
popular. However to arrive at an approximate conclusion
about the functional contribution of a certain cortical part,
many subtraction Žand correlation. experiments must be
done with different reference or control conditions.
5. Intersection analysis
Whereas it is hazardous to claim a particular functional
specialization of a field on the basis of a subtraction
96
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
paradigm applied on a single group of subjects, repeated
applications of subtraction paradigms Žwith different control conditions, andror test conditions, all containing the
functional aspect searched for. provide, if a consistent set
of fields is identified, a stronger evidence for the functional contributions of the isolated field. Thus put in
another way, the set of fields expected to subserve a
certain functionrfunctional contribution should be consistent across groups and conditions including that
functionrfunctional contribution. This is why this approach first was called consistence analysis w107x. A more
appropriate name is intersection analysis, because the functional contributions of fields are revealed by Boolean
intersections. Intersection analysis provides an opportunity
for radically different experiment designs. Let us assume
that conditions A and B differ in all respects except one.
Then one can formulate the hypothesis that a certain set of
fields will be active in both conditions A and B. The
isolation of these fields will appear in the cluster images of
the subtractions, condition A y control A l condition B
y control B . A Boolean intersection between the two
statistical images showing the statistically significant fields
in A y control A and B y control B shows the statistically
significant fields activated in both condition A and condition B. Similarly, Boolean intersection can be applied to
statistical Žcluster. images showing correlation between the
DrCBF and an external vector Žfor example a measure of
psychophysical performance.. Especially when applied to
independent statistical Žcluster. images, intersection analysis increases statistical power. An alternative approach
having some similarities to intersection analysis, building
on the general linear model was presented by Price and
Friston w116x.
By multiplying cluster images from different subtractions one obtains an image in which all voxels are zero
except for those that contain a non-zero value in the
images. This image thus shows the overlaps. Previously
one has defined a priori the criterion that two activated
fields X i,A and X i,B originating from the respective cluster
images A and B reflect activity in roughly the same
synaptic cortical field. Let the estimated volumes and
centers of gravities of X i,A and X i,B be Vi,A and Vi,B , and
a i,cog and bi,cog respectively. Let Vi,A - Vi,B . The activated
fields X i,A and X i,B reflect activity in roughly the same
synaptic cortical field if, on multiplying the cluster images
A and B
X i ,A l X i ,B s Q; VQ G 0.5Vi ,A ; a i ,cog g Q and bi ,cog g Q
Ž 2.
This means that two fields reflect activation in the same
location if the overlap is equal to or greater than half the
volume of the smallest of the two fields, and the centers of
gravity of the two fields are included in the overlap.
Overlaps which satisfy these arbitrary formal criteria are
major overlaps. It should be emphasized that these criteria
are arbitrary. Furthermore, this practical and formal approach does not address the issue of getting overlaps
which are minor, nor the issue of the probability by chance
to get such an overlap. Since the extent of significant
clusters in current approaches depends on threshold and
filter width, the two cluster images must come from images which have been filtered to the same extent and
thresholded to the same extent to give the same significance limit a.
In the case that the cluster images A and B are obtained
from statistical images which are correlated Žfor example
A y control A and B y control A . it is only possible to give
an upper bound for the probability of a major overlap. The
upper bound is the equal to the significance level of X i,A .
The probability of overlaps between cluster images
originating from two statistically independent Žcluster.
images can be calculated w117x. Further, one can estimate
the probability that an overlap of a certain volume occurs
by chance w117x. This makes it possible to use intersection
analysis more generally and examine the overlap quantitatively.
5.1. Intersection analysis, examples
Intersection analysis was used by Ledberg and Roland
w93x to demonstrate three somatosensory functional areas
in the parietal operculum, by Roland et al. w117x to identify
a new functional somatosensory area, IPA, lining the
anterior portion of the intraparietal sulcus, and recently by
Gulyas et al. w118x studying color perception in two independent groups of subjects.
The first group of subjects discriminated colored random dot patterns, the control being discrimination of similar black and white random dot patterns w45x. The second
group of subjects detected color changes under isoluminant
Ganzfield conditions, the control being attention to color
changes in a condition in which no such changes occurred
w118x. The third group detected color changes under isoluminant non-Ganzfield conditions. The control was rest,
closed eyes. In addition, visual areas V1 and V2 were
delimited in five postmortem brains. The location and
extent of the cytoarchitectural areas V1 and V2 were
transformed into standard anatomical space w119x.
The intersection of the images showing statistically
significant fields of activation for each of the three groups
revealed major overlaps in V1, V2, the fusiform and
occipital lateral gyri.
Hypotheses about the functional contribution of a field
built on results from a single subtraction paradigm run a
high risk of being short lived. It takes only one experiment
activating the same field Žor more than half the area of the
field. and permitting an alternative interpretation, to kill
such hypotheses. The point about intersection analysis is
exactly that hypotheses about functional contributions of
fields are made on all available information and new
information can easily be added.
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
6. Another example of intersection analysis of short
term and long term memory representations
Although fields are the largest functional elements of
the cortex, this does not imply that fields are modular. In
other words, the appearance of a functional field Žobtained
with PET and fMRI. marks the afferent and intrinsic
synaptic activity, and since the same region may receive
different combinations of active afferents in different brain
tests, it is not guaranteed that all combinations of active
afferents will cover the exact same cortical territory. On
one hand one expects fields appearing during a particular
97
test to be reproducible if that test is repeated in the same
population of subjects. Small, presumably insignificant
differences may appear in field extent if the test is repeated
in another subject population, especially if the tested populations are small. This is likely to be due to the individual
variance in location and extent of fields. Secondly the
exact extent of fields must be determined in an unambiguous and unique manner w80,81x, which has first become
possible at this moment. For this reason, there are no data
available analyzed rigorously.
Important criteria for field reproducibility are the localization i.e. center of gravity and the extent of the field.
Fig. 6. ŽA. One of the patterns to be learned. ŽB. The seven targets to be kept in working memory.
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
98
Since the extent of the field depends on the threshold used
in generating the statistically significant fields from the
statistical image Žoften a t-image., fields from two different groups of subjects are comparable only if the same
thresholds are used and the material in other respects is
comparable. With these reservations, we have re-analyzed
two materials with an intersection analysis, that of motor
preparation for pointing w120x, and that of recall of learned
geometrical patterns w47x.
With the purpose of localizing the cortical fields in the
human brain representing a spatial visual pattern being
kept in working memory w120x, as well as those representing recall of previously learned geometrical patterns w47x,
the rCBF was measured in two groups of 10 normal
healthy, male volunteers with positron emission tomography ŽPET.. The first group did a delayed pointing task
which required the targets for the pointing to be kept in
working memory in the delay period between target exposure and actual pointing. The PET measurement was made
in this delay period. The second group learned 10 large
colored, stationary field visual patterns over a period of 1
h, and were then requested to recall all learned patterns
during a PET measurement. The only common denominators of these two tasks were that the stimuli were colored,
stationary, visual patterns occupying a large part of the
visual field ŽFig. 6., and that internal visual representations
were necessary to solve the tasks.
Control experiments were run to make sure the subjects
were keeping the targets in their short term memory and
were able to point to the targets. Similarly that the second
group actually had learned the 10 patterns and were able to
imagine these.
Table 1 and Fig. 7 describe the results. Obviously the
two tasks engaged partly the same fields in the posterior
parietal cortex despite the different nature of the tasks. If
one applies the criterion of the center of gravity Žc.o.g.. of
the field in motor preparation and pattern recall, the c.o.g.
of the field in the left posterior superior parietal cortex and
the c.o.g. of the field in the left angular gyrus in pattern
recall as well as in preparation for pointing were all within
the two overlap areas ŽFig. 7.. However, as can be seen
from the volume of overlap, this was in both cases less
than 50% of the smallest field ŽTable 1.. The large field in
the left angular gyrus appearing in preparation for pointing
undoubtedly consists of several functional fields. This
indicates that although the fields overlapped considerably,
the underlying synaptic active populations differed slightly
in the top regions of left angular gyrus and the posterior
part of the superior parietal lobule. On the other hand, the
intersection of fields indicates that the working memory of
large visual field images and the recall of large visual field
geometrical patterns shares synaptic and neuronal populations in the top region of the angular gyrus and the
posterior extreme of the superior parietal lobule.
One could argue that these consistent activations may
be due to the task instructions being kept in a working
memory rather than to internal visual representations.
However, in the group keeping targets in working memory
the instruction was to use the targets for subsequent pointing, in the recall group the instruction was to recall already
learned patterns. Also with respect to the internal representations the instructions differed: the first group of subjects
were requested to keep one single large visual field image
showing spatial relations between targets in mind until
they had completed their pointing. The second group was
requested to recall 10 different geometrical patterns repetitively, which they did with a frequency of one per 6 s.
Thus the instructions for the two tasks differed in almost
all respects and neither task activated any areas associated
with language representations. Each task was associated
with other increases in the parietal and temporal and
frontal lobe. These activations might well be also associated with representations of the internal visual images
indicating that the angular gyrus and the superior parietal
lobule were not the cortical fields exclusively representing
the spatial and geometrical images. However, the only
common denominator of the two tasks was the internal
visual representations of large visual field, complex, stationary colored patterns, and this consistently activated the
angular gyrus, and the superior parietal lobule. For these
reasons we conclude the rCBF increases in the angular
gyrus, the superior parietal lobule, and the posterior inferior temporal cortex to reflect increased synaptic activity
associated with recall and working memory of internal
representations of the large field patterns. Areas corresponding to the location of these fields have been activated
Table 1
Localization and extent of parietal fields in preparation for pointing, recall of geometrical patterns and the intersection of these fields
Preparation for pointing
Left intraparietal post.
Left post. superior parietal
Left angular gyrus
Right intraparietal post.
Recall of pattern
Preparation l recall
x
y
z
mm3
x
y
z
mm3
x
y
z
mm3
39
19
52
y38
y63
y71
y46
y59
34
44
39
42
1070
2220
7780
1950
45
17
50
y39
y67
y73
y47
y58
y41
39
35
39
1150
1810
1860
830
44
17
49
y38
y65
y71
y46
y57
38
43
40
42
180
580
620
60
post.: posterior.
Coordinates are centers of gravity in the HBA w119x and Talairach coordinate system w140x. The statistical images of preparation and recall were
thresholded at t s 2.00 and cluster size 800 mm3. The mean t values of the above regions in all cases ) 2.7.
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
99
Fig. 7. Left shows the parietal lobe activations of the 10 subjects recalling geometrical patterns from their long-term memory. Middle shows the parietal
activations of the eight subjects keeping the target field in their working memory. Right shows the intersection between the two cluster images.
when subjects were recalling full visual field scenes in
color from long term memory w121x.
This experiment thus was inconclusive with respect to
whether the field extent was reproducible. One reason
could be that the measurements were done on two different
PET scanners, and the data transformed from the format of
the atlas of Greitz et al. w122x to the Human Brain Atlas
ŽHBA.. More compelling data showing major overlaps and
centers of gravity included in intersections were produced
by Ledberg et al. w123x and Roland et al. w117x. However,
only the application of an unambiguous mathematical description of the extent and localization of fields w81,82x
will resolve the issue of whether fields are exactly reproducible in shape extent and localization of c.o.g.
7. Formulation of the structural–functional problem in
cortical parcellation
As stated, a cortical area is defined by multiple criteria
on the microstructural domain. For the cerebral cortex of
man, one important microstructural criterion is — for
practical reasons — missing, that of connectivity. Cortical
areas defined by microstructure vary in localization, extent
and topography among individuals. In the functional domain, changes in activity of the cerebral cortex manifest
themselves as functional fields. Similarly, the localization
and extent of functional fields vary among individuals.
Functional domains with particular functional contributions may be isolated by field intersection analysis. The
results of the microstructural analysis are population maps
indicating the probability that all members of the population are represented by a particular area, defined by quantitative microstructural analysis. Similarly, the localization
and extent of the functional fields and functional domains
must be given in the form of population maps. Although it
is usually assumed that there is a close spatial correspondence between the extent of functional fields and microstructurally defined cortical areas, data supporting this
are sparse w124x. Therefore the correspondence between
functional domains and cortical areas is only a hypothesis,
and as such it must be tested experimentally. Depending
on which probability of occurrence one settles for Že.g.,
0.5, 0.75, 0.95., the hypotheses of equiterritorial microstructural and functional domains can be tested region
by region for different microstructural modalities andror
combinations thereof. For example V2 defined by cyto-,
myelo and receptor differences to neighboring regions is
an example of an area defined by multiple microstructural
criteria. The intersection analysis allows functional domains to be described as intersections of fields originating
from multiple subtractions, or multiple correlations with
other fields or with external vectors, for example psychophysical variables. Since the functional anatomical
techniques of PET and fMRI in particular show changes in
physiological and biochemical variables related to afferent
and intrinsic synaptic activity of the field, these methods
should be complemented with electrophysiological recordings such as EEG, MEG to investigate the spatial concordance of changes in electrical, physiological and biochemical variables.
8. The interference principle
If it is so that neurons in the cerebral cortex activate in
large distinct populations or fields, and if the field in
essence consists of neurons performing cooperative computation as described in Section 4, it follows that two
different brain tasks requiring the participation of one or
several identical fields cannot be performed simultaneously. This is the interference principle: ‘‘If two different
brain tasks make use of one or several identical fields, they
cannot be performed simultaneously. From knowledge of
field activations one can predict which types of tasks will
interfere’’ w2x.
The interference of two tasks thus is dependent on
whether fields occupy the same neuronal and synaptic
territory and since all neurons within the field are, under
the strong version of the CFH, one way or another engaged
100
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
in the local computation, the neurons within the field
cannot simultaneously be engaged in two tasks. For example, if fields executing brain task A totally overlap one or
more fields engaged in executing brain task B, tasks A and
B will interfere. It is important to bethink that one or more
major overlaps might be sufficient to provoke an interference w125x. The interference principle is fundamental for
the field activation hypothesis. If two tasks can be executed simultaneously without any increase in reaction time
or number or errors, despite that the two tasks rely on one
or more identical or mainly identical fields, the strong
version of the CFH fails. On the other hand, if the neurons
within the fields perform cooperative computation and
major overlaps occur between fields belonging to different
brain tasks, this is a limitation of the parallel processing in
the brain. The question thus is not whether tasks containing almost exactly the same perceptual and information
processing constraints will interfere, for example, two
color discrimination tasks, but is more general and pertain
to tasks which share one or more fields.
So far, the interference principle has been examined in
three studies. In one study Passingham w126x found interference, i.e. increase in reaction times for two tasks both
engaging the supplementary motor area and between two
other tasks both engaging the superolateral prefrontal cortex and the anterior cingulate cortex. Since working memory of material from different sensory modalities activates
almost identical fields in the superolateral part of the
prefrontal cortex w102x, one would expect that two working
memory tasks performed simultaneously would interfere.
Indeed this was so, an auditory working memory task and
a visual working memory task when performed simultaneously increased the reaction times and the error rates,
compared to conditions in which the two tasks were
performed separately w125x. In this example, the interference was associated with multiple overlaps, a few of these
were major overlaps in the prefrontal cortex and anterior
cingulate cortex and inferior parietal lobule.
These studies however, did not address the question of
whether simultaneous performance of two tasks which
interfere would be associated with activations of additional
functional fields. For example that the brain during dual
task performance would recruit fields in excess of those
engaged in each of the two tasks. If this was so it would
indicate that the brain had alternative and in some sense
parallel modes of solving dual task performance. The same
auditory working memory task was then performed single
in one PET scan, the visual working memory task was
performed single in another PET scan, and in a third
condition both working memory task were performed simultaneously during PET scans. The fields associated with
dual task performance were the spatial sum of the fields
present in the single task conditions w103x. In other words,
no additional fields appeared during dual task performance. Thus the brain does not always — or in general
perform parallel processing. Non-parallel processing, ex-
emplified by partial and major overlaps of fields and
presumably partial and major sharing of neuronal population is associated with limitations in performance.
9. Parallel processing in the human brain
A prerequisite of a demonstration of parallel processing,
under the strong version of the CFH, is to have two tasks
and show that the fields executing the brain part of the
tasks would be non-overlapping. This, as repetitively argued in this report, requires a mathematically unambiguous
measure of field shape and extent and a high sensitivity of
the statistical methods. A further requirement is that the
neurons at one set of fields increase their firing rate or
synchronize exclusively in the one task, whereas the neurons at another set of fields increase their firing rate or
synchronize exclusively at the other task. Thus a demonstration of parallel processing in the primate and human
brain requires an experimental effort which is not yet
realistic.
Using functional imaging techniques, researchers have
shown that the point of maximal changes Žin blood flow or
BOLD signal. occur at disparate locations for the visual
processing of motion and color w104x, the visual processing
of faces and spatial relations w127–129x, the visual attention to shape, color and velocity w130x, and the visual
discrimination of shape from color and shape from motion
w131x. What has really been demonstrated by these studies
is that certain loci, exemplified by the coordinates of
points of maximal intensity, centers of gravity or clusters,
have been more active in one condition Že.g. color. than
another Že.g. motion. and vice versa. This, of course, is not
identical to having proven that nowhere in the brain are the
same neurons responding to both color and visual motion.
One may argue that it might be that one population of
neurons is responding in tasks of color perceptionrdiscrimination and motion perceptionrdiscrimination, but in
these instances the responses, although task related, are not
related to the color processing or motion processing per se.
This might well be so, but in that case the suggestions for
parallel processing are limited to the visual cortices, and
do not pertain to the question of parallel processing in the
brain in general. The proposal of parallel processing in
visual cortices relies on neuroanatomical and neurophysiological data from non-human primates as well w132–135x.
To our knowledge there has been no study of parallel
processing within the auditory and somatosensory cortices.
This makes the hypotheses of parallel processing, with a
few prefrontal exceptions, w136,137x to rely mainly on
visual data.
9.1. A case for parallel processing in the somatosensory
cortex
When feeling surfaces and objects tactually one can
distinguish two domains: microgeometry and macrogeome-
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
try. A surface may contain small deviations and vaults
which cannot be singled out, but give the impression of a
texture. This is microgeometry. At a slightly larger scale
one can begin to single out local surface curvatures, single
vaults, edges; and at an even larger scale single objects.
This is macrogeometry. The border between microgeometry and macrogeometry is at a scale around some 500 mm
between the deviations w96x. In man, lesions of the parietal
operculum interfere with the discrimination of microgeometry Žroughness, texture., whereas lesions of the cortex in
the anterior parts of the parietal lobules interfere with
discrimination of macrogeometry Žcurvature, shape, length.
w79x. This indicates that the somatosensory cortices necessary for normal tactile perception of microgeometry differ
from those necessary for normal perception of macrogeometry. In addition Ledberg et al. w123x found the lateral
parietal operculum more activated by roughness discrimination than by length discrimination. This prompted us to
examine a possible difference in functional contributions
of the somatosensory association area located in the parietal operculum ŽLPO. and that or those areas the anterior
parts of the parietal lobules.
The rCBF was measured with positron emission tomography ŽPET. in two independent groups of normal volunteers using their right hand for all tests. One group discriminated the shapes of a series of ellipsoids w96x. The rCBF
was measured also in a control state in which the subjects
were motionless and fixating on a cross. The subjects in
the other group were blindfolded and asked to discriminate
the length of smooth cylinders with their right hand, and in
a separate rCBF measurement they discriminated the
roughness of macrogeometrically similar but rough cylinders. The rCBF was also measured in a motor control state
in which the subjects reproduced the same scanning movements, but without any cylinder to feel. All tasks were
matched for tactile contact time and task difficulty, measured as percent correct.
The LPO’s were significantly more activated in discrimination of roughness than in the discrimination of shape.
No fields posterior to the postcentral sulcus were activated
in the comparison of roughness with length or shape. A
Boolean intersection on the cluster images showing the
fields more activated in roughness than length and roughness than shape wi.e. ŽrCBF roughy rCBF length. l ŽrCBF
rough y rCBF shape.x revealed that the left LPO was
significantly and consistently activated more by discrimination of microgeometry than by discrimination of macrogeometry Žlength and shape. w117x.
The IPA, on the contrary, was significantly more activated both when length discrimination Žlengthy roughness.
and when shape discrimination Žshape y roughness. was
compared to roughness discrimination. This was the only
area which was significantly more activated in both length
and shape discrimination compared to roughness discrimination as demonstrated by the Boolean intersection Žshape
y roughness l lengthy roughness.. Furthermore the IPA
101
Fig. 8. Projections on the surface of the standard brain of the IPA region
and LPO region. The figure shows the extent of the intersection between
shape-roughness l length-roughness, shape-control l length-control for
the IPA, and the intersection roughness-length l roughness-shape for the
LPO w117x.
was the only area which was consistently and significantly
more activated in discrimination of macrogeometry both
when compared to discrimination of microgeometry and
when compared to the different controls Ži.e. shape y
roughness l lengthy roughness l shape y control l
length y control. w117x.
Thus a double dissociation of function exists between
LPO and IPA in the activation by roughness on one side
and length and shape on the other ŽFig. 8.. Although the
precise mechanisms underlying somatosensory form perception and roughness perception are still unknown, the
differences in functional contributions of the LPO and IPA
indicate that separate processing streams exist for the
different somatosensory submodalities of microgeometry
and macrogeometry. Parallel processing of visual submodalities is likely within the somatosensory areas and
visual areas respectively. Indications of parallel processing
between modalities may also exist. For example, the somatosensory processing of shape is segregated from the
visual processing of shape in the respective modality specific areas — even when tactile shapes must be matched
with visual shapes and vice versa w138x.
10. Conclusions
1. Microstructural parcellation of the human cerebral cortex must be made on the basis of: Quantitative measurements and multiple criteria. Such microstructurally defined areas appear as population maps.
2. The organisation of cerebral cortex into functional
columns may apply for koniocortices — elsewhere the
102
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
active synaptic populations take the form of larger
blobs or patches. The columns, patches or blobs combine in cortical functional fields, postulated to be the
largest functional entities of the cerebral cortex w1,2x.
Multiple fields provide the functional contributions of
the cerebral cortex.
3. The functional parcellation of the human cerebral cortex must be based on field-population maps, at best
from intersection analysis across groups.
4. There is a separation of functional contributions of the
lateral parietal operculum ŽLPO. and the cortex lining
the anterior part of the intra-parietal sulcus ŽIPA.. The
LPO engaging in somatosensory analysis of microgeometry, the IPA engaging in analysis of macrogeometry. This indicates that parallel processing streams may
appear in the somatosensory cortices. Further observations suggest that parallel processing may also exist
between the somatosensory and visual modalities.
5. There is now some experimental evidence for the interference principle: If two brain tasks make use of one or
several identical Žor major overlapping. fields, they
cannot be performed simultaneously. The brain is not
capable of parallel processing in all circumstances.
Non-parallel processing, exemplified by partial and major overlaps in engaged neuronal and synaptic populations are associated with limitations in performance.
Acknowledgements
This work was supported by EU Biotech programme
and the Swedish Medical Research Council. We appreciate
the help of Jonas Larsson, Anders Ledberg, Torkel Klingberg, Katrin Amunts and Aleksandar Malikovic who kindly
provided some of the figures and data.
References
w1x P.E. Roland, Application of imaging of brain blood flow to behavorial neurophysiology: The cortical field activation hypothesis, in: L.
Sokoloff ŽEd.., Brain Imaging and Brain Function, Raven, New
York, 1985, pp. 87–106.
w2x P.E. Roland, Brain Activation, Wiley, New York, 1993.
w3x V.B. Mountcastle, An oranizing principle for cerebral function: the
unit module and the distributed system, in: The Mindful Brain:
Cortical Organisation and the Group-Selective Theory of Higher
Brain Function, The MIT Press, Cambridge, 1978, pp. 7–50.
w4x C. von Economoo, G.N. Kosknina, Die cytoarchitektonik der hirnrinde des erwachsenen menschen, Springer, Wien, 1925.
w5x A.W. Campbell, Histological Studies on the Localization of Cerebral Function, University Press, Cambridge, 1905.
w6x G.E. Smith, A new topographical survey of the human cerebral
cortex, being an account of the distribution of the anatomically
distinct cortical areas and their relationship to the cerebral sulci, J.
Anat. 41 Ž1907. 237–254.
w7x K. Brodmann, Neue Ergebnisse uber
vergleichende histologische
¨
Lokalisation der Grosshirnrinde mit besonderer Berucksichtigung
¨
des Stirnhirns, Anat. Anz. 41 Ž1912. 157–216.
w8x P. Bailey, G. von Bonin, The Isocortex of Man, University of
Illinois Press, Urbana, 1951.
¨
w9x I.N. Filimonoff, Uber
die Variabilitat
¨ der Grosshirnrindenstruktur.
Mitteilung II: Regio occipitalis beim erwachsenen Menschen, J.
Psychol. Neurol. 44 Ž1932. 1–96.
w10x S.S. Stensaas, D.K. Eddington, W.H. Dobelle, The topography and
variability of the primary visual cortex in man, J. Neurosurg. 40
Ž1974. 747–755.
w11x P.E. Roland, K. Zilles, Brain Atlases — A new research tool,
Trends Neurosci. 17 Ž1994. 458–467.
w12x K. Zilles, A. Schleicher, Cyto- and myeloarchitecture of human
visual cortex and the periodical GABA A receptor distribution, in B.
Gulyas, D. Ottoson, P.E. Roland ŽEds.., Functional Organisation of
the Human Visual Cortex, Pergamon Press, Oxford, 1993, pp.
111–121.
w13x S. Geyer, A. Ledberg, A. Schleicher, S. Kinomura, T. Schormann,
J. Larsson, U. Simon, K. Zilles, P.E. Roland, Two different areas
within the primary motor cortex of man, Nature 282 Ž1996. 805–
807.
w14x P.E. Roland, K. Zilles, The developing European computerized
human brain database for all imaging modalities, NeuroImage 4
Ž1996. S39–S47.
w15x U. Simon, S. Geyer, K. Zilles, T. Schormann, A. Dabringhaus, A.
Schleicher, P.E. Roland, Architectonic and receptor autoradiographic mapping of the human primary somatosensory cortex,
Human Brain Mapping 1 ŽSuppl.. Ž1995. 259.
w16x K. Zilles, M. Qu,
¨ A. Schleicher, Regional distribution and heterogeneity of a-adrenoceptors in the rat and human central nervous
systeim, J. Hirnforschung 34 Ž1993. 123–132.
w17x K. Zilles, G. Schlaug, S. Geyer, P.E. Roland, Mapping of human
supplementary and cingulate motor areas, Human Brain Mapping 1
Žsuppl.. Ž1995. 285.
w18x K. Zilles, G. Schlaug, S. Geyer, G. Luppino, M. Matelli, M. Qu,
¨ A.
Schleicher, T. Schormann, Anatomy and transmitter receptors of
the supplementary motor areas in the human and non-human
ŽEd.., Adv. Neuro. vol 70, Suppleprimate brain, in H.O. Luders
¨
mentary sensorimotor area, Lippincott-Raven, Philadelphia, 1995,
pp. 29–43.
w19x K. Zilles, G. Schlaug, M. Matelli, G. Luppino, A. Schleicher, M.
Qu,
¨ A. Dabringhaus, R.J. Seitz, P.E. Roland, Mapping of human
and macaque sensorimotor areas by integrating architectonic, transmitter receptor, MRI and PET data, J. Anat. 187 Ž1995. 515–537.
w20x A. Schleicher, K. Zilles, A. Wree, A quantitative approach to
cytoarchitectonics: Software and hardware aspects of a system for
the evaluation and analysis of structural inhomogeneities in nervous tissue, J. Neurosci. Meth. 18 Ž1986. 221–235.
w21x A. Schleicher, K. Amunts, S. Geyer, U. Simon, K. Zilles, P.E.
Roland, A method of observer-independent cytoarchitectonic mapping of the human cortex, Human Brain Mapping 1 Žsuppl.. Ž1995.
77.
w22x T. Schormann, A. Dabringhaus, K. Zilles, Statistics of deformations in histology and application to improved alignment with MRI,
IEEE Transactions on Medical Imaging 14 Ž1995. 25–35.
w23x T. Schormann, S. Henn, K. Zilles, A new approach to fast elastic
alignment with applications to human brains, Lect. Notes Comput.
Sci. 1131 Ž1996. 337–342.
w24x Kononova, in: S.A. Sarkisov, J.N. Filimonov, and N.S. Preobrashenskoya ŽEds.., Cytoarchitecture of the human cortex cerebri,
Medgiz, Moscow, 1949, pp. 236–336.
w25x G. Rajkowska, P.S. Goldman-Rakic, Cytoarchitectonic definition
of prefrontal areas in the normal human cortex: I. Remapping of
areas 9 and 46 using quantitative criteria, Cereb. Cortex 5 Ž1995.
307–322.
w26x G. Rajkowska, P.S. Goldman-Rakic, Cytoarchitectonic definition
of prefrontal areas in the normal human cortex: II. Variability in
locations of areas 9 and 46 and relationship to the Talairach
coordinates system, Cereb. Cortex 5 Ž1995. 323–337.
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
w27x P.T. Fox, S. Mikiten, G. Davis, J.L. Lancaster, in: R.W. Thatcher,
M. Hallett, T. Zeffiro, R.E. John ŽEds.., For functional neuroimaging: Technical foundations, Academic Press, New York, 1994, pp.
259–262.
w28x T. Paus, N. Otaky, Z. Caramanos, D. Macdonald, A. Zijdenbos, D.
D’Avirro, D. Gutmans, C. Holmes, F. Tomaiuolo, A. Evans, In
vivo porphometry of the intrasulcal gray matter in the human
cingulate paracingulate, and superior-rostral sulci: Hemipheric
asymmetries, gender differences and probability maps, J. Comp.
Neurol. 376 Ž1996. Ž1996. 664–673.
w29x P.E. Roland, S. Geyer, K. Amunts, T. Schormann, A. Schleicher,
A. Malikovic, K. Zilles, Cytoarchitectural maps of the human brain
in standard anatomical space, Human Brain Mapping 5 Ž1997.
222–227.
w30x A. Schleicher, K. Zilles, A quantitative approach to cytoarchitectonics: Analysis of structural inhomogeneities in nervous tissue
using an image analyser, J. Microscopy 157 Ž1990. 367–381.
w31x W.J. Dixon, BMDP statistical software manual, University of
California Press, Berkeley, 1988.
w32x T. Schormann, M. von Matthey, A. Dabringhaus, K. Zilles, Alignment of 3-D brain data sets originating from MR and histology,
Bioimaging 1 Ž1993. 119–128.
w33x P.E. Roland, C.J. Graufelds, J. Wahlin,
L. Ingelman, M. Anders˚
˚
son, A. Ledberg, J. Pedersen, S. Akerman,
A. Dabringhaus, K.
Zilles, Human Brain Atlas: For high resolution functional and
anatomical mapping, Human Brain Mapping 1 Ž1994. 173–184.
w34x V.B. Mountcastle, T.P.S. Powell, Central nervous mechanisms
subserving position sense and kinesthesis, Bull. John Hopkins
Hosp. 105 Ž1959. 173–200.
w35x D.H. Hubel, T.N. Wiesel, Receptive fields, binocular interaction
and funcional architecture in the cat’s visual cortex, J. Physiol.
ŽLondon. 160 Ž1962. 106–154.
w36x C. Kennedy, O. Sakurada, M. Shinohara, J. Jehle, L. Sokoloff,
Local cerebral glucose utilization in the normal conscious macaque
monkey, Ann. Neurol. 4 Ž1978. Ž1978. 293–301.
w37x S.L. Juliano, P.J. Hand, B.L. Whitsel, Pattern of increased metabolic
activity in somatosensory cortex of monkeys, Macaca fascicularis,
subjected to controlled cutaneous stimulation-A 2-deoxyglucos
study, J. Neurophys. 46 Ž1981. 1260–1284.
w38x P.S. Goldman-Rakic, L.D. Selemon, M.L. Schwartz, Dual pathways connecting the dorsolateral prefrontal cortex with the hippocampal formation and parahippocampal cortex in the rhesus
monkey, Neuroscience 12 Ž1984. 719–743.
w39x D. Purves, D.R. Riddle, A.S. LaMantia, Iterated patterns of brain
circuitry Žor how the cortex gets its spots, Trends Neurosci. 15
Ž1992. 362–368.
w40x P.E. Roland, L. Eriksson, L. Widen,
´ S. Stone-Elander, Changes in
regional cerebral oxidative metabolism induced by tactile learning
and recognition in man, Eur. J. Neurosci. 1 Ž1989. 3–18.
w41x P.E. Roland, G. Blomqvist, R.J. Seitz, S. Stone-Elander, E.
Schwenner, H. Kraft, C. Halldin, H. Boshagen,
Physiological mod¨
ulation of Ca2q channels in the living brain studied with PET, Soc.
Neurosci. Abstr. 16 Ž1990. 1275.
w42x P.E. Roland, G. Blomqvist, R.J. Seitz, S. Stone-Elander, E.
Schwenner, H. Kraft, C. Halldin, H. Boshagen,
In vivo function of
¨
the Ca2q-L-channel in man, Eur. J. Neurosci., ŽSuppl.. 3 Ž1990.
262.
w43x B. Gulyas, P.E. Roland, Cortical fields participating in form and
color discrimination in the human brain, NeuroReport 2 Ž1991.
585–588.
w44x B. Gulyas, P.E. Roland, Binocular disparity detection in human
visual cortex: Functional anatomy by positron emission tomography, Proc. Natl. Acad. Sci. USA 91 Ž1994. 1239–1243.
w45x B. Gulyas,
´ P.E. Roland, Processing and analysis of form, colour,
and binicular disparity in the human brain: functional anatomy by
positron emission tomography, Eur. J. Neurosci. 6 Ž1994. 1811–
1828.
103
w46x P.E. Roland, B. Gulyas,
´ Visual imagery and visual representation;
Visual representations of scenes and objects: retinotopical or nonretinotopical? Trends Neurosci. 17 Ž1994. 281–297.
w47x P.E. Roland, B. Gulyas, Visual memory, visual imagery and visual
recognition of large field patterns by the human brain. Functional
anatomy, by positron emission tomography, Cerebr. Cortex 5 Ž1995.
79–93.
w48x W.N. Ross, Changes in intracellular calcium during neuron activity, Annu. Rev. Physiol. 51 Ž1989. 491–506.
w49x R. Yuste, L.C. Katz, Control of postsynaptic Ca2q influx in
developing neocortex by excitatory and inhibitory neurotransmitters, Neuron 6 Ž1991. 333–344.
w50x R. Yuste, D.W. Tank, D. Kleinfeld, Funcitonal study of the rat
cortical microcircuitry with voltage-sensitive dye imaging of neocortical slices, Cereb. Cortex 7 Ž1997. 546–558.
w51x T. Sawaguchi, Modular activation and suppression of neocortical
activity in the monkey revealed by optical imaging, NeuroReport 6
Ž1994. 185–189.
w52x L. Sokoloff, M. Reivich, C. Kennedy, M.H. Des Rosiers, C.S.
Patlak, K.D. Pettigrew, O. Sakurada, M. Shinohara, The w 14 Cx
deoxyglucose method for the measurement of local cerebral glucose utilisation: Theory, procedure, and normal values in the
conscious and anesthetized albino rat, J. Neurochem. 28 Ž1977.
897–916.
w53x L. Sokoloff, Localization of functional activity in the central
nervous sytem by measurement of glucose utilization with radioactive deoxyglucose, J. Cereb. Blood Flow Metab. 1 Ž1981. 7–36.
w54x S.L. Juliano, P.J. Hand, B.L. Whitsel, Patterns of metabolic activity
in cytoarchitectural area SII and surrounding cortical fields of the
monkey, J. Neurophys. 50 Ž1983. 961–980.
w55x P.S. Goldman-Rakic, Modular organization of prefrontal cortex,
Trends Neurosci. 7 Ž1984. 419–424.
w56x H.R. Friedman, P.S. Goldman-Rakic, Coactivation of prefrontal
cortex and inferior parietal cortex in working memory tasks revealed by 2DG functional mapping in the rhesus monkey, J.
Neurosci. 14 Ž1994. 2775–2788.
w57x G. Wang, K. Tanaka, M. Tanifuji, Optical imaging of functional
organisation in the monkey interotemporal cortex, Science 272
Ž1996. 1665–1668.
w58x M.M. Haglund, G.A. Ojemann, D.W. Hochman, Optical imaging
of epileptiform and functional activity in human cerebral cortex,
Nature 358 Ž1992. 668–671.
w59x A.P. Georgopoulos, J.F. Kalaska, R. Caminiti, J.T. Massey, On the
relations between the direction of two-dimensional arm movements
and cell discharge in primate motor cortex, J. Neurosci. 2 Ž1982.
1527–1537.
w60x A.P. Georgopoulos, J.F. Kalaska, M.D. Crutcher, R. Caminiti, J.T.
Massey, The representation of movement direction in the motor
cortex: Single cell and population studies, in: G.M. Edelman, W.E.
Gall, W.M. Cowan ŽEds.., Dynamic aspets of neocortical function,
Wiley Interscience, New York, 1984, pp. 501–524.
w61x J.F. Kalaska, R. Caminiti, A.P. Georgopoulos, Cortical mechanisms related to the direction of two-dimensional arm movements:
relations in parietal area 5 and comparison with motor cortex, Exp.
Brain Res. 51 Ž1983. 247–260.
w62x J.F. Kalaska, D.A.D. Cohen, M. Prud’Homme, M.L. Hyde, Parietal
area 5 neuronal activity encodes movement kinematics, not movement dynamics, Exp. Brain Res. 80 Ž1990. 351–364.
w63x J.F. Kalaska, D.J. Crammond, Deciding not to Go: Neuronal
correlates of response selection in a GOrNOGO task in primate
premotor and parietal cortex, Cereb. Cortex 5 Ž1995. 410–428.
w64x P.R. Roelfsema, A.K. Engel, P. Konig,
W. Singer, Visuomotor
¨
integration is associated with zero time-lag synchronization among
cortical areas, Nature 385 Ž1997. 157–161.
w65x J.S. Lund, T. Yoshioka, J.B. Levitt, Comparison of intrinsic connectivity in different areas of macaque monkey cerebral cortex,
Cereb. Cortex 3 Ž1993. 148–162.
104
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
w66x M.L. Pucak, J.B. Levitt, J.S. Lund, D.A. Lewis, Patterns of intrinsic and associational circuitry in monkey prefrontal cortex, J.
Comp. Neurol. 376 Ž1996. 614–630.
w67x L. Griffin, The intrinsic geometry of the cerebral cortex, J. Theor.
Biol. 166 Ž1994. 261–273.
w68x S.H.C. Hendry, H.D. Schwark, E.G. Jones, J. Yan, Numbers and
proportions of GABA-immunoreactive neurons in different areas of
monkey cerebral cortex, J. Neurosci. 7 Ž1987. 1503–1519.
w69x J.C. Houk, S.P. Wise, Distributed modular architectures linking
basal ganglia, cerebellum, and cerebral cortex: their role in planing
and controlling action, Cereb. Cortex 5 Ž1995. 95–110.
w70x H. Nakamura, R. Gattass, R. Desimone, L. Ungerleider, The
modular organization of projections from areas V1 and V2 to areas
V4 and TEO in macaques, J. Neurosci. 13 Ž1993. 3681–3691.
w71x K.S. Rockland, Configuration, in serial reconstruction of individual
axons projecting from area V2 to V4 in the macaque monkey,
Cereb. Cortex 2 Ž1992. 353–374.
w72x K.S. Rockland, Morphology of individual axons projecting from
area V2 to MT in the macaque, J. Comp. Neurol. 355 Ž1995.
15–26.
w73x K.S. Saleem, K. Tanaka, K.S. Rockland, Specific and columnar
projection from area TEO to TE in the macaque inferotemporal
cortex, Cereb. Cortex 3 Ž1993. 454–464.
w74x E.G. Jones, S.P. Wise, Size. laminar and columnar distribution of
efferent cells in the sensory-motor cortex of monkeys, J. Comp.
Neurol. 175 Ž1977. 391–438.
w75x G.M. Innocenti, General organisation of callosal connections in the
cerebral cortex, in: E.G. Jones, A. Peters ŽEds.., Cerebral Cortex,
Plenum, New York, 1986, pp. 291–353.
w76x K.S. Rockland, G.W. Drash, Collateralized divergent feedback
connections that target multiple cortical areas, J. Comp. Neurol.
373 Ž1996. 529–548.
w77x C. Cavada, P.S. Goldman-Rakic, Posterior partietal cortex in rhesus
monkey: II. evidence for segregated corticocortical networks linking sensory and limbic areas with the frontal lobe, J. Comp.
Neurol. 287 Ž1989. 422–445.
w78x C.L. Colby, R. Gattass, C.R. Olson, C.G. Gross, Topographical
organization of cortical afferents to extrastriate visual area PO in
the macaque: A dual tracer study, J. Comp. Neurol. 269 Ž1988.
392–413.
w79x P.E. Roland, An anatomical and physiological study of somatosensory detection of microgeometry, macrogeometry and kinaesthesia
in man, Brain Res. Rev. 12 Ž1987. 43–94.
w80x T. Lindeberg, Scale-space theory in computer vision: The kluwer
international series in engineering and computer science, Kluwer,
Dordrecht, 1994.
w81x P. Lidberg, T. Lindeberg, P.E. Roland, Analysis of brain activation
patterns using a three-dimensional scale-space primal sketch, Proceedings SSAB’97, 1997, pp. 61–65.
w82x T. Lindeberg, P. Lidberg, P.E. Roland, Analysis of brain activation
patterns using a 3-D scale-space primal sketch, Human Brain
Mapping Ž1997. Žsubmitted..
w83x J.B. Poline, B.M. Mazoyer, Analysis of individual positron emission tomography activation maps by detection of high signal-tonoise-ratio pixel clusters, J. Cereb. Blood Flow Metab. 13 Ž1993.
425–437.
˚
w84x P.E. Roland, B. Levin, R. Kawashima, S. Akerman,
Three-dimensional analysis of clustered voxels in 15O-butanol brain activation
images, Human Brain Mapping 1 Ž1993. 3–19.
w85x K.J. Friston, K.J. Worsley, R.S.J. Frackowiak, J.C. Mazziotta, A.C.
Evans, Assessing the significance of focal activations using their
spatial extent, Human Brain Mapping 1 Ž1994. 210–220.
w86x K.J. Worsley, S. Marrett, P. Nelin, A.C. Vandal, K.J. Friston, A.C.
Evans, A unified statistical approach for determining significant
signals in images of cerebral activation, Human Brain Mapping 4
Ž1996. 58–73.
w87x T. Shallice, P. Fletcher, C.D. Frith, P. Grasby, R.S.J. Frackowiak,
w88x
w89x
w90x
w91x
w92x
w93x
w94x
w95x
w96x
w97x
w98x
w99x
w100x
w101x
w102x
w103x
w104x
w105x
w106x
w107x
R.J. Dolan, Brain regions associated with acquisition and retrieval
of verbal episodic memory, Nature 368 Ž1994. 633–635.
L.M. Parsons, P.T. Fox, H.J. Downs, T. Glass, T.B. Hirsch, C.C.
Martin, P.A. Jerabek, J.L. Lancaster, Use of implicit motor imagery
for visual shape discrimination as revealed by PET, Nature 375
Ž1995. 54–58.
D.L. Schacter, E. Reiman, A. Uecjer, M.R. Polster, S.Y. Lang,
L.A. Cooper, Brain regions associated with retrieval of structurally
coherent visual information, Nature 376 Ž1995. 587–590.
G. Bottini, R. Corcoran, R. Sterzi, E. Paulesu, P. Schenone, P.
Scarpa, R.S.J. Frackowiak, C.D. Frith, The role of the right hemisphere in the interpretation of figurative aspects of language, A
positron emission tomography activation study, Brain 117 Ž1994.
1241–1253.
R.B.H. Tootell, J.B. Reppas, A.M. Dale, R.B. Look, M.I. Sereno,
R. Malach, T.J. Brady, B.R. Rosen, Visual motion aftereffect in
human cortical area MT revealed by functional magnetic resonance
imaging, Nature 375 Ž1995. 139–141.
K.J. Worsley, A.C. Evans, S. Marrett, P. Nelin, A three-dimensional analysis for CBF activation studies in human brain, J. Cereb.
Blood Flow Metab. 12 Ž1992. 900–918.
A. Ledberg, P.E. Roland, Intersubject variation of cerebral activations in a somatosensorydetection task, NeuroImage 3 Ž1996. S330.
S. Lai, A.L. Hopkins, E.M. Haacke, D. Li, B.A. Wasserman, P.
Buckley et al., Identification of vascular structures as a major
source of signal contrast in high resolution 2D and 3D functional
activation imaing of the motor cortex at 1.5T: Preliminary results,
Magn. Reson. Med. 30 Ž1993. 387–392.
C. Segebarth, V. Belle, C. Delon, R. Massarelli, J. Decety, J.F. Le
Bas, M. Decorps,
A.L. Benabid, Functional MRI of the human
´
brain: predominance of signals from extracerebral veins, NeuroReport 5 Ž1994. 813–816.
P.E. Roland, E. Mortensen, Somatosensory detection of microgeometry, macrogeometry and kinaesthesia in man, Brain Res. Rev.
12 Ž1987. 1–42.
S. Kinomura, R. Kawashima, P.E. Roland, The access of sensory
information to the motor areas in the human brain: A PET study,
Human Brain Mapping 1 ŽSuppl.. Ž1995. 288.
P.E. Roland, B. Larsen, Focal increase of cerebral blood flow
during stereognostic testing in man, Arch. Neurol. 33 Ž1976.
551–558.
B. Gulyas,
´ P.E. Roland, Cortical fields participating in spatial
frequency and orientation discrimination: Functional anatomy by
positron emission tomography, Human Brain Mapping 3 Ž1995.
133–152.
R. Kawashima, B.T. O’Sullivan, P.E. Roland, Closing the ‘Mind’s
Eye’, Proc. Natl. Acad. Sci. USA 92 Ž1995. 5969–5972.
R. Kawashima, P.E. Roland, B.T. O’Sullivan, Functional anatomy
of reaching and visuomotor learning-A positron emission tomography study, Cereb. Cortex 5 Ž1995. 111–122.
T. Klingberg, R. Kawashima, P.E. Roland, Activation of multimodal cortical areas underlies short-term memory, Eur. J. Neurosci.
8 Ž1996. 1965–1971.
T. Klingberg, The neurophysiology of working memory-functional
mapping of the human brain with positron emission tomography,
1997, Thesis.
S. Zeki, D.G. Watson, C.J. Lueck, K.J. Friston, C. Kennard, R.S.J.
Frackowiak, A direct demonstration of functional specialization in
human visual cortex, J. Neurosci. 11 Ž1991. 641–649.
S.E. Petersen, P.T. Fox, M.I. Posner, M.A. Mintun, M.E. Raichle,
Positron emission tomographic studies of the cortical anatomy of
single-word processing, Nature 331 Ž1988. 585–589.
C.J. Lueck, S. Zeki, K.J. Friston, M.-P. Deiber, P. Cope, V.J.
Cunningham, A.A. Lammertsma, C. Kennard, R.S.J. Frackowiak,
The colour centre in the cerebral cortex of man, Nature 340 Ž1989.
386–389.
P.E. Roland, R. Kawashima, B. Gulyas, B.T. O’Sullivan, Positron
P.E. Roland, K. Zillesr Brain Research ReÕiews 26 (1998) 87–105
w108x
w109x
w110x
w111x
w112x
w113x
w114x
w115x
w116x
w117x
w118x
w119x
w120x
w121x
w122x
w123x
w124x
emission tomography in cognitive neuroscience: Methodological
constraints, strategies and examples from learning and memory, in:
M. Gazzaniga ŽEd.., The Cognitive Neurosciences, MIT Press,
1995, pp. 781–788.
K.J. Friston, C.J. Price, P. Fletcher, C. Moore, R.J. Frackowiak,
R.J. Dolan, The trouble with cognitive subtraction, NeuroImage 4
Ž1996. 97–104.
B. Gulyas, J. Larsson, K. Amunts, K. Zilles, P.E. Roland, Cortical
regions in the human brain systematically participating in the
processing and analysis of colour, NeuroImage 5 Ž1997. S2.
D.J. Felleman, D.C. van Essen, Distributed hierarchical processing
in the primate cerebral cortex, Cerebr. Cortex 1 Ž1991. 1–47.
M.P. Young, Objective analysis of the topological organisation of
the primate cortical visual system, Nature 358 Ž1992. 152–154.
W. Singer, Synchronization of cortical activity and its putative role
in information processing and learning, Annu. Rev. Physiol. 55
Ž1993. 349–374.
M.E. Phelps, J.C. Mazziotta, D.E. Kuhl, M. Nuwer, J. Packwood,
J. Metter, J. Engel, Tomographic mapping of human cerebral
metabolism: Visual stimulation and deprevation, Neurology 31
Ž1981. 517–529.
B. Larsen, E. Skinhoj,
¨ N.A. Lassen, Variations in regional cortical
blood flow in the right and left hemishperes during automatic
speech, Brain 191 Ž1978. 192–209.
E. Matthew, P. Andreason, R.E. Carson, P. Herscovitch, K. Pettigrew, R. Cohen, C. King, C.E. Johanson, S.M. Paul, Reproducibility of resting cerebral blood flow measurements with H 15
2 0 positron
emission tomography in humans, J. Cereb. Blood Flow Metab. 13
Ž1993. 748–754.
C.J. Price, K.J. Friston, Cognitive conjunction: A new approach to
brain activation experiments, NeuroImage 5 Ž1997. 261–270.
P.E. Roland, B. O’Sullivan, R. Kawashima, A. Ledberg, Shape and
roughness activate different somatosensory areas in the human
brain, Proc. Natl. Acad. Sci. USA Ž1997. in press.
B. Gulyas,
´ J. Larsson, P.E. Roland, Cortical areas in the human
brain engaged by colour and luminance detection, Human Brain
Mapping 1 ŽSuppl.. Ž1995. 56.
P.E. Roland, C.J. Graufelds, J. Wahlin,
L. Ingelman, M. Anders˚
˚
son, A. Ledberg, J. Pedersen, S. Akerman,
A. Dabringhaus, K.
Zilles, Human Brain Atlas: For high resolution functional and
anatomical mapping, Human Brain Mapping 1 Ž1994. 173–184.
J. Decety, R. Kawashima, B. Gulyas,
´ P.E. Roland, Preparation for
reaching: A PET study of the participating structures in the human
brain, NeuroReport 3 Ž1992. 761–764.
P.E. Roland, L. Eriksson, S. Stone-Elander, L. Widen,
´ Does mental
activity change the oxidative metabolism of the brain?, J. Neurosci.
7 Ž1987. 2373–2389.
T. Greitz, C. Bohm, S. Holte, L. Eriksson, A computerized brain
atlas: Construction, anatomical content, and some applications, J.
Comput. Assist. Tomogr. 15 Ž1991. 26–38.
A. Ledberg, B.T. O’Sullivan, S. Kinomura, P.E. Roland, Somatosensory activations of the parietal operculum of man, A PET study,
Eur. J. Neurosci. 7 Ž1995. 1934–1941.
K.A. Macko, C.D. Jarvis, C. Kennedy, M. Miyaoka, M. Shinohara,
L. Sokoloff, M. Mishkin, Mapping the primate visual system with
Ž2- 14 C. deoxyglucos, Science 218 Ž1982. 394–397.
105
w125x T. Klingberg, P.E. Roland, Interference between two concurrent
tasks is associated with activation of overlapping fields in the
cortex, Cogn. Brain Res. 6 Ž1997. 1–8.
w126x R.E. Passingham, Attention to action, Phil. Trans. R. Soc. Lond. B
351 Ž1996. 1473–1479.
w127x J.V. Haxby, C.L. Grady, B. Horwitz, L.G. Ungerleider, M. Mishkin,
R.E. Carson, Dissociation of object and spatial visual processing
pathways in human extrastriate cortex, Proc. Natl. Acad. Sci. USA
88 Ž1991. 1621–1625.
w128x J.V. Haxby, C.L. Grady, B. Salerno, J. Horwitz, L.G. Ungerleider,
M. Mishkin, M.B. Schapiro, Dissociation of object and spatial
visual processing pathways in human extrastriate cortex, in: B.
Gulyas, D. Ottoson, P.E. Roland ŽEds.., Functional Organisation of
the Human Visual Cortex, Pergamon Press, Oxford, 1993, pp.
329–340.
w129x J.V. Haxby, B. Horwitz, L.G. Ungerleider, Jm. Maisog, P. Pietrini,
C.L. Grady, The functional organization of human extrastriate
cortex: A PET-rCBF study of selective attention to faces and
locations, J. Neurosci. 14 Ž1994. 6336–6353.
w130x M. Corbetta, M. Francis, F.M. Miezin, S. Dobmeyer, G.L. Shulman, E. Steven, S.E. Petersen, Selective and divided attention
during visual discriminations of shape, color, and speed: functional
anatomy by positron emission tomograhy, J. Neurosci. 11 Ž1991.
2383–2402.
w131x B. Gulyas, C.A. Heywood, D.B. Popplewell, A. Cowey, P.E.
Roland, Visual form discrimination from colour or motion cues:
Functional anatomy by positron emission tomography, Proc. Natl.
Acad. Sci. USA 91 Ž1994. 9965–9969.
w132x L.G. Ungerleider, M. Mishikin, Two cortical visual systems, in:
D.J. Ingle, M.A. Goodale, R.J.W. Mansfield ŽEds.., Analysis of
Visual Behaviour, The MIT Press, Cambridge, MA, 1982, pp.
549–586.
w133x M. Livingstone, D. Hubel, Segregation of form, color, movement,
and depth: Anatomy, physiology, and perception, Science 240
Ž1988. 740–749.
w134x E.A. DeYoe, D.C. Van Essen, Concurrent processing streams in
monkey visual cortex, Trends Neurosci. 11 Ž1988. 219–226.
w135x W.H. Merigan, J.H.R. Manusell, How parallel are the primate
visual pathways, Annu. Rev. Neurosci. 16 Ž1993. 369–402.
w136x F.A.W. Wilson, S.P.O. Scalaidhe, P.S. Goldman-Rakic, Dissociation of object and spatial processing domains in primate prefrontal
cortex, Science 260 Ž1993. 1955–1958.
w137x P.S. Goldman-Rakic, Topography of cognition: parallel distributed
networks in primate association cortex, Ann. Rev. Neurosci. 11
Ž1988. 137–156.
w138x N. Hadjikhani, P.E. Roland, Cross-modal transfer of information
between the tactile and the visual representations in the human
brain-a PET study, J. Neurosci. Ž1997. in press.
w139x K. Brodmann, Vergleichende Lokalisationslehre der Grosshirnrinde
in ihren Prinzipien dargestellt auf Grund des Zellenbaues, Barth,
Leipzig, 1909.
w140x J. Talairach, P. Tournoux, Co-planar Stereotactic Atlas of the
Human Brain: 3-Dimensional Proportional System: An Approach
to Cerebral Imaging, Georg Thieme, Stuttgart, 1988.