The present study explored differences in dendritic/spine extent
across several human cortical regions. Specifically, the basilar
dendrites/spines of supragranular pyramidal cells were examined
in eight Brodmann’s areas (BA) arranged according to Benson’s
(1993, Behav Neurol 6:75–81) functional hierarchy: primary cortex
(somatosensory, BA3-1-2; motor, BA4), unimodal cortex (Wernicke’s
area, BA22; Broca’s area, BA44), heteromodal cortex (supple-
mentary motor area, BA6β; angular gyrus, BA39) and supramodal
cortex (superior frontopolar zone, BA10; inferior frontopolar zone,
BA11). To capture more general aspects of regional variability,
primary and unimodal areas were designated as low integrative
regions; heteromodal and supramodal areas were designated as high
integrative regions. Tissue was obtained from the left hemisphere of
10 neurologically normal individuals (Mage= 30 ± 17 years; five
males, five females) and stained with a modified rapid Golgi
technique. Ten neurons were sampled from each cortical region(n =
800) and evaluated according to total dendritic length, mean
segment length, dendritic segment count, dendritic spine number
and dendritic spine density. Despite considerable inter-individual
variation, there were significant differences across the eight
Brodmann’s areas and between the high and low integrative regions
for all dendritic and spine measures. Dendritic systems in primary
and unimodal regions were consistently less complex than in
heteromodal and supramodal areas. The range within these rankings
was substantial, with total dendritic length in BA10 being 31%
greater than that in BA3-1-2, and dendritic spine number being 69%
greater. These findings demonstrate that cortical regions involved in
the early stages of processing (e.g. primary sensory areas) generally
exhibit less complex dendritic/spine systems than those regions
involved in the later stages of information processing (e.g. prefrontal
cortex). This dendritic progression appears to reflect significant
differences in the nature of cortical processing, with spine-dense
neurons at hierarchically higher association levels integrating a
broader range of synaptic input than those at lower cortical levels.
The cerebral cortex has historically been parceled according to
cyto- and myelo-architectonic criteria (Brodmann, 1909; Vogt,
1910), with little attention to potential regional variation in
dendritic systems. Although the qualitative characteristics of cor-
tical pyramidal neurons have been relatively well documented
(Ramón y Cajal, 1909, 1911; Ramón-Moliner, 1962), much less is
known about their quantitative variation across cortical areas
because most studies have focused on only one region at a time.
Nevertheless, in the last decade, since Scheibel et al. (Scheibel et
al., 1985, 1990) suggested a positive relationship between
dendritic extent and functional complexity in human cortex,
several quantitative neuromorphological investigations have
begun to document regional dendritic variation. The present
study extends this concept of regional cortical variability by
exploring the degree to which the basilar dendritic and spine
systems of supragranular pyramidal neurons vary across eight
regions of the human cerebral cortex.
Both human and non-human animal research indicate that
regional dendritic variation may be extensive, with profound
functional implications for cortical processing. In humans,
Jacobs et al. (Jacobs et al., 1997) noted that the basilar dendrites
and associated spines in Brodmann’s area (BA) 10 were
significantly more extensive than those in BA18. Functionally,
the more limited dendritic systems in BA18 neurons appear to
correspond with more discrete sampling of afferent information
(i.e. smaller receptive fields). In contrast, the more complex
dendritic arrays in BA10 neurons may facilitate a broader
sampling of afferent information, thereby potentially increasing
their integrative capacity. In an extensive series of studies on
hierarchically arranged visual pathways in monkeys, Elston and
Rosa (Elston et al., 1996; Elston and Rosa, 1997, 1998a,b) have
documented a caudal–rostral progression in dendritic field size
and spine number, suggesting a more extensive input sampling
by dendritic systems at higher levels. These data correspond with
demonstrated size increases in intrinsic axonal clusters across
the visual cortical hierarchy (Amir et al., 1993).
The present investigation provides a broader view of regional
variability in humans than past research by incorporating a more
extensive sampling of cortical areas. To this end, the current
study explores the morphological underpinnings of the func-
tional cortical hierarchy proposed by Benson (Benson, 1993,
1994). Benson’s hierarchy draws heavily on the sensory-fugal
gradients of cortical connectivity proposed by Mesulam
(Mesulam, 1985), which have recently undergone considerable
elaboration (Mesulam, 1998). In Benson’s useful hierarchical
schema, the cerebral cortex is classified into four divisions
based on clinical/anatomical correlations. Each of these cortical
types represents a progressively more complex level of neural
processing: primary cortex is involved in the initial processing
of sensory impulses, or the final output stage for motor func-
tions; unimodal regions discriminate, categorize and integrate
information within a single modality to form a percept of the
same modality; heteromodal cortex compares a particular
percept with previously experienced percepts from other
modalities to form complex multimodal percepts; and supra-
modal association regions are involved in executive control of
Although these divisions and their anatomical boundaries
are far from absolute, especially given the interconnectional
complexity of cortical circuitry, these categories do provide
an initial framework for examining dendritic/spine systems
vis-à-vis a functional hierarchy. In the current study, two
Regional Dendritic and Spine Variation in
Human Cerebral Cortex: a Quantitative
Bob Jacobs, Matthew Schall1, Melissa Prather2, Elisa Kapler,
Lori Driscoll3, Serapio Baca4, Jesse Jacobs, Kevin Ford, Marcy
Wainwright5and Melinda Treml
Laboratory of Quantitative Neuromorphology, Department of
Psychology, The Colorado College, 14 E. Cache La Poudre,
Colorado Springs, CO 80903,1UniFocus, 1330 Capital Parkway,
Carrolton, TX 75006,2Department of Neuroscience, University
of California, Davis, CA 95616,3Department of Psychology,
Cornell University, Ithaca, NY 14853,4Neurosciences Graduate
Program, University of California, San Diego, CA 92093-0357
and5Department of Neurobiology and Anatomy, University of
Texas, Houston, TX 77030, USA
Cerebral Cortex Jun 2001;11:558–571; 1047–3211/01/$4.00
© Oxford University Press 2001. All rights reserved.
Brodmann’s areas were chosen to represent each level of
Benson’s schema: primary cortex (somatosensory, BA3-1-2;
motor, BA4), unimodal cortex (Wernicke’s area, BA22; Broca’s
area, BA44), heteromodal cortex (supplementary motor area,
BA6β; angular gyrus, BA39) and supramodal cortex (superior
frontopolar zone, BA10; inferior frontopolar zone, BA11). In
order to capture more general aspects of regional cortical vari-
ation, primary and unimodal areas were designated as relatively
low integrative regions; heteromodal and supramodal areas were
designated as relatively high integrative regions. Based on the
above evidence and on preliminary research (Baca et al., 1995;
Prather et al., 1997), dendritic measures were expected to
increase in a relatively consistent manner from primary to
supramodal cortical regions, with dendritic/spine systems in
high integration regions being significantly more complex than
those in low integration regions.
Materials and Methods
Tissue was obtained from 10 neurologically normal subjects (Mage= 30 ±
17 years; five males: Mage= 38 ± 20 years; five females: Mage= 23 ± 9 years;
see Table 1). Autolysis time (AT) averaged 12 ± 6 h (range: 1–22 h); all
brains were immersion fixed in 10% neutral buffered formalin for an
average of 32 ± 16 days prior to staining. Tissue was provided by (i) Dr D.
Bowerman, of the El Paso County coroner’s office (six brains); and (ii) Drs
E. Orsini, W. Tyson and S. Caldwell, of Denver’s Children’s Hospital (four
brains). All possible historical information (e.g. age, gender, cause of
death, agonal period) on each subject was obtained from autopsy reports
and medical records. Tissue was excluded if there were any signs of
trauma, cerebral edema or chronic illness with central nervous system
involvement. The research protocol was approved by The Colorado
College Human Subjects Review Board (#H94-004).
Tissue Selection and Processing
A tissue block (1–2 cm along the long axis of the gyrus) was removed from
the left hemisphere in each of the following regions: BA3-1-2, BA4, BA22,
BA44, BA6β, BA39, BA10 and BA11 (see Fig. 1). The relative location and
anatomical characteristics of each of these regions is briefly described
BA3,1,2 and BA4
BA3-1-2 and BA4 were removed from adjacent regions of the post- and
pre-central gyri, respectively (∼2–3 cm from the midline along the
dorsolateral convexity). This location generally represented the arm/
hand region in the classical somotopically organized homunculi maps
(Penfield and Boldrey, 1937), and appears consistent with modern three-
dimensional characterizations of this region (Sastre-Janer et al., 1998).
Sections removed from thepost-central gyrus were notfurthersubdivided
by cytoarchitectonic criteria, although, given that cells were sampled
from the crown of the gyrus, it is likely that most neurons were from BA1
and BA2. The agranular, pre-central gyrus sections often contained
diffuse aggregations of Betz cells.
BA22 constituted classical Wernicke’s area (area TA1, von Economo and
Koskinas, 1925), and was removed according to established criteria: the
anterior edge of the block was adjacent with the posterior edge of
the antero-lateral tip of the primary transverse gyrus of Heschl (Jacobs
and Scheibel, 1993). Cytoarchitectonically, this area represents typical
six-layered isocortex and is characterized by large pyramidal cells in layer
IIIc (Braak, 1980).
Body weight (inkg) Autolysis time (in h)Causeof death Occupation (education)b
acute myelogenous leukemia
mucopolysacchridosis type VI
motor vehicle accident
motor vehicle accident
middle school student(MS)
high school student(HS)
high school student(HS)
Computer programmer (UNI)
aSubjects are referred to by gender (M = male; F = female) and by age in years.For example, F11 refers to an 11-year-old female. Note also thatsome tissuesamples (namely, BA10) from all subjects
except F11 and F15 were used previously (Jacobs et al., 1997).
bAbbreviations: MS = middle school; HS = high school; UNI = university.
the relative position of tissue blocks (listed by Brodmann’s areas, BA) in the present
sample. BA3-1-2, BA4, BA22 and BA44 were classified as low integration regions;
BA6β, BA10, BA11 and BA39 were classified as high integration regions. Note that
these depictions do not capture the considerable morphological variation that typifies
individual brains. d = dorsal, l = lateral.
Cerebral Cortex Jun 2001, V 11 N 6 559
Classical Broca’s area is composed of both BA44 and BA45, which appear
to be heterogeneous in terms of architectonic and functional criteria. The
present sample was limited to tissue from the opercular portion of the
inferior frontal gyrus, corresponding to the agranular fronto-opercular
zone of Sanides (Sanides, 1962). This sampling is consistent with recent
quantitative mapping of this region (Amunts et al., 1999).
BA6β constituted the supplementary motor region, and was removed
from the superior frontal gyrus anterior to the paracentral lobule on the
medial surface of the hemisphere. This midline area thus represents the
superior portion of Braak’s (Braak, 1980) frontal magnopyramidal region.
BA39 constituted the angular gyrus, and was removed from the lobule
surrounding the ascending posterior segment of the parallel sulcus.
Cytoarchitectonically, this region is characterized by a narrow granular
layer IV and a relative ‘clearing’ of layers IIIb and V, which provide it with
its overall eulaminate quality and distinct horizontal lamination (Eidelberg
and Galaburda, 1984).
BA10 and BA11
Both BA10 and BA11 constitute association isocortex with a clear inner
granular layer. BA10 was removed superiorly from the frontal pole,
∼1.5 cm from the midline and 3–4 cm superior to the orbitomedial
surface. BA11 was removed more inferiorly, ∼1.5 cm lateral from the
midline and along the anterior-most portion of the lateral orbital gyrus. It
should be noted that Brodmann’s (Brodmann, 1909) exploration of these
regions was not as detailed as modern analyses (Cavada et al., 2000; Van
Hoesen et al., 2000). In the present study, BA10 and BA11 correspond
respectively to the superior and inferior portions of the granular
frontopolar zone of Sanides (Sanides, 1962). In more recent terminology,
the present sampling of BA11 corresponds to area FP in the map of
Hof et al. (Hof et al., 1995), and to area 10o in the schema of Öngür and
Price (Öngür and Price, 2000). However, for consistency, Brodmann’s
nomenclature is used throughout the present analysis.
Tissue blocks were coded to prevent experimenter bias, trimmed to 3–5
mm in antero-posterior thickness, and processed by a modified rapid
Golgi technique (Scheibel and Scheibel, 1978). To be consistent with
previous research (Jacobs et al., 1997), processed tissue was serially
sectioned at 120 µm with a vibratome such that the preparation was
vertical to the pial surface and perpendicular to the long axis of the gyrus.
Adjacent tissue blocks were sectioned at 50 µm and stained with a
modified cresyl echt violet technique (Gridley, 1960), which permitted
cytoarchitectonic comparisons for routine control purposes and for
measures of laminar depth (expressed as the mean of five sample
measurements taken across the crown of the gyrus).
Cell Selection Criteria and Dendritic/Spine Quantification
Ten relatively isolated supragranular pyramidal cells per tissue block
(i.e. 80 cells per brain) were randomly chosen for analysis following
previously established criteria (Jacobs et al., 1997). Briefly, selected
neurons appeared fully impregnated, and relatively complete, with the
soma located centrally within the 120 µm section depth and the apical
dendrite perpendicular to the pial surface. To assure a relatively
homogeneous cell population, all cells were sampled no further than
1.5 cm vertically from the tip of the gyral crown, with a running average
of soma depth from the pial surface maintained as cells from each cortical
area were drawn. Magnopyramidal neurons were not traced, nor was a
distinction made between subtypes of pyramidal neurons.
Cells were quantified along x-, y- and z-coordinates on a Neurolucida
system (Microbrightfield, Inc.) interfaced with an Olympus BH-2
microscope under a planachromat ×40 (0.70) dry objective. Tracings
began with the soma, which was traced at its widest point in the
two-dimensional plane to provide an estimate of its cross-sectional area.
After drawing the apical shaft, basilar dendrites were traced in their
entirety along with all visible spines. No distinction was made between
different spine types. Dendritic processes were not followed into
adjacent sections nor was dendritic diameter examined. Broken tips
and unclear terminations were indicated as incomplete endings. Of the
43 954 dendritic segments quantified, 45% were intermediate segments.
With regard to terminal segments, 45% were complete and 55% were
incomplete. Sectioned segments were not differentially analyzed because
elimination of cells with incomplete segments would have biased the
sample towards smaller neurons (Uylings et al., 1986).
Cells were traced by 12 individuals. Intrarater reliability was
determined by having each rater trace the same dendritic system
(including somata and spines) 10 times over a 2–4 day period. There was
little variation in tracings; the average coefficient of variation (CV) across
all raters for soma size, total dendritic length and dendritic spine number,
was 5, 3 and 6%, respectively. To further test intrarater reliability, a split
plot design (α = 0.05) compared the first five tracings with the second
five tracings; no significant difference was found within raters for any of
these measures. To maximize interrater reliability, all raters were normed
before quantification by comparing their tracings over a 1–2 week period
to those of the primary investigator (B.J.). In tracings of 10 different
dendritic systems, Pearson product correlations across soma size, total
dendritic length and dendritic spinenumber averaged 0.93, 0.99 and 0.97,
respectively. The tested agreement among raters was further evaluated by
using an analysis of variance (ANOVA; α = 0.05), which indicated no
significant difference among raters on these measures. Finally, all tracings
of neurons were re-examined by the primary investigator to assure quality
Dependent Dendritic/Spine Measures
Dendritic systems were quantified according to a centrifugal nomen-
clature (Bok, 1959; Uylings et al., 1986): dendritic branches arising from
the soma are first-order segments until they bifurcate into second-order
segments, which branch into third-order segments, and so on. Five
measures (represented as mean ± SEM) characterized each cell’s dendritic
system. Total dendritic length (TDL) refers to the summed length of
dendritic segments. Mean segment length (MSL) represents the mean
length of dendritic segments. Dendritic segment count (DSC) refers to the
number of dendritic segments. Dendritic spine number (DSN) refers to
the sum of all spines on dendritic segments. Dendritic spine density
(DSD) represents the average number of spines per micron of dendritic
length. It should be noted here that many of these measures are inter-
related (e.g. TDL is the product of MSL and DSC values).
Independent Variables and Statistical Analyses
Individual Brodmann’s areas provided one independent measure for the
present study. In addition, to capture more general aspects of regional
variation, Brodmann’s areas were grouped in the following manner: areas
representing primary and unimodal cortices (BA3-1-2, 4, 22 and 44) were
designated as low integrative regions; areas representing heteromodal and
supramodal cortices (BA6β, 10, 11 and 39) were designated as high
The raw dendritic data set was aggregated by neuron (Cell). Separate
analyses subsequently evaluated the effects of (i) Brodmann’s areas
(Brodmann) and (ii) integration level (Integration) on each of the five
dependent measures by using a nested ANOVA design (Proc Nested; SAS,
6.08 for UNIX). In this model, Cell was nested within Brodmann or
Integration, each of which was nested within Brain. Briefly, this is
ostensibly a nested, repeated measures design, whereby each dependent
measure is afforded its own nested analysis, thereby increasing the ability
to identify how much each independent variable contributes to the values
found for the dependent measures. Because this design analyzed one
dependent variable at a time, a Bonferroni–Dunn correction (α = 0.01)
was used to maintain an experimentwise alpha of 0.05.
Summary of Neuronal Sample
Golgi-impregnated tissue did not exhibit the autolytic changes
(e.g. irregular varicose enlargements, constriction of dendrites)
described by Williams et al. (Williams et al., 1978). Moreover,
there were no significant correlations between autolysis time
and any of the dependent measures [TDL: r(800) = –0.003;
560 Regional Dendritic Variation • Jacobs et al.
MSL: r(800) = 0.053; DSC: r(800) = 0.039; DSN: r(800) = –0.108;
DSD: r(800) = –0.106]. Several measures taken during data
collection served as guidelines to minimize variability in the
neuronal sampling procedure: (i) soma depth from pial surface;
(ii) soma size; and (iii) laminar thickness. To explore further the
relationship among these measures, two-tailed Pearson product–
moment correlations based on the 800 sampled neurons were
calculated [because of multiple correlations, a Bonferroni–Dunn
correction (α = 0.001) was used to maintain an experimentwise
alpha of 0.05].
The average soma depth for sampled neurons varied only
slightly among Brodmann’s areas (from 800.1 µm in BA10 to
852.7 µm in BA44), and was thus very similar between the two
integration levels (Table 2). This sampling placed the majority
of somata in upper layer III. Soma size increased slightly, but
significantly with soma depth [r(800) = 0.26, P < 0.0001]. There
was also a slight positive relationship between soma depth and
TDL [r(800) = 0.15, P < 0.0001], underscoring the necessity of
controlling soma depth.
Most sampled neurons were small- to medium-sized pyramidal
cells, with the mean soma size being slightly larger in high than
in low integration regions (Table 2). As soma size increased, so
did TDL [r(800) = 0.31, P < 0.0001], and DSC [r(800) = 0.32, P <
0.0001], indicating that larger somata generally exhibited more
complex dendritic arbors.
Overall laminar and cortical thickness appeared roughly
comparable between the two integration levels (Table 2). The
cortex was thickest in BA4 (3.8 mm) and thinnest in BA10
(3.1 mm). The present laminar values are generally consistent
with those of previous cytoarchitectonic findings for these
regions (Brodmann, 1909; von Economo and Koskinas, 1925).
Dendrite and Spine Systems
Photomicrographs of selected Golgi preparations indicate the
overall quality of the stain (Fig. 2). In general, despite con-
siderable interindividual variability, the present results indicate
significant differences among Brodmann’s areas in the predicted
direction, with high integration regions being significantly more
complex than low integration regions. Within this classification,
overall TDL variability within the four high integration regions
(CV = 27%) was slightly less pronounced than TDL variability
within the four low integration regions (CV = 32%), which was
expected given that the low integration grouping contained
not only homotypical but also heterotypical isocortical areas.
These regional hierarchical differences are depicted in sample
Neurolucida tracings from two individuals: an 11-year-old female
(Fig. 3) and a 50-year-old male (Fig. 4). A more detailed analysis of
each dependent measure is provided below.
There was a significant difference for both TDL [F(70,720) =
3.66, P < 0.0001] and MSL [F(70,720) = 3.17, P < 0.0001] across
Brodmann levels. A considerable spread in TDL obtained among
Brodmann’s areas (Fig. 5A), with the most complex region
(BA10; 4,193 µm/cell) being 31.4% higher than the least
complex region (BA3-1-2; 3,191 µm/cell). Differences between
Brodmann’s areas were somewhat attenuated for MSL (Fig. 6A),
with the most complex region (BA6β; 73 µm/cell) being
only 14.4% greater than the least complex region (BA3-1-2;
64 µm/cell). In terms of the initial classification of these regions,
the following ranking obtained for both TDL and MSL: primary <
unimodal < supramodal < heteromodal. The values for hetero-
modal and supramodal regions were nearly identical.
The High Integration level was significantly more complex
than the Low Integration level for both TDL [F(10,180) = 8.95,
P < 0.0001] and MSL [F(10,180) = 5.97, P < 0.0001]. For TDL,
the high integration regions were 17.2% higher than the low
integration regions (Fig. 5A). For MSL, the difference was smaller
at 5.6% (Fig. 6A). Moreover, MSL was the only measure where a
low integration region (namely, BA44) was more complex than a
high integration region (namely, BA11). Dendritic envelopes for
TDL (Fig. 5B) and MSL (Fig. 6B) indicated that high integration
regions were consistently more complex across almost all
dendritic orders, with the greatest difference appearing in third-
and fourth-order segments for TDL.
There was a significant difference across Brodmann levels
for DSC [F(70,720) = 2.272, P < 0.0001]. As illustrated in Figure
7A, the most complex region (BA10; 59 segments/cell) was
19.9% higher than the least complex region (BA3-1-2; 49
segments/cell). The relative complexity of primary, unimodal,
heteromodal and supramodal regions was in the predicted
direction. The High Integration level was significantly more
complex (by 11.1%) in terms of DSC than the Low Integration
level [F(10,180) = 5.58, P < 0.0001]. As with dendritic length, the
highest values were exhibited by the middle of the dendritic
envelope (especially segment orders 3 and 4), where clear
differences emerged between the integration levels (Fig. 7B).
The most marked regional cortical differences emerged in
dendritic spine measures. There was a significant difference
across Brodmann levels for both DSN [F(70,720) = 7.39, P <
0.0001] and DSD [F(70,720) = 8.31, P < 0.0001]. A substantial
spread in DSN was evident among Brodmann’s areas (Fig. 8A),
with the most complex region (BA10; 1,378 spines/cell) being
68.9% higher than the least complex region (BA3-1-2; 816
spines/cell). Although differences were smaller for DSD (Fig.
9A), BA10 (0.26 spines/µm) was still 36.8% higher than BA3-1-2
(0.19 spines/µm). The relative complexity of primary, unimodal,
heteromodal and supramodal regions was in the predicted
direction for both DSN and DSD.
The High Integration level was significantly more complex
than the Low Integration level for both DSN [F(10,180) = 19.79,
P < 0.0001] and DSD [F(10,180) = 17.68, P < 0.0001]. For DSN,
the high integration regions were 32.2% greater than the low
integration regions (Fig. 8A). For DSD, the difference was smaller
Laminar and sampledsoma depths (µm) and somasize (µm2)a
Low integrationregions High integrationregions
Layer I/II junction
Layer II/III junction
Sampled soma depth
Sampled soma size
Layer III/IV junction
306 ± 70
569 ± 102
838 ± 173
266 ± 95
1499 ± 254
3517 ± 527
314 ± 56
562 ± 95
816 ± 181
290 ± 89
1471 ± 170
3304 ± 504
aValues representmean ± SD.
Cerebral Cortex Jun 2001, V 11 N 6 561
at 15% (Fig. 9A). Much the same as TDL (Fig. 5B), DSN values
were highest in the middle of the dendritic envelope, which also
most clearly differentiated the high and low integration regions
(Fig. 8B). In contrast, DSD values closely follow the same pattern
as MDL measures (Fig. 6B), namely increasing somatofugally
until the fifth-order segment, and remaining at relatively high
levels thereafter (Fig. 9B).
Because of the age range of the present subjects (58 years),
correlations between age and the dependent measures were
determined. All correlations except TDL were significant [TDL:
r(800) = 0.014; MSL: r(800) = –0.183, P < 0.001; DSC: r(800) =
0.138, P < 0.001; DSN: r(800) = –0.440, P < 0.001; DSD: r(800) =
–0.586, P < 0.001], indicating marked decreases in spine
The present investigation revealed significant, progressive in-
creases in dendritic/spine extent among hierarchically arranged
cortical regions of the human brain. Although the exact
sequence of individual Brodmann’s areas depended somewhat
on the particular aspect of the dendritic tree examined (as
illustrated in Figs 5A–9A), clear patterns did emerge. Consistent
with the observation that (i) sensory information undergoes
extensive elaboration and modulation during the integration
process (Mesulam, 1998), and (ii) the processing demands
placed on dendritic systems in various cortical regions sub-
stantially influence their ultimate expression (Ramón y Cajal,
1894; Diamond et al., 1964), dendritic/spine systems in the
present analysis were generally less complex in low integration
regions (primary and unimodal cortices) than in high inte-
gration regions (heteromodal and supramodal cortices). Before
Figure 2. Photomicrographs of supragranular pyramidal cells illustrating the sharp contour and intact quality of the Golgi impregnated dendritic systems. Several cases and
Brodmann’s areas are represented: (A) M32 (= 32-year-old male), BA39; (B) F11, BA11; (C) F15, BA44; and (D) F11, BA3-1-2. For A and B, scale bars = 50 µm; for C and D, scale
bars = 10 µm.
562 Regional Dendritic Variation • Jacobs et al.
discussing these results in detail, several methodological issues
need to be addressed.
The implications of the present results are tempered by
well-known methodological limitations (e.g. the practical
constraints of human research, Golgi stains, small sample sizes
and post-mortem delay effects), most of which have been
addressed elsewhere (Williams et al., 1978; de Ruiter 1983;
Flood, 1993; Jacobs and Scheibel, 1993; Jacobs et al., 1993b,
1997). A primary limitation of the present study is that it did
not address potential hemispheric differences, all of which
Figure3. SampletracingsofsupragranularpyramidalcellsfromeachBrodmannarea(BA)forF11. Theseregionshavebeenarrangedtorepresenttherelativeorderofdendritic/spine
complexity for this individual, with cells in BA3-1-2 being the least complex and cells in BA10 being the most complex. Overall, cells in the low integration regions (BA3-1-2, BA4,
BA22 and BA44) exhibited less dendritic branching than cells in the high integration regions (BA6, BA11, BA39 and BA10). Scale bars = 100 µm.
Figure 4. Sample tracings of supragranular pyramidal cells from each Brodmann area (BA) for M50. These regions have been arranged to represent the relative order of
dendritic/spine complexity forthis individual. As with F11 (Fig. 3), cells in BA3-1-2 were theleast complex and cells in BA10 were the most complex. However, note that the relative
order of regions within low (BA3-1-2, BA22, BA44, and BA4) and high integration regions (BA39, BA11, BA6, and BA10) differs from F11, illustrating the interindividual differences
that characterize human tissue. Scale bars = 100 µm.
Cerebral Cortex Jun 2001, V 11 N 6 563
clearly affect certain aspects of cortical function/organization
(Anderson and Rutledge, 1996). Three additional issues require
elaboration: (i) dendritic and spine quantification; (ii) individual
morphological/cytoarchitectonic variability; and (iii) hier-
archical classification of cortical tissue.
Dendritic and Spine Quantification
The present quantification technique provided only a limited
view of cortical neuropil. As noted by Jacobs et al. (Jacobs et al.,
1997), 120 µm sections result in sectioned dendrites. As such,
the present dendritic values represent underestimations of actual
dendritic extent, particularly with regard to the more distal
segments (i.e. sixth-order and higher). Importantly for the
present study, the number of incomplete segments (mostly
due to sectioning) in high integration regions was 6.6% higher
than in low integration regions. Similarly, insofar as spines
cannot be visualized directly above or below dendrites with
light microscopy, the present spine measures also represent
underestimates, especially for thicker dendrites (Horner and
Arbuthnott, 1991). Although correction equations (Feldman
and Peters, 1979) and three-dimensional reconstructions of
dendrites (Belichenko and Dahlström, 1995) may compensate
Figure 5. (A) Bar graph of the relative total dendritic length (TDL, µm) for each Brodmann area (BA), arranged from lowest (BA3-1-2) to highest (BA10). Areas have been
lines. Note the relatively higher TDL values for the high integration regions over the low integration regions. Error bars represent SEM. (B) Dendritic envelopes for low and high
integration regions indicating the greatest differences in third-, fourth- and fifth-order dendritic branches.
Figure 6. (A) Bar graph of the relative mean segment length (MSL, µm) for each Brodmann area (BA), arranged from lowest (BA3-1-2) to highest (BA6). Areas have been
characterized as low (BA3-1-2, BA22, BA4 and BA44) and high integration regions (BA11, BA39, BA10 and BA6). MSL is slightly higher for the high integration regions over the low
increase in MSL values with each successive dendritic order — until the sixth — for both low and high integration regions.
564 Regional Dendritic Variation • Jacobs et al.
for this underestimation, such techniques were not feasible here
because spines were quantified along the entire basilar dendritic
array rather than along short (i.e. 50 µm) segments of uniform
diameter. Importantly, the present underestimation of spines is
likely to be greatest in those regions exhibiting more complex
(and perhaps thicker) dendritic systems. Thus, the observed
regional differences in both dendritic and spine measures may
actually be greater than reported.
Individual Morphological/Cytoarchitectonic Variability
The eight regions examined in the current analysis were chosen
not only to represent different levels in Benson’s (Benson, 1993,
1994) hierarchy, but also because they could be identified
consistently on the basis of anatomical landmarks. Neverthe-
less, extensive interindividual variability characterizes (human)
brain tissue (Bartley et al., 1997). This variation is particularly
confounding when attempting to establish structure–function
relationships because classical cytoarchitectonic maps (i) can
be misinterpreted (Zeki, 1979); (ii) do not necessarily map
isomorphically with gyral–sulcal morphology (Loftus et al.,
1995); (iii) are typically based on qualitative rather than
quantitative criteria; and (iv) are limited to observations of only
Figure 7. (A) Bar graph of therelative dendritic segment count (DSC) foreach Brodmann area (BA), arranged from lowest (BA22) to highest (BA10). Areas have been characterized
as low (BA11, BA3-1-2, BA4 and BA44) andhigh integrationregions (BA11,BA39, BA6 andBA10), withthe average DSCvalueforeach grouping indicated by thedotted lines. Note
indicating the greatest differences in third- and fourth-order dendritic branches.
Figure 8. (A)Bargraphofthedendriticspinenumber(DSN)foreachBrodmannarea(BA),arrangedfromlowest(BA3-1-2)tohighest (BA10). Areashavebeen characterized aslow
(BA3-1-2, BA22, BA4 and BA44) and high integration regions (BA6, BA11, BA39 and BA10), with the average DSN value for each grouping indicated by the dotted lines. Note the
considerablyhigherDSNvaluesforthehigh integrationregionsoverthelowintegrationregions.Errorbarsrepresent SEM.(B) Dendriticenvelopesforlowandhigh integrationregions
indicating the greatest differences in third-, fourth- and fifth-order dendritic branches.
Cerebral Cortex Jun 2001, V 11 N 6 565
a few or even a single brain (Rajkowska and Goldman-Rakic,
1995a,b). Although modern quantitative, histochemical tech-
niques have recently permitted objective mapping of cortical
areas [e.g. BA9 and BA46 (Rajkowska and Goldman-Rakic,
1995a); BA44 and BA45 (Amunts et al., 1999)], a definitive map
of the human cortex has not been established. Laminar
examination of Nissl stains from the sampled regions of the
present study indicated these areas were, in general, cytoarchi-
tectonically consistent with their designation. Nevertheless,
without quantitative cytoarchitectonic verification of every
region from every brain, the areas examined in the present
investigation should be seen as close approximations to their
designated Brodmann classification.
Hierarchical Classification of Cortical Tissue
The present study adopted Benson’s (Benson, 1993, 1994)
hierarchical schema, which was derived from clinical and
anatomical correlations. This schema is generally consistent with
earlier, anatomy-based hierarchies (e.g. Pandya and Kuypers,
1969; Jones and Powell, 1970) outlining a stepwise cortico-
cortical progression along sensory-fugal gradients (Mesulam,
1998). However, such a serial perspective of cortical organiza-
tion remains an acknowledged oversimplification because
extensive parallel cortico-cortical and subcortico-cortical con-
nections are crucial to information processing (Selemon and
Goldman-Rakic, 1988). The complexity of such interconnections
has been demonstrated most clearly in the visual system, where
the original hierarchical configuration (Hubel and Wiesel, 1962)
has been substantially modified with the addition of recursively
interactive parallel networks (Felleman and Van Essen, 1991).
Ultimately, these distributed systems are critical for a compre-
hensive mapping of all hierarchical configurations, including
Benson’s proposed schema (Bressler, 1995).
Accepting that complete incorporation of subcortical, limbic
and parallel networks into the cortical hierarchy is beyond the
scope of the present analysis, the primary and supramodal
regions of the present schema are typically more clearly
categorized than are the intervening regions. In the present
study, for example, BA22 and BA44 could have been categorized
as heteromodal rather than as unimodal regions—indeed, Benson
(Benson, 1994) himself indicates their potentially polymodal
nature. Unfortunately, such a determination cannot be made
with certainty without detailed functional mapping of each
cortical region. Thus, the current hierarchical schema should be
viewed as a simplified heuristic for cortical processing; as such,
progression along a proposed hierarchy may not be paralleled
exactly by concomitant increases in dendritic field area.
Age and Gender Issues
The present sample included individuals of both genders over a
relatively broad age range. Although there is little reason to
expect significant gender-related differences in dendritic extent
(Jacobs and Scheibel, 1993), significant age-related changes in
dendritic/spine systems — similar to those in the present study —
have been extensively documented (Jacobs et al., 1997). In
examining the dendritic/spine values of each brain in the
present study, we noted no appreciable gender or age-related
differences in regional patterns, suggesting that the obtained
regional patterns are quite robust and may be present by early
adolescence. Indeed, developmental positron emission tom-
ography research indicates that the adult pattern (not adult
values) of local cerebral metabolic rates for glucose, which
appear closely associated with dendritic extent (Jacobs et al.,
1995), are typically obtained around the first year of life
(Chugani et al., 1987). We are currently examining regional
dendritic/spine variation in infant tissue to determine when the
adult pattern emerges.
Regional Differences in Dendritic/Spine Systems
The present results are generally consistent with the findings
of Elston and Rosa, who documented progressive increases in
basal dendritic complexity along hierarchically arranged visual
pathways in the monkey (Elston et al., 1996; Elston and Rosa,
1997, 1998a,b). Specifically, their findings suggest a stepwise
Figure 9. (A) Bargraph ofthedendriticspine density(DSD)foreach Brodmann area (BA),arranged from lowest(BA3-1-2) tohighest(BA10). Areashavebeencharacterizedaslow
(BA3-1-2, BA44, BA22 and BA4) and high integration regions (BA6, BA11, BA39 and BA10), with the average DSD value for each grouping indicated by the dotted lines. Note the
relatively higher DSD values for the high integration regions over the low integration regions. Error bars represent SEM. (B) Dendritic envelopes for low and high integration regions
indicating somatofugal increases in DSD, with the greatest differences exhibited in more distal segments.
566 Regional Dendritic Variation • Jacobs et al.
progression in dendritic/spine complexity, with the more
rostrally located, spine-dense neurons integrating a wider range
of modulatory input than the more caudally located, sparsely
spined dendritic trees. Moreover, the present findings elaborate
considerably on those of Jacobs et al. (Jacobs et al., 1997), who
documented in humans significantly more complex dendritic/
spine systems in BA10 over BA18, again suggesting that the
dendritic/spine systems of cortical areas involved in the initial
stages of information processing are not as complex as those
involved later in the processing stream. Below, we provide a
brief overview of each region examined in the present analysis.
The overview is far from exhaustive, but provides a general
indication of the relationship between dendritic/spine extent
and the general integrative nature of each area.
Low Integration Regions
As the initial cortical area for discriminating incoming somato-
sensory information, BA3-1-2 receives most of its input from the
ventral posterior nuclei of the thalamus (Clark and Boggon,
1935), with additional, limited input from adjacent sensorimotor
cortices (Jones and Powell, 1970). In an investigation of intrinsic
axon collaterals within this region, Porter (1997) has noted that
proximal basilar dendrites in BA2 pyramidal cells appear to be
the primary target of the projections from BA3a. Insofar as
proximal dendritic systems mature before more distal systems,
these proximal connections are probably crucial to the relatively
early functional maturation of this cortical region (Chugani et al.,
1987). As such, it is not surprising that BA3-1-2 in the present
study generally exhibited the least complex dendritic/spine
system of all regions examined. Consistent with the hierarchical
predictions, the measures for BA3-1-2 (e.g. TDL = 3191 µm/cell;
DSN = 816 spines/cell) were also less than those observed in
BA18 (e.g. TDL = 3563 µm/cell; DSN = 933 spines/cell) of a
comparable human population, namely the younger group (Mage
= 34 years) of Jacobs et al. (Jacobs et al., 1997).
In contrast to BA3-1-2, BA4 appears to be more richly
interconnected, receiving projections from the ventrolateral
thalamus (Wiesendanger and Wiesendanger, 1985) and syn-
thesizing information from several cortical areas, including
somatosensory, premotor, supplementary motor, parietal
association and prefrontal cortices (Jones et al., 1978). A more
complex dendritic array in BA4 neurons would facilitate the
synthesis of various sources of input, including processed
proprioceptive and tactile information from layer II and III
pyramidal neurons in BA3-1-2 (Porter, 1997), prior to initiating
smooth voluntary movements. Indeed, BA4 neurons exhibited
dendritic/spine systems of comparable or greater complexity
than those of BA18 in Jacobs et al. (Jacobs et al., 1997), or BA22
in the present study. Thus, BA4 neurons appear to be slightly
more complex than predicted by Benson’s functional hierarchy,
suggesting that one should not group primary sensory and motor
cortices together when comparing them at the morphological
In terms of unimodal regions, neurons in BA22 were slightly
less complex than expected, ranking above only primary
somatosensory cortex in complexity for most of the dendritic/
spine measurements. The present dendritic measures for BA22
(e.g. TDL = 3302 µm/cell; DSN = 903 spines/cell) are greater than
in previous reports for this region [TDL = 2672 µm/cell (Jacobs
et al.,1993a); TDL = 2589 µm/cell (Jacobs et al., 1993b); and TDL
= 2239 µm/cell and DSN = 526 spines/cell (Anderson and
Rutledge, 1996)], presumably because of differing subjects and
quantification techniques. Dendritic/spine values for BA22 were
nevertheless similar to those obtained by Jacobs et al. (Jacobs et
al., 1997) for another unimodal region, namely BA18 (TDL =
3563 µm/cell; DSN = 933 spines/cell). One possible explanation
for the relatively low complexity of this region is that the present
sampling criteria for BA22 placed it adjacent to primary auditory
cortex, thus probably limiting it more to unimodal, auditory
processing. Conceivably, a section more posterior along the
superior temporal gyrus would be involved in synthesizing a
greater proportion of polymodal information, especially given
that the sensory speech region receives a wide sampling of
cortical and subcortical input (Jones and Powell, 1970; Seldon
1985). In turn, a more posterior region might also have exhibited
more complex dendritic ensembles.
Significantly connected with BA22 by the superior longi-
tudinal and arcuate fasciculi (Krieg, 1963; Petrides and Pandya,
1988) is classical Broca’s area, which was the most dendritically
complex of all low integration regions. In fact, BA44 was the only
low integration region to surpass a high integration region on
one of the dendritic measures (recall Fig. 6A). Classification of
this region as unimodal is itself problematic, however. Indeed,
Benson (1994) classified BA44 as unimodal and the more
anterior BA45 as heteromodal. Given the variability of these
regions (Amunts et al., 1999), it is possible that the sampled area
in the present study was multimodally responsive, receiving not
only auditory, but also visual, somesthetic and limbic input
(Geschwind, 1965). The relatively complex dendritic ensembles
in this region would thus assist in integrating the polymodal
information required for generating the motor sequences
involved in language output.
High Integration Regions
According to Mesulam (Mesulam, 1998), heteromodal regions
represent ‘epicenters’ for large networks, with each epicenter
potentially interacting with several other networks. The two
heteromodal regions examined in the present study, BA6β and
BA39, certainly fit this description and exhibited very similar
levels of dendritic/spine complexity. The supplementary motor
area constitutes a convergence zone for projections from pri-
mary and secondary sensorimotor cortices, parietal association
cortex, the anterior cingulate gyrus and indirectly from the basal
ganglia (Damasio et al., 1981; Schell and Strick, 1984). This
pattern of input suggests that BA6β integrates multimodal
sensory information [except that of a visual nature (Pandya and
Kuypers, 1969)] and limbic-mediated input in order to prepare,
inhibit and/or modify (learned) motor programs for internally
driven behavior (Orgogozo and Larsen, 1979). Synthesizing such
a constellation of diverse input presumably requires the rela-
tively complex dendritic/spine systems exhibited by this region.
Similarly, BA39 has been found to contain primarily multi-
modally responsive neurons involved in integrating information
from several surrounding transitional fields and from subcor-
tical regions (Hyvärinen and Shelepin, 1979; Eidelberg and
Galaburda, 1984). At the cortical level, BA39 receives ipsilateral
association fibers from frontal (motor, premotor and prefrontal),
parietal, temporal and occipital lobes, and from the contralateral
inferior parietal lobule (Hyvärinen, 1982a,b). Subcortical
afferents originate in the thalamus (anterior and posterior nuclei
and pulvinar), hippocampus, basal forebrain nuclei, claustrum,
substantia nigra, locus coerueus and Raphé nuclei (Hyvärinen,
1982a,b). Neurons in this region are thus involved not only in
polymodal integration, but in more abstract cognitive and
symbolic functions such as spatial referencing, mental arith-
metic and semantic memory (Démonet et al., 1992; Roland,
Cerebral Cortex Jun 2001, V 11 N 6 567
1993). The true integrative nature of this region is revealed
through lesions, which result in complex constellations of
deficits such as Balint’s and Gerstmann’s syndromes.
In contrast to the two heteromodal regions, the two supra-
modal areas differed considerably from each other in terms
of dendritic/spine complexity. For all measures (especially DSN
and DSD), BA11 neurons were substantially less complex than
BA10. Indeed, BA11 was the least complex of all high integration
regions. BA10, however, exhibited the highest dendritic/spine
values of all regions examined, values that were very similar to
those reported by Jacobs et al. (Jacobs et al., 1997). The reason
for this discrepancy between BA10 and BA11 is unclear,
although one could speculate that it reflects underlying
differences in the type and/or degree of connectivity. Whereas
the dorsolateral portion of the prefrontal cortex (including
BA10) receives dense projections from the dorsal parietal cortex,
the ventrolateral portion (including BA11) receives dense
projections from inferotemporal cortex (Wilson et al., 1993;
Barbas, 1995). Moreover, given that the present sampling of
BA11 borders the orbitomedial division of the prefrontal lobe, it
is likely that it shares more direct anatomical connections with
the amygdala and periamygdaloid cortical regions than BA10
(Cavada et al., 2000; Öngür and Price, 2000), and thus may
contribute particularly to the emotional evaluation of sensory
experiences (Morecraft et al., 1992; Barbas, 1995).
Regardless of potential differences in the interconnectivity
patterns of BA10 and BA11, the dorsolateral prefrontal cortex
remains one of the most richly integrative regions of the primate
brain, receiving not only input from subcortical structures such
as the dorsomedial thalamus (Goldman-Rakic and Porrino, 1985;
Barbas et al., 1991), but also higher order sensory information
from parietal, temporal and occipital association areas (Jacobson
and Trojanowski, 1977; Preuss and Goldman-Rakic, 1991). In
particular, the majority of prefrontal supragranular pyramidal
cells appear to receive substantial excitatory input from other
prefrontal pyramidal neurons through long-distance axon
collaterals (Melchitzky et al., 1998). Moreover, these intrinsic
connections appear to be greater than in posterior cortical
regions (McGuire et al., 1991; Lund et al., 1993). Although
frontopolar regions such as BA10 do not exhibit consistent
co-activation patterns with specific behaviors or sensory input,
these regions are crucially involved in higher-order orchestra-
tions of cortical networks (Roland, 1984). Thus, dorsolateral
prefrontal neurons, which display the selective, delayed firing
crucial to maintaining internal representations of behaviorally
relevant cues (Fuster 1973; González-Burgos et al., 2000), are
in a unique position to orchestrate top-down (i.e. sensory-petal)
control of memory and attention mechanisms relevant to mul-
tiple perceptual and cognitive domains (Cabeza and Nyberg,
1997; Mesulam, 1998).
Complex dendritic/spine systems, particularly as expressed
by BA10 pyramidal neurons, provide the necessary surface area
for transmodal integration of such a broad spectrum of
information. Indeed, the high spine values for BA10 in the
present study are consistent with recent research in monkeys,
which indicates that the basilar dendrites of prefrontal neurons
are significantly more spiny than those in occipital, temporal or
parietal cortices (Elston, 2000). Metabolic evidence provides
additional support for the existence of complex dendritic arbors
in the dorsolateral prefrontal region. Given that up to 80% of
glucose utilization is devoted to maintenance of the Na+/K+ion
exchange across cell membranes (Sokoloff, 1977), and that
dendrites may constitute >90% of a neuron’s receptive surface
area (Schadé and Baxter, 1960), it is not surprising that the
dorsolateral prefrontal cortex (including BA10) typically exhibits
higher metabolism and regional cerebral blood flow in the
normal, resting state than do other cortical areas (Roland, 1984).
Dendritic Integration and Functional Implications
As indicated in the present study, the increase in total dendrite
length in humans from primary cortical regions to supramodal
areas is approximately one-third, and that of total spine number
is about two-thirds. Both structural and physiological research
suggests that such quantitative morphological differences may
contribute to the robust qualitative differences assumed to exist
between processing mechanisms in different cortical regions.
Structurally, pyramidal cells communicate predominantly with
each other through vertically recurrent collaterals and horizontal
long-distance intrinsic projections (Winfield et al., 1981;
Douglas et al., 1995). The primary targets of these intracortically
derived connections appear to be the basilar dendrites (Globus
and Scheibel, 1967a,b,c). Thus, progressive increases in basilar
dendritic extent, as documented in the present study for high
integration regions, might reflect the neurons’ increased
exposure to intracortical influences and participation in cortical
networks, a relationship that intuitively meets assumptions
about information convergence and increasingly more complex
Physiologically, dendritic systems appear to be highly com-
partmentalized, independent subunits, whose morphological
and membrane characteristics crucially determine a cell’s
input–output transformations (Helmchen, 1999; Spruston et al.,
1999). These dendrites dynamically sample surrounding areas
for correlated activity (Katz et al., 1989; Kossel et al., 1995),
and integrate it with a complex repertoire of nonlinear, active,
electrochemical responses that provide considerable computa-
tional flexibility (Quartz and Sejnowski, 1997). These active
characteristics boost the effect of (distal) synaptic input,
contributing significantly to synaptic integration locally and
across the entire neuron. Consequently, the fine, outermost
branches of the dendrite ensemble — such as those in high
integration regions — may assume physiological importance out
of proportion to the modest fraction of the dendritic ensemble
they represent (Magee and Cook, 2000). Moreover, dendritic
spines, which are generally more dense on distal than proximal
segments (recall Fig. 9B), are crucial to this integration process
insofar as the most peripheral spines are thought to be particu-
larly effective in adjusting synaptic potency (Shepherd et al.,
1985). Thus, the excitability of an entire dendrite may be
disproportionately regulated by changes in distal spine density
(Jaslove, 1992). In summary, both anatomical and physiological
evidence suggests that the dendritic/spine variations observed
across diverse cortical regions in the present study contribute
to the formation of integrative neural networks underlying
complex cognitive processes (Knudsen, 1994; Yuste and Tank,
Quantitative neuromorphological investigations such as the
present study substantially enhance our understanding of the
dendritic ensembles first described in qualitative observations
and set the stage for the development of a quantitative dendritic
map of the cerebral cortex. Despite considerable inter-individual
variation and inherent design limitations, clear regional
differences in the predicted direction were revealed by the
present quantitative analysis. Dendritic/spine systems in primary
568 Regional Dendritic Variation • Jacobs et al.
(BA3-1-2 and BA4) and unimodal (BA22 and BA44) regions were
consistently less complex than in heteromodal (BA6β and BA39)
and supramodal (BA10 and BA11) areas. Dendritic/spine systems
were thus significantly more complex in high integration regions
than in low integration regions. Highest dendritic/spine values
were in BA10, which exhibited 31% greater total dendritic length
and 69% greater dendritic spine number than the least complex
region, namely BA3-1-2. It seems likely that these regional
variations reflect significant differences in the nature of cortical
processing. Many other factors are undoubtedly involved in
determining the range of computational strategems as one
moves from first-level sensory representations to the highest
associational levels. Nonetheless, the quantitative characteristics
of the receptive dendritic membrane of individual neuronal
elements and their variations along the length of the dendritic
shaft appear to represent central issues in cortical computation
and behavioral flexibility.
Partial support for this work was provided by the National Science
Foundation’s Division of Undergraduate Education grant (DUE-
#9550790), the Hughes Foundation, the John D. and Catherine T.
MacArthur Professorship, and The Colorado College’s divisional research
funds. Preliminary reports of some of these results have appeared in
abstract form (Baca et al., 1995; Prather et al., 1997). We gratefully
acknowledge David Bowerman, Al Correl, Richard Sherwin, Leroy
Fischer, Edmund Orsini and Wes Tyson for their generous assistance with
this project. We also thank several students who participated in data
collection: Sherry Bekhit, Becca Kernan, Birgit Fisher, Jennifer Ferguson,
Jon Driscoll and Kelly Courns. Finally, we dedicate this work with
admiration to Dr. Arnold B. Scheibel.
Address correspondence to Bob Jacobs, Laboratory of Quantitative
Neuromorphology, Department of Psychology, The Colorado College,
14 E. Cache La Poudre, Colorado Springs, CO 80903, USA. Email:
Amir Y, Harel M, Malach R (1993) Cortical hierarchy reflected in the
organization of intrinsic connections in macaque monkey visual
cortex. J Comp Neurol 334:19–64.
Amunts K, Schleicher A, Bürgel U, Mohlberg H, Uylings HB, Zilles K
(1999) Broca’s region revisited: cytoarchitecture and intersubject
variability. J Comp Neurol 412:319–341.
Anderson B, Rutledge V (1996) Age and hemisphere effects on dendritic
structure. Brain 119:1983–1990.
Baca S, Larsen L, Fisher B, Kernan R, Schall M, Jacobs B (1995) Dendritic
and spine analyses across hierarchically arranged areas of human
neocortex: a quantitative Golgi study. Soc Neurosci 182.9 [Abstract].
Barbas H, Henion THH, Dermon CR (1991) Diverse thalamic projections
to the prefrontal cortex in the rhesus monkey. J Comp Neurol
Barbas H (1995) Anatomic basis of cognitive–emotional interactions in
the primate prefrontal cortex. Neurosci Biobehav Rev 19:499–510.
Bartley AJ, Jones DW, Weinberger D (1997) Genetic variability of human
brain size and cortical gyral patterns. Brain 120:257–269.
Belichenko PV, Dahlström A (1995) Studies of the 3-dimensional
architecture of dendritic spines and varicosities in human cortex by
confocal laser scanning microscopy and Lucifer Yellow micro-
injections. J Neurosci Methods 57:55–61.
Benson DF (1993) Prefrontal abilities. Behav Neurol 6:75–81.
Benson DF (1994) The neurology of thinking. New York: Oxford Press.
Bok ST (1959) Histonomy of the cerebral cortex. Amsterdam: Elsevier.
Braak H (1980) Architectonics of the human telencephalic cortex. Berlin:
Bressler SL (1995) Large-scale cortical networks and cognition. Brain Res
Brodmann K (1909) Vergleichende Lokalisationlehre der Grosshirnrinde
in ihren Prinzipien dargestellt auf Grund des Zellenbaues. Leipzig: J.A.
Cabeza R, Nyberg L (1997) Imaging cognition: an empirical review of PET
studies with normal subjects. J Cogn Neurosci 9:1–26.
Cavada C, Compañy T, Tejedor J, Cruz-Rizzolo RJ, Reinoso-Suárez F
(2000) The anatomical connections of the Macaque monkey
orbitofrontal cortex. A review. Cereb Cortex 10:220–242.
Chugani HT, Phelps ME, Mazziotta JC (1987) Positron emission
tomography study of human brain functional development. Ann
Clark WEL, Boggon RH (1935) The thalamic connections of the parietal
and frontal lobes of the brain in the monkey. Phil Trans R Soc Lond
Damasio AR, Van Hoesen GW, Vilensky J (1981) Limbic-motor pathways
in the primate: a means for emotion to influence motor behavior.
de Ruiter JP (1983) The influence of post-mortem fixation delay on the
reliability of the Golgi silver impregnation. Brain Res 266:143–147.
Démonet J-F, Chollet F, Ramsay S, Cardebat D, Nespoulous JL, Wise R,
Rascol A, Frackowiak R (1992) The anatomy of phonological and
semantic processing in normal subjects. Brain 115:1753–1768.
Diamond MC, Krech D, Rosenzweig MR (1964) The effects of an enriched
environment on the histology of the rat cerebral cortex. J Comp
Douglas RJ, Kock C, Mahowald M, Martin KAC, Suarez HH (1995)
Recurrent excitation in neocortical circuits. Science 269:981–985.
Eidelberg D, Galaburda AM (1984) Inferior parietal lobule: divergent
architectonic asymmetries in the human brain. Arch Neurol 41:
Elston GN (2000) Pyramidal cells of the frontal lobe: all the more spinous
to think with. J Neurosci 20:1–4.
Elston GN, Rosa MGP (1997) The occipitoparietal pathway of the
macaque monkey: comparison of pyramidal cell morphology in layer
III of functionally related cortical visual areas. Cereb Cortex
Elston GN, Rosa MGP (1998a) Complex dendritic fields of pyramidal cells
in the frontal eye field of the macaque monkey: comparison with
parietal areas 7a and LIP. NeuroReport 9:127–131.
Elston GN, Rosa MGP (1998b) Morphological variation of layer III
pyramidal neurones in the occipitotemporal pathway of the macaque
monkey visual cortex. Cereb Cortex 8:278–294.
Elston GN, Rosa MGP, Calford Mbyte (1996) Comparison of dendritic
fields of layer III pyramidal neurons in striate and extrastriate visual
areas of the marmoset: a Lucifer Yellow intracellular injection study.
Cereb Cortex 6:807–813.
Feldman ML, Peters A (1979) Technique for estimating total spine
numbers on Golgi-impregnated dendrites. J Comp Neurol 118:
Felleman DJ, Van Essen DC (1991) Distributed hierarchical processing in
the primate cerebral cortex. Cereb Cortex 1:1–47.
Flood DG (1993) Critical issues in the analysis of dendritic extent in aging
humans, primates, and rodents. Neurobiol Aging 14:649–654.
Fuster JM (1973) Unit activity in prefrontal cortex during delayed-
response performance: neuronal correlates of transient memory. J
Geschwind N (1965) Disconnection syndromes in animals and man. Brain
Globus A, Scheibel AB (1967a) Pattern and field in cortical structure: The
rabbit. J Comp Neurol 131:155–172.
Globus A, Scheibel AB (1967b) Synaptic loci on visual cortical neurons of
the rabbit: the specific afferent radiation. Exp Neurol 18:116–131.
Globus A, Scheibel AB (1967c) Synaptic loci on parietal cortical neurons:
termination of corpus callosum fibers. Science 156:1127–1129.
Goldman-Rakic PS, Porrino LJ (1985) The primate mediodorsal (MD)
nucleus and its projections to the frontal lobe. J Comp Neurol
González-Burgos G, Barrionuevo G, Lewis DA (2000) Horizontal synaptic
connections in monkey prefrontal cortex: an in vitro electrophysio-
logical study. Cereb Cortex 10:1047–3211.
Gridley M (1960) Manual of histologic and special staining techniques,
2nd edn. New York: McGraw-Hill.
Helmchen F (1999) Dendrites as biochemical compartments. In:
Dendrites (Stuart G, Spruston N, Häusser M, eds), pp. 161–192. New
York: Oxford University Press.
Hof PR, Mufson EJ, Morrison JH (1995) Human orbitofrontal cortex:
cytoarchitecture and quantitative immunohistochemical parcellation.
J Comp Neurol 359:48–68.
Cerebral Cortex Jun 2001, V 11 N 6 569
Horner CH, Arbuthnott E (1991) Methods of estimation of spine density —
are spines evenly distributed throughout the dendritic field? J Anat
Hubel DH, Wiesel TN (1962) Receptive fields, binocular interaction and
functional architecture in the cat’s visual cortex. J Physiol (Lond)
Hyvärinen J (1982a) Posterior parietal lobe of the primate brain. Physiol
Hyvärinen J (1982b) The parietal cortex of monkey and man. Berlin:
Hyvärinen J, Shelepin Y (1979) Distribution of visual and somatic
functions in the parietal association area of the monkey. Brain Res
Jacobs B, Batal HA, Lynch B, Ojemann G, Ojemann L, Scheibel AB (1993a)
Quantitative dendritic and spine analyses of speech cortices: a case
study. Brain Lang 44:239–253.
Jacobs B, Schall M, Scheibel AB (1993b) A quantitative dendritic analysis
of Wernicke’s area in humans. II. Gender, hemispheric, and environ-
mental factors. J Comp Neurol 327:97–111.
Jacobs B, Chugani HT, Allada V, Chen S, Phelps ME, Pollack DB, Raleigh
MJ (1995) Developmental changes in brain metabolism in sedated
rhesus macaques and vervet monkeys revealed by positron emission
tomography. Cereb Cortex 5:222–233.
Jacobs B, Driscoll L, Schall M (1997) Life-span dendritic and spinechanges
in areas 10 and 18 of human cortex: a quantitative Golgi Study. J Comp
Jacobs B, Scheibel AB (1993) A quantitative dendritic analysis of
Wernicke’s area in humans. I. Lifespan changes. J Comp Neurol
Jacobson S, Trojanowski JQ (1977) Prefrontal granular cortex of the
rhesus monkey. I. Intrahemispheric cortical afferents. Brain Res
Jaslove SW (1992) The integrative properties of spiny distal dendrites.
Jones EG, Powell TPS (1970) An anatomical study of converging sensory
pathways with the cerebral cortex of the monkey. Brain 93:793–820.
Jones EG, Coulter JD, Hendry SHC (1978) Intracortical connectivity of
architectonic fields in the somatic sensory, motor, and parietal cortex
of monkeys. J Comp Neurol 181:291–348.
Katz LC, Gilbert CD, Wiesel TN (1989) Local circuits and ocular
dominance columns in monkey striate cortex. J Neurosci 9:
Knudsen EI (1994) Supervised learning in the brain. J Neurosci 14:
Kossel A, Lowel S, Bolz J (1995) Relationships between dendritic fields
and functional architecture in striate cortex of normal and visually
deprived cats. J Neurosci 15:3913–3926.
Krieg WJW (1963) Connections of the cerebral cortex. Evanston: Brain
Loftus WC, Tramo MJ, Gazzaniga MS (1995) Cortical surface modeling
reveals gross morphometric correlates of individual differences. Hum
Brain Mapp 3:257–270.
Lund JS, Yoshioka T, Levitt JB (1993) Comparison of intrinsic connectivity
in different areas of macaque monkey cerebral cortex. Cereb Cortex
Magee JC, Cook EP (2000) Somatic EPSP amplitude is independent of
synapse location in hippocampal pyramidal neurons. Nature Neurosci
McGuire BA, Gilbert CD, Rivlin PK, Wiesel TN (1991) Targets of
horizontal connections in macaque primary visual cortex. J Comp
Melchitzky DS, Sesack SR, Pucak ML, Lewis DA (1998) Synaptic targets of
pyramidal neurons providing intrinsic horizontal connections in
monkey prefrontal cortex. J Comp Neurol 390:211–224.
Mesulam M-M (1985) Principles of behavioral neurology. Philadelphia:
Mesulam M-M (1998) From sensation to cognition. Brain 121:1013–1052.
Morecraft RJ, Geula C, Mesulam M-M (1992) Cytoarchitecture and neural
afferents of orbitofrontal cortex in the brain of the monkey. J Comp
Öngür D, Price JL (2000) The organization of networks within the orbital
and medial prefrontal cortex of rats, monkeys and humans. Cereb
Orgogozo JM, Larsen B (1979) Activation of thesupplementary motor area
during voluntary movement in man suggests it works as a supramotor
area. Science 206:847–850.
Pandya DN, Kuypers HG (1969) Cortico-cortical connections in the
rhesus monkey. Brain Res 13:13–36.
Penfield W, Boldrey E (1937) Somatic motor and sensory representation
in the cerebral cortex of man as studied by electrical stimulation.
Petrides M, Pandya DN (1988) Association fiber pathways to the frontal
cortex from the superior temporal region in the rhesus monkey.
J Comp Neurol 273:52–66.
Porter LL (1997) Morphological characterization of a cortico-cortical relay
in the cat sensorimotor cortex. Cereb Cortex 7:100–109.
Prather M, Treml M, Driscoll L, Schall M, Jacobs B (1997) Regional
variation in dendritic and spine complexity: a quantitative Golgi
analysis of human cerebral cortex. Soc Neurosci 87.13 [Abstract].
Preuss TM, Goldman-Rakic PS (1991) Ipsilateral cortical connections of
granular frontal cortex in the strepsirhine primate Galago, with
comparative comments on anthropoid primates J Comp Neurol
Quartz SR, Sejnowski TJ (1997) The neural basis of cognitive
development: a constructivist manifesto. Behav Brain Sci 20: 537–596.
Rajkowska G, Goldman-Rakic PS (1995a) Cytoarchitectonic definition of
prefrontal areas in the normal human cortex: I. Remapping of areas 9
and 46 using quantitative criteria. Cereb Cortex 5:307–322.
Rajkowska G, Goldman-Rakic PS (1995b) Cytoarchitectonic definition of
prefrontal areas in the normal human cortex: II. Variability in locations
of areas 9 and 46 and relationship to the Talairach coordinate system.
Cereb Cortex 5:323–337.
Ramón y Cajal S (trans. by L. Azoulay) (1894) Les nouvelles ideés sur la
structure du système nerveux chez l’homme et chez les vertébrés.
Ramón y Cajal S (trans. by L. Azoulay) (1909, 1911) Histologie du système
nerveux de l’homme et des vertébrés, 2 vols. Paris: Maloine.
Ramón-Moliner E (1962) An attempt at classifying nerve cells on the basis
of their dendritic patterns. J Comp Neurol 119:211–227.
Roland PE (1984) Metabolic measurements of the working frontal cortex
in man. Trends Neurosci 7:430–435.
Roland PE (1993) Brain activation. New York: John Wiley & Sons.
Sanides F (1962) Die Architektonik des menschlichen Stirnhirns. Berlin:
Sastre-Janer FA, Regis J, Belin P, Mangin J-F, Dormont D, Masure M-C,
Remy P, Frouin V, Samson Y (1998) Three-dimensional reconstruction
of the human central sulcus reveals a morphological correlate of the
hand area. Cereb Cortex 8:641–647.
Schadé JP, Baxter CF (1960) Changes during growth in the volume and
surface area of cortical neurons in the rabbit. Exp Neurol 2:158–178.
Scheibel ME, Scheibel AB (1978) The methods of Golgi. In: Neuro-
anatomical research techniques (Robertson RT, ed.), pp. 89–114. New
York: Academic Press.
Scheibel AB, Paul LA, Fried I, Forsythe AB, Tomiyasu U, Wechsler A, Kao
A, Slotnick J (1985) Dendritic organization of the anterior speech area.
Exp Neurol 87:109–117.
Scheibel AB, Conrad T, Perdue S, Tomiyasu U, Wechsler A (1990) A
quantitative study of dendrite complexity in selected areas of the
human cerebral cortex. Brain Cogn 12:85–101.
Schell GR, Strick PL (1984) The origin of thalamic inputs to the arcuate
premotor and supplementary motor areas. J Neurosci 4:539–60.
Seldon HL (1985) The anatomy of speech perception: Human auditory
cortex. In: Cerebral cortex (Peters A, Jones EG, eds), vol. 4, pp.
273–327. New York: Plenum Press.
Selemon LD, Goldman-Rakic PS (1988) Common cortical and subcortical
targets of the dorsolateral prefrontal and posterior parietal cortices in
the rhesus monkey: evidence for a distributed neural network
subserving spatially guided behavior. J Neurosci 8:4049–4068.
Shepherd GM, Brayton RK, Miller JP, Segev I, Rinzel J, Rall W (1985)
Signal enhancement in distal cortical dendrites by means of inter-
actions between active dendritic spines. Proc Natl Acad Sci USA
Sokoloff L (1977) Relation between physiological function and energy
metabolism in the central nervous system. J Neurochem 29:13–26.
Spruston N, Stuart G, Häusser M (1999) Dendritic integration. In:
Dendrites (Stuart G, Spruston N, Häusser M, eds), pp. 231–270. New
York: Oxford University Press.
Uylings HBM, Ruiz-Marcos A, van Pelt J (1986) The metric analysis of
570 Regional Dendritic Variation • Jacobs et al.
three-dimensional dendritic tree patterns: a methodological review. Download full-text
J Neurosci Methods 18:127–151.
Van Hoesen GW, Parvizi J, Chu C-C (2000) Orbitofrontal cortex pathology
in Alzheimer’s disease. Cereb Cortex 10:243–251.
Vogt O (1910) Die myeloarchitektonische Felderung des menschlichen
Stirnhirns. J Psychol Neurol 15:221–232.
von Economo C, Koskinas GN (1925) Die Cytoarchitektonik der
Hirnrinde des erwachsenen Menschen. Berlin: Springer.
Wiesendanger R, Wiesendanger M (1985) Cerebello-cortical linkage in the
monkey as revealed by transcellular labeling with the lectin wheat
germ agglutinin conjugated to the marker horseradish peroxidase.
Exp Brain Res 59:105–117.
Williams RS, Ferrante RJ, Caviness VS Jr. (1978) The Golgi rapid
method in clinical neuropathology: morphological consequences of
suboptimal fixation. J Neuropathol Exp Neurol 37:13–33.
Wilson FA, Scalaidhe SP, Goldman-Rakic PS (1993) Dissociation of object
and spatial processing domains in primate prefrontal cortex. Science
Winfield DA, Brooke RNL, Sloper JJ, Powell TPS (1981) A combined
Golgi-electron microscopic study of the synapses made by the
proximal axon and recurrent collaterals of a pyramidal cell in the
somatic sensory cortex of the monkey. Neuroscience 6: 1217–1230.
Yuste R, Tank DW (1996) Dendritic integration in mammalian neurons, a
century after Cajal. Neuron 16:701–716.
Zeki SM (1979) Zu Brodmann Area 18 und 19. Exp Brain Res 36:195–197.
Cerebral Cortex Jun 2001, V 11 N 6 571