Intersubject Variability in Cortical Activations
during a Complex Language Task
Jinhu Xiong,* Shobini Rao, Paul Jerabek, Frank Zamarripa, Marty Woldorff,
Jack Lancaster, and Peter T. Fox
Research Imaging Center, The University of Texas Health Science Center, San Antonio, Texas; and *Department of Clinical Psychology,
National Institute of Mental Health and Neurosciences, Bangalore, India
Received May 23, 2000
Intersubject variability in the functional organization
of the human brain has theoretical and practical impor-
tance for basic and clinical neuroscience. In the present
study, positron emission tomography (PET) and anatom-
ical magnetic resonance imaging (MRI) were used to
study the functional anatomy of language processes. In-
tersubject variability in task-induced activations in six
brain regions was assessed in 20 normal subjects (10
men and 10 women) for frequency of occurrence, loca-
tion, intensity, and extent. A complex, but well-studied
task (overt verb generation) was compared to a simple
baseline (visual fixation) to induce activations in brain
areas serving perceptual, motoric, and cognitive func-
tions. The frequency of occurrence was high for all se-
lected brain areas (80 –95%). The variability in response
location in Talairach space, expressed as the standard
deviation along each axis (x, y, z), ranged from 5.2 to 9.9
mm. This variability appears to be uniformly distributed
across the brain, uninfluenced by regional differences in
the complexity of gyral anatomy or mediated behavior.
The variability in response location, expressed as the
average Euclidean distances (averaged across subjects)
about mean locations of activations, varied from 9.40 to
13.36 mm and had no significant differences by region
(P > 0.05, 0.20). Intensity variability was also rela-
tively small and homogenous across brain regions. In
contrast, response extent was much more variable both
across subjects and across brain regions (0.79 to 1.77,
coefficient of variation). These findings are in good
agreement with previous PET studies of intersubject
variability and bode well for the possibility of using
functional neuroimaging to study neural plasticity sub-
sequent to congenital and acquired brain lesions. © 2000
Academic Press
Key Words: PET; functional human brain mapping;
functional anatomy; plasticity; verb generation.
INTRODUCTION
Individual variability in the functional organization
of the human brain is an issue of relevance for basic
neuroscience, for clinical neuroscience and clinical ap-
plications of brain-mapping methods. The nature, pat-
terns, and magnitude of intersubject variability in
functional pacellation undoubtedly reflect important
properties and features of brain development. To in-
vestigate the mechanisms underlying the development of
cortical functional organization, knowledge of the aver-
age organizational pattern is insufficient. The nature and
magnitude of normal variations must also be established.
Despite the clear need for studies of variability in normal
subjects, few such studies have been published.
Functional mapping of the normal human brain is
most readily performed with positron emission tomog-
raphy (PET) and functional magnetic resonance imag-
ing (fMRI). A large and rapidly growing literature re-
ports the brain locations of a wide variety of cognitive,
perceptual, motoric, and emotional processes. Of the
PET studies to date, the majority have relied on inter-
subject averaging and have reported grand-mean ef-
fects. Rather than analyzing and reporting the func-
tional organization for each individual, the individual
mappings were averaged across subjects to increase
the detection of very subtle (2–3%) neural activations
(Fox et al., 1988; Friston et al., 1991). Intersubject
averaging is done by spatially normalizing individual
brains to a standard orientation, shape, and size,
thereby placing all brain images in a common, three-
dimensional, stereotactic coordinate space (Talairach
and Tournoux, 1988; Fox et al., 1985b). Intersubject
averaging has been instrumental in advancing our un-
derstanding of the functional organization of the hu-
man brain. The grand-averaging strategy, however,
extracts only the collective effects of neural activations
that are spatially coincident across subjects, providing
no information about individual variability in func-
tional organization of the human brain.
Neuroimaging studies that report individual sub-
jects—and, therefore, individual variability— do exist.
Unfortunately, a substantial fraction of these studies
has not reported their observations within a standard-
ized coordinate space. This is particularly true for
NeuroImage 12, 326 –339 (2000)
doi:10.1006/nimg.2000.0621, available online at http://www.idealibrary.com on
326
1053-8119/00 $35.00
Copyright © 2000 by Academic Press
All rights of reproduction in any form reserved.