ASSESSING BRAIN DEVELOPMENT USING NEUROPHYSIOLOGIC AND

BEHAVIORAL MEASURES
KATHLEEN M. THOMAS, PHD

One critical aspect of pediatric research is the assessment of outcome measures after treatment or intervention. Behavioral
measures of physical growth, school achievement, and general intelligence have proven to be important scales for assessing gross
developmental outcome and differences between pediatric treatment groups. However, more subtle and sophisticated measures
may be required to assess finer grained differences in brain development at the structural and functional levels. Advances in
noninvasive brain imaging techniques over the past decade have improved our ability to link specific cognitive functions to
changes in brain structure and function in healthy infants and children. This paper highlights some of the ways that
electrophysiologic and functional magnetic resonance imaging methods have been combined with behavioral measures of
cognitive and emotional function to advance our understanding of brain-behavior relations. Such combined neurophysiologic
and behavioral methods may help to identify the role specific interventions have on long-term developmental outcomes in
childhood. (J Pediatr 2003;143:S46-S53)

ediatric clinical research frequently includes assessments of short- and long-term

P developmental outcome subsequent to particular disease states, developmental

insults, or experimental treatments or interventions. For instance, studies of long-
term developmental outcome in children born prematurely or with various birth com-
plications tend to focus on gross measures of physical development, school achievement,
and general intelligence. Whereas these standardized measures may illuminate subtle shifts
in mean performance between groups, they are less helpful in identifying the specific
deficits that accompany particular developmental histories. Although a downward shift in
mean intelligence has significance at the population level, such results rarely inform our
understanding of the most appropriate treatments or interventions for improving the long-
term outcome of individual children. More subtle and sophisticated measures may be
required to assess finer grained differences in brain development at the structural and
functional levels. For example, although it is difficult to relate general intelligence to
specific brain systems, necessary neural systems are being identified for more specific skills
such as working memory, behavioral inhibition, unconscious learning, or emotion pro-
cessing. The application of such targeted cognitive measures may improve our un-
derstanding of the deficits incurred, as well as the competencies spared in response to early
brain insult, or the subtle effects of specific interventions or treatments.
Technological advances over the past two decades have resulted in a number of
noninvasive methods appropriate for examining brain structure and function in vivo in the
developing child. Whereas these tools can aid clinicians in diagnosing a structural or neuro-
chemical abnormality, they become even more powerful when used in conjunction with From Sackler Institute for Develop-
behavioral measures of specific cognitive functions. The combination of neuropsychologic mental Psychobiology, Weill Medical
College of Cornell University, Ithaca,
assessments or carefully controlled experimental paradigms, with structural and functional New York.
brain imaging techniques has significantly advanced the field of adult cognitive Reprint requests: Kathleen M. Thomas,
neuroscience. However, it is only throughout the past 5 to 10 years that our field has PhD, Institute of Child Development,
University of Minnesota, 51 E River
applied these methods to child development in a noninvasive way. This paper presents the Rd, Minneapolis, MN 55455. E-mail:
thoma114@umn.edu.
Copyright ª 2003 Mosby, Inc. All rights
reserved.
EEG Electroencephalogram MRI Magnetic resonance imaging 0022-3476/2003/$30.00 + 0
ERP Event-related potential RF Radiofrequency
10.1067/S0022-3476(03)00401-3

S46
Fig 1. Examples of ERP responses to familiar and novel stimuli in children and adults. The P300 component indicated by the black arrow
is typical of rare, target, task-relevant stimuli (to which the participant makes some response) and tends to show maximal amplitude at
electrode locations over central and posterior scalp regions. As demonstrated in these data, the majority of ERP responses are broader
(of longer duration) for children compared with adults, and tend to appear as large, slow voltage changes in infants (see Fig 2).

basic principles behind several available brain imaging
techniques. This paper is not intended as a comprehensive
review of the pediatric neuroimaging literature nor as a tutorial
for all available neuroimaging methods, but rather surveys the
primary techniques that we and others have used to combine
behavioral and neuroimaging measures to advance our un-
derstanding of brain-behavior relations during childhood.
METHODS
Event-Related Potentials (ERP)
Electrophysiologic techniques provide one method for
examining online brain function during development. Active
neurons generate electric signals that travel through brain
tissue and can be measured by electrodes placed on the scalp
surface, as in a standard clinical electroencephalogram (EEG).
This scalp-recorded activity is thought to reflect the in-
termittent synchronization of small populations of predom-
inantly cortical neurons.1 Voltage is measured at one or more
electrodes relative to a reference electrode, with activity closer
to the electrode location or from especially activated brain
regions presumably contributing the largest share of the
measured signal. The density of electrode placement over the
scalp determines the spatial sensitivity of the recorded signal.
Although in previous years, studies used as few as three
electrodes placed along the midline, current techniques use as
many as 256 electrodes evenly distributed across the entire
scalp surface. This high density arrangement provides a more
sensitive measure of the surface topography of the voltage
distribution, and is more amenable to mathematical tech-
niques for modeling the brain generators of the observed
signal (dipole source localization). Electrophysiologic mea-
surements require a conducting agent to allow the brain-
generated signal to reach the surface electrode. Although most
EEG systems use a conducting gel at each electrode site and
may require minor scalp abrasion, some systems require only
a salt water solution, improving the utility of this technique
with very young children.
To understand the brain activity associated with
cognitive processes, investigators have typically examined the
event-related potential (ERP), which reflects EEG activity
time-locked to particular sensory and/or cognitive events. The
ERP reflects the average response to an event over multiple
presentations, and provides a very fine scale for determining
the timing and temporal sequencing of particular neural
events. Specific sensory modalities (eg, visual or auditory) or
basic cognitive processes (eg, visual attention) produce unique
components in the event-related waveform. These compo-
nents traditionally are labeled according to their polarity
(positive peak or negative peak) and timing with respect to the
stimulus, either in order of appearance (eg, P1, P2, P3) or in
milliseconds (eg, N200, P300).2 The average waveforms can
be compared across multiple stimulus types or cognitive
events, and across study groups, and then analyzed for changes
in the amplitude or latency of these identified components.
Figure 1 illustrates a typical ERP waveform produced by
eight-year-old children compared with adults at a single
electrode location in response to a recognized visual stimulus.3
The large positive peak, called the P300 or P3b component, is
thought to reflect updating of working memory after the
presentation of a stimulus. As evident from the brief time
scale, the onset and peak magnitude of individual components
can be reliably timed to within tens of milliseconds. Individual
components demonstrate unique scalp topography and re-
sponse characteristics. For example, the P300 component is
largest at parietal compared with frontal or temporal electrode

Assessing Brain Development Using Neurophysiologic and Behavioral Measures S47

Fig 2. Example method of recording the EEG or ERP with infants. A, Example of ERP waveforms from a 6-month-old infant in
response to familiar and novel stimuli. B, Example data showing voltage across all scalp locations at a single point in time (eg, 200 msec
after stimulus presentation). These data represent changes in the voltage measured at a single electrode over time.

locations. The amplitude of the P300 component is larger for
visual stimuli that require a behavioral response (targets) than
for those that do not (standards), and is larger for rare stimuli
than for frequently repeating stimuli. Not all components are
evident at all points in development. For example, the P300
response changes in shape (morphology) across early child-
hood and does not reach its adult form until early adolescence.
In fact, the P300 component is not evident in infant ERP
waveforms (eg, Fig 2). To date, it is unclear to what degree the
change in morphology reflects structural and physiologic brain
development or development of the cognitive functions
producing the response.
In addition to these classic plots displaying voltage at
a single electrode over time, ERP data can be viewed as a map
of the scalp surface at a single point in time (eg, Fig 2). These
voltage plots, or their mathematically transformed cousin, the
current source density map, provide further information regard-
ing the spatial distribution of electrical activity. However, this
improved spatial information should not be confused with
brain localization. Both temporal and spatial voltage plots
reflect only the scalp-recorded distribution of the generated
activity and sophisticated analysis techniques are required to
infer information regarding the brain systems generating this
activity. Thus far, very few studies have used mathematical
models such as dipole source localization to estimate the brain
generators of electrophysiologic data from infants or children.
Such techniques require a large number of behavioral trials and
very low noise levels, both of which are more difficult to obtain
with young subjects.
ERP methods have proven useful in examining pediatric
group differences in specific cognitive domains such as covert
attention, recognition memory, or long-term recall.4-6 In
addition, behavioral and ERP results obtained from specific
cognitive probes can be examined for relationships with gross
cognitive outcome measures such as intelligence or school
achievement. For example, Molfese and Molfese have exam-
ined whether early language skills as measured by electro-
physiologic methods during the first year of life are predictive
of later intelligence or language function.7,8 Their data
demonstrated that an electrophysiologic measure of speech
sound discrimination (/bi/ vs /gi/) in infancy could be used in
a discriminant function analysis to predict which children
would show higher (>100) or lower (<100) verbal IQ scores at
age 5 years. Measures such as these may eventually prove

S48 Thomas The Journal of Pediatrics  October 2003

useful in identifying children who could benefit from specific
early intervention and remediation efforts in an attempt to
improve their long-term global outcome.
Physiologic measures can sometimes provide an index of
functional brain differences in the face of no behavioral
difference or impairment. For example, Nelson et al examined
memory development in infants of diabetic mothers using
both behavioral and electrophysiologic measures.5 Given the
physiologic effects of maternal diabetes during pregnancy,
some infants have chronic fetal hypoxia, reactive hypoglyce-
mia, and/or fetal iron deficiency, placing them at risk for
altered hippocampal structure and potentially for hippocam-
pal-related cognitive deficits.9-11 Nelson et al reported that 6-
month-old infants of diabetic mothers showed no difference
from control infants on recognition memory tests as assessed
by standard behavioral looking time measures, or on Bayley
standardized developmental examinations at 12 months of
age.5 However, ERP measures at 6 months of age indicated
neurophysiologic differences between groups during a recog-
nition memory task. Whereas control infants showed ERP
waveforms that differed for old and new stimuli, infants of
diabetic mothers showed no difference in brain generated
electrical activity in response to old items compared with new
items. In this particular study, the stimuli were pictures of the
infant’s mother or a stranger. Given that all of the infants
showed everyday behavior to indicate that they recognized
their mother, these ERP results suggest that memory function
may tap a different neural mechanism in these infants com-
pared with the control infants, or that the electrophysiologic
response to memory is altered in the infants of diabetic mothers.
ERP methods are particularly useful for examining very
brief cognitive events and for identifying the temporal order of
cognitive processes, but are limited in their ability to identify
the precise brain regions generating the measured signal. ERP
methods have been particularly useful with very young
subjects, including infants, given the relatively robust signal
in the face of motion artifact. Other neuroimaging techniques,
particularly the magnetic resonance imaging (MRI) methods
described later, are not appropriate for cognitive imaging in
very young children who cannot remain completely still on
command. When used in combination with structural and
functional MRI methods, electrophysiologic measures may
provide an additional layer of information regarding the
temporal dynamics of brain activity. Currently, ERP recor-
dings remain one of the few methods feasible for assessing
ongoing brain activity correlates of cognitive function in infants
and very young children.
Structural and Functional MRI
Another technique that is now being applied to
developmental research is MRI (Fig 3, A). The MRI signal
arises from the fact that charged protons behave differently in
the presence of a strong magnetic field. In the presence of an
ambient magnetic field, the constantly spinning and moving
protons tend to become aligned along the main axis of the
field. Images of the brain are generated by applying a series of
Assessing Brain Development Using Neurophysiologic and Behavioral Measures
brief energy pulses, radio frequency (RF) pulses, through the
tissue. Each RF pulse contains enough energy to momentarily
disrupt or tilt the protons in a particular plane. Measurements
of the resultant change from one energy state to another are
used to create images of the tissue. High resolution structural
images are constructed based on measures of the energy
released as pulsed protons relax back to their aligned state.
Different body tissue types demonstrate different relaxation
times, with lipids showing different relaxation times than
water.12,13 These relaxation differences can be mapped to
produce structural images of the brain emphasizing differences
between regions high in water compared with regions high in
fat (eg, T1-weighted or T2-weighted images).
One means of examining structural brain development is
to carefully quantify the size or volume of specific brain
structures using hand tracing or semi-automated methods for
segmenting the structural images into distinct brain regions.
Giedd et al have published a large body of research examining
normative changes in brain structure across development. For
example, they report decreases in the volume of subcortical
gray matter structures like the basal ganglia during child-
hood,14 whereas cortical gray matter does not appear to show
volume decreases until puberty.15 Group differences in re-
gional volume may also provide a method for identifying net-
works that are disrupted following injury or in various disease
states. For example, De Bellis et al showed that children with
generalized anxiety disorder have larger amygdala volumes
than healthy children, specifically on the right side.16 This
right volume difference was specific to the amygdala, a region
thought to be critical in processing emotional stimuli. For
instance, children with anxiety disorders did not differ from
controls in hippocampal volume. This same group of resea-
rchers reported that children with a history of post-traumatic
stress disorder (PTSD) subsequent to abuse did not show
regional volume decreases in the hippocampus as expected from
the adult PTSD literature.17 Instead, these children showed
evidence of an overall decrease in total cerebral volume, sug-
gesting a global rather than a local effect of the trauma.
MRI can be used to link brain structure to behavior
by examining correlations among MRI-based anatomical or
morphometric measures and measures of behavior. For
example, Casey et al hypothesized that anterior cingulate
cortex plays an important role in human attention.18 Two
separate behavioral measures of attention (automatic and
controlled processing) were assessed in 5- to 16-year-old
children in parallel with MRI-based anatomical measures of
the anterior cingulate cortex. After controlling for age, IQ, and
total cerebral volume, correlation analyses showed that volume
of the right anterior cingulate cortex was significantly
correlated with performance on the predominantly controlled
attention task, but not the predominantly automatic attention
task. This effect was nonsignificant for left anterior cingulate
cortex. Results like these demonstrate that very precise
cognitive probe tasks may be critical in determining the
relations among brain structure and function.
In addition to structural information, MRI techniques
can be used to assess brain activity as well. Functional MRI
S49
Fig 3. Example method of MRI measures with children. Example data are provided from four different MRI techniques. A, Structural
MRI scan through a coronal slice of the brain at the level of the lateral ventricles and basal ganglia. The white tracing provides an example of
one method used to quantitatively measure the volume of the caudate nucleus. B, An example indicating functional activity in the inferior
frontal gyrus and insular cortex during performance of a cognitive task. Regions of significant functional activation are superimposed on
a structural magnetic resonance scan in the coronal plane. C, An example of white matter fiber tracks produced from diffusion tensor imaging
measures of the degree of anisotropy in water diffusion across the brain. (Image courtesy of Dr. Richard Watts, Weill Medical College of
Cornell University). D, Typical neurometabolite concentrations in a single brain region as evidenced by phosphorus (31P) MRS (top graph)
and proton (1H) MRS (lower graph).

(fMRI) relies on the assumption that a change in brain activity
is associated with a local hemodynamic change in blood flow
and oxygenation.19-21 The most frequent type of functional
imaging, blood oxygenation level dependent (BOLD) imag-
ing, is based on the observation that oxygenated and
deoxygenated hemoglobin behave differently when pulsed by
an RF magnetic field. Deoxygenated hemoglobin is para-
magnetic and causes distortions in the local magnetic field,
leading the spinning protons in that region to drop out of
phase with one another. This change in energy state from the
in-phase to the out-of-phase spin states (called T2* relaxation)
can then be mapped to indicate the relative neuronal activity,
as indexed by changes in blood oxygenation, across the brain
(Fig 3, B). This hemodynamic or blood oxygenation change is
not instantaneous; that is, the observed signal change is
delayed in time from the presumed neuronal activity. In fact,
the peak measurable signal change would be expected to occur
as late as 5 to 6 seconds after the presumed event-related
neuronal response.
As with structural MRI, fMRI can be used to examine
correlations between brain and behavior. In the case of fMRI,
correlations are based on changes in the magnetic resonance
signal measured in a given region rather than the size or
volume of a structure. In addition, magnetic resonance signal
changes and behavioral changes can be assessed simulta-
neously as well as in separate sessions. For example, in one of
the first published pediatric fMRI studies, Casey et al
examined functional brain activity during an inhibitory control
task in children.22 Examination of the functional imaging data
revealed a number of brain regions that were more active
during inhibition than during the control task. However,
activity in only two of these regions showed a correlation with
behavioral performance on the inhibition task. Specifically,
when the task was difficult (subjects made more errors),

S50 Thomas The Journal of Pediatrics  October 2003

activity increased in anterior cingulate cortex. In contrast,
when children were performing well (making few errors),
activity was increased in orbitofrontal cortex. These results
suggest that these two regions may be heavily involved in the
development of inhibitory control, and are consistent with the
anatomical findings of a correlation between attentional
control and volume of the anterior cingulate cortex.18
In a similar manner, we have used fMRI to examine
amygdala function in healthy children and children with
anxiety or depression.23,24 Adult imaging studies have
suggested that processing of facial expressions of emotion,
particularly fearful expressions, activates the amygdala in
healthy adults.25 We found that healthy children showed
a different pattern of amygdala response than healthy adults,
showing greater amygdala responses to neutral faces than
fearful faces.23 Children with anxiety showed an exaggerated
amygdala response to fearful expressions compared with
healthy children of the same age.24 Interestingly, the degree
of hyper-responsiveness to fearful stimuli was positively
related to the child’s everyday level of anxiety as measured by
a self-report scale outside the scanner environment. This
brain-behavior correlation is consistent with the structural
group differences in amygdala volume reported earlier for
children with generalized anxiety.16
One benefit of MRI methods over the electrophysio-
logic methods discussed earlier is the ability to precisely
localize the brain region showing structural and/or functional
change between groups or between behavioral conditions. The
ability to acquire both structural and functional measures of
brain development within the same session provides a clear
advantage of MRI over other measurement techniques.
However, given the relatively slow time course of the
hemodynamic response associated with neuronal activity,
fMRI measures have relatively poor temporal resolution (on
the order of seconds) compared with ERP measures. The up-
front cost of MRI is still quite high at most institutions,
making this an unrealistic method for smaller research groups,
and MRI is not ideal for every subject population. Participants
must lie in a loud, confined space and remain still for the
duration of the scan (as long as one hour, in the case of fMRI).
Even minor head motion (on the order of 3-5 mm) can result
in significant blurring of the images and potentially un-
recoverable data. These parameters may be too restrictive for
certain groups, particularly very young children who may be
unable to stay sufficiently still to acquire artifact-free images.
Other MRI Techniques
In addition to volumetric and hemodynamic changes,
MRI can be used to assess other aspects of brain structure and
physiology. Two techniques that are gaining attention in the
pediatric neuroimaging arena are diffusion tensor imaging
(DTI) and magnetic resonance spectroscopy (MRS). Both can
be acquired with the same magnetic resonance scanner used
for traditional structural and functional imaging but index
quite different aspects of brain development.
Diffusion tensor imaging is a measure of the relative
degree of water diffusion in various brain regions. Although
Assessing Brain Development Using Neurophysiologic and Behavioral Measures
this may sound quite removed from questions of behavioral
development or cognitive performance, water diffusion can be
used as a marker for neuronal axon myelination, which we
assume to be related to the efficiency of neural communication
and therefore, cognitive functions. In an unrestricted system,
such as a bucket of water, diffusion is isotropic or essentially
equal in all directions. However, in the brain, structural
barriers prevent water from diffusing equally in all directions.
Most significantly, water diffuses more easily along the length
of myelinated fibers than across the myelin sheath. This
unequal pattern is termed anisotropic diffusion. DTI measures
can be used to estimate diffusion in many directions across
various regions of the brain. These measures of diffusion can
be compared across groups of persons as an index of the degree
of myelination. For example, patients with multiple sclerosis
show decreased anisotropy compared with age-matched
healthy adults, presumably reflecting decreases in mye-
lination.26 Likewise, DTI has been used to assess the de-
velopmental progression of myelination from infancy into
adulthood27-30 and to identify potential disruptions in white
matter efficiency or connectivity in pediatric patient groups,
such as children with spastic quadriplegic cerebral palsy.31
DTI also can be used to trace the path of individual fibers
through the brain to assess connectivity among various brain
regions (Fig 3, C).32 Most recently, several groups are using
DTI to understand the relationship between regional
anisotropy and cognitive function. For example, Klingberg et
al demonstrated that adults with a history of developmental
dyslexia showed a decrease in anisotropy in temporo-parietal
brain regions compared with healthy adults.33 Even more
interesting, however, was the finding that quantitative
measures of anisotropy were positively correlated with reading
ability. That is, across all participants, poor reading was
associated with more isotropic diffusion, presumably reflecting
relatively less efficient processing and potential disruptions in
fiber connectivity.
Whereas diffusion tensor imaging and traditional
magnetic resonance methods image tissue water content (in
the form of hydrogen protons), MRS can assess the
concentration of additional biologic compounds, particularly
metabolites related to cell viability and membrane biochem-
istry.34-36 Regional changes in the concentration of common
metabolites may provide information regarding neuronal and
glial processes, membrane phospholipids, and high-energy
phosphates. Proton (or 1H) spectroscopy yields primary
resonance frequency peaks for N-acetyl aspartate (NAA),
creatine (with phosphocreatine), choline (with glycerophos-
phocholine), and, under certain imaging parameters, myo-
inositol and glutamate/glutamine (eg, Fig 3, D). In the adult
brain, NAA is found only in mature neurons, and increases in
NAA are associated with neuronal damage, or neuronal or
axonal loss.37 In contrast, active demyelination may be
reflected by high levels of choline. Creatine and myo-inositol
are compounds more prevalent in glial cells than in neurons.
Therefore, increased concentration of these compounds may
be associated with gliosis.36 In a similar fashion, phosphorus
(31P) spectroscopy provides an assessment of membrane
S51
phospholipids and high-energy metabolism. 31P spectra
include primary resonance peaks for phosphomonoesters
(predominantly precursors of membrane phospholipids) and
phosphodiesters (predominantly breakdown products of
membrane phospholipids), as well as phosphocreatine (Fig
3, D).38 Measures of regional metabolite concentration and
membrane phospholipids during development may be useful
in assessing normative changes in brain biochemistry as well as
the effects of various interventions. For example, NAA
concentration is known to be low early in development (first
1-2 years) with corresponding high levels of choline and myo-
inositol.39 Likewise, membrane phospholipid precursors are
high early in postnatal brain development, and concentrations
of phospholipid breakdown products are low.40,41 These
biochemical techniques are particularly relevant for the current
special issue addressing potential changes in neurode-
velopment after dietary supplementation with long-chain
polyunsaturated fatty acids. To date, although many studies
have reported regional metabolite differences for patient
populations with known cognitive impairments, few studies
have addressed whether empirical measures of cognitive
performance are related to these metabolite differences.
However, a recent study of children with attention deficit-
hyperactivity disorder suggests that performance on a sustained
attention task may be correlated with concentrations of
creatine/phosphocreatine.42 This finding is particularly in-
teresting because no group differences were observed in
neurometabolite concentrations between children with and
without attention deficit-hyperactivity disorder, suggesting
that in this study, MRS measures of creatine may be related
more to behavioral symptoms across persons than to di-
agnostic category boundaries.
CONCLUSIONS
Developmental neuroimaging methods offer several
advantages over behavioral measures alone by identifying
neurophysiologic correlates of cognitive differences between
groups, or alternatively, by demonstrating structural and/or
functional brain differences between groups displaying
equivalent cognitive performance. Both the fields of de-
velopmental psychology and cognitive neuroimaging have
advanced sufficiently as to allow for more precise and focused
measures of developmental outcome after insult or interven-
tion. Although global outcome variables can inform our
understanding of broad shifts in mean performance for
a group, precise cognitive, emotional, or motor tasks must
be applied in a hypothesis driven way if we are to understand
the role that specific changes in brain structure or function play
in behavioral performance or vice versa.
The current pediatric brain imaging literature provides
a few basic lessons. First, a lack of behavioral difference
between groups, or before and after intervention, does not
necessarily imply a lack of functional change in brain activity or
organization. Many of our current behavioral measures are too
broad to capture the subtle deficits present for one group but
not another. Although these deficits are subtle on a cognitive
S52 Thomas
level, they may have great impact on the child’s everyday
function and long-term developmental outcome. Any means
that we have for identifying these deficits early in life improves
our chances of successful intervention. The same imaging
methods that aid in identifying a specific group difference may
also aid in assessing the efficacy of intervention techniques.
Second, as in other research domains, the combination
of multiple converging methods of assessment will be critical
for understanding normal human brain development and for
evaluating the impact of specific interventions or clinical trials.
Within the neuroimaging literature itself, it is clear that the
most sophisticated studies will include detailed behavioral
measures as well as both temporal and spatial assessments of
the brain systems involved in the specific behavioral process of
interest. Given the state of the current literature, descriptive
methods of research are unlikely to be sufficient. Behavioral as
well as neuroimaging measures should be designed and
implemented in the context of broader hypotheses regarding
the mechanisms of action of particular disease states or
interventions, as well as hypotheses regarding the neural
systems underlying specific behavioral functions.
REFERENCES
1. Andreassi JL. Psychophysiology: Human behavior and physiological
response. 2nd ed. Hillsdale (NJ): Lawrence Erlbaum Associates; 1989.
2. Polich J. Cognitive brain potentials. Curr Dir Psychol Sci 1993;2:175-9.
3. Thomas KM, Nelson CA. Age-related changes in the electrophysio-
logical response to visual stimulus novelty: a topographical approach.
Electroencephal Clin Neurophysiol 1996;98:294-308.
4. Carver LJ, Bauer PJ, Nelson CA. Associations between infant brain
activity and recall memory. Dev Sci 2000;3:234-46.
5. Nelson CA, Wewerka S, Thomas KM, Tribby-Walbridge S, deRegnier
RA, Georgieff MK. Neurocognitive sequelae of infants of diabetic mothers.
Behav Neurosci 2000;114:950-6.
6. Richards JE. Localizing the development of covert attention in infants
using scalp event-related potentials. Dev Psych 2000;36:91-108.
7. Molfese DL, Molfese VJ. Electrophysiological indices of auditory
discrimination in newborn infants: the basis for predicting later language
performance? Infant Behav Dev 1985;8:197-211.
8. Molfese DL, Molfese VJ. Discrimination of language skills at five years
of age using event-related potentials recorded at birth. Dev Neuropsychol
1997;13:135-56.
9. Barks JD, Sun R, Malinak C, Silverstein FS. gp120, an HIV-1 protein,
increases susceptibility to hypoglycemic and ischemic brain injury in perinatal
rats. Exp Neurol 1995;132:123-33.
10. Nelson C, Silverstein FS. Acute disruption of cytochrome oxidase
activity in brain in a perinatal rat stroke model. Ped Res 1994;36:12-9.
11. Rao R, DeUngria M, Sullivan D, Wu P, Wobken JD, Nelson CA, et al.
Perinatal iron deficiency increases the vulnerability of rat hippocampus to
hypoxic ischemic insult. J Nutr 1999;129:667-75.
12. Bloch F. Nuclear induction. Physiol Rev 1946;70:460-74.
13. Young SW. Magnetic resonance imaging: basic principles. New York
(NY): Ravens Press; 1988.
14. Giedd JN, Snell JW, Lange N, Rajapakse JC, Casey BJ, Kozuch PL,
et al. Quantitative magnetic resonance imaging of human brain development:
ages 4-18. Cereb Cortex 1996;6:551-60.
15. Giedd JN, Blumenthal J, Jeffries NO, Castellanos FX, Lui H, Zijdenbos
A, et al. Brain development during childhood and adolescence: a longitudinal
MRI study. Nat Neurosci 1999;2:861-3.
16. De Bellis MD, Casey BJ, Dahl R, Birmaher B, Williamson D, Thomas
KM, et al. A pilot study of amygdala volumes in pediatric generalized anxiety
disorder. Biol Psychiatry 2000;48:51-7.
The Journal of Pediatrics  October 2003
17. De Bellis MD, Keshavan MS, Clark DB, Casey BJ, Giedd JN, Borin
AM, et al. Developmental traumatology part II: brain development. Biol
Psychiatry 1999;45:1271-84.
18. Casey BJ, Trainor R, Giedd J, Vauss Y, Vaituzis CK, Hamburger S. The
role of the anterior cingulate in automatic and controlled processes:
a developmental neuroanatomical study. Dev Psychobiol 1997;30:61-9.
19. Ogawa S, Lee TS, Nayak AS, Glynn P. Oxygenation-sensitive contrast
in magnetic resonance image of rodent brain at high magnetic fields. Magn
Reson Med 1990;26:68-78.
20. Turner R, Le Bihan D, Moonen CTW, Despres D, Frank J. Echo-
planar time course MRI of cat brain oxygenation changes. Magn Reson Med
1991;22:159-66.
21. Kwong KK, Belliveau JW, Chesler DA, Goldberg IE, Weisskoff RM,
Poncelet BP, et al. Dynamic magnetic resonance imaging of human brain
activity during primary sensory stimulation. Proc Natl Acad Sci U S A
1992;89:5675.
22. Casey BJ, Trainor RJ, Orendi JL, Schubert AB, Nystrom LE, Giedd JN,
et al. A developmental functional MRI study of prefrontal activation during
performance of a go-nogo task. J Cog Neuro 1997;9:835-47.
23. Thomas KM, Drevets WC, Whalen PJ, Eccard CH, Dahl RE, Ryan
ND, et al. Amygdala response to facial expressions in children and adults. Biol
Psychiatry 2001;49:309-16.
24. Thomas KM, Drevets WC, Dahl RE, Ryan ND, Birmaher B, Eccard
CH, et al. Amygdala response to fearful faces in anxious and depressed
children. Arch Gen Psychiatry 2001;58:1057-63.
25. Whalen PJ, Rauch SL, Etcoff NL, McInerney SC, Lee MB, Jenike MA.
Masked presentations of emotional facial expressions modulate amygdala
activity without explicit knowledge. J Neurosci 1998;18:411-8.
26. Guo AC, MacFall JR, Provenzale JM. Multiple sclerosis: diffusion
tensor MR imaging for evaluation of normal-appearing white matter.
Radiology 2002;222:729-36.
27. Klingberg T, Vaidya CJ, Gabrieli JD, Moseley ME, Hedehus M.
Myelination and organization of the frontal white matter in children:
a diffusion tensor MRI study. NeuroReport 1999;10:2817-21.
28. McGraw P, Liang L, Provenzale JM. Evaluation of normal age-related
changes in anisotropy during infancy and childhood as shown by diffusion
tensor imaging. Am J Roentgenol 2002;179:1515-22.
29. Mukherjee P, Miller JH, Shimony JS, Philip JV, Nehra D, Snyder AZ,
et al. Diffusion-tensor MR imaging of gray and white matter develop-
ment during normal human brain maturation. Am J Neuroradiol 2002;
23:1445-56.
30. Schmithorst VJ, Wilke M, Dardzinski BJ, Holland SK. Correlation of
white matter diffusivity and anisotropy with age during childhood and
Assessing Brain Development Using Neurophysiologic and Behavioral Measures
adolescence: a cross-sectional diffusion-tensor MR imaging study. Radiology
2002;222:212-8.
31. Hoon AH, Lawrie WT, Melhem ER, Reinhardt EM, Van Zijl PC,
Solaiyappan M, et al. Diffusion tensor imaging of periventricular leukomalacia
shows affected sensory cortex white matter pathways. Neurology 2002;
59:752-6.
32. Poupon C, Mangin J, Clark CA, Frouin V, Regis J, Le Bihan D, et al.
Towards inference of human brain connectivity from MR diffusion tensor
data. Med Image Analysis 2001;5:1-15.
33. Klingberg T, Hedehus M, Temple E, Salz T, Gabrieli JD, Moseley ME,
et al. Microstructure of temporo-parietal white matter as a basis for reading
ability: evidence from diffusion tensor magnetic resonance imaging. Neuron
2000;25:493-500.
34. Howe FA, Maxwell RJ, Saunders DE, Brown MM, Griffiths JR. Proton
spectroscopy in vivo. Magn Reson Quarterly 1993;3:31-59.
35. Ross BD, Michaelis T. Clinical applications of magnetic resonance
spectroscopy. Magn Reson Quart 1994;10:191-247.
36. Urenjak J, Williams SR, Gadian DG, Noble M. Proton nuclear
magnetic resonance spectroscopy unambiguously identifies different neural
cell types. J Neurosci 1993;13:981-9.
37. Petroff OA, Graham GD, Blamire AM, al-Rayess M, Rothman DL,
Fayed PB, et al. Spectroscopic imaging of stroke in humans: histopathology
correlates of spectral changes. Neurology 1992;42:1349-54.
38. Stanley JA, Pettegrew JW, Keshavan MS. Magnetic resonance
spectroscopy in schizophrenia: methodological issues and findings–part I.
Biol Psychiatry 2000;48:357-68.
39. Kreis R, Ernst T, Ross BD. Development of the human brain: in vivo
quantification of metabolite and water content with proton magnetic
resonance spectroscopy. Magn Reson Med 1993;30:424-37.
40. Buchli R, Martin E, Boesiger P, Rumpel H. Developmental changes of
phosphorus metabolite concentrations in the human brain: a 31P magnetic
resonance spectroscopy study in vivo. Ped Res 1994;35:431-5.
41. Hanaoka S, Takashima S, Morooka K. Study of the maturation of the
child’s brain using 31P-MRS. Ped Neurol 1998;18:305-10.
42. Yeo RA, Hill DE, Campbell RA, Vigil J, Petropoulos H, Hart B, et al.
Proton magnetic resonance spectroscopy investigation of the right frontal lobe
in children with attention-deficit/hyperactivity disorder. J Am Acad Child
Adol Psychiatry 2003;42:303-10.
43. Thomas KM, King SW, Franzen PL, Welsh TF, Berkowitz AL, Noll
DC, et al. A developmental fMRI study of spatial working memory.
Neuroimage 1999;10:327-38.
S53