INTRO-PAULS STUFF [Link]

htm

Functional magnetic resonance imaging (fMRI):

A window into the brain

[Link] and [Link]

Copyright 2002 A. Parry and P.M. Matthews

Download Word version here

Centre for Functional Magnetic Resonance Imaging of the Brain

Department of Clinical Neurology

University of Oxford

The John Radcliffe Hospital

Headley Way

Headington, Oxford OX3 9DU UK

Correspondence to Prof. P. M. Matthews

Tel: 0044 1865 222729
Fax: 0044 1865 222717
Email: paul@[Link]

Abstract
Functional MRI (fMRI) has enabled scientists to look for the first time into the human brain in vivo,
to literally watch it while it works .



This has revealed exciting insights into the spatial and

temporal changes underlying a broad range of brain functions, such as how we see, feel, move,
understand each other and lay down memories. The technique is safe, allowing repeated
examinations to probe time-dependent changes such as those involved in learning. fMRI also is
improving our understanding of a variety of brain pathologies. Some, such as the addictive

1 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

behaviours of gambling or drug abuse, are without structural brain changes. In other cases in which
there are clear structural changes, fMRI has shown that functional changes occur in widely distributed
regions of the normal appearing brain. Understanding these and the adaptive potential that they confer
promises to guide rational approaches to improved rehabilitation strategies. Here we briefly review
these and related applications of the technique for better understanding of the healthy and diseased
brain.

2 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

Magnetic resonance imaging: seeing brain structure

What connection could an ancient Greek shepherd have with modern brain science? The answer is
magnetism. Two thousand years ago a shepherd named Magnes came across a rock with very
peculiar qualities. It attracted bits of iron and seemed to orient itself as if magically when held from a
string. The rock was magnetite and the observations were the first in a long history of physical
studies of magnetism that extend to this day.

The connection with brain science comes because magnetism allows us to see the living brain. You
may not realise this, but your brain is composed of billions upon billions of tiny magnets. Each of the
hydrogen atoms in each of the molecules of water in your brain is a tiny magnetic dipole. There are
over 100,000,000,000,000,000,000,000 in even a teaspoon-size scoop of your brain! When these tiny
magnets in the brain tissue are placed inside of a very strong magnetic field, they align with the field
just as a compass aligns with the earth s magnetic field. Of course, a very strong magnetic field is
needed to make this happen. Typically, a supercooled (to a temperature of about 270o below zero)
magnet over 50,000 times more powerful than of the earth s magnetic field is used. A short pulse
of radiofrequency energy perturbs these tiny magnets from their preferred alignment. As they
subsequently return to their original position they give off small amounts of energy that can be
detected and amplified with an antenna or receiver coil placed directly around the head.



The
detected signal allows us to distinguish water molecules and to distinguish the relative amounts in
each part of the head.

Like other tissues in the body, over 70% of brain is composed of water. Thus, finding out how the
water is distributed in the head is just like localising brain tissue. Different parts of the brain have
slightly different amounts of water. Nerve cells, for example, are relatively rich in water, whereas the
fatty coating around the long nerve fibres and cells, called myelin, has less. This generates contrast
between the surface cortex and the underlying white matter of the brain that can be used to provide
exquisite details of structure by the technique of magnetic resonance imaging or MRI (Fig. 1).

Figure 1. A. A sagittal MRI image from a healthy brain. Different structures in the brain can be
defined with exquisite detail. B. MRI imaging contrast is based primarily on the distribution of water
and differences in its physical environment in different tissues. Here the differences between signals
from the cortical grey and the underlying white matter are illustrated (Images courtesy of Mr. P.
Hobden and Dr. S. Clare).

3 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

The brain is a very unprepossessing organ at first glance. It has a grey and rather uniform
appearance. Through the window of MRI, however, a rich internal structure can be defined.



MRI can provide a detailed picture of the brain with resolution of less than 0.5 mm, about the size of
the tip of a typical ballpoint pen. Even structures buried deep within the centre of the brain
immediately can be made apparent. The rich network of arteries, including those penetrating deep
into the substance of the brain that are as small in diameter as a hair, also can be seen with MRI

However whilst MRI has provided us with a detailed knowledge of the structure of the brain, it does
not tell us how the brain functions. We would like to understand the neural basis for our behaviour,
such as how we are able to learn new skills, remember important facts and integrate and make sense
of the large amount of ever-changing stimuli in our enviroment. With the development of functional
imaging, we are now able to look inside the brain of humans in real time and appreciate the



neural mechanisms behind our behaviour, rather than just observing the consequences of these
effects.

A short history of mapping brain functions



The unassuming character of the surface of the brain led scientists to underestimate its importance until relatively late in

the history of medicine. Even after its significance was appreciated in the 17th century, the way in which it was organised

was debated hotly. In keeping with the relatively uniform appearance of the brain, many scientists believed in the theory

of mass action which stated that areas of the brain worked together for every task without much regional functional

specialization 1 .

4 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

In contrast, the phrenology school of the Viennese physician Thomas Gall argued that specific types
of brain function were associated with particular regions of the brain that could be localised by
palpating the overlying bumps on the skull surface. The notion that particular regions of the brain
acted as specific functional modules formed the foundation for the concept of discreet cortical



localization. Gall and his collaborators however further assumed that the structure of the brain was
linked directly to character , with individual brain regions determining individual traits 2. Sadly



for Gall, most of his specific theories proved to be incorrect. However, the notion of cortical
functional localisation remains supported by current science, although we now think more in terms of
specific types of computational processes being performed by individual brain regions.

Phrenology gave way to a different localised approach to mapping the mind based on lesion studies.
Many scientists were impressed by the observation that damage to different areas of the brain
produced relatively specific deficits in patients. One particularly well-known 19 th century example
was a patient made famous by the eminent French neurologist Charcot and known to us as Tan



because this was the only sound that he could utter. Charcot was impressed that Tan s inability to
speak was not associated with either mental deficiency, problems of movement, or impairment of
language comprehension. When his patient finally died, Charcot found a single large tumour pushing
on and damaging the lower left side of the front of the brain. He therefore hypothesised this area of
the brain had a specific function for spoken language 3.
Although human lesion studies such as these provide valuable information concerning the brain
regions necessary to perform a task, they are limited in several respects. From a practical perspective
there are few instances in which a lesion (e.g., a brain tumour) is confined only to one discreet brain
region. Most lesions spill out into adjacent regions, making it difficult to interpret the results
unequivocally. In addition, the presence of focal abnormal brain tissue may lead to changes in the
functioning of adjacent normal looking tissue by indirect effects.



For example, an expanding

tumour could cause bruising of the adjacent brain tissue necessary for a specific task with



subsequent deterioration in brain function, the region occupied by the lesion itself not actually being
necessary for task performance.

A whole new appreciation of the ways in which the brain becomes structurally specialised for
sensation, cognition and direction of action is developing with use of the extraordinary ability of
magnetic resonance imaging to resolve small changes in the healthy living brain that can be related to
behaviour. Surprisingly reminiscent of aspects of the phrenologists theories, there are now

5 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

intriguing examples of situations in which small, local variations in brain shape (rather than skull
shape!) are correlated directly with changes in brain function. For example, it has been reported that
there is a relationship between the size of the brain region which controls hand function and the
number of years of practice of a musical skill, suggesting that the structural variability in this brain
region occurs as a consequence of experience or use rather than reflecting individual differences
within the population 4. Similarly, a recent study reported that London taxi drivers have an increased
size of the part of the brain involved in the type of memory used for map reading 5. The limitations
of these cross-sectional studies however are that they do not allow us yet to conclude confidently that
structural variation arises as a consequence of practiced behaviour.

Functional magnetic resonance imaging: seeing brain function

Functional magnetic resonance imaging (fMRI) is a relatively recent imaging technique that is able to
avoid some of the problems in the interpretation of lesion or purely structural imaging studies. It
aims to determine the neurobiological correlate of behaviour by identifying the brain regions (or

functioning modules ) that become active during the performance of specific tasks



in vivo.
In addition, the non-invasive and relative safety of the technique allow repeated studies to be carried
out within a given subject so that important questions, such as the relationship between
experience-dependent use and changes in brain structure or function can be addressed 6. To
appreciate how fMRI exploits the demands of neural activity to provide spatial and temporal
information on brain function it is important to first understand the basic functional unit of the brain.

The basic functional unit of the brain is a cell called a neuron. A typical neuron has a cell body (the
central command centre ), from which project short processes called dendrites that receive



message inputs from other neurons and then relay the information back to the cell body.


A
neuron also has a long process called an axon that conducts electrical impulses away from the cell
body towards the dendrites of other neurons. The region where the axon from one neuron and the
dendrites from a second make contact is called the synapse. When electrical impulses travelling
down the first axon reach a certain threshold they cause the release of local chemical
neurotransmitters from the end of the axon into the synapse and onto receptors on the nearby
dendrites of the neighbouring neuron. The contact of neurotransmitter with these receptors may then
trigger the transmission of a nerve impulse down the second neuron and then onto further neurons
until that a large number are activated coherently in a functional network .



Although the most

interactions are excitatory , some synapses are inhibitory and limit subsequent neuronal




excitation. Importantly, after the neurotransmitters have been released into the synapse, they are

6 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

recycled and re-enter the neuron in a process that requires energy. FMRI makes indirect use of this
local need for energy to image the regions of the brain recruited into the activated neural network 6,
7.

Blush at the thought!



Like so many old phrases, there are many levels on which this admonition
can be appreciated. In fact, we do blush with thought.



In the last decade of the nineteenth

century, the Oxford physiologist Charles Sherrington made the observation that when a small area of
exposed cat brain was stimulated electrically (an exaggerated form of the changes that occur when the
region becomes active in thought) there is a flush of red blood 8. This increases the blood supply to
the brain locally, ensuring an adequate supply of oxygen to regions working harder in thinking.

This effect can be used to precisely map areas of the brain involved in brain function. Fortuitously,
the properties of the small bar magnets of water molecules in the brain change slightly between areas
that are near blood with its oxygen exhausted relative to those near freshly oxygenated blood 9. The
local increase in energy requirements arising as a consequence of neuronal firing is largely met
through an increase in oxygen-based metabolism with the increased demand for oxygen being
delivered seconds later by an increase in the local blood flow (the hemodynamic response) 10.
Changes in the oxygenation level of the blood therefore occur as a consequence of neuronal activity
and so the magnitude of change in signal intensity can be used as an indirect measure of excitatory
input to neurons which is generally related closely to the cell firing rate 11, 12, 13. Thus, when there
is a blush from increased activity, there is a small increase in signal intensity on the MRI scan.



If the MRI experiment is done while a mental task is given to a subject, a so-called functional
magnetic resonance image (or fMRI) image is generated. On such an image we can see how different
tasks activate different parts of the brain relatively specifically. When we listen to music, for
example, a specialised area in the so-called auditory cortex along the sides of the brain shows
increased signal 14. Vision activates a region in the back of the brain (the occipital cortex), localised
precisely to regions of the visual field 15 (Fig. 2). Touch brings increased signal along the side of the
brain, particularly in the side of the brain opposite to the part of the body that is touched 16 and
movements activate regions in the front and the top of the brain in cortex specialised for motor
control 17.

Figure 2. A healthy subject was asked to listen to sentences being spoken while watching a screen

7 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

with a flashing checkerboard presented. The sentences started and stopped at slightly different times
than the flashing picture was turned on and off. On the basis of the differences in timing activation in
the brain, the areas responsible for hearing (in the middle of the brain in grey) and vision (in the back
of the brain in white) could be localised by functional magnetic resonance imaging (fMRI) (Image
courtesy of Dr. S. Smith from [Link]).

FMRI therefore is able to identify the brain regions that become more active during specific task
performance. However, there are several caveats to interpretation of the increase in fMRI signal. It
is likely that the magnitude and extent of the hemodynamic response is a measure of more than a
single energy-requiring process. It may reflect not only the frequency of local excitatory synaptic
input, but also the extent of the post-synaptic depolarization (neuronal firing) 12. Whilst it may be
less energy-consuming, the firing of inhibitory neurons also could contribute to an increase in local
energy requirement and subsequent hemodynamic response 18, 19. Thus, regions that appear to be

activated on fMRI as a consequence of task could sometimes be selectively inhibited.



What
may be more important is the varying inhibitory input could change the excitatory input necessary for
post-synaptic depolarization and thus the magnitude of the fMRI response (reflecting the local
metabolic cost ).



A more fundamental issue in interpretation is that it is not the neuronal response that is monitored
directly, but a surrogate , the hemodynamic response.



The hemodynamic response (occurring

8 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

over seconds) is much slower than the neuronal response (occurring over tens to hundreds of
milliseconds). Although there is typically a delay of between 4-6 seconds between the onset of the
task and the peak of the hemodynamic response (the latency of the response), the temporal resolution
of fMRI as a technique would not be compromised if the shape of the hemodynamic response
function was fixed and was simply convolved with the time course of the task-induced neural
activity. However, there is some evidence that characteristics of the hemodynamic response may
vary between different individuals 20, different brain regions, with the nature of the task 21 or in the
presence of disease 22. The unknown potential variability in the latency of the hemodynamic
response limits precise interpretations of the magnitude and the temporal resolution of the signal 23.

The latter problem however does not generally limit uses of fMRI as a mapping technique.



For
such experiments, the time-course of the response is itself not the most critical issue. A standard
method of task presentation uses a box-car or block design where blocks of the stimulus or




task are presented typically for 20-30 seconds, alternating with periods of rest or a control



condition. The control task is chosen carefully such that it activates all of the neural processes

common to the stimulus-task with the exception of the cognitive process of interest. By subtracting
the brain regions recruited during the performance of the control task from the brain regions recruited
during the test condition, the areas of the brain whose activity is associated specifically with the
cognitive process of interest can be identified.

An alternative experimental approach is to present stimuli as isolated brief events separated in time so
that the individual response to single events can be identified. The principle advantage of this

event-related approach is that it avoids the potential confounding factors of habituation or



fatigue which may arise in a block design as a consequence of the presentation of repeated identical
stimuli. While this approach can preserve much of the temporal information in the hemodynamic
response, it often is not used.

Using fMRI to investigate the normal and diseased brain

By using a combination of these and other experimental approaches fMRI has provided important
insights into the neural basis of a multitude of normal human cognitive processes, such as our ability
to memorise information 24 or recognise faces 25, how we experience pain 26 and how subjective
experiences can be modulated by the context of the stimulus presentation 27. FMRI also has
provided a greater understanding of the neurobiological basis for problems of drug addiction or

9 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

gambling! 28, 29.

One of the main advantages of fMRI is that the same subject can be studied a number of times in a
longitudinal design. This approach has been used to investigate both how healthy subjects learn new
complex tasks 30 and also to understand how people with initially disabling neurological disease,
such as a stroke, improve functionally over time 31. The clinical relevance of fMRI also has been
strengthened by recent experiments which have explored the neurobiological basis of specific
rehabilitation strategies associated with the recovery of function 32. Other clinical applications of
fMRI such as the pre-surgical mapping of eloquent cortex 33 and new approaches to the early
diagnosis of disease 34 also suggest that fMRI will soon play an important role in the routine
management of patients with neurological disease.

How do we remember?
Most people have sat in an exam room and been unable to remember a specific fact or figure, despite
knowing that they had previously revised the topic area well. How is it that we are able to remember
some facts and not others? Is it a random process related to the failure to retrieve the information
from our memory bank , or is it directly related to how well we encoded the information when



we were revising for the exam?

Unfortunately, simple behavioural tests of memory are not be able to distinguish these possibilities as
the assessment of memory encoding is limited by the necessity to retrieve the previously encoded
facts 35. However, the ability of event-related (a series of single trials) fMRI to break down the
complex cognitive processes involved in memory into the separate components of memory encoding
and retrieval, avoids this confound.
Brewer and colleagues 36 scanned subjects whilst they were shown a series of pictures and had to
decide whether each picture represented either an inside or an outside scene. Thirty minutes after
completion of the test, subjects were given an unexpected memory test and had to indicate whether a
picture presented had been included in the original picture list. By comparing the brain activation
during the presentation of a picture in the original test which the subjects later correctly identified as
being included in the original test, to the brain activation occurring when subjects were shown a
picture in the original test which they subsequently did not recall as being in the original picture list,
the experimenters were able to demonstrate that how well a picture was remembered was related to
the amount of activity in a particular bit of the brain. This suggests that the likelihood of us correctly
remembering a piece of information is related directly to how well we absorb the information

10 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

initially, or the strength of the encoding process.

In addition to remembering facts we learnt weeks or months previously, we sometimes have to

remember a number of different facts in a short space of time and apply them to solving immediate
problems. How do we juggle and retain these facts in our mind so that we can remember them all at
the appropriate time? Braver and colleagues 37 investigated the neural basis for this type of memory
using fMRI with a parametric design. In essence, parametric experimental designs aim to identify
brain regions where the pattern of activation changes in accordance with the different demands of the
task across several levels of difficulty, thus increasing the likelihood that the brain regions identified
are specific to the cognitive processes involved in the task. The scientists used a sequential letter task
in which the difficulty of the task varied in an incremental manner, thereby in effect generating a

dose-response curve for the neural activity relative to task and demonstrated that two specific


regions in the front of the brain appeared important for correct task performance.

Recognising faces and objects

Even for a given sense, such as vision, the patterns of brain activation can be highly specific, because
different parts of the brain work together in order to generate the complexity of thought. For
example, while the primary visual cortex in the back of the brain is the most active region with simple
visual stimulus such as a flashing checkerboard, if we see a familiar face, a very specific area along
the base of the back of the brain in the so called fusiform gyrus becomes active 25. This is an area
that appears specialised for complex types of visual experience about which we have some form of

expert knowledge.


Andrews and colleagues provided a lovely illustration of this in a study of brain activity during
observation of the Rubin vase ambiguous character. This black and white figure is perceived
alternatively as either an old woman s face or a vase. Using a special trick, Andrews was able to
study subjects whose attention had been arrested in one state or the other perception. He found that
when subjects viewed the same stimulus as a face, the fusiform gyrus at the base of the brain in the
back became active, but when they viewed it as a vase, this activity disappeared- although the image
had not itself changed! 38

A similar, exquisite functional specialisation holds in other areas of the brain. For example, in areas
of the lateral temporal lobe, on the lower, middle part of the brain, there are regions specialised for
processing of sound and different regions become active depending on the precise nature of the sound

11 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

being heard. Narain and colleagues have found that meaningless or noise sound is perceived more
towards the front of the so-called superior temporal sulcus, whereas meaningful sound interpreted as
speech is heard behind this area 39

Everything depends on context

Depending on the context, even the same stimulus can be perceived very differently. Most of us have
experienced the downside of gluttony, the glorious first taste of our favorite pudding changing to the
horrible sense of over-satiation when we have over-indulged!

Edmund Rolls and his colleagues have studied just this phenomenon. They presented the taste of
bananas to subjects while they were being studied with fMRI. All of the subjects found the taste
pleasant and activated in an area along the middle of the lower part of the front of the brain in the
orbital frontal cortex. The experiment then was repeated after the subjects had been asked to eat half
a dozen bananas. But now when they had the banana taste in their mouth, with full stomachs and the
memory of bananas they had just eaten, the taste was no longer pleasant and activated the orbital-
frontal cortex in a different way 40
This experiment also illustrates that fMRI can be used as a window on entirely subjective, individual
experience. One particularly exciting application for this is using the techniques to tell us about the
experience of pain. Tracey and colleagues have been studying the way in which the brain processes
pain. They have used a novel way of testing in which they train subjects that when they see one
particular colour of light, they will experience a painful burning sensation. By studying the brain
response that follows the coloured lights separately from the brain responses following the painful
stimuli, Tracey s group have been able to differentiate the distinct aspects of anticipation and
response to pain.

They found that the anticipation involves areas of the brain known to be important in alerting



responses. The pain response involves areas known to respond to touch. However, what may be the
most important is that activation of emotional centres of the brain in the limbic system was found
only with anticipation of pain or pain itself 26.

Also activated by pain is a small region deep in the lower part of the brain known as the
periaqueductal grey. This is a very primitive part of the brain, through which sensory information
from the body must pass to go to the brain. Although it is well known that the perception of pain can
be modulated by psychological factors such as stress, the neural basis for this was poorly understood.

12 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

Tracey and her colleagues aimed to define the neuroanatomical location where attention could
modulate the perception of pain. During painful stimuli, subjects were asked either not to pay
attention to the pain or to focus on the pain. When the subjects were distracted, the response to pain
in the periaqueductal grey was changed 39. It appeared as though this area in the lower part of the
brain (the brainstem), between the spinal cord input and the cerebral cortex, was gating the
responses. With such experiments we can begin to build up a model circuit in which functionally
distant areas in the brain interact in order to produce full subjective responses.

The modulation of our perception of pain therefore may explain common phenomena, such as why
when we are emotionally stressed that we feel pain much more. Similarly, these experiments may
help us to understand the mechanism behind reports of humans claiming not to perceive pain from a
real injury when their attention is focused on a more immediate danger e.g., rescuing their child from
a burning house. The definition of such networks is important, because it is a first step towards
designing ways of controlling their responses. With a direct readout of the response of the brain



to pain using fMRI, it is possible to measure the effects of pain-relieving drugs and to develop new
strategies for pain relief.

Why do gamblers gamble?

FMRI also may cast a window into the neural mechanisms underlying certain maladaptive
behaviours. Most people at some point have placed a monetary bet on a horse or on a game. If we
won we might be tempted to place another bet, especially if the odds looked favorable. However if
we lost a lot of money we would be unlikely to keep gambling for fear of losing even more. However
some people are unable to stop, and seem almost oblivious to consequences of the loss they are
incurring.

O Doherty and colleagues used fMRI to probe the neural basis of monetary reward and punishment
using a gambling experiment.

Subjects were presented with two stimuli, one that was associated
with a high monetary reward and comparatively low monetary punishment, and an alternative stimuli
where the converse was true. Subjects had to use trial and error to determine which stimulus was the
more profitable to choose, to keep track of the profitability and to reverse their choice of profitable
stimuli when the experimenters reversed the rewards and punishments associated with each type of
stimuli.

The scientists found that particular areas of the brain were associated with reward and punishment

13 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

respectively and that there was a correlation between the magnitude of the brain activation and the
magnitude of the rewards and punishments received, thus suggesting that these regions were involved
in the representation of the size of rewards or punishments. They concluded that damage to this
region of the brain may lead to the inability to represent the size losses or gains incurred and
consequently result in difficulty in judging the degree to which a particular stimulus choice is
advantageous on the basis of cumulative monetary gain 28.

Learning how we learn

One of the main advantages of fMRI is that as a non-invasive and safe technique, studies can be
performed again and again in the same individuals. This has allowed scientists to monitor directly the
ability of the brain to adapt with time 41. The reason that most of you are currently reading this is
article is that you are trying to learn. Learning involves strengthening new connections in the brain to
change the relationship between brain input and output.

In studies, of how a motor skill is learned, for example, fMRI allows the way the brain changes
functionally over time to be followed, contributing to an understanding of which brain circuits are
involved in developing the skills 42. As one example, Heal and colleagues (R. Heal, unpublished
data) have been looking at what happens when people begin to learn a complex button pressing
sequence which is rather like learning to play a simple piano tune. Initially, the sequence is unknown
and subjects have to attend very closely, remembering carefully each finger movement before going
on to the next. This creates large demands in the prefrontal cortex, part of which is involved in
processing for short-term memories, as well as in areas more traditionally recognised as important for
motor functions. A surprising observation has been that when subjects are trying to learn something
they use motor areas on both sides of the brain, while when they perform the same button pressing
task under conditions in which they are not required to be learning the task, they activate a much
smaller region, mostly in the hemisphere on the opposite side of the brain to that in which the hand is
moved. Thus, the regions of the brain that becomes active for the same movement task depend upon
the context in which the movement has been performed (unpublished observations). It also may be
relevant that under the conditions of learning in which the movement is taxing, a greater amount of
brain becomes active. It is as if the effortfulness of the task is related to the amount of brain
involved.

Similar studies have demonstrated that increases in activation in the main motor areas persist for
several weeks after training is stopped 30. The authors suggest that the repeated practise of a new

14 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

task may lead to modifications in how the relevant brain regions recruited in the task behave. One
possible mechanism for this may include the strengthening of links between different motor neurons
as a consequence of repeated correlated firing. Structural changes also may occur as a consequence
of repeated activity 30 and be mediated via a change in the strength of the synaptic connections 43
which may be associated with changes in the morphology of dendrites, the extensively branching
arms of neurons on which the synapses form 44.



Similarly, numerous fMRI experiments have demonstrated that repeated perception of a stimulus may
lead to functional reorganization in the area of the brain (the somatosensory cortex) which detects
sensations, e.g., touch to the hand. The representation of the digits of the skilled hand is expanded in
string musicians 45 and also in blind Braille readers in comparison to the unskilled hand 46 in this
region of the brain, suggesting use-dependent reorganization.

Watching the brain heal itself

The experiments outlined above have suggested that cortical activity may be modulated by learning
and experience. Analogous to the changes in brain function that arise during the learning of a
complex skill in a healthy subject, could patients with damage (e.g., caused by a stroke) to a part of a
neural network used to perform a specific task regain the lost function by recruiting a new area



of the brain to take over? Patients who have had an acute stroke have been shown to make
spontaneous recovery of function beyond the period where healing of the damaged tissue appears to
have reached a plateau 47. This suggests that mechanisms other than resolution of the initial injury
may contribute to recovery. If such changes could be defined with fMRI, then results could be used
to harness the mechanisms to maximise recovery after injury, e.g., with rehabilitative physiotherapy.

A number of fMRI experiments have now provided exciting evidence that the brain may be capable
of functional re-organization after injury. Similar to the way that healthy subjects learn a complex
task, a number of studies have shown that subjects who have recovered successfully after a stroke
show increased activity in motor areas on the same side as the affected hand, rather than just in the
opposite side of the brain normally used to move the hand (Fig. 3) 48, 49, 50. This suggests that
parallel motor pathways from both sides of the brain, normally kept relatively inhibited in the healthy
individual, are activated to compensate for injury to one of them.

Figure 3. The pattern of brain activation associated with a simple hand movement changes after a
stroke causing arm weakness. FMRI activity with movement of the affected hand shows recruitment

15 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

of both sides of the brain as an adaptive response to the injury, rather than primarily just the side of
the brain opposite the hand moved, as with the unaffected hand (Image courtesy of Ms. H.
Johansen-Berg).

Subtle changes in the location of brain activity observed in the motor areas controlling the affected
hand also have been reported. As one example, Pineiro and colleagues studied the movement of the
right hand in right-handed healthy controls and also in patients who had suffered a stroke affecting
the function of that hand. In addition to showing a greater relative increase in the extent of the
activation in the motor area on the same side as the affected hand, they also noted that the centre of
activation in the main motor area opposite to the side of the affected hand in patients was shifted
posteriorly relative to that of control subjects, thus confirming altered patterns of activation in both
sides of the brain 31. This suggests that modulation of widely distributed parts of the cortical
network for motor control are associated with changes in the pattern of brain activation that may
contribute to functional recovery after acute injury.

We have found similar results in patients with multiple sclerosis, a chronic neurological disease with
superimposed acute exacerbations, radiologically characterised by the presence of multiple widely
distributed and predominantly white matter lesions on MRI. FMRI can be combined with more
conventional MRI approaches to provide simultaneous information on both structural and functional

16 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

pathology. Studies have shown that patterns of brain function are altered by the lesions and that the
degree to which cortical activation is altered is correlated with the severity of the underlying brain
injury 51. Recruitment of new brain areas may be adaptive and able to limit the progression of



manifestations of disease 52.

Providing compelling evidence for this remains challenging. A critical question is whether the
altered patterns of activation in patients truly reflect an adaptive response that contributes to the
recovery process, or are only a bystander reaction of a damaged brain? To address this problem, the
ability to use fMRI to perform serial examinations can be exploited to relate dynamic changes in
altered patterns of fMRI activation both to the resolution of brain injury and to the extent of
functional recovery. For example, Reddy and colleagues serially examined a patient with multiple
sclerosis during an acute clinical exacerbation causing impaired hand function, which occurred
secondary to a new large lesion in the brain. In addition to serial clinical assessment and fMRI
measures of changes in cortical activation, the severity of the damage in the brain was quantified (by
measuring the amount of the chemical, N-acetyl-aspartate, a marker of axonal injury). It was
demonstrated that functional recovery from disability preceded the recovery of axonal function (the
amount of N-acetyl-aspartate) around the lesion. However, functional recovery was associated with
early increased recruitment of motor cortex (on the same side as the affected hand), consistent with
the hypothesis that parallel neural pathways may become unmasked during injury to take over task
performance and thus enable functional recovery 53.

The clinical potential of fMRI as a tool to evaluate the effects of clinical interventions also has been
explored. Rehabilitation initiated relatively long after the occurrence of a stroke has still been shown
to produce functional gains, further supporting the concept that compensatory reorganization can
occur in response to injury and that this may be enhanced by a focused physiotherapy approach.
Johansen-Berg and colleagues have been trying to explore the neural basis underlying functional
improvement during a focused physiotherapy programme by relating changes in fMRI cortical
activity to changes in behaviour. Specifically, Johansen-Berg et al tried to test whether enhancement
of activity in the premotor area (a region situated directly in front of the primary motor cortex) is



responsible for the improvements that come with successful rehabilitation therapy. She and her
colleagues took a small group of patients who had suffered from stroke and gave them an intensive
2-week period of hand training. This involved restraint of the unaffected arm to force the subjects to
use the affected hand, as well as a graded series of exercises for the affected hand. After the therapy
period ended, some subjects showed significant improvements in their ability to use the previously

17 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

impaired hand. Correlated with the degree of improvement, was increased activity in the so-called
premotor cortex on both sides of the brain (Fig. 4).

Figure 4. Rehabilitation after a stroke can improve function by enhancing activation of the brain in
intact regions. For this study, patients with weakness on one side were studied before and after a two
week period of intensive rehabilitation. After the rehabilitation period, in patients whose arm
function improved there was an associated increase in brain activation with movement- a

learned recruitment of additional areas. This was not found with patients who did not show


functional improvements, confirming that the brain changes are related directly with the behavioural
improvements (Image courtesy of Ms. H. Johansen-Berg).

To confirm the functional significance of this special area, Johansen-Berg then investigated what
would happen if the newly recruited pre-motor regions were transiently inactivated. Of course, it is
not possible to take out an area of the brain just to test its importance. However, magnetic fields can
be used in another exciting way. If a small electrical coil is placed over the brain and a brief pulse of
electricity pushed through it, a local magnetic field is generated for what is known as transcranial
magnetic stimulation (TMS). This magnetic field can transiently (over periods of fractions of a
second only) interfere with the function of nerves in the underlying cortex. By using special

18 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

techniques, the position of the TMS coil could be correlated with the underlying brain anatomy as
determined from the MRI scanning and the areas of activation known from the fMRI experiment.
The scientists showed that the pre-motor areas in the unaffected side of the brain become particularly
important to functions of the impaired hand in patients after injury. The degree to which they become
important seems related directly to the extent of the activation seen in the fMRI experiment 50

Helping to guide the surgeon s knife

Currently surgery is often used in the management of patients with brain tumours or
pharmacological-resistant epileptic foci. However, there is a trade-off between the margin of
excision used to ensure complete removal and the potential loss of function that may arise as a
consequence of removing normal surrounding brain tissue. There are several invasive approaches
used in neurosurgery to define eloquent areas of the cortex prior to surgical excision. One approach
is to perform electrophysiological mapping of the cortex in the awake patient at the time of the
operation. A second approach routinely used prior to surgery for epilepsy is the Wada test, where the
predominant side of the brain used for language and memory is identified by the invasive sequential
injection of sodium amytal (which transiently stops the brain from working) into each of the two
main blood vessels supplying different sides of the brain, during standard neuropsychological
testing.

An alternative non-invasive approach is to use fMRI to localise activations associated with important
tasks such as limb movement or speech production. A number of groups have now reported the
effectiveness of fMRI in correctly identifying the localisation of the main motor strip or language
area pre-operatively in patients with lesions near these eloquent regions 33, 54, 55. The cost of
performing such an fMRI study is at least three times cheaper than admitting a patient into hospital
for preoperative assessment using the Wada test. Further refinement of these techniques therefore
may lead to the discontinuation of the former expensive and invasive approaches.

Seeing disease before symptoms

The role of fMRI as a diagnostic tool for neurological conditions also has been explored. One such
example is Alzheimer s disease, a progressive neurodegenerative condition that is characterised
clinically by progressive memory loss thought to arise secondary to pathological change in the brain
regions important for memory, such as the temporal and frontal lobe 56. A number of fMRI studies
have shown differing patterns of activation in the brains of patients with Alzheimer s disease
relative to healthy age-matched controls, consistent with localised pathological dysfunction 57.

19 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

However, the insidious onset of the memory disturbance often leads to a delay in the diagnosis of the
disease. Similarly, the characteristic pathological changes found at post-mortem have been detected
before the onset of symptoms. This delay in the diagnosis of the disease may be important if there
are therapeutic agents that can slow the progression of the disease. Bookheimer and colleagues used
fMRI to examine whether this technique could be used as an early diagnostic tool and predict
subsequent memory decline in patients genetically at risk for the disease 34. Specifically, the
experimenters wanted to see firstly, if the pattern of brain activation in patients at risk for
Alzheimer s disease was different to those not at risk when doing a memory task and secondly to
determine whether the baseline brain activation pattern was correlated with subsequent memory
decline.

The scientists compared the patterns of brain activation in genetically at-risk individuals to those of
healthy controls when performing a memory test. Although none of the subjects at-risk for
Alzheimer s disease scored abnormally on standard tests of memory performed outside of the
scanner, Bookheimer et al found that the magnitude and extent of brain activation in the regions of
the brain commonly affected by Alzheimer s disease were greater amongst the genetically at-risk
group. This suggested that increased brain activation in these brain areas may be an early marker for
late-onset disease. To test this hypothesis, the experimenters reexamined the subjects 2 years later
and found that the magnitude of the baseline brain activation correlated with the degree of decline in
memory over the following two years. The authors interpreted the baseline differences in activation
as reflecting the recruitment of extra neural and cognitive processes to compensate for the decline in
function of the necessary neural network normally used for the memory task.

If the differences in observed brain activation are true markers of pre-symptomatic disease, then
fMRI may provide a sensitive index of sub-clinical neural dysfunction and allow the identification
and monitoring of asymptomatic disease progression .



FMRI also may prove to be a sensitive

tool for assessing the impact of preventative and therapeutic interventions.

Defining disease based on patterns of brain function

FMRI could be used to define the range of functional characteristics or phenotypes of patients



with a clinical syndrome. Defining such phenotypes is often a first step in moving from descriptions
of syndromes to specific diseases (i.e., entities with a specifically defined cause and progression).
This may be especially relevant in diseases such as dyslexia, which are difficult to classify on the

20 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

basis of clinical assessment and for which the conventional tests may not have identified
abnormalities. In addition to providing a sensitive marker of the functional trait of the disease, fMRI
studies may provide further insights into the pathophysiology of the disease process.

A number of observations using fMRI have supported the notion that patients with dyslexia may have
abnormalities in a specific part of their visual system which leads to deficits in the processing of
moving stimuli. Eden et al 58 presented a moving stimulus to control subjects and patients with
dyslexia and demonstrated that in the patients, this stimulus did not activate a particular region of the

motion-sensitive visual cortex which receives specific inputs from the



magnocellular pathway,
whereas the presentation of stationary stimuli did not result in any abnormal pattern of brain
activation. Subsequent studies have extended this observation and suggest that deficits in the
integrity of the magnocellular pathway could be used as an objective clinical marker of the disorder
59. The functional definition of sub-groups of patients should facilitate the understanding of the

disorder at the molecular and genetic level.

Conclusions
Together these studies illustrate how magnetism has been used in exciting ways to learn about both
the structure and the function of the brain. With the exquisite detail provided by modern MRI
techniques, we are beginning to learn how brain structure varies between individuals and how it is
related to different behaviours. With fMRI it is possible to link information about brain function
directly to information about brain structure to begin to understand the brain basis of mind. Not only
should this lead to new insights into fundamental problems such as the nature of consciousness, but it
also points the way towards developing new strategies for understanding and treating disease.

Acknowledgements

The authors wish to thank their many colleagues and collaborators in the Oxford Centre for

Functional Magnetic Resonance Imaging of the Brain for continuing to freely share information and

excitement in this new area for cognitive neuroscience. They gratefully acknowledge support for

work in the Oxford Centre particularly from the Medical Research Council and the MS Society of

Great Britain and Northern Ireland.

LITERATURE CITED

21 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

1. S. FINGER: Minds Behind the brain - A History of the Pioneers and Their Discoveries ,
(2000), Oxford, Oxford University Press.

2. S. ZOLA-MORGAN: Annu. Rev. Neurosci., 1995, 118, 359 - 383.

3. J. C. HORTON: Nature, 2000, 406, 565.

4. K. AMUNTS, G. SCHLAUG, L. JANCKE, H. STEINMETZ, A. SCHLEICHER, A.

DABRINGHAUS, and K. ZILLES: Human Brain Mapping., 1997, 5[3], 206-215.

5. E. A. MAGUIRE, C. J. MUMMERY, and C. BUCHEL: Hippocampus, 2000, 10, 475 - 482.

6. A. GJEDDE: Brain energy metabolism and the physiological basis of the haemodynamic

response , in: Functional MRI: an introduction to methods , (ed. [Link] , [Link] , and

P.M. Matthews.), 37-66, (2001), Oxford, Oxford University Press.

7. R. G. SHULMAN: Am. J Psychiatry, 2001, 158, 11 - 20.

8. C. SHERRINGTON and C. ROY: J. Physiol. (London), 1890, 11, 85 - 108.

9. S. OGAWA, T. M. LEE, A. R. KAY, and D. W. TANK: Proc. Natl. Acad. Sci. USA. 1990, 87,
9868 - 9872.

10. [Link] and [Link]: J Cereb. Blood Flow Metab, 1997, 17, 64 - 72.

11. S. OGAWA, R. S MENON, D. W. TANK, S. G. KIM, H. MERKLE, J. M. ELLERMANN, and

K. UGURBIL: Biophys. J., 1993, 64, 803 - 812.

12. N. K. LOGOTHETIS, J. PAULS, M. AUGATH, T. TRINATH, and A. OELTERMANN:

22 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

Nature, 2001, 412, 150 - 157.

13. G. REES, K. FRISTON, and C. KOCH: Nat. Neurosci., 2000, 3, 716 - 723.

14. M. J. SCHLOSSER, N. AOYAGI, R. K. FULBRIGHT, J. C. GORE, and G. MCCARTHY:

Hum. Brain Mapp., 1998, 6, 1 - 13.

15. R. B. TOOTELL, A. M. DALE, M. I. SERENO, and R. MALACH: Trends Neurosci., 1996, 19,
481 - 489.

16. R. KURTH, K. VILLRINGER, B. M. MACKERT, J. SCHWIEMANN, J. BRAUN, G. CURIO,

A. VILLRINGER, and K. J. WOLF: Neuroreport, 1998, 9, 207 - 212.

17. S. G. KIM, J. ASHE, A. P. GEORGOPOULOS, H. MERKLE, J. M. ELLERMANN, R. S.

MENON, S. OGAWA, and K. UGURBIL: J Neurophysiol., 1993, 69, 297 - 302.

18. R. J. NUDO and R. B. MASTERTON: J Comp Neurol, 1986, 245, 553 - 565.

19. A. GJEDDE: The Relation Between Brain Function and Cerebral Blood Flow and
Metabolism , in: Cerebrovascular Disease , (ed. H. H. Batjer, L. R. Caplan, L. Friberg, R.
G. Greenlee, T. A. Kopitnik, and W. L. Young), 23-40, 1997, New York, Lippincott-Raven.

20. G. K. AGUIRRE, E. ZARAHN, and M. D’ESPOSITO: Neuroimage., 1998, 8, 360 - 369.

21. J. C. RAJAPAKSE, F. KRUGGEL, J. M. MAISOG, and D. Y. VON CRAMON: Hum. Brain

Mapp., 1998, 6, 283 - 300.

22. R. PINEIRO and P. M. MATTHEWS: Rev. Neurol., 2001, 33, 701 - 708.

23. P. A. BANDETTINI, The temporal resolution of Functional MRI , in : Functional

Magnetic resonance Imaging , (ed. C. Moonen and P. A. Bandettini), 1999, Berlin, Springer.

23 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

24. A. M. OWEN: Curr. Biol., 1998, 8, R850 - R852.

25. N. KANWISHER, J. MCDERMOTT, and M. M. CHUN: J Neurosci., 1997, 17, 4302 - 4311.

26. A. PLOGHAUS, I. TRACEY, J. S. GATI, S. CLARE, R. S. MENON, P. M. MATTHEWS, and

J. N. RAWLINS: Science, 1999, 284, 1979 - 1981.

27. S. E. LONGE, R. WISE, S. BANTICK, D. LLOYD, H. JOHANSEN-BERG, F. MCGLONE, and

I. TRACEY: Neuroreport, 2001, 12, 2021 - 2025.

28. J .O’DOHERTY, M. L. KRINGELBACH, E. T. ROLLS, J. HORNAK, and C. ANDREWS: Nat.

Neurosci., 2001, 4, 95 - 102.

29. H. C .BREITER and B. R. ROSEN: Ann. N. Y. Acad. Sci., 1999, 877, 523 - 547.

30. A. KARNI, G. MEYER, C. REY-HIPOLITO, P. JEZZARD, M. M. ADAMS, R. TURNER, and

L. G. UNGERLEIDER: Proc. Natl. Acad. Sci. U. S. A, 1998, 95, 861 - 868.

31. R. PINEIRO, S. PENDLEBURY, H. JOHANSEN-BERG, and P. M. MATTHEWS: Stroke,

2001, 32, 1134 - 1139.

32. H. JOHANSEN-BERG, H. DAWES , C. GUY, S. M. SMITH, and D. T. WADE: Submitted,

2002.

33. J. R. BINDER, S. J. SWANSON, T. A. HAMMEKE, G. L. MORRIS, W. M. MUELLER, M.

FISCHER, S. BENBADIS, J. A. FROST, S. M. RAO, and V. M. HAUGHTON: Neurology,
1996, 46, 978 - 984.

34. S. Y. BOOKHEIMER, M. H. STROJWAS, M. S. COHEN, A [Link], M. A.

PERICAK-VANCE, J. C. MAZZIOTTA, and G. W. SMALL: N. Engl. J Med., 2000, 343, 450 -
456.

24 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

35. A. M. OWEN, R. EPSTEIN, AND I. S. JOHNSRUDE: FMRI: applications to cognitive

neuroscience , in: Functional MRI: an introduction to methods. , (ed. P. Jezzard, P. M.
Matthews, and S. M. Smith), 311 - 328, 2001, Oxford, Oxford University Press.

36. J. B. BREWER, Z. ZHAO, J. E. DESMOND, G. H. GLOVER, and J. D. GABRIELI: Science,

1998, 281, 1185 - 1187.

37. T. S. BRAVER, J. D. COHEN, L. E. NYSTROM, J. JONIDES, E. E. SMITH, and D. [Link]:

Neuroimage., 1997, 5, 49 - 62.

38. T. J. ANDREWS, D. SCHLUPPELE, D. HOMFRAY, P. M. MATTHEWS and C.

BLAKEMORE: submitted, 2002.

39. I. TRACEY, A. PLOGHAUS, J. S. GATI, S. CLARE, S. SMITH, R. S. MENON, and P. M.

MATTHEWS: submitted, 2002.

40. J. O’DOHERTY, E. T. ROLLS, S. FRANCIS, R. BOWTELL, F. MCGLONE, G. KOBAL, B.

RENNER, and G. AHNE: Neuroreport, 2000, 11, 399 - 403.

41. J. DOYON: Int. Rev. Neurobiol., 1997, 41, 273 - 294.

42. I. TONI, N. RAMNANI, O. JOSEPHS, J. ASHBURNER, and R. E. PASSINGHAM:

Neuroimage., 2001, 14, 1048 - 1057.

43. G. HESS, C. D. AIZENMAN, and J. P. DONOGHUE: J Neurophysiol., 1996, 75, 1765 - 1778.

44. T. [Link], R. J. RACINE, and B. KOLB: Synapse, 2000, 37, 16 - 22.

45. T. ELBERT, C. PANTEV, C. WIENBRUCH, B. ROCKSTROH, and E. TAUB: Science, 1995,

270, 305 - 307.

46. A. PASCUAL-LEONE and F. TORRES: Brain, 1993, 116 ( Pt 1), 39 - 52.

25 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

47. D. T. WADE: Curr. Opin. Neurol Neurosurg., 1993, 6, 78 - 82.

48. S. C. CRAMER, G. NELLES, R. R. BENSON, J. D. KAPLAN, R. A. PARKER, K. K. KWONG,

D. N. KENNEDY, S. P. FINKLESTEIN, and B. R. ROSEN: Stroke, 1997, 28, 2518 - 2527.

49. L. CAO, M. C. KIRK, L. U. COWARD, P. JACKSON, and J. N. WHITAKER: Arch. Biochem.

Biophys., 2000, 377, 9 - 21.

50. H. JOHANSEN-BERG: An fMRI study of the neurobiological basis of motor rehabilitation

after stroke , Submitted as DPhil thesis. 2002, University of Oxford.

51. M. LEE, H. REDDY, H. JOHANSEN-BERG, S. PENDLEBURY, M. JENKINSON, S. SMITH,

J. PALACE, and P. M. MATTHEWS: Ann. Neurol., 2000, 47, 606 - 613.

52. H. REDDY, S. NARAYANAN, R. ARNOUTELIS, M. JENKINSON, J. ANTEL, P. M.

MATTHEWS, and D. L. ARNOLD: Brain, 2000, 123 (Pt 11), 2314 - 2320.

53. [Link], S. NARAYANAN, P. M. MATTHEWS, R. D. HOGE, G. B. PIKE, P. DUQUETTE,

J. ANTEL, and D. L. ARNOLD: Neurology, 2000, 54, 236 - 239.

54. S. W. ATLAS, R. S. HOWARD, J. MALDJIAN, D. ALSOP, J. A. DETRE, J. LISTERUD, M.

D’ESPOSITO, K. [Link], E. ZAGER, and M. STECKER: Neurosurgery, 1996, 38, 329 - 338.

55. G. L. MORRIS, W. M. MUELLER, F. Z. YETKIN, V. M. HAUGHTON, T. A. HAMMEKE, S.

SWANSON, S. M. RAO, A. JESMANOWICZ, L. D. ESTKOWSKI, P. A. BANDETTINI:
Epilepsia, 1994, 35, 1194 - 1198.

56. P. V. ARRIAGADA, K. MARZLOFF, and B. T. HYMAN: Neurology, 1992, 42, 1681 - 1688.

57. K. R. THULBORN, C. MARTIN, and J. T. VOYVODIC: AJNR Am. J Neuroradiol., 2000, 21,
524 - 531.

26 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

58. G. F. EDEN, J. W. VANMETER, J. M. RUMSEY, J. M. MAISOG, R. P. WOODS, and T. A.

ZEFFIRO: Nature, 1996, 382, 66 - 69.

59. J. B. DEMB, G. M. BOYNTON, and D. J .HEEGER: Proc. Natl. Acad. Sci. U. S. A, 1997, 94,
13363 - 13366.

27 of 28 11/12/2004 01:29 PM
INTRO-PAULS STUFF [Link]

28 of 28 11/12/2004 01:29 PM