BRAIN AND LANGUAGE 62, 298–308 (1998)
ARTICLE NO. BL981951

NOTES AND DISCUSSION
Different Neural Circuits Subserve Reading before and after
Therapy for Acquired Dyslexia
Steven L. Small,* Diane Kendall Flores,† and Douglas C. Noll‡
*Department of Neurology, University of Maryland; †Department of Communication
Science and Disorders, University of Pittsburgh; and ‡Department of Radiology,
University of Pittsburgh Medical Center
Rehabilitative measures for stroke are not generally based on basic neurobiological principles, despite evidence from animal models that certain anatomical and
pharmacological changes correlate with recovery. In this report, we use functional
magnetic resonance imaging (fMRI) to study in vivo human brain reorganization
in a right handed patient with an acquired reading disorder from stroke. With phonological dyslexia, her whole-word (lexical) reading approach included inability to
read nonwords and poor reading of function words. Following therapy, she was
able to read nonwords and function words, and preferred a decompositional (sublexical) strategy in general. fMRI was performed during a reading task before and
after treatment. Prior to therapy, her main focus of brain activation was in the left
angular gyrus (area 39). After therapy, it was instead in the left lingual gyrus (area
18). This result suggests first that it is possible to alter brain physiology with therapy
for acquired language disorders, and second, that two reading strategies commonly
used in normal reading use distinct neural circuits, possibly reconciling several conflicting neuroimaging studies of reading.  1998 Academic Press

INTRODUCTION

Rehabilitative measures for stroke are generally based on empirical concepts of functional outcome rather than on basic neurological principles.
While there is some knowledge about neuropharmacological and neuroanatomical changes that take place after stroke and during stroke recovery, clinical rehabilitation does not generally use this information. Some of the atThe authors gratefully acknowledge the support of NIH-NIDCD under Grant CD R01-337801. Special thanks also to R. Berndt, M. Burton, D. Davis, D. Emge, M. Forbes, C. Genovese,
R. Gullapalli, A. Haendiges, P. Hlustik, G. Hoffman, M. McNeil, C. Mitchum, A. Solodkin,
and R. Wellington.
298
0093-934X/98 $25.00

Copyright  1998 by Academic Press
All rights of reproduction in any form reserved.

NOTES AND DISCUSSION

299

tempts to alter the course of neurobiological recovery in animal models
(Feeney, Gonzalez, & Law, 1982) and humans (Crisostomo, Duncan, Propst,
Dawson, & Davis, 1988; McNeil, Doyle, Spencer, Goda, Flores, & Small,
1997; Small, 1994) have shown promise, but have not yet demonstrated the
potential to have a major impact.
The advent of functional neuroimaging permits in vivo neuroanatomical
investigation in humans, thus opening up many possibilities for stroke rehabilitation research that were not previously available. One potentially important application of this new method is in the study of treatment, where the
state of the art depends on empirical assessments of functional outcome,
rather than on any theory of neurobiological mechanism. In the present single
case study, we applied this technology to the investigation of neuroanatomical reorganization accompanying treatment of an acquired reading disorder
in a woman with stroke.
From the framework of cognitive neuropsychology, reading syndromes
include those primarily affecting visual processing and others that involve
‘‘central’’ neural processes subsequent to vision. These central syndromes
include surface dyslexia, phonological dyslexia, and deep dyslexia (Friedman, 1988; McCarthy & Warrington, 1990).
In surface dyslexia, patients read by using the spelling to sound rules of
their language. This becomes noticeable when they use this strategy on irregularly spelled words that do not follow the rules. In phonological dyslexia,
patients do the opposite, and read whole words at a time, without invoking
spelling to sound correspondences. This becomes evident when patients try
to read uncommon words, or in the psychology laboratory, when they are
asked to read non-words that have spellings that are consistent with the rules
of their language. Since they are only capable of reading whole words at a
time, uncommon words and non-words are difficult (or impossible) to read.
Many of these patients also have difficulties with words without straightforward semantic interpretations, such as function words. Deep dyslexia combines semantic paralexia with the phonological problems. The anatomy of
the central dyslexias remains uncertain, despite a number of case reports
including neuropathological or neuroradiological findings (McCarthy &
Warrington, 1990).
In this report, we discuss a functional neuroimaging study of a patient
with phonological dyslexia and her treatment for spelling to sound conversion deficits (see Flores et al. (1997) for a discussion of the treatment
method). The hypothesis was that successful treatment of phonological dyslexia with supervised training of word and nonword reading is accompanied
by changes in the neuroanatomical substrate of reading. Imaging was conducted with functional magnetic resonance imaging both before and after
therapy. Significant changes took place both in the patient’s behavior and
in the brain, thus representing an exciting example of how behavioral rehabil-

300

NOTES AND DISCUSSION

TABLE 1
Schedule of fMRI Scans and Relationship to Therapy
fMRI
Scan

Time
(days)

Baseline testing and
treatment time points

1
2

t⫽0
t ⫽ 14

3

t ⫽ 38

Beginning of baseline testing
End of baseline testing and beginning
of treatment
End of treatment

itative therapy, motivated by theoretically elaborated cognitive neuropsychological models, can have a direct effect on functional brain organization after
stroke.
METHODS
A 42-year-old right-handed woman had a large left fronto-temporal ischemic stroke at the
age of 25 with severe aphasia and hemiplegia. Despite significant recovery in spoken language,
reading remained a major problem. The patient demonstrated a pattern of reading consistent
with phonological dyslexia; her reading of common (high frequency) words was better than
uncommon words, and she could not read nonwords. She read function words poorly. She did
not make semantic paralexic errors. We performed therapy using supervised (teacher assisted)
learning of grapheme to phoneme correspondences (GPCs) through exemplars. During therapy, the patient learned to read nonwords. She was observed by the therapist to decompose
words into parts in order to read them, and to apply this strategy more generally than actually
necessary (Flores et al., 1997).
fMRI is a task-dependent activation method, and thus requires comparison of two functional
brain states, an ‘‘active’’ condition and a ‘‘control’’ condition (Donders, 1868/1969; Sergent,
Zuck, Lévesque, & MacDonald, 1992; Sternberg, 1969). The active condition was a reading
version of the Token Test (De Renzi & Vignolo, 1962; McNeil & Prescott, 1978), specifically
designed for functional imaging studies, in which the subject decides whether or not a visually
presented sentence (e.g., ‘‘The red square is above the green circle’’) describes a picture containing an array of colored tokens. The passive condition consisted of false font text and (the
same) visually presented tokens, in which the subject was asked to respond true if the picture
showed any squares above any circles and false otherwise. fMRI scans were performed before
and after treatment as shown in Table 1. Informed consent was obtained from the subject in
accordance with the policies and with the approval of the Institutional Review Board for
Biomedical Research of the University of Pittsburgh.
During each trial, active and control tasks were presented alternately for 60 seconds each
for 8 minutes. During each 60 second presentation of one task, eight functional images were
collected, for a total of 64 images in each trial. Of these, 32 were collected during the main
task and 32 during the control task. Each scanning session consisted of four separate trials
of stimulus presentation, in which two different sets of sentences and pictures were employed.
These corresponded to subtests 6 and 8 of the Revised Token Test (McNeil & Prescott, 1978),
adjusted for use in reading and for neuroimaging experiments. The total data analyzed thus
consisted of 128 scans acquired during performance of the main task and 128 scans acquired
during performance of the control task. The actual time series of scan acquisition consists of
8 scans of one task followed by 8 scans of the other, in a repetitive pattern for a total of 16
condition pairs.
Functional imaging employed the spiral k-space method (Noll, Cohen, Meyer, & Schneider,

NOTES AND DISCUSSION

301

1995) on a 1.5 T GE Signa scanner with a standard clinical head coil. One complete multishot (four spirals) scan of the entire brain (20 slices) was performed in 7.5 seconds. Images
were obtained in the sagittal plane with a 7 mm interslice interval and a 3.7 mm resolution
in the imaging plane. Movement artifacts were minimized at scan time with head immobilization. Part of the image analysis pathway also employed image registration (Woods, Cherry, &
Mazziotta, 1992) for further compensation of movement artifact as well as MR angiography,
which was used to exclude the presence of vascular artifacts.
Each imaging session consisted of four eight-minute experimental runs, yielding four time
series of data, consisting of 64 time points at each voxel location in the image (24 slices).
Since each time series was collected at a different time (between 8 and 24 minutes apart),
the MRI scanner could experience signal drift and the subject could change location. Thus
the first analysis step was to normalize each time series to a zero mean and to apply a detrending
procedure to remove changes in the baseline signal (Cox, 1997), and the second step was to
co-register all the images together using the 3D version of the automated image registration
algorithm (Woods et al., 1992).
The four resulting time series were concatenated together, and a reference sinusoidal
waveform was cross-correlated with the concatenated time series (Bandettini, Jesmanowicz,
Wong, & Hyde, 1993). The cross-correlation was performed with sine waves offset by 0, 7.5,
15, 22.5, and 30 degrees (0, 2.5, 5, 7.5, 10 seconds) and the best match retained to account
for hemodynamic lag (Bandettini et al., 1993).
A correlation coefficient threshold of r ⫽ 0.45 was used for all further analyses of the data
of all three imaging sessions. In order to reduce false positives and to minimize the effects
of multiple comparisons, a minimum cluster size threshold (using a 3D neighborhood of 28
voxels) of 4 voxels (4 voxels ⫽ 98.4 mm3) was used. The active regions shown in the tables
all meet this criterion. The total activation numbers, used to show fractional activations, are
shown including all voxels regardless of minimum cluster size.

RESULTS

The main results of the study are summarized in Tables 2 and 3 and Figs.
1–3. Several features of these results are important for the understanding of
phonological dyslexia and its treatment. First, the primary site of activation
at the two time points prior to treatment is the angular gyrus of the left
inferior parietal lobule (Brodmann’s area 39). This activation is highly robust
and is consistent with historical literature on visual language processing (Dejerine, 1891). The fractional activation in this region increased during the
period of baseline testing, when the patient was exposed to a large number
of words but not given any instruction on sublexical reading. The center of
the left angular gyrus activation during this period of whole word (lexical)
reading was at Talairach coordinates (⫺51, ⫺64, 36) (Scan 1).
Second, although this inferior parietal site of activation remains after training, it is much less prominent, with the most active site of activation being
the area of the left lingual gyrus (BA 18). This region of activation is very
reliable over the course of four experimental runs, and is consistent with
PET studies of single word processing (Petersen, Fox, Posner, Mintun, &
Raichle, 1988). The center of the left lingual gyrus activation during this
period of phonological (sub-lexical) reading was at Talairach coordinates
(⫺16, ⫺79, ⫺10) (Scan 3).

302

NOTES AND DISCUSSION

TABLE 2
Characteristics of Brain Activation Patterns in Phonological Dyslexia before and after
Learning Grapheme Phoneme Correspondences
Hemisphere

Sign

L
L
L

⫹
⫹
⫹

L
L
L
L
L
R
R

⫹
⫹
⫹
⫹
⫹
⫹
⫺

L
L
L
L
L
L
R
R

⫹
⫹
⫹
⫹
⫹
⫹
⫹
⫺

Anatomical region

Volume

Fraction

Brodmann

0.17
0.33
0.50

6/m
6/l
39

Scan 2: After baseline testing and before therapy
Supplementary motor area
0.22
0.12
Lateral premotor area
0.37
0.21
Pre-cuneate area
0.17
0.10
Angular gyrus
0.57
0.32
Lingual gyrus
0.37
0.21
Prefrontal cortex
0.10
0.05
Middle temporal gyrus
0.15
0.08

6/m
6/l
7
39
18li
46
22

Scan 1: Before baseline testing
Supplementary motor area
0.12
Lateral premotor area
0.25
Angular gyrus
0.37

Scan 3: After therapy
Precentral gyrus
0.10
Lateral premotor area
0.25
Pre-cuneate area
0.10
Angular gyrus
0.10
Cuneate gyrus
0.17
Lingual gyrus
0.69
Cuneate gyrus
0.15
Middle temporal gyrus
0.37

0.06
0.16
0.06
0.06
0.11
0.44
0.10
0.24

4
6/l
7
39
18cu
18li
18cu
22

Note. Key: Hemisphere: Left (L) or Right (R); sign: Positive activation (⫹) or negative
activation (⫺); anatomical region: Gyrus and/or functional region; volume: shown in cubic
centimeters; for conversion to voxels, divide by 0.0246 cc3 or 24.6 mm3, the size of one voxel
(1.875 ⫻ 1.875 ⫻ 7); Fraction: Fraction of total positive activation in the brain (negative
activation is also expressed in terms of the total positive activation for perspective); Brodmann:
Brodmann’s areas.

TABLE 3
Left Hemisphere Activation Volume by Regions
Brodmann area

Scan 1

Scan 2

Scan 3

Areas 6m/6l
Area 7
Area 18 cu
Area 18 li
Area 39
Other (7, 8, 9, 46)

0.37 (0.50)
0.00 (0.00)
0.00 (0.00)
0.00 (0.00)
0.37 (0.50)
0.00 (0.00)

0.59 (0.33)
0.17 (0.10)
0.00 (0.00)
0.37 (0.21)
0.57 (0.32)
0.00 (0.00)

0.25 (0.16)
0.10 (0.06)
0.17 (0.11)
0.69 (0.44)
0.10 (0.06)
0.10 (0.06)

Note. Volumes are the regional volumes in cc3 (see previous legend); in parentheses are
the fractions of total positive brain activation. Brodmann’s areas 6m and 6l refer respectively
to the medial and lateral portions of premotor area 6; areas 18 cu and 18 li refer respectively
to the cuneate and lingual portions of visual area 18.

NOTES AND DISCUSSION

303

FIG. 1. Graph showing changes in fractional activation of five regions of interest over
the course of baseline testing (scans 1–2) and treatment (scans 2–3). Numbers shown in the
legend are Brodmann areas, with area 6 including medial and lateral portions, and area 18
divided into cuneate and lingual portions.

DISCUSSION

The present study demonstrates that in the case of a specific neuropsychologically defined syndrome and a theoretically motivated treatment method,
functional MRI can assess the neuroanatomical changes that accompany successful behavioral therapy in aphasia. In the case of our patient with phonological dyslexia, fMRI showed that learning how to make grapheme to phoneme correspondences led to functional neuroanatomical changes. With
decreased use of whole word (lexical) reading strategies came decreased

FIG. 2. Activation in the inferior parietal lobule (a) before baseline testing; (b) after baseline testing and before therapy; and (c) after therapy. All regions have cluster size ⱖ 4 and
r ⱖ 0.45 and are indicated by arrows.

304

NOTES AND DISCUSSION

FIG. 3. Activation in the lingual gyrus (a) before baseline testing; (b) after baseline testing
and before therapy; and (c) after therapy. All regions have cluster size ⱖ 4 and r ⱖ 0.45 and
are indicated by arrows (except for (a) which has no activation in that region). Note that scan
(b) also shows activation in lateral premotor cortex.

reliance on the dominant left angular gyrus (50% of total activation in scan
1; 6% of total activation in scan 3); and with greater use of decompositional
(phonological; sublexical) reading strategies came increased reliance on the
dominant left lingual gyrus (44% of total activation in scan 3; 0% of total
activation in scan 1).
In the primary sensory and motor cortices of animals, relevant experience
changes cortical structure/function relationships. The present study shows
an analogous change in human language reorganization. Several common
features of anatomy and physiology underlie these changes in both cases:
First is that functional encodings in the cerebral cortex are dynamic; second
is that functional anatomy changes by unmasking existing circuits and altering synaptic connectivity; and third is that experience of the subject can
induce these changes.
The dynamic nature of these functional encodings is best demonstrated
by the experience-induced functional anatomical changes that occur rapidly
(Gilbert, 1992). These changes take place through the ‘‘unmasking’’ of circuits that follow pre-existing anatomical connections (Jacobs & Donoghue,
1991), and have been postulated to explain fast reorganization in motor,
sensory, and visual systems after stroke (Das & Gilbert, 1995; Jacobs &
Donoghue, 1991; Kaas, Krubitzer, Chino, Langston, Polley, & Blair, 1990;
Keller, Weintraub, & Miyashita, 1996). On the longer time frame (i.e.,
months or years), axonal sprouting and synaptogenesis may mediate more
substantial anatomical changes and concomitant improvements in recovery
(Gilbert, 1992).
Functional cortical reorganization seems to depend on the nature of practice, rather than simply on its presence (Nudo, Wise, SiFuentes, & Milliken,
1996; Taub, Miller, Novack, Cook, Fleming, Nepomuceno, et al., 1993). Our
patient with phonological dyslexia was trained using a particular method of
supervised training with exemplars of words and nonwords. This training

NOTES AND DISCUSSION

305

successfully taught her to read by a sublexical route (Flores et al., 1997),
and with fMRI, it was possible to show that the functional neuroanatomy of
her reading accompanied this change in behavior.
Within the framework of the cognitive theories of reading, (at least) two
mechanisms are postulated, one involving visual decomposition and conversion of small visual components to their verbal syllabic counterparts (grapheme to phoneme conversion), and the second incorporating whole word recognition (Coltheart, Curtis, Atkins, & Haller, 1993). The patient described
here was able to read well using the whole word route prior to any therapy,
but was not able to use the decompositional route. After therapy, she was
able to decompose words into constituent graphemes and successfully read
these strings that are not supported by either frequent occurrence or lexical
semantics. In fact, after therapy, the patient was noted by the research therapist to do this as a general strategy when trying to read (Flores et al., 1997).
The nature of this functional reorganization might have been expected
on the basis of neuropsychological theory regarding the central dyslexias.
Previous functional neuroanatomical research on single word reading has
implicated two different areas in the processing of single words presented
visually. The first such study, performed using PET technique, suggested
that the left medial extrastriate cortex played a special role in processing
words and pseudo-words and not false font strings or consonant clusters
(Petersen et al., 1988; Petersen, Fox, Snyder, & Raichle, 1990). A subsequent
study, incorporating a task design aimed at eliminating the contributions of
phonological processing, suggested that the left posterior superior temporal
gyrus, adjacent to ‘‘the classical location of visual word forms, namely the
angular gyrus’’, plays the primary role in visual word processing (Howard,
Patterson, Wise, Brown, Friston, Weiller, et al., 1992). This reaffirmed a
long-standing belief in neurology about the localization of ‘‘visual word
forms’’ (Dejerine, 1891), and was subsequently replicated using fMRI technique (Small, Noll, Perfetti, Hlustik, Wellington, & Schneider, 1996).
An important finding of the present study is that it may help reconcile
these two apparently conflicting PET studies of single word reading. The
patient described here undoubtedly used a whole word reading strategy prior
to therapy. This is the only strategy available to patients with phonological
dyslexia, and was used exclusively by this patient at the time of the first two
fMRI scans. It is possible that between the first and second scans, when the
patient was exposed to many words as part of baseline testing of reading,
she developed some subtle improvement in phonological reading. However,
based on formal testing and intuitive observation, this was not a profound
improvement.
The area of activation present during reading prior to therapy is very close
to the area implicated in single word reading in neurological tradition
(Dejerine, 1891; Dejerine, 1892) and neuroimaging studies (Howard et al.,
1992; Small et al., 1996). This area showed decreased activation after ther-

306

NOTES AND DISCUSSION

apy—could it be particularly important to the whole word reading mechanism? The occipital area present after therapy includes significant overlap
with the left medial extrastriate region that activated in the original PET
study of single word reading (Petersen et al., 1988) and was found later to
become active only for strings in which a graphemic decomposition was
possible (Petersen et al., 1990). Since our patient preferred a decompositional
strategy at all times after therapy (often using her finger or an index card to
mark visually the graphemic components of presented strings), the imaging
results suggest that the functional brain reorganization in this patient led to
unmasking of circuits in the occipital cortex that are ordinarily employed as
the initial stages in phonological (decompositional) reading.
Two additional findings will not be discussed at length. First is the involvement of medial and lateral supplementary motor areas in the present reading
task, which declined over time. Although these areas have not generally been
present in studies of single word reading, they may reflect the processing of
entire sentences when timing issues play a larger role. Second is the increasing de-activation of the right middle temporal gyrus over time. It might be
speculated that deactivation of some right hemisphere regions homologous
to important language areas in the left hemisphere is required for successful
phonological reading.
The present study demonstrates that therapy for phonological dyslexia,
which changed reading from a whole word to a decompositional approach,
leads to specific changes in the pattern of brain activation, predicted by
neuropsychological theory and previous neuroimaging studies of normal
subjects. For aphasia therapy, this result suggests that specific therapeutic
methods can alter the neurophysiology of language. For basic cognitive neuroscience, this result may help understand two neuroimaging studies of single
word reading in terms of current theory in the neuropsychology of reading.
Finally, this study suggests a fundamental relationship among cognitive neuropsychological theory, aphasia therapy, and basic neuroscience.
REFERENCES
Bandettini, P. A., Jesmanowicz, A., Wong, E. C., & Hyde, J. S. 1993. Processing strategies
for time-course data sets in functional MRI of the human brain. Magnetic Resonance in
Medicine, 30, 161–73.
Coltheart, M., Curtis, B., Atkins, P., & Haller, M. 1993. Models of reading aloud: Dual-route
and parallel-distributed-processing approaches. Psychological Review, 100, 589–608.
Cox, R. 1997. Improving the task-to-signal correlation in fMRI by detrending slow components. In Proceedings of the ISMRM 5th Scientific Meeting (p. 1668). Vancouver, Canada.
Crisostomo, E. A., Duncan, P. W., Propst, M., Dawson, D. V., & Davis, J. N. 1988. Evidence
that amphetamine with physical therapy promotes recovery of motor function in stroke
patients. Annals of Neurology, 23, 94–97.
Das, A., & Gilbert, C. D. 1995. Long-range horizontal connections and their role in cortical

NOTES AND DISCUSSION

307

reorganization revealed by optical recording of cat primary visual cortex. Nature, 375,
780–4.
De Renzi, E., & Vignolo, L. A. 1962. The Token Test: A sensitive test to detect receptive
disturbance in aphasics. Brain, 85, 665–678.
Dejerine, J. 1891. Sur un cas de cécité verbale avec agraphie, suivi d’autopsie. Comptes Rendus
Hebdomodaires des Séances et Mémoires de la Société de Biologie, 3(Series 9), 197–
201.
Dejerine, J. 1892. Contribution à l’Etude Anatomo-Pathologique et Clinique des Differentes
Variétés de Cécité Verbale. Comptes Rendus Hebdomodaires des Séances et Mémoires
de la Société de Biologie, 4(Series 9), 61–90.
Donders, F. C. 1868/1969. On the speed of mental processes. Acta Psychologica, 30, 412–
431.
Feeney, D. M., Gonzalez, A., & Law, W. A. 1982. Amphetamine, haloperidol, and experience
interact to affect rate of recovery after motor cortex injury. Science, 217, 855–857.
Flores, D., McNeil, M. R., & Small, S. L. 1997. Rule-based treatment for acquired dyslexia.
Aphasiology, in press.
Friedman, R. 1988. Acquired alexia. In F. Boller & J. Grafman (Ed.), Handbook of Neuropsychology: Volume 1. Amsterdam: Elsevier Science Publishers. Pp. 377–391.
Gilbert, C. D. 1992. Horizontal Integration and Cortical Dynamics. Neuron, 9, 1–13.
Howard, D., Patterson, K., Wise, R., Brown, W. D., Friston, K., Weiller, C., & Frackowiak,
R. 1992. The cortical localization of the lexicons: Positron emission tomography evidence. Brain, 115, 1769–1782.
Jacobs, K. M., & Donoghue, J. P. 1991. Reshaping the cortical motor map by unmasking
latent intracortical connections. Science, 251, 944–947.
Kaas, J. H., Krubitzer, L. A., Chino, Y. M., Langston, A. L., Polley, E. H., & Blair, N. 1990.
Reorganization of retinotopic cortical maps in adult mammals after lesions of the retina.
Science, 248, 229–231.
Keller, A., Weintraub, N. D., & Miyashita, E. 1996. Tactile experience determines the organization of movement representations in rat motor cortex. NeuroReport, 7(14), 2372–2378.
McCarthy, R. A., & Warrington, E. K. 1990. Cognitive Neuropsychology: A Clinical Introduction. San Diego: Academic Press.
McNeil, M. R., Doyle, P. J., Spencer, K. A., Goda, A. J., Flores, D., & Small, S. L. 1997.
A double-blind, placebo-controlled study of pharmacological and behavioural treatment
of lexical-semantic deficits in aphasia. Aphasiology, 11(4/5), 385–400.
McNeil, M. R., & Prescott, T. E. 1978. Revised token test. Austin, TX: Pro-Ed.
Noll, D. C., Cohen, J. D., Meyer, C. H., & Schneider, W. 1995. Spiral K-space MRI of cortical
activation. Journal of Magnetic Resonance Imaging, 5, 49–56.
Nudo, R. J., Wise, B. M., SiFuentes, F., & Milliken, G. W. 1996. Neural substrates for
the effects of rehabilitative training on motor recovery after ischemic infarct. Science,
272(5269), 1791–1794.
Petersen, S. E., Fox, P. T., Posner, M. I., Mintun, M. A., & Raichle, M. E. 1988. Positron
emission tomographic studies of the cortical anatomy of single-word processing. Nature,
331, 585–589.
Petersen, S. E., Fox, P. T., Snyder, A. Z., & Raichle, M. E. 1990. Activation of extrastriate
and frontal cortical areas by visual words and word-like stimuli. Science, 249, 1041–
1043.
Sergent, J., Zuck, E., Lévesque, M., & MacDonald, B. 1992. Positron emission tomography
study of letter and object processing: Empirical findings and methodological considerations. Cerebral Cortex, 2(1), 68–80.

308

NOTES AND DISCUSSION

Small, S. L. 1994. Pharmacotherapy of aphasia: A critical review. Stroke, 25(6), 1282–1289.
Small, S. L., Noll, D. C., Perfetti, C. A., Hlustik, P., Wellington, R., & Schneider, W. 1996.
Localizing the lexicon for reading aloud: Replication of a PET study using fMRI.
NeuroReport, 7(4), 961–965.
Sternberg, S. 1969. The discovery of processing stages: Extensions of the Donders’ method.
Acta Psychologica, 30, 276–315.
Taub, E., Miller, N. E., Novack, T. A., Cook, E. W., 3d, Fleming, W. C., Nepomuceno,
C. S., Connell, J. S., & Crago, J. E. 1993. Technique to improve chronic motor deficit
after stroke. Archives of Physical Medicine and Rehabilitation, 74(4), 347–354.
Woods, R. P., Cherry, S. R., & Mazziotta, J. C. 1992. Rapid automated algorithm for aligning
and reslicing PET images. Journal of Computer Assisted Tomography, 16, 620–633.