## Abstract

Somatosensory evoked fields (SEFs) to repetitive tactile stimuli were recorded from eight dyslexic and eight normal-reading adults. Three successive stimuli, produced by diaphragms driven by compressed air, were delivered to thumb, index finger and thumb in sequence, with stimulus-onset asynchronies (SOAs) of 100 and 200 ms in different runs. Both hands were stimulated alternatingly with an intertrain interval of 1 s, and the responses were recorded with a whole-scalp neuromagnetometer. Whereas the primary somatosensory cortex responses to the first stimuli of the trains did not differ between dyslexics and controls, responses to the second stimuli (and the ratios of second to first responses) were significantly smaller in dyslexic than in control subjects in the right hemisphere (differences 41 and 28% for response amplitudes at the 100 and 200 ms SOAs). The results agree with the proposed pansensory nature of temporal processing deficits in dyslexia, specifically demonstrating abnormal response recovery in the right somatosensory cortex.

## Introduction

During the last decades, developmental dyslexia has been convincingly linked, in addition to difficulties of phonological processing (Lundberg et al., 1980; Bradley and Bryant, 1983; Liberman and Shankweiler, 1985), to problems of sensory temporal processing that is not directly related to reading acquisition. Dyslexic adults face difficulties in, for example, processing non-speech auditory stimuli and acoustic changes that follow each other in rapid succession (Hari and Kiesilä, 1996; Helenius et al., 1999; Nagarajan et al., 1999; Renvall and Hari, 2002). Such data are in line with earlier findings of auditory discrimination disorders in dyslexic children (Tallal, 1980).

We have recently tried to bind up findings of sensory temporal processing deficits in dyslexia by proposing that the subjects, due to weakened and sluggish stimulus-triggered attentional mechanisms, have prolonged ‘input chunks’ that lead into distorted processing of rapid stimulus sequences and impair the proper development of cortical representations needed for reading acquisition (Hari and Renvall, 2001). This hypothesis envisages similar sluggishness in all senses, and probably in motor output as well. The pansensory temporal processing deficit is supported by recent psychophysical findings that dyslexic children and adults are impaired in the perception of rapidly presented visual and tactile stimuli (Laasonen et al., 2000, 2001). In the present study, we tested the pansensory deficit in dyslexic subjects at brain signal level, by recording neuromagnetic signals to tactile stimuli presented, in groups of three, to fingers of either the left or the right hand. The results imply abnormal response recovery in the right hemisphere of dyslexic individuals. Preliminary results have been reported in abstract form (Renvall et al., 2002).

## Materials and Methods

### Subjects

We studied, with informed consent, eight adult dyslexic individuals (mean age ± SEM = 28 ± 2 years; five females, three males; all right-handed) and eight normal-reading control subjects (29 ± 2 years; five females, three males; seven strongly right-handed and one ambidextrous with a laterality index of −9 in the Edinburgh handedness test, in which the left- versus right-handedness ranges from −100 to +100). The dyslexic subjects were selected on the basis of definite childhood history of difficulties in learning to read. All of them had participated in special tutoring at school age, and all except the oldest subject had a diagnosis of dyslexia stated by a special teacher, speech therapist or psychologist. Two dyslexic subjects had a university degree, one had an academic-level professional degree and three were university students. The study had received prior approval by the Ethics Committee of the Helsinki Uusimaa Hospital District.

One dyslexic subject did not participate in the above Finnish-language tests because her mother tongue was Swedish, the second official language in Finland. However, she had a convincing history of childhood problems in learning to read, and had also been thoroughly behaviourally tested at the age of 19. On the basis of her own report, she was still (at the age of 24) slow in reading, and made mistakes in writing and grammar.

### Stimulation and Recording

Tactile stimuli were delivered to the palmar skin of the distal phalanges of thumb and index finger, ∼1.5 cm from the fingertip, with balloon diaphragms driven by compressed air (Mertens and Lütkenhöner, 2000). The pressure was kept the same for all subjects, and it resulted in a percept of a clear touch at an area of ∼0.8 cm2. Trains of three stimuli were delivered in a sequence of thumb → index finger → thumb, alternatingly to the left and right hands with an intertrain interval (from the beginning of the third stimulus to the beginning of the next first stimulus) of 1 s. The stimulus-onset asynchronies (SOAs) within each train were 100 and 200 ms in separate runs, and the order of runs was randomized across subjects within both groups.

Because of the inherently long duration of our stimulus (rise time = 30 ms, peak pressure duration = 100 ms, fall time = 150 ms; see Fig. 3), the third stimulus in the 100 ms SOA train started during the fall phase of the first stimulus presented to the same finger. However, at that point the pressure of first stimulus was <30% of the peak pressure, and thus the potential interaction can be considered minor, especially because the stimuli were perceived distinct up to stimulation frequencies of at least 20 Hz. During the measurement, the subject was watching a movie without any further task.

Somatosensory evoked fields (SEFs) were recorded in a magnetically shielded room while the subject was sitting with the head supported against the helmet-shaped bottom of the 306-channel Vectorview™ (Neuromag Ltd., Helsinki, Finland) neuromagnetometer. The device contains 102 identical triple sensors, comprising two orthogonal planar first-order gradiometers and one magnetometer, each of them coupled to a SQUID (Superconducting QUantum Interference Device). Four head-position-indicator coils were attached to the scalp, and their positions were measured with a three-dimensional digitizer; the head coordinate frame was anchored to the two periauricular points and the nasion. The head position with respect to the sensor array was determined by feeding current to the marker coils.

The recording passband was 0.03–172 Hz and the signals were digitized at 600 Hz. The averaged signals were digitally band-pass filtered through 2–90 Hz to remove sustained fields that occur during stimulus trains; this procedure simplified amplitude measurements of the transient responses. Both horizontal and vertical electro-oculograms were recorded to discard data contaminated by eye blinks and eye movements. For both left- and right-sided stimulation, responses to odd- and even-numbered trains were averaged to two different bins, and the replicability of the responses was ensured by visual inspection. A minimum of 140 responses was averaged per stimulus train.

### Signal Analysis

The signals were analysed using several methods. First, to obtain a crude idea of the main features of the data, the response latencies and amplitudes were measured from the maximum channel in each hemisphere. For statistical analysis, areal vector sums at the site of the maximum signal were calculated, by first computing vector sums

$$\sqrt{\left(\frac{{\partial}B_{z}}{{\partial}x}\right)^{2}{+}\left(\frac{{\partial}B_{z}}{{\partial}y}\right)^{2}}$$
of the two orthogonal gradients for each channel pair, and then averaging signals across 6–9 channel pairs. Compared with amplitude measurements from single channels, the vector sums simplify the analysis when the orientation of the neural current changes drastically during the analysis period, with minor accompanying changes in the source location. Such behaviour was expected for responses to stimulation of two nearby fingers. Note that Figures 1 and 3 depict the original responses.

Figure 1.

Evoked responses of a normal-reading control subject C4 to left-hand stimulation at the 200 ms SOA. The upper and lower traces in each location correspond to the latitudinal and longitudinal gradients of the radial magnetic field component Bz. The inserts show enlarged responses from the maximum channel over the right hemisphere before (A) and after (B) high-pass filtering at 2 Hz. The insert (C) depicts the areal vector sum of a subset of 14 channels encircled with the gray line.

Figure 1.

Evoked responses of a normal-reading control subject C4 to left-hand stimulation at the 200 ms SOA. The upper and lower traces in each location correspond to the latitudinal and longitudinal gradients of the radial magnetic field component Bz. The inserts show enlarged responses from the maximum channel over the right hemisphere before (A) and after (B) high-pass filtering at 2 Hz. The insert (C) depicts the areal vector sum of a subset of 14 channels encircled with the gray line.

In all subjects, the first stimulus of the train produced a prominent response. If no clear response was detected to the following stimuli at the predicted latency, the average noise in the two orthogonal sensors (typically ∼7 fT/cm) was used as the amplitude value (for details, see Results).

Hemispheric lateralization of response amplitudes was quantified by calculating lateralization index (LI) between the right (R) and left (L) hemispheres: LI = (R − L)/(R + L). LI values range from −1 (left-hemisphere activation only) to 1 (right-hemisphere activation only); the 0-value refers to hemispheric symmetry.

We concentrated, for simplicity, on the latency and amplitude measures of the areal vector sums. However, we also identified the cerebral sources of the evoked responses using dipole models: equivalent current dipoles (ECDs) were searched by a least-squares fit to explain responses of 14–32 gradiometer channels over the sensorimotor cortex contralateral to the stimulated hand (Hämäläinen et al., 1993). An ECD represents the location, orientation and strength of the net current in the activated brain area. Only ECDs explaining >85% of the local field variance during the response peaks were accepted for further analysis.

For source analysis, the head was modelled as a homogeneous sphere. The model parameters were optimized for the intracranial space obtained from MR images that were available for all subjects of the control group and for two dyslexic individuals; the average of these 10 subjects' head models was used for the analysis of the remaining six dyslexic subjects.

The signals were first divided into several time periods, and during each of them one ECD was found at the main response peak. In many cases, the whole response sequence was not adequately explained by a single ECD, and thus three dipoles (the first for response to thumb stimuli, the second for response to index finger stimuli and the third for the late response) were identified in each hemisphere; the single ECDs were used to explain the data only during the corresponding response peaks. The ECDs were identified for the strongest responses that typically were obtained at the 200 ms SOA, and the same sources were used to explain responses at the other SOA. One normal-reading and one dyslexic subject were discarded from the source-analysis group data because no satisfactory sources were found.

### Statistical Analysis

Two-tailed t-tests were used for the between-subjects statistical comparisons of the behavioural data, source locations, source strengths and response latencies, and for the within-subjects comparisons of the response latencies and lateralization indices. The response amplitudes were compared with mixed-model analysis of variance [ANOVA; Subject Group as the between-subjects factor, and SOA, Response (first, second and third) and Hemisphere as within-subjects factors]. To test the effect of SOA on the responses, the 100 ms versus 200 ms response ratios were compared with mixed-model ANOVA. Lateralization indices were also compared between subjects with Mann–Whitney U-test.

## Results

The dyslexic subjects had normal performance in general linguistic tasks (ranges 104–140, 89–134 and 86–122, in Comprehension, Similarities, and Digit Span of WAIS-R, respectively). Table 1 presents results of the behavioural tasks and also gives statistical significances for group differences. Compared with a group of 39 age-matched control subjects, the dyslexic subjects were significantly (P < 0.01) slower in oral reading and rapid naming (mean differences 62 words/min and 219 ms/item, respectively), whereas the normal readers (6/8 tested) did not differ from this larger control population (P > 0.4). Dyslexic subjects were significantly slower at recognizing both real words and pseudowords (mean differences 314 and 571 ms) than the normal-reading subjects (7/8 tested).

Table 1

The behavioural profile of normally-reading control subjects and of dyslexic individuals in reading-related tests

Control subjects

Dyslexic subjects

P

Mean ± SEM

n

Mean ± SEM

n

Oral reading (words/min) 157 ± 15 97 ± 10 0.009
Rapid naming (ms/item) 437 ± 33 689 ± 47 0.002
Word recognition (ms)
Real words 533 ± 26 847 ± 76 0.006
Pseudowords

601 ± 29

7

1172 ± 155

7

0.01

Control subjects

Dyslexic subjects

P

Mean ± SEM

n

Mean ± SEM

n

Oral reading (words/min) 157 ± 15 97 ± 10 0.009
Rapid naming (ms/item) 437 ± 33 689 ± 47 0.002
Word recognition (ms)
Real words 533 ± 26 847 ± 76 0.006
Pseudowords

601 ± 29

7

1172 ± 155

7

0.01

### MEG Responses

Figure 1 illustrates the spatial distribution of somatosensory evoked fields of control subject C4 to left-sided stimulus trains presented at the 200 ms SOA. The strongest responses occur over the right sensorimotor cortex and consist of three prominent transient deflections, each peaking ∼50–90 ms after the onset of a finger stimulus; this triplet of transients is followed by a smaller fourth response. The inserts show these responses enlarged, both with the 0.03–90 Hz passband (Fig. 1A) and after high-pass filtering at 2 Hz (Fig. 1B). The earliest response peaks at 55 and 75 ms, and is followed by a larger, but similar response, with peaks at 256 and 273 ms (56 and 73 ms after the onset of the second stimulus). The third response is broader, with peaks at 453 and 488 ms (53 and 88 ms after onset of the third stimulus). The small fourth response peaks at 689 ms, 289 ms after the onset of the third stimulus and 201 ms after the previous transient response. Figure 1C shows the areal vector sum of a subset of 14 channels, encircled with a gray line in the figure; the statistical analysis was based on this type of vector sums. The response peaks that are seen at the single channels can be detected also in the areal vector sum.

### Sources of Responses

Figure 2 shows for control subject C2 the ECD for the first response superimposed on her MR images. The source is located in the posterior wall of the central fissure, corresponding to the hand area of the primary somatosensory cortex, SI. The peak latencies of source waveforms were in good agreement with latencies measured from the areal vector sums. Table 2 shows that the source locations did not differ significantly between the subject groups. The source strengths of the first responses were, on average, 17–23 nA·m, and did not differ significantly between the groups.

Figure 2.

The equivalent current dipole of the first response in subject C2 superimposed on her coronal and axial MR images.

Figure 2.

The equivalent current dipole of the first response in subject C2 superimposed on her coronal and axial MR images.

Table 2

The mean source coordinates ± SEM (in mm) for the first and second responses, and for the fourth response at the 200 ms SOA

Left hemisphere

Right hemisphere

x

y

z

x

y

z

Controls
First response −37 ± 2 3 ± 4 86 ± 3 40 ± 2 8 ± 4 85 ± 2
Second response −38 ± 1 3 ± 4 87 ± 2 39 ± 2 7 ± 4 86 ± 2
Late response −41 ± 2 6 ± 3 83 ± 3 40 ± 3 11 ± 3 82 ± 3
Dyslexics
First response −36 ± 2 4 ± 2 82 ± 2 42 ± 3 10 ± 2 83 ± 2
Second response −35 ± 2 7 ± 3 84 ± 2 42 ± 3 7 ± 1 85 ± 2
Late response

−39 ± 2

8 ± 3

81 ± 3

41 ± 3

12 ± 2

80 ± 3

Left hemisphere

Right hemisphere

x

y

z

x

y

z

Controls
First response −37 ± 2 3 ± 4 86 ± 3 40 ± 2 8 ± 4 85 ± 2
Second response −38 ± 1 3 ± 4 87 ± 2 39 ± 2 7 ± 4 86 ± 2
Late response −41 ± 2 6 ± 3 83 ± 3 40 ± 3 11 ± 3 82 ± 3
Dyslexics
First response −36 ± 2 4 ± 2 82 ± 2 42 ± 3 10 ± 2 83 ± 2
Second response −35 ± 2 7 ± 3 84 ± 2 42 ± 3 7 ± 1 85 ± 2
Late response

−39 ± 2

8 ± 3

81 ± 3

41 ± 3

12 ± 2

80 ± 3

The coordinate system is based on external landmarks on the skull: the x axis goes through periauricular points from left to right, y axis from the back of the head to nasion, and z axis points towards the vertex.

### Response Amplitudes in Dyslexic versus Control Subjects

Figure 3 depicts responses of the contralateral SI region in three control and three dyslexic subjects. The dashed horizontal lines show, for each individual, the peak amplitude of the first ∼50 ms response. At the 200 ms SOA (Fig. 3, top), the second response tends to be larger than the first in the control subjects, and the ratio of the second and first response is either slightly smaller in the left than in the right hemisphere (subjects C2 and C4), or similar between the hemispheres (C7). However, in the dyslexic subjects, the corresponding second/first response ratio is smaller in the right than in the left hemisphere. Note that, in all subjects, the responses typically are double-peaked, and the latencies and amplitudes of the first peak were used in the vector sum analysis.

Figure 3.

Evoked responses from the maximum channels in both hemispheres for three control and three dyslexic subjects at the 200 ms (top) and 100 ms (bottom) SOAs. The dashed lines mark for each subject the level of first response peak. The responses to second stimuli are shaded. The horizontal white and shaded bars below the traces refer to the stimuli to the first and second finger, respectively.

Figure 3.

Evoked responses from the maximum channels in both hemispheres for three control and three dyslexic subjects at the 200 ms (top) and 100 ms (bottom) SOAs. The dashed lines mark for each subject the level of first response peak. The responses to second stimuli are shaded. The horizontal white and shaded bars below the traces refer to the stimuli to the first and second finger, respectively.

The responses to second and third stimuli were smaller at the 100 ms than at the 200 ms SOA in all six subjects (Fig. 3, bottom). Whereas the first and third responses (both to thumb stimuli) at the 200 ms SOA were of equal size in these subjects, at the 100 ms SOA the third response was clearly smaller than the first. Interestingly, the fourth response seemed to be less affected by the SOA.

Figure 4 shows the mean (± SEM) response amplitudes and latencies across all subjects, calculated from the areal vector sums at both SOAs. In control subjects, the response amplitudes behave in a similar way in both hemispheres: the first and second responses are practically equal at the 100 ms SOA, and the second response is larger than the first at the 200 ms SOA.

Figure 4.

The mean (± SEM) response amplitudes and latencies for all subjects at both SOAs measured from the areal vector sums. Note that the error bars for latencies are typically so small that they are covered by the symbols.

Figure 4.

The mean (± SEM) response amplitudes and latencies for all subjects at both SOAs measured from the areal vector sums. Note that the error bars for latencies are typically so small that they are covered by the symbols.

In the right hemisphere, Subject Group × Response interaction was statistically significant [F(2,28) = 5.5, P < 0.01, ANOVA]: the responses to second stimuli were at both SOAs smaller in dyslexic than control subjects (100 ms SOA: mean difference 41%, P < 0.01; 200 ms SOA: 28%, P < 0.03). In the left hemisphere, no similar Subject Group × Response interaction emerged (P = 0.8, ANOVA).

Figure 5 illustrates the lateralization indices calculated for the second/first response ratios in all subjects at the 200 ms SOA. The LIs of control subjects are centered to the middle of the left–right axis (mean ± SEM = 0.04 ± 0.05; non-significant difference with P = 0.38 compared with zero). In contrast, the LIs of dyslexics were clustered to the left side of the axis (−0.21 ± 0.05; P < 0.005). The LI distributions differed significantly between the two groups (P < 0.005; Mann–Whitney U-test). The corresponding LIs were symmetrical at 100 ms SOA in both groups, as were the LIs for the third/first response ratios at both SOAs.

Figure 5.

The lateralization indices of second/first response ratios in both subject groups at the 200 ms SOA.

Figure 5.

The lateralization indices of second/first response ratios in both subject groups at the 200 ms SOA.

The second and third responses were, in both subject groups and in both hemispheres, significantly smaller at the 100 ms than at the 200 ms SOA (P < 0.001, ANOVA; by 43 ± 5% in dyslexics and by 38 ± 7% in controls). At the 100 ms SOA, the second response could not be detected in the left hemisphere of one control and of two dyslexic subjects, and in the right hemisphere of one dyslexic subject (see Materials and Methods). The third response was not detected in either hemisphere of one control subject, nor in the left hemisphere of another control subject and of three dyslexic subjects.

### Response Latencies

The response latencies did not differ between the groups or hemispheres. The responses peaked up to 35 ms later to the third than the first stimulus (i.e. to the repeated stimulus at the thumb) in both subject groups at the 100 ms SOA (P < 0.01).

### Later Transient Responses

At the 200 ms SOA, an additional fourth response was seen in all subjects. This response peaked 319 ± 4 ms after the third stimulus, and it was of similar strength and latency in both subject groups (see Fig. 4). At the 100 ms SOA, two additional responses (the fourth and fifth transients), peaking at 200 ± 5 ms and 340 ± 9 ms after the third stimulus, respectively, were detected in seven control and in seven dyslexic subjects at least in one hemisphere. The SOA affected the second, the third and the latest responses (the fourth at 200 ms and the fifth at 100 ms SOA) in a different manner [F(2,28) = 3.7, P < 0.04, ANOVA]: the late responses were significantly less affected (P < 0.02 and P = 0.06 compared with the second and third responses, respectively), without differences between the subject groups.

## Discussion

The present MEG recordings revealed significant differences between dyslexic and normal-reading subjects in the reactivity of their right somatosensory cortex to repetitive tactile stimuli: with SOAs of 100 and 200 ms, the 50 ms responses to second stimuli of the trains were significantly smaller in dyslexic than control subjects. Consequently, the recovery cycles were clearly asymmetric in dyslexic subjects. Responses to the first stimuli of the trains were of similar strength between groups and hemispheres, suggesting that the observed effect was specific to stimulus repetition within the train.

The primary somatosensory cortex, SI, is known to respond strictly time-locked to tactile stimuli. We focused on the 50-ms tactile responses that are likely to be generated in the cytoarchitectonic area 3b, and to correspond, e.g. in their current direction, to the 30-ms responses elicited by electric median nerve stimuli (Simões et al., 2001). Area 3b has a clear somatotopic organization but with significant functional overlap of representations for fingers of the same hand (Simões et al., 2001) so that the observed interaction of thumb and index finger stimulation in the present study might be due to the inhibitory surround of the activated thumb area.

The strength of an evoked response depends strongly on the interstimulus interval, and the response recovery also varies according to the cortical area (Hari et al., 1993; Uusitalo et al., 1996). The auditory 100 ms response to the second sound of a pair is, at short SOAs, smaller in dyslexic than normal-reading adults, suggesting prolonged post-stimulus suppression and prolongation of the auditory recovery cycle (Nagarajan et al., 1999). Similarly, our present results could be related to prolongation of the tactile recovery cycle in dyslexic adults, but only in their right hemispheres.

The diminished responses in the right SI cortex to rapidly presented tactile stimuli agree with the pansensory nature of the temporal processing deficit in dyslexic subjects, as proposed by several authors (Stein and Walsh, 1997; Hari and Renvall, 2001). Earlier findings on tactile processing in dyslexic subjects are surprisingly sparse. Language-learning-impaired children had difficulties in identifying which two fingers of the same hand were touched simultaneously (Johnston et al., 1981; Tallal et al., 1985). Moreover, dyslexic adults were impaired in detecting 3 Hz, but not 30 or 300 Hz, vibratory stimuli in the index finger of the writing hand (Stoodley et al., 2000), and their tactile discrimination thresholds for the orientation and ridge-width of gratings were enhanced in both hands (Grant et al., 1999). In addition, segregation of rapidly presented tactile, auditory and visual stimuli was impaired in dyslexic children and adults (Laasonen et al., 2000, 2001, 2002).

Although the present study addresses very basic somatosensory processing mechanisms, the right-sided parietal lobe abnormality is intriguing in the framework of the recently reported left-sided visual ‘mini-neglect’ in dyslexic adults (Hari et al., 2001; see also Stein et al., 1989; Riddell et al., 1990; Facoetti and Turatto, 2000). Lesions of the right parietal cortex can produce unilateral neglect, often associated with stimulus extinction, i.e. a failure to detect left-sided visual, tactile or auditory stimuli when they are presented simultaneously with right-sided stimuli (for a review, see Vallar, 1998). Against this background, the present hemispheric asymmetry in the recovery cycles of somatosensory responses could predict decreased perceptual salience of left-sided stimuli during rapid bilateral stimulation in dyslexic individuals.

Although only three tactile stimuli were presented, four (at 200 ms SOA) or five (at 100 ms SOA) transient responses were observed within the analysis interval. A control experiment on one individual indicated that single stimuli evoke transient off-responses ∼300 ms after the stimulus onset, and therefore the fourth response for the 200 ms SOA trains is likely an off-response to the third stimulus, and the fourth and fifth responses for the 100 ms SOA trains reflect off-responses to the second and third stimuli, respectively.

Sensory processing deficits and their severity differ strikingly between dyslexic individuals, without necessarily predicting the subjects' phonological or reading skills that are affected by individual compensatory mechanisms, as well as by the orthography of the language. Many of the dyslexic subjects' sensory and motor temporal processing deficits reported in the literature are likely to play only a limited role in the development of the reading impairment. However, the existence of such minor abnormalities in many sensory modalities suggests a general problem, and therefore may give an important insight into the genesis of the reading impairment as well. Our data, demonstrating abnormal recovery of somatosensory cortical responses in the right hemisphere of young dyslexic adults, further support the proposed pansensory nature of temporal processing deficit in dyslexic individuals.

We thank Tiina Parviainen for collecting the behavioural data, and Tiina Parviainen and Topi Tanskanen for comments on the manuscript. This work was supported financially by the Academy of Finland and the Sigrid Jusélius Foundation.

## References

1983
) Categorizing sounds and learning to read — a causal connection.
Nature

301
:
419
–421.
Facoetti A, Turatto M (
2000
) Asymmetrical visual fields distribution of attention in dyslexic children: a neuropsychological study.
Neurosci Lett

290
:
216
–218.
Grant AC, Zangaladze A, Thiagarajah MC, Sathian K (
1999
) Tactile perception in developmental dyslexia: a psychophysical study using gratings.
Neuropsychologia

37
:
1201
–1211.
Hämäläinen M, Hari R, Ilmoniemi RJ, Knuutila J, Lounasmaa OV (
1993
) Magnetoencephalography — theory, instrumentation, and applications to noninvasive studies of the working human brain.
Rev Mod Phys

65
:
413
–497.
Hari R, Karhu J, Hämäläinen M, Knuutila J, Salonen O, Sams M, Vilkman V (
1993
) Functional organization of the human first and second somatosensory cortices: a neuromagnetic study.
Eur J Neurosci

5
:
724
–734.
Hari R, Kiesilä P (
1996
) Deficit of temporal auditory processing in dyslexic adults.
Neurosci Lett

205
:
138
–140.
Hari R, Renvall H (
2001
) Impaired processing of rapid stimulus sequences in dyslexia.
Trends Cogn Sci

5
:
525
–532.
Hari R, Renvall H, Tanskanen T (
2001
) Left minineglect in dyslexic adults.
Brain

124
:
1373
–1380.
Helenius P, Uutela K, Hari R (
1999
) Auditory stream segregation in dyslexic adults.
Brain

122
:
907
–913.
Johnston RB, Stark RE, Mellits ED, Tallal P (
1981
) Neurological status of language-impaired and normal children.
Ann Neurol

10
:
159
–163.
Laasonen M, Tomma-Halme J, Lahti-Nuuttila P, Service E, Virsu V (
2000
) Rate of information segregation in developmentally dyslexic children.
Brain Lang

75
:
66
–81.
Laasonen M, Service E, Virsu V (
2001
) Temporal order and processing acuity of visual, auditory, and tactile perception in developmentally dyslexic young adults.
Cogn Affect Behav Neurosci

1
:
394
–410.
Laasonen M, Service E, Virsu V (
2002
) Crossmodal temporal order and processing acuity in developmentally dyslexic young adults.
Brain Lang

80
:
340
–354.
Liberman IY, Shankweiler D (
1985
) Phonology and the problems of learning to read and write.
Remed Spec Educ

6
:
8
–17.
Lundberg I, Olofsson A, Wall S (
1980
) Reading and spelling skills in the first school years predicted from phonemic awareness skills in the kindergarten.
Scand J Psychol

21
:
159
–173.
Mertens M, Lütkenhöner B (
2000
) Efficient neuromagnetic determination of landmarks in the somatosensory cortex.
Clin Neurophysiol

111
:
1478
–1487.
Nagarajan S, Mahncke H, Salz T, Tallal P, Roberts T, Merzenich MM (
1999
) Cortical auditory signal processing in poor readers.

96
:
6483
–6488.
Renvall H, Hari R (
2002
) Auditory cortical responses to speech-like stimuli in dyslexic adults.
J Cogn Neurosci

14
:
757
–768.
Renvall H, Lehtonen R, Hari R (
2002
) Abnormal recovery cycle in the right primary somatosensory cortex of dyslexic adults. Hum Brain Mapp 2002:abstr. 877.
Riddell P, Fowler S, Stein J (
1990
) Spatial discrimination in children with poor vergence control.
Percept Mot Skills

70
:
707
–718.
Simões C, Mertens M, Forss N, Jousmäki V, Lütkenhöner B, Hari R (
2001
) Functional overlap of finger representations in human SI and SII cortices.
J Neurophysiol

86
:
1661
–1665.
Stein J, Walsh V (
1997
) To see but not to read; the magnocellular theory of dyslexia.
Trends Neurosci

20
:
147
–152.
Stein J, Riddell P, Fowler S (
1989
) Disordered right hemisphere function in developmental dyslexia. In: Brain and reading (von Euler C, Lundberg I, Lennerstrand G, eds), pp. 139–157. Houndmills: Macmillan.
Stoodley CJ, Talcott JB, Carter EL, Witton C, Stein JF (
2000
) Selective deficits of vibrotactile sensitivity in dyslexic readers.
Neurosci Lett

295
:
13
–16.
Tallal P (
1980
) Auditory temporal perception, phonics, and reading disabilities in children.
Brain Lang

9
:
182
–198.
Tallal P, Stark R, Mellits D (
1985
) Identification of language-impaired children on the basis of rapid perception and production skills.
Brain Lang

25
:
314
–322.
Uusitalo M, Williamson SJ, Seppä MT (
1996
) Dynamical organisation of the human visual system revealed by lifetimes of activation traces.
Neurosci Lett

213
:
149
–152.
Vallar G (
1998
) Spatial hemineglect in humans.
Trends Cogn Sci

2
:
87
–97.

## Author notes

1Brain Research Unit, Low Temperature Laboratory, Helsinki University of Technology, FIN-02015 HUT, Espoo, Finland and 2Department of Clinical Neurophysiology, Helsinki University Central Hospital, FIN-00290 Helsinki, Finland