OUP user menu

Cross-modal plasticity revealed by electrotactile stimulation of the tongue in the congenitally blind

Maurice Ptito, Solvej M. Moesgaard, Albert Gjedde, Ron Kupers
DOI: http://dx.doi.org/10.1093/brain/awh380 606-614 First published online: 5 January 2005

Summary

In sensory substitution, information acquired with one sensory modality is used to accomplish a task which is normally subserved primarily by another sensory modality. We used PET to study cross-modal plasticity in the congenitally blind, using electrotactile stimulation of the tongue. Blind (n = 6) and sighted blindfolded controls (n = 5) were scanned before and after they were trained to use their tongue in a Snellen orientation detection task. Results showed that both groups of subjects learned the discrimination orientation task after seven 1 h training sessions. Before training, no significant changes in regional cerebral blood flow (rCBF) were observed in the occipital cortex in either group. In sharp contrast, activity in the occipital cortex increased after practice for the blind, but not for the sighted, providing evidence for training-induced plasticity in the blind. An inter-regional correlation analysis showed that task-related rCBF changes in left posterior parietal cortex were positively correlated with rCBF changes in the occipital area of the trained blind. These data reveal that cross-modal plasticity in the blind develops rapidly and that the occipital cortex is part of a functional neural network for tactile discrimination in conjunction with the posterior parietal cortex. Our data further show that the tongue can act as a portal to convey somatosensory information to visual cortex.

  • sensory substitution
  • cross-modal plasticity
  • vision
  • blind
  • brain imaging
  • LOtv = lateral occipital tactile-visual area
  • rCBF = regional cerebral blood flow
  • TMS = transcranial magnetic stimulation
  • TDU = Tongue Display Unit

Introduction

Early brain lesions can lead to altered connectivity patterns in the adult brain. For example, the destruction of the superior colliculus and ascending auditory pathways leads to the formation of permanent and novel retinal projections to non-visual thalamic nuclei such as the medial geniculate nucleus that normally mediates audition (Sur et al., 1988; Frost et al., 2000). These abnormal retinal projections are retinotopically organized and form functional synapses (Frost and Métin, 1985; Campbell and Frost, 1988). Neurons in the auditory cortex of these rewired animals possess visual receptive field properties similar to those of neurons in primary visual cortex (Ptito et al., 2001). The newly formed pathway from the retina to the auditory cortex can mediate visually guided behaviours in the absence of the normal retino-geniculate pathway (Frost et al., 2000; Von Melchner et al., 2000).

Cross-modal plasticity has also been demonstrated in humans with various forms of sensory deficits. One of the earliest examples stems from brain imaging studies showing that Braille reading activates primary and secondary visual cortical areas in blind subjects (Sadato et al., 1996, 1998; Buchel et al., 1998; Cohen et al., 1999; Burton et al., 2002a). The functional role of these activations is underscored by the observation that transcranial magnetic stimulation (TMS) of the occipital cortex interferes with Braille reading performance in the blind (Cohen et al., 1997). In deaf subjects, visual stimuli activate the auditory cortex (Finney et al., 2001) whereas auditory localization tasks activate the visual cortex of the congenitally blind (Weeks et al., 2000), lending more support to remarkable cortical plasticity.

Recently, many efforts have been devoted to the development of devices to convey visual information in the blind. Non-invasive sensory substitution techniques replacing vision with audition (De Volder et al., 1999; Arno et al., 2001) or somesthesis (Kajimoto et al., 2001; Bach-y-Rita and Kercel, 2003) have been tested successfully. Highly invasive techniques include electrical stimulation of the retina (Humayun et al., 1999), optic tract (Veraart et al., 2003) and visual cortex (Schmidt et al., 1996; Normann et al., 1999), and intraocular implantation of stem cells (Fine et al., 2003). We have used a newly developed human–machine interface, the Tongue Display Unit (TDU), that has proven its efficiency in blind subjects (Bach-y-Rita et al., 1998; Sampaio et al., 2001; Bach-y-Rita and Kercel, 2003). This device uses the tongue as a substrate for electrotactile stimulation because of its excellent conductance for electrical stimulation, high sensitivity, good accessibility and large cortical representation within the somatosensory cortex (Picard and Olivier, 1983; Nakamura et al., 1998). We utilized PET to identify the cerebral correlates of cross-modal plasticity that develops following training with the TDU device. PET methodology was chosen simply because the TDU equipment currently is not yet compatible with functional MRI. Congenitally blind and sighted blindfolded controls were tested pre- and post-training in a task in which they had to detect the orientation of a Snellen pattern applied to the tongue. The results revealed significant increases of regional cerebral blood flow (rCBF) in visual cortical areas after training in the blind but not in the sighted controls.

Material and methods

Subjects

Six congenitally blind (three males and three females, mean age 38 ± 7.3 years; Table 1) and five sighted controls (two males and three females, mean age 29 ± 3.2 years) participated in the study. Structural brain MRIs of the blind appeared normal. Sighted controls had normal neurological examinations and normal visual acuity. All subjects gave written informed consent and the study was approved by the Ethics Committee of Aarhus county.

View this table:
Table 1

Characteristics of the blind subjects

SubjectSexAge (years)Onset of blindnessCause of blindnessYears of Braille reading
B.F.F41BirthRetinotopy of prematurity35
A.H.F23BirthRetinotopy of prematurity17
S.N.F357 yearsLeber's congenital amaurosis30
O.B.M41BirthRetinotopy of prematurity35
C.T.M33BirthRetinotopy of prematurity26
J.B.M42BirthRetinotopy of prematurity35

Electrotactile tongue stimulation

The tactile vision substitution system comprises the TDU, a laptop computer with custom-made software (imagiTact; http://www.wicab.com) and an electrode array (3 × 3 cm) consisting of 144 gold-plated contacts each with a 1.55 mm diameter arranged in a 12 × 12 matrix. The subject explores, with the help of a computer mouse, a tumbling T (stimulus size: 270 × 225 pixels or 2.7 × 2.3 cm) presented on the laptop (Fig. 1 top). The stimulus is sampled and reduced to the 12 × 12 resolution of the tongue display with an update rate of 14–20 frames/s. An electrical pulse (40 μs positive pulse) is generated when the cursor is superposed on a pixel forming the T. The number of electrodes active at any particular time depends on the superposition of the cursor with the image on the screen (Fig. 1 bottom). Only when the cursor is positioned completely over the image are all the pixels of the image simultaneously converted into electrical pulses and the subject will perceive the whole T (a). When the cursor partly overlaps the image, only those pixels of the T that are covered by the cursor are converted into electrical pulses (b and c). If the cursor is outside the image, none of the electrical contacts are active (d). At the beginning of each trial, the cursor is always centred over the T.

Fig. 1

Experimental set-up and procedure. Top: the TDU and its component parts (see text for details). Bottom: in (a), the cursor is perfectly positioned over the stimulus and all pixels forming the T are electrically activated (dark shading). In (b), the cursor is moved to the left and only the pixels forming the left part of the T are activated. In (c), the cursor is moved upright and only the pixels forming the horizontal part of the T are activated and in (d), the cursor is far from the T and no pulses are delivered to the tongue.

Behavioural training

Both blind subjects and controls were blindfolded during the experiments. They were trained during seven consecutive days using the TDU. Each session lasted ∼1 h and comprised 32 trials. During training, subjects learned to use the TDU to discriminate the orientation of the letter T that was presented on a laptop. Ts were presented randomly in one of four possible orientations (up, right, left or down). Subjects had a maximum of 60 s to explore each image and they received immediate feedback about the correctness of their response. Both the reaction time and response accuracy were measured. Performance criterion was set at 85% correct responses on two consecutive days. Statistical analysis of the behavioural data was carried out using the general linear model for repeated measures (Statistical Package for the Social Sciences). Values of P < 0.05 were considered as statistically significant.

PET scanning

Before the first PET scanning session, subjects were familiarized with the apparatus. Cerebral blood flow was measured with an ECAT Exact HR47+ PET camera in 3D mode following intravenous bolus injections of 400 MBq of 15O-labelled water. A single 60 s frame was acquired, starting at 60 000 true counts/s. Successive scans were separated by at least 10 min. Subjects were scanned at rest (the electrode array was placed on the tongue but no stimulation was given), during the presentation of random dots (no pattern was present in the stimulus) and during an orientation discrimination task. In the orientation task, the letter T was presented in different orientations and subjects had to detect its orientation. The presentation of the tumbling Ts started simultaneously with tracer injection and was continued throughout the rest of the scanning time. Each condition was repeated four times in a semi-randomized order. The PET images were reconstructed after correction for scatter and measured attenuation from a 68Ga transmission scan. The 47 3.1 mm slices were filtered to 14 mm full width at half maximum (FWHM) isotropic (Hanning filter cut-off frequency 15 cycles/s). PET volumes were realigned using the Automated Image Registration (AIR) software to correct for head movements between scans. The first PET image was co-registered with the subject's T1-weighted MRI brain volume and mapped into standardized stereotactic space. Statistical maps were calculated after a pixel-by-pixel subtraction of PET volumes. Activation maps were computed with respect to rest and random dots, but only the former are presented in the Results.

Results

Both blind and control subjects learned the orientation task. Figure 2 illustrates the learning curves for percentage correct responses (Fig. 2A) and reaction times (Fig. 2B). A statistical analysis of the time×group interactions yielded no significant differences in the percentage of correct responses (F = 1.62; P > 0.05) and reaction times (F = 1.82; P > 0.05), indicating that learning occurred as rapidly in both groups of subjects. Before training, rCBF did not increase in occipital cortex during the orientation task in either group. Activations pre-training were limited to parietal and prefrontal (motor) cortices (Table I). After training, blind subjects activated large areas of occipital (cuneus, inferior, medial and lateral occipital cortex), occipito-parietal and occipito-temporal (fusiform gyrus) cortices (Fig. 3A). Activated brain areas (Brodmann areas and Talairach coordinates) for blind and control subjects pre- and post-training are reported in Table 2. In sharp contrast, controls showed no activation of visual cortex (Fig. 3B), despite behavioural performance equal to the blind.

Fig. 2

Performance curves of congenitally blind and blindfolded controls in the orientation discrimination task.

Fig. 3

PET data (group analysis) showing brain activations following electrotactile stimulation of the tongue following training for the blind (A) and sighted controls (B). Note the rCBF increases in visual cortex of the trained blind only; rCBF changes were correlated with task performance. The left side of the image corresponds to the left side of the brain. The values below refer to the z-coordinate of the slices.

View this table:
Table 2

Brain areas activated before and after training in the orientation discrimination task for blind and sighted blindfolded controls (TDU baseline)

Congenitally blindSighted controls
Anatomical area of activationTalairach coordinatesAnatomical area of activationTalairach coordinates
xyztxyzt
Before training
    Cerebellum0−76−205DIPSA−38−37577.7
32−54−244.642−38515.4
    DIPSA−40−40514.7Cerebellum8−57−156.8
    DIPSM−26−59653.7Postcentral gyrus66−16326.4
    Inferior temporal gyrus50−61−173.7Medial frontal gyrus−28−4626.8
    Medial frontal gyrus−113593.03253213.7
Medial frontal gyrus−4−1605.1
Superior frontal gyrus (SMA)88664.8
Thalamus8−784.0
After training
    Cerebellum10−62−1611.5Inferior frontal gyrus6411276.4
31−60−228.7−59−3344.0
    DIPSM−20−56658.9Cerebellum0−62−126.4
17−59647.4DIPSA46−40665.9
    Inferior temporal gyrus34−61−218.7−38−43644.2
−22−57−245.2Medial frontal gyrus3254235.3
46−71−86.4Anterior insula381975.3
    DIPSA−38−38528.0−2722114.9
33−38526.4Postcentral gyrus69−19385.2
    Precentral gyrus−58−11357.3Thalamus−3−985.1
56−9324.312−14124.8
    Cuneus0−92116.4Anterior cingulate cortex717485.0
    Superior occipital gyrus−16−78405.7Postcentral gyrus68−16384.9
20−81393.9Inferior parietal lobule69−33284.9
    Superior frontal gyrus−14−7666.4Superior frontal gyrus (SMA)374654.4
27−1604.6−174604.3
    Lateral occipital sulcus−40−83114.7DIPSM−27−57684.4
−48−74−63.9Inferior temporal gyrus42−50−234.4
    Medial occipital gyrus38−69−66.2
  • DIPSA = anterior dorsal intraparietal sulcus area; DIPSM = medial dorsal intraparietal sulcus area.

Control subjects activated primary somatosensory and motor cortex, right dorsolateral prefrontal and anterior cingulate cortices, right anterior insula and thalamus. Controls also showed significant rCBF increases in right posterior parietal cortex. It should be noted that following training, rCBF increases in superior parietal cortex were significantly stronger in the blind (left: x = −20; y = −56; z = 65; t = 8.9; right: x = 17; y = −59; z = 64; t = 7.4) compared with controls (left: x = –27; y = −57; z = 68; t = 4.4; right: x = 46; y = −40; z = 66; t = 5.91). A time × group interaction analysis further revealed that rCBF in occipital cortex increased significantly (P < 0.001) in the blind after practice, but not in sighted controls (Fig. 4).

Fig. 4

Time × group interactions in four regions of interest: cuneus (x = 0, y = −91, z = 11), superior occipital (x = −16, y = −78, z = 41), medial dorsal intraparietal sulcus area (DIPSA; x = −20, y = −55, z = 65) and anterior dorsal intraparietal sulcus area (DIPSM; x = −37, y = −38, z = 53). Graphs show changes in rCBF during orientation discrimination compared with rest. Open bars represent changes in rCBF before training; grey shaded bars represent rCBF changes after training (mean ± SEM). Before training, significant rCBF increases occurred in the DIPSM and DIPSA in both groups. Following training, blind subjects (but not controls) showed significant activation of the cuneus and superior occipital gyrus. Training also induced significant rCBF increases in DIPSM in both groups. No significant training-induced rCBF increase was observed in DIPSA. Between-group comparisons revealed that activity in the cuneus and the occipital cortex increased significantly in the blind compared with controls after practice (P < 0.05). *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

A single subject analysis revealed that all six blind subjects activated the visual cortex following training. Figure 5 shows the individual results on axial slices, taken at the level of superior occipital cortex and cuneus. Single subject analysis further revealed that none of the control subjects showed an occipital activation following training (data not shown).

Fig. 5

PET data (single subject analysis) showing significant rCBF increases in the superior occipital cortex and the cuneus (indicated by the red arrows) in the six trained blind subjects.

In the control condition, using non-patterned stimulation (random noise), the visual cortex was significantly less activated than during the orientation task. After training, blind subjects showed only a moderate rCBF increase in cuneus and fusiform gyrus but not in superior occipital or parieto-occipital cortices. Control subjects did not show any rCBF increases in the random noise condition.

Discussion

Our data show that following a 7 day period of intensive training with a sensory substitution device, congenitally blind subjects activate their visual cortex when performing a simple orientation discrimination task whereas trained blindfolded sighted controls do not despite equal task performance. These data add to a growing amount of evidence for a functional role for the occipital cortex in the blind. Earlier brain imaging studies showed occipital activation in the blind during Braille reading (Sadato et al., 1996, 1998; Buchel et al., 1998; Cohen et al., 1999; Burton et al., 2002a), auditory localization (Weeks et al., 2000), lexical and phonological tasks (Burton et al, 2002b), haptic exploration of man-made objects (Amedi et al., 2001; James et al., 2002; Pietrini et al., 2004) and memory processing (Amedi et al., 2003). In addition, abnormally high rates of glucose metabolism, rCBF and oxygen consumption in visual cortex of the early blind further argue for a functional role for this area (De Volder et al., 1997). Since we used a novel task, we could compare rCBF changes in the group of early blind subjects with those in a group of sighted controls that were equally naive on the behavioural task. We were therefore able to dissociate group or group× time effects from simple performance effects. Without the inclusion of the control group that matched the blind group in terms of performance, it would have been impossible to discern whether the rCBF changes post-training reflect cross-modal plasticity or are merely the result of improvement in performance. The absence of statistically significant differences in the performance of both groups strengthens the argument that observed rCBF differences are not a result of differences in task difficulty or in task performance.

In line with our hypothesis, we did not find any significant activation of visual areas during the orientation discrimination task in our non-trained blind and control subjects. This is in contrast to some recent preliminary findings by Sadato et al. (2004) in two late blind subjects who were naive to Braille. These authors reported a small, yet significant, increase in occipital activity in both subjects during a Braille discrimination task, and they concluded that cross-modal plasticity in the late blind is not learning dependent. The discrepancy between the results of both studies might be explained by differences in methodology. For example, our subjects were congenitally blind with no visual experience and no residual light perception, whereas those of Sadato et al. (2004) were blinded late in life and still had some light perception. They therefore had acquired a large visual repertoire that could have been used to recruit the occipital cortex. Moreover, the paradigm in the study of Sadato et al. also involved a memory component that may explain the occipital activation (Amedi et al., 2003).

Interestingly, control subjects showed a significant deactivation of the visual cortex during the orientation discrimination task before and after training. This is in line with results from earlier functional neuroimaging studies (Haxby et al., 1994; Kawashima et al., 1995) showing that attention directed towards a particular sensory modality is associated with significant decreases in activity in cortical areas processing input from other sensory modalities. For instance, large blood flow decreases in primary auditory and somatosensory cortices have been reported during tasks of visual processing (Haxby et al., 1994; Kawashima et al., 1995; Shulman et al., 1997). By decreasing activity in these task-irrelevant modalities that might compete for attention, selectivity processing is enhanced. Therefore, the significant deactivation of visual cortex in the control subjects can be explained by the increased attentional resources directed towards the somatosensory input.

In sharp contrast to the lack of activation in visual areas before training, blind (but not control) subjects activated large parts of the visual cortex following training. The activation pattern in the blind following training shows remarkable similarities with that observed in normal seeing subjects during the performance of a visual orientation task. Activation of anterior and posterior intraparietal sulcus in a visual orientation discrimination task was previously reported in normal subjects using fMRI (Faillenot et al., 2001; Shikata et al., 2001). The stereotactic coordinates of the activation sites in our blind subjects are very close to those reported in normal subjects by Faillenot et al. (2001). Their coordinates for the left anterior and posterior part of the intraparietal sulcus are, respectively, −40, −40, 56 and −24, −60, 64 compared with ours, −38, −38, 52 and −20, −56, 65. In addition, our blind subjects also activated the cuneus, lateral occipital, posterior fusiform and inferior temporal cortex, as reported in seeing control subjects during visual orientation discrimination (Faillenot et al., 2001).

How can we explain this pattern of rCBF increases in visual cortex in the blind following training? It might be argued that mental imagery is at the basis of the activation of visual cortex. For several reasons, this seems to be an unlikely explanation. First, our blind subjects never had visual experiences and, during the debriefing following the experiments, they did not report that they had engaged in visual imagery during the orientation detection task (see also Pietrini et al., 2004). Secondly, if mental imagery would be the basis of the activation in visual cortex, sighted controls should also have activated this area which was clearly not the case.

It could also be argued that the occipital activation is a task-specific epiphenomenon triggered by the mere fact of sensory or motor activation. We have some direct and indirect arguments against this hypothesis. First, there is evidence that TMS of the occipital cortex disrupts Braille reading in the early blind but not in normal seeing control subjects (Cohen et al., 1997). In the latter group, TMS over somatosensory cortex interfered with task performance. In addition, a recent PET study showed that in contrast to Braille reading, pure sensory stimulation or motor tasks involving the hand used for Braille reading do not activate the occipital cortex in blind subjects (Gizewski et al., 2003). Accordingly, our control condition in which non-patterned (random noise) electrotactile stimulation was applied to the tongue also did not reveal occipital activation, further supporting the task specificity of the activation (see also Sadato et al., 2002).

A possible explanation for the observed occipital activation in the blind is that it is mediated by the superior parietal lobe (Neal et al., 1990). Tactile information is processed in the anterior part of the superior parietal cortex (area 7a), whereas visual information is processed more caudally, in area 7b. Therefore, tactile information may reach the visual system through increased connectivity between areas 7a and 7b after loss of vision, as suggested by a stronger activation of superior parietal areas in the blind. This hypothesis is in line with data of single unit recordings in blind-raised monkeys by Hyvarinen et al. (1981). These authors reported a dramatic increase in the percentage of neurons responding to active object manipulation in parietal association (BA7) and extrastriate visual (BA19) cortices in blind-raised monkeys compared with control monkeys (Hyvarinen et al., 1981). A connectivity analysis (Horwitz et al., 1998; Weeks et al., 2000), performed a posteriori with the activated areas of the parietal cortex as the reference region, indicated increased connectivity between anterior and posterior parietal cortex and between these areas and the visual cortex (Fig. 6). This supports our hypothesis that tactile information may reach the occipital cortex through the parietal association cortex. This hypothesis is supported by the results of a recent study by Wittenberg et al. (2004). These authors used repetitive TMS combined with PET to probe the connections between primary somatosensory and occipital cortex in early blind, late blind and sighted control subjects. They reported that only the early blind subjects showed activation of early visual areas when repetitive TMS was delivered over primary somatosensory cortex, further supporting the idea of increased connectivity between somatosensory and occipital cortex in the early blind.

Fig. 6

Within-task inter-regional correlations for three reference regions: left DIPSA (−38, −38, 52), left DIPSM (−20, −56, 65) and midline cuneus (0, −92, 11) following training. We calculated for each subject the relationship between blood flow for a single voxel in a reference region and for all other brain voxels. A correlation coefficient across subjects was determined in order to define the strength of the connection between regions: the higher the correlation coefficient, the stronger the functional connectivity with the reference region (40). The statistical threshold was set at P < 0.05. The maps shown are for the blind group only, since no significant inter-regional correlations for the chosen reference regions were found for the blindfolded controls. Correlation maps indicate increased connectivity between parietal and occipital regions following training (scale bar red, r = 0.90; purple, r = 0.60). Numbers below refer to the z-level in Talairach space. DIPSM = medial dorsal intraparietal sulcus area; DIPSA = anterior dorsal intraparietal sulcus area.

Alternatively, a pathway involving the lateral occipital tactile-visual (LOtv) region may play a role in the activation of the visual cortex (Amedi et al., 2002). This area is activated by visual and haptic exploration of objects (Amedi et al., 2002; James et al., 2002). The centre of area LOtv (x = −47, y = −62, z = −10) is very close to the coordinates of the present lateral occipital activation (x = 38, y = −69, z = −6). Our results are therefore also consistent with the hypothesis that somatosensory stimulation in the blind activates the occipito-temporal region and further expands into earlier retinotopic areas.

Blind subjects also showed a significant training-induced rCBF increase in the cerebellum. This is in line with earlier results showing strong cerebellar activation during Braille reading (Buchel et al., 1998; Sadato et al., 1998; Burton et al., 2002a; Gizewski et al., 2004). Although the cerebellum is traditionally considered to be a structure involved in motor coordination, recent evidence seems to indicate that it may also play a purported role in a variety of non-motor functions, including sensory discrimination, working memory, verbal learning, attention and time perception (see Schmahmann and Sherman, 1998 for a review). It could therefore be argued that increased cerebellar activity may be involved in performance improvement after training. We do not think this is the case. First, our control subjects also learned to perform the orientation discrimination task and they did not show an increase in cerebellar activity post-training. Secondly, the stereotactic coordinates of our cerebellar activation coincide with those reported for right hand movement (Allen et al., 1997; Mima et al., 1999). We therefore believe that the cerebellar activation in the blind reflects increased motor output of the right hand used to manipulate the mouse rather than training-induced changes in sensory discrimination of the tongue. Alternatively, the rCBF response in superior-anterior cerebellum may rely on increased attentional demands (Allen et al., 1997; Rees et al., 1997) or occipital input (Schmahmann and Pandya, 1993). Anatomical studies have shown that the cerebellum receives input from dorsal peristriate regions via the medial pontine pathway (Fries, 1990; Schmahmann and Pandya, 1993). Our superior occipital activation may therefore be at the basis of the cerebellar activation. Taken together, the most parsimonious explanation for our findings is that the improvement in performance in our blind subjects is mediated by the activation in occipital cortex.

Although blind and sighted subjects showed comparable performance in the behavioural tasks, only the blind recruited the visual cortex. It therefore seems that in sighted subjects, correct task performance does not necessitate a contribution of the visual cortex but rather of the somatosensory cortex. When sensory input is interrupted early in life, as in congenital blindness, the visual cortex becomes more responsive to somatosensory and auditory input. Enhanced effective connections between primary somatosensory cortex and occipital areas are therefore likely to mediate the transfer of tactile information to the visual cortex. This could occur through strengthening or unmasking of existing cortico-cortical connections that are not recruited in sighted individuals learning tactile tasks (silent visual cortical areas) but only in early blind subjects.

In conclusion, the present results show that occipital brain areas are recruited for tactile discrimination in the blind and that the tongue can act as a portal to the visual cortex. Moreover, we demonstrated that cross-modal plasticity in the blind develops rapidly, after only 1 week of intensive training. Electrotactile stimulation of the tongue could serve as a valuable supplement to Braille reading, with the advantage of being hands-free and, more importantly, the capacity to convey sensory input from stimuli distal to the egocentric space. The latter constitutes a major improvement in the quality of life of the blind.

Acknowledgments

The authors are indebted to Professor Paul Bach-y-Rita (University of Wisconsin) for introducing us to the TDU and to Professor Eliana Sampaio (Laboratoire Brigitte Frybourg, CNAM, Paris) for lending us the equipment. This study was supported by the Danish Medical Research Council.

References

View Abstract