This finding suggests that the differential effect is due to activation during. Braille reading in the late blind subjects and not deactivations in the other conditions.
Brain (1998), 121, 409–419
Different activation patterns in the visual cortex of late and congenitally blind subjects Christian Bu¨chel, Cathy Price, R. S. J. Frackowiak and Karl Friston The Wellcome Department of Cognitive Neurology, Institute of Neurology, London, UK
Correspondence to: Christian Bu¨chel, The Wellcome Department of Cognitive Neurology, Institute of Neurology, 12 Queen Square, London WC1N 3BG, UK
Summary A key issue in developmental neuroscience is the role of activity-dependent mechanisms in the epigenetic induction of functional organization in visual cortex. Ocular blindness and ensuing visual deprivation is one of the rare models available for the investigation of experience-dependent cortical reorganization in man. In a PET study we demonstrate that congenitally blind subjects show taskspecific activation of extrastriate visual areas and parietal association areas during Braille reading, compared with auditory word processing. In contrast, blind subjects who lost their sight after puberty show additional activation in
the primary visual cortex with the same tasks. Studies in blind-raised monkeys show that crossmodal responses in extrastriate areas can be elicited by somatosensory stimulation. This is consistent with the crossmodal extrastriate activations elicited by tactile processing in our congenitally blind subjects. Since primary visual cortex does not show crossmodal responses in primate studies, the differential activation in late and congenitally blind subjects highlights the possibility of reciprocal activation by visual imagery in subjects with early visual experience.
Keywords: cortical plasticity; blindness; Braille; functional imaging Abbreviations: BA 5 Brodmann area; ERP 5 event related potentials; FDG 5 [18F]fluorodeoxyglucose; MEG 5 magnetoencephalography; rCBF 5 regional cerebral blood flow
Introduction Blindness due to ocular or anterior visual pathway pathology permits the study of cortical reorganization after sensory deprivation in man. The extent of reorganization depends upon the timing of onset of blindness (Hubel and Wiesel, 1977). Visual experience is important for early organization of the visual cortex (Price et al., 1994; Blakemore, 1991), suggesting that early reorganization depends upon the age of onset of deprivation. It has been shown previously that within the first 3 months after birth, there is a considerable capacity for functional reorganization but there are also other neurodevelopmental factors, besides age, that may affect the organization of visual cortex. For example, puberty has been found to be an important milestone in terms of eliminating supernumerary synapses in the visual cortex of monkeys (Rakic et al., 1986; Bourgeois and Rakic, 1993). In terms of organization, synaptic pruning can be seen as a refinement of connectivity within and between different cortical areas. The importance of puberty as a milestone for visual cortex development in humans is suggested by [18F]fluorodeoxy© Oxford University Press 1998
glucose (FDG) PET studies demonstrating elevated metabolism in the visual cortex of early blind but not in late (i.e. after puberty) blind subjects (Wanet-Defalque et al., 1988; Veraart et al., 1990). However, the augmented glucose metabolism in visual cortex of early blind subjects compared with sighted blindfolded and late blind subjects, as shown by these FDG studies was not task-specific (i.e. tactile and auditory) and was even present at rest (Wanet-Defalque et al., 1988; Veraart et al., 1990). Recent studies using event related potentials (ERPs) have found similar responses in the occipital cortex during tactile and auditory tasks, pointing to a non-specific activation of visual cortex (Ro¨sler et al., 1993). Activation of posterior brain regions in blind subjects by auditory processing has also been reported using magnetoencephalography (MEG) (Kujala et al., 1995) and ERP (Kujala et al., 1992). A recent H215O-PET activation study demonstrated activation of primary visual cortex in a heterogeneous group
410
C. Bu¨chel et al.
of blind subjects (Sadato et al., 1996). In contrast to evidence suggesting that early visual experience influences the functional development of the visual cortex, the authors reported no differences in the activation pattern in relation to the onset of blindness, when they compared two congenitally blind and six late blind subjects. We have studied sighted subjects, and congenitally and late blind subjects with H215O-PET, comparing Braille reading, visual reading and auditory word processing to address two issues. (i) Has visual cortex that has been deprived of visual input shifted its allegiance sufficiently to be activated by other sensory modalities and, related to this, does the age of onset of blindness influence such reorganization? (ii) Comparing two sensory modalities (i.e. auditory and tactile) directly with each other, also allowed us to demonstrate modality specific activations. This is an important issue, as ERP and FDG-PET studies have suggested non-specific activation (i.e. during Braille reading and auditory processing) of the occipital cortex in blind subjects (Ro¨sler et al., 1993). In this paper we also address structural changes in the occipital cortex of blind subjects, using voxel based morphometry (i.e. statistical tests of differences in regional grey matter using structural MRI).
Methods Subjects
Six male congenitally blind subjects (mean age 49.2 6 12 years) and three late blind subjects (two male and one female; mean age of onset of complete blindness 18.3 6 3.8 years; mean age 45 6 7.6 years) were studied. All were righthanded and proficient Braille readers. The causes of blindness included retinopathy of prematurity (two), optic nerve abnormalities (three) and congenital glaucoma (one). All late blind subjects experienced progressive visual impairment prior to the onset of complete blindness, however, all had been able to read print. The initial pathology in the late blind group comprised iridocyclitis, trauma and microphthalmia. Subsequently, complete blindness was due to glaucoma and/ or cataract. Knowing that their vision was likely to deteriorate, they started to learn Braille at the age of 7.3 6 4 years (5, 5 and 12 years of age). At the time of our study, clinical examination (pupillary light response) and conscious awareness (‘Do you have any light perception when you are facing the sun?’) showed both groups to be blind without residual light perception. This is important, since we had to reject 50% of our blind volunteers during the recruitment phase of the study due to residual light perception. Informed written consent was obtained prior to scanning. Although the blind subjects included in our study used both hands to read Braille it was found more suitable to read with one hand in the scanning environment. Eight blind subjects were able to read Braille, at the presentation rate used during scanning with the right hand alone; one congenitally blind subject used both hands. This subject was more accurate at the high
presentation rate used when tested before scanning, using his left hand as a guide. Our study also included a control group of six sighted, right-handed, male subjects (mean age 26.8 6 4 years). Permission to administer 5 mSv of 15O-labelled water was obtained from the Administration of Radioactive Substances Advisory Committee of the UK. Ethical approval was obtained from the joint National Hospital of Neurology and Neurosurgery/Institute of Neurology medical ethics committee.
Experimental design Each subject underwent 12 scans. In each scanning session we presented either real words or non-words in either of two modalities (Braille non-words: consonant letter strings, auditory non-words: reversed words). Reversed words comprised digitally recorded words played backwards. Sixdot non-contracted Braille strings (one Braille character per letter) were presented on an electronic Braille display. In the Braille and auditory conditions stimuli were presented every 2.5 s (presentation time 2300 ms). The task in all conditions was a feature-detection task, involving word processing, as used in Price et al. (1996). Word processing in conjunction with a feature-detection task allowed us to control the reading speed and accuracy of subjects behaviourally, and provided us with indices of performance. The instructions given to the subjects were ‘Read the word and if you find a target, press the button’. The presentation time was determined by psychophysical testing in a separate session with a sample of different blind subjects to ensure an accuracy of feature detection .80%. In the Braille conditions, the target was an elevated dot number 6 (the right lower dot) in one of the characters. In the auditory conditions a short high pitch tone after a word constituted the target. The subjects were instructed to read or listen to the word or non-word and determine whether a target was present. Responses were made by pressing a button with the thumb of the right hand. Words and nonwords were matched for the number of characters (5.6 6 1.3) and occurrence of targets (48%). The group of sighted control subjects was scanned under identical auditory conditions. Instead of Braille we presented single words or consonant letter strings on a computer monitor 50 cm distant. Targets were descenders as found in the letters ‘q’, ‘y’ and ‘p’. Presentation time was shorter (1000 ms) with the assumption that parallel processing of print reading is faster than the serial processing of Braille. This assumption was confirmed by the behavioural data. The stimulus frequency (one word every 2.5 s) was the same as for the blind subjects.
PET scanning Scans were performed on a CTI EXACT HR1 32 slice scanner (CTI Inc., Knoxville, Tenn., USA) with retracted collimating septa covering a field of view of 15.5 cm. Subjects received an intravenous bolus of H215O infused over
Activation patterns in blind subjects 20 s followed by a 20-s saline flush. There were 12 successive administrations of H215O each separated by 8 min. Images were reconstructed with a Hanning filter of 0.5 mm to a full width at half maximum of 6.5 mm. Data were acquired in a 90-s scan frame after injection of 8–10 mCi of H215O. Each task began 10 s before image acquisition. The order of presentation of tasks was counterbalanced.
Data analysis Images were realigned and co-registered with the subject’s structural MRI (Friston et al., 1995a). Volumes were smoothed with an isotropic Gaussian kernel (full width at half maximum, 16 mm). Statistical analysis was performed with SPM96 (Wellcome Deptartment of Cognitive Neurology, London, UK). Specific effects were tested by applying appropriate linear contrasts to the parameter estimates for each condition, resulting in a t-statistic for each and every voxel. These t-statistics (transformed to Z-statistics) constitute statistical parametric maps (SPM{Z}), that are then interpreted by referring to the probabilistic behaviour of Gaussian random fields (Friston et al., 1995b). The threshold adopted was P , 0.05 (corrected for multiple comparisons).
Results Reaction times No significant difference detection was found at the 1% level between late blind (1809 6 364ms) and congenitally blind (1605 6 281ms) subjects’ reaction times to target. Accuracy was .85%, and performance was not significantly different (at P , 0.01) in the late and congenitally blind groups. The reaction times for the sighted group reading print compared with those for both blind groups reading Braille confirmed that the serial process of Braille reading is slower than reading print. The mean reaction time for the sighted group was 552 ms (6116 ms) and 1679 ms (6326 ms) for the combined blind groups (unpaired t test, P , 0.01)
Morphometric comparison Comparison of late blind, congenitally blind and sighted subjects raises the issue of differences in structural anatomy of the visual cortex. To address this we performed voxelbased morphometric comparisons of the MRI scans between the three groups. The spatially normalized, smoothed (4 and 8 mm full width half maximum) grey matter segmented T1weighted MRI images were submitted to a voxel-based ANCOVA testing for significant differences in grey matter density using SPM96 (Wright et al., 1995). This analysis revealed no significant differences at this macroscopic level of structural anatomy, in the occipital cortex, between the three groups at P , 0.05 (corrected) for either degree of smoothing.
411
Functional anatomy To assess the effect of Braille reading relative to auditory processing we compared all the auditory tasks with all the Braille tasks, pooling across words and non-words. These comparisons were intended to reveal the main (simple) effects of Braille reading relative to auditory processing. As these effects were estimated in both late and congenitally blind subjects, we can also assess the interaction between group and modality, asking whether there are differential activations in Braille reading, relative to auditory word processing, in the late and congenitally blind groups. Finally we looked at the interaction between auditory processing and blindness per se, by comparing auditory specific activation in blind and sighted subjects.
Main effect of Braille versus auditory processing In the congenitally blind group Braille stimuli, unlike auditory stimuli, activated an extensive parieto-occipital system (Fig. 1A). Bilaterally activated areas included parietal association areas (area 7), superior visual association cortex (superior area 19) and the cerebellum. Additional activations in the left hemisphere were found in primary sensorimotor cortex, parietal area 40, inferior visual association cortex (inferior area 19) and the occipito-temporal junction (Table 1). Subjects who had lost their sight after puberty showed a similar pattern of activation (Fig. 1B), with two striking differences. First, in the late blind group there was an activation in primary visual cortex, which was absent in the congenitally blind subjects (Table 2). The second significant difference was found in the right medial occipital gyrus. Comparing reading print with auditory processing in sighted individuals showed only striate and extrastriate activations (Fig. 1C).
Interaction A direct comparison of the differences in Braille-specific activations between the late and congenitally blind groups showed a statistically significant group-by-condition interaction in the primary visual cortex and the right medial occipital cortex (Figs 1D and 2). A plot of regional cerebral blood flow (rCBF) in V1 shows higher values during Braille reading in the late blind group than in auditory processing for either of the blind groups and higher values than Braille reading in the congenitally blind group (Fig. 2B). This finding suggests that the differential effect is due to activation during Braille reading in the late blind subjects and not deactivations in the other conditions. There were no significant differences in activations between late and congenitally blind subjects when we looked for greater activations in auditory processing, relative to Braille reading.
412
C. Bu¨chel et al.
Fig. 1 Activations for congenitally blind subjects (A) and late blind subjects (B) during a feature decision task involving Braille reading, compared with an auditory feature decision task (i.e. controlling for implicit word processing) are shown. Column C shows activations for the group of sighted subjects reading print, compared with auditory processing. Activations are rendered on a template brain viewing the superior, posterior, left and medial surfaces. The activations rendered are thresholded at Z . 4.5. Activation of primary visual cortex is evident for the late blind subjects (B) whereas the congenitally blind subjects show no activation in area 17 (A). Column D shows the interaction between group and modality, looking for a significant difference comparing Braille with auditory processing in the late blind relative to the congenitally blind subjects. In accord with the visible differences between (A) and (B) the interaction shows significantly different activations in primary visual cortex and the right medial occipital gyrus. All images were normalized to the standardized space defined by Talairach and Tournoux (1988) using the structural MRI of each subject (Friston et al., 1995a). Note the similarity of activations for the late blind and sighted groups (B and C) although the former are reading by touch and the latter visually.
Auditory activations in blind versus sighted subjects The comparison of the effects due to auditory processing in blind relative to sighted subjects revealed greater activation in bilateral temporal regions, extending more posteriorly than in the sighted group (Fig. 3). The posterior part of the activated region in the right temporal lobe showed similar levels of activity for print reading and auditory processing for the sighted group, whereas both blind groups showed a marked activation for auditory processing (Fig. 3B).
Discussion Macroscopic anatomy of the visual cortex in blind subjects Most studies investigating either metabolic or electrical activation of the occipital cortex of blind subjects assume that the anatomy of the occipital cortex is the same as that in sighted subjects. Wanet-Defalque et al. (1988) looked for anatomical differences in the visual cortex of blind subjects using MRIs. This analysis was purely descriptive and relied on a subjective judgement.
Activation patterns in blind subjects
413
Table 1 Significant activations (P , 0.05 corrected) comparing Braille reading with auditory processing in congenitally blind subjects Left side x, y, z (mm)
Right side Z
Area
x, y, z (mm)
Z
Area
Occipital –18, –80, 32 –18, –94, 8
9.7 6.5
BA 19 (superior) BA 18/19
22, –76, 36 28, –88, 20
10.3 6.0
BA 19 (superior) BA 19
Parietal –46, –34, 36
9.6
44, –34, 36
8.7
–32, –52, 48
6.6
BA 40 (inferior parietal lobe) BA 7
14, –68, 52 30, –44, 44
8.6 7.0
BA 40 (inferior parietal lobe) BA 7 BA 7
44, –64, –24
5.3
Hemisphere
Temporal –46, –58, –20
BA 19/37 (posterior fusiform gyrus)
Fronto-parietal –46, –2, 32
6.0
BA 3/4 (sensorimotor)
Cerebellum –28, –52, –24 –14, –66, –28
9.1 6.0
Hemisphere Dentate nucleus
Coordinates are according to Talairach and Tournoux (1988).
Table 2 Significant activations (P , 0.05 corrected) comparing Braille reading with auditory processing in late blind subjects Left side x, y, z (mm) Occipital –10, –100, 0 –2, –90, –4 –32, –90, 0 –18, –82, 32 –40, –74, –12
Right side Z
Area
x, y, z (mm)
Z
Area
6.2 7.1 8.0 8.4
BA 17 BA 17 BA 18 BA 19 (superior occipital gyrus) BA 19 (fusiform gyrus)
6, –86, –8
7.0
BA 18/(17)
22, –72, 40
8.5
BA 19 (superior)
28, –84, 8
9.1
BA 19 (medial occipital)
44, –46, 40
7.8
BA 40 (inferior parietal lobe)
14, –60, –24 22, –64, –24
6.3 5.8
Dentate nucleus Hemisphere
8.0
Parietal –40, –44, 44
6.3
–48, –38, 36 –16, –72, 44
6.9 7.4
Temporal –38, –50, –24
BA 40 (inferior parietal lobe) BA 40 BA 7 BA 37 (fusiform gyrus)
Fronto-parietal –46, –2, 32
6.0
BA 3/4 (sensori-motor)
Cerebellum –2, –74, –24 –32, –56, –24
7.2 5.6
Vermis Hemisphere
Coordinates are according to Talairach and Tournoux (1988).
Using a statistical morphometric analysis, we were able to test for significant differences in anatomy. However, we were unable to reject the null hypothesis of no morphological difference in the visual cortex of blind and sighted subjects. This supports earlier reports of similar macroscopic
appearances of the occipital cortex, even in congenitally blind subjects (Wanet-Defalque et al., 1988). There is evidence from the non-human primate literature that early visual deprivation leads to changes in the structural anatomy of visual cortex at the microscopic level (Dehay et al., 1989;
414
C. Bu¨chel et al.
Fig. 2 The interaction between group and modality, looking for a significant difference comparing Braille with auditory processing in the late blind relative to the congenitally blind subjects is shown. This can be considered as the difference in activations engendered by Braille reading (relative to auditory processing) between the late and congenitally blind groups (see also Fig. 1D). The statistical parametric map for this comparison is superimposed on the sagittal and transverse section of a T1-weighted structural mean MRI from all subjects (A). The intersection of the lines indicates a voxel in the primary visual cortex (x 5 2 mm, y 5 –86 mm and z 5 –4 mm) with a Z-value of 4.7 (P , 0.05 corrected for multiple dependent comparisons). A plot of the activity at this voxel is shown in (B). Activations are arbitrarily adjusted to a whole brain mean of 50 ml/dl/min. It is evident that the significant difference between late and congenitally blind subjects is due to a high activity during Braille reading in the late blind group.
Bourgeois and Rakic, 1996). These studies showed a smaller surface of area 17 and a shift of the histologically identifiable boundary between area 17 and 18. In the context of functional imaging experiments it is important to test for any macroscopic differences in structural anatomy which might moderate the interpretation of functional data. It should be noted that our failure to reject the null hypothesis does not mean that differences do not exist (i.e. our analyses might not be sensitive enough to detect the changes found in the histological studies above). However, we can say that, from a neuroimaging perspective, they are substantially smaller than the functional differences we were able to detect.
Common activations in congenitally and late blind subjects during Braille reading The activation of parietal area 7 during a tactile task (Braille reading) is consistent with the crossmodal tactile responses detected in this parietal multi-modal association area in the monkey (Carlson et al., 1987). Activation of area 7 in humans has been demonstrated in motor control guided by sensory cues (Deiber et al., 1991; Grafton et al., 1992). Activation of the left sensorimotor cortex in an area of the hand
representation and adjacent Brodmann area (BA) 40 (rostral inferior parietal cortex) is in accord with the fact that blind subjects read Braille with their right hand. The activation of the inferior parietal lobule might reflect a correlate of motor preparation, which has also been shown by Deiber et al. (1991) and Stephan et al. (1995). Activation of the cerebellar hemispheres as seen in Braille reading was expected given the involvement of the cerebellum in tactile discrimination and motor control (Glickstein, 1992). Activation of the dentate nucleus during tactile discrimination has been shown using functional MRI by Gao et al. (1996). Comparing Braille reading with auditory processing raises the question whether the activation seen in this contrast could be due to deactivations during auditory processing. Figure 2B shows that the levels of rCBF during auditory processing are similar for congenitally and late blind subjects. The only difference is an increase of rCBF for Braille reading in the late blind group. On the other hand one might argue that activation of the network we ascribe to Braille reading could be similarly observed in sighted people doing a tactile task. However, Sadato et al. (1996) have shown that tactile processing, compared with rest, leads to deactivation of the occipital cortex of sighted subjects.
Activation patterns in blind subjects
415
Fig. 3 (A) A statistical parametric map (SPM{Z}) as a maximum intensity projection showing the activations due to auditory processing relative to Braille reading, comparing the congenitally blind and late blind groups (combined) with the group of sighted subjects. Views are from the right, top and behind. Voxels exceeding a threshold of P , 0.001 (uncorrected) are shown. A plot of the activity at a voxel in the posterior part of the right temporal lobe (x 5 52 mm, y 5 –60 mm, z 5 8 mm) is shown in (B). Activations are arbitrarily adjusted to a whole brain mean of 50 ml/dl/min. Note that the activation is higher for both blind groups in this region.
Differential activations in congenitally and late blind subjects during Braille reading A direct comparison of the differences in Braille-specific activations between late and congenitally blind groups showed a statistically significant group-by-condition interaction in the primary visual cortices and the right medial occipital cortex. In other words, late blind subjects showed a significant activation in primary visual cortex, whereas such an activation was absent in congenitally blind subjects. This finding suggests that the absence of early visual stimulation prevents responsiveness of the primary visual cortex to meaningful touch stimuli. This notion is supported by data from blindraised animals. Early visual deprivation in lid-sutured monkeys results in crossmodal reorganization of extrastriate and parietal areas (Hyvarinen et al., 1981a; Carlson et al., 1987). After reopening the sutured lids 20% of cells studied in area 19 responded exclusively to tactile stimuli. Reflecting the fact that cortical association areas are more likely to be subject to cortical reorganization (Rauschecker et al., 1992; Rauschecker, 1995), recordings from cells in the centre of area 17 showed no response to tactile stimuli, though a few neurons at the border between areas 17 and 18 responded weakly to tactile stimulation. These results predict crossmodal reorganization of extrastriate cortical areas in man, without primary visual cortical involvement. This hypothesis is consistent with the activation pattern due to tactile discrimination in our congenitally blind group. Although non-primary sensory areas may show more
plastic potential than primary areas in response to changed sensory input, there are also studies suggesting changes in primary sensory areas in man (Neville, 1990) and nonhuman primates (Rauschecker, 1995). However, in most of these studies, including one study of reorganization in amputees (Kew et al., 1994), plastic changes in primary sensory areas were only found within the native modality of the cortex in question. In this paper we have tested blindness as a perturbation model to address neurodevelopmental plasticity. The ensuing cross-modal plasticity is likely to reflect different mechanisms from those associated with altered sensory input in the post-developmental period. This is because experience and activity-dependent mechanisms may have profoundly different roles when interacting with epigenetic factors during development. The remaining questions are: Why is metabolic activity of visual cortex increased in early blind subjects as revealed by FDG-PET (Wanet-Defalque et al., 1988; Veraart et al., 1990) and why, on the other hand, does primary visual cortex not show task, activation in these subjects? Studies of early binocular deprivation in cats show a decreased density of cells sending projections from area 17 to area 18 (Price et al., 1994) but a high synaptic density in visual cortex (Huttenlocher et al., 1982) due to impaired elimination of supernumerary synapses (Bourgeois and Rakic, 1993). Although the emergence of crude clusters (demonstrated with intrinsic retrograde labelling) in area 17 is not affected by binocular lid suture, an abnormal
416
C. Bu¨chel et al.
distribution of intrinsic axon collaterals, reflecting the lack of cluster refinement has been shown (Callaway and Katz, 1991). A recent study by Bourgeois and Rakic (1996) showed that, although certain basic parameters of synaptic development in the supra- and infragranular layers of the striate cortex develop normally in the absence of both retinae, the maturation of certain features of connectivity in the thalamorecipient sublayers is dependent on retinal input. In accord with these findings, visual deprivation from birth results in an alteration of receptive field properties in the neurons of area 17 (Wiesel and Hubel, 1965; Singer and Tretter, 1976). Berman (1991) studied the effect of early enucleation on visual cortical connections in cats and found increased projections of thalamic nuclei (medial septal nucleus and intralaminar nuclei) that provide non-specific input to areas 17 and 18. In summary, these findings may explain the high glucose metabolism (an index of presynaptic activity) seen in visual cortex (Wanet-Defalque et al., 1988; Veraart et al., 1990) on the one hand and they further point to a lack of functional responsiveness as demonstrated by our findings.
Activation of primary visual cortex in late blind subjects The differential activation of primary visual cortex in different subjects has to be explained on the basis of differences between those subjects. The major difference between late and congenitally blind subjects is visual experience early in life. This visual experience has also influenced their mental representations of the environment. Although now blind, they still have the potential faculty to relate new sensory information to visually experienced items, a phenomenon that is directly linked to mental imagery. The possibility that increased metabolism of the primary visual cortex might mask activations elicited by Braille reading seems unlikely, given that the FDG-PET studies reported increased metabolism not only for primary visual cortex but also prestriate visual areas (Wanet-Defalque et al., 1988; Veraart et al., 1990). All our late blind subjects reported that they immediately transform tactile and auditory cues into a visual representation. All late blind subjects learnt to read print prior to learning Braille. This leads to an interesting phenomenon in tasks involving word processing. One subject reported that when doing crosswords she ‘sees’ embossed Braille dots on top of printed characters in her mind’s eye. She also reported that she uses the information of the printed letters to solve the crossword and not the Braille characters in which the crossword is presented to her. This subjective report suggests that the late blind subjects we studied can visualize the Braille dot pattern and the related print characters while they read. This strategy is denied the congenitally blind subjects and may contribute to difficulties in spatial orientation in some of these subjects. The subjective report of imagery of
the Braille dot pattern, rather than the concept depicted by the word is in accord with the fact that we observed no significant difference in V1 comparing either Braille or auditory words with non-words in the late blind group. Therefore it cannot be ruled out that primary visual cortex activation in late blind subjects is due to mental imagery, which has been shown to activate early components of the visual system (Kosslyn et al., 1995) down to the level of the lateral geniculate nuclei (Chen et al., 1996). This explanation raises the question as to why mental imagery activates primary visual cortex in blind subjects but fails to do so in vivid visual hallucinations (Silbersweig et al., 1995) and other mental imagery studies (Roland and Gulyas, 1994)? A possible answer is that the retrograde activation of primary visual cortex is facilitated by the absence of competing thalamic inputs. Our findings are in accord with those of Sadato et al. (1996) in the prestriate cortex, but differ substantially where striate cortex is concerned. Sadato et al. (1996) examined a mixed group of congenitally and late blind subjects (mean age of onset of blindness 4.3 6 5.5 years). The authors emphasized activation in primary visual cortex but rejected the idea that it was visual imagery that may have mediated this activation; they interpreted their results in terms of crossmodal reorganization of primary visual cortex. Although these authors reported no differences between two congenitally blind subjects and the rest of this group, a failure to demonstrate an interaction cannot be used to discount such a differential activation. In addition to the visual imagery associated with tactile discrimination in acquired blindness, contributions to primary visual cortical activation might relate to details of experimental design (the authors compared Braille reading and tactile discrimination with rest). Comparing Braille reading with rest does not properly control for the many explicit and implicit components of the tasks employed (e.g incidental phonological processing, implicit activation of structural and semantic memories and visual imagery). A further explanation for V1 activation in early blind subjects in the Sadato et al. (1996) study may be an interaction of these cognitive components with abnormal thalamic afferent activity (Berman, 1991). The analysis of congenitally and late blind subjects described in this paper, allowed us to address the issue of differential activations explicitly. Although, like Sadato et al. (1996) we employed word and non-word stimuli, the tasks we used were different. In the study of Sadato et al. (1996), subjects had to detect words in a stream of non-words and vice versa, whereas in our study the items presented in a block were always either words or non-words and subjects had to monitor for a target in every word. However, it is very unlikely that the discrepancy of primary visual cortex activations can be attributed to the different tasks, as late and congenitally blind subjects in our study were involved in the same task. Further, the tasks were similar in so far as blind subjects read (mostly)
Activation patterns in blind subjects words during some conditions and (mostly) non-words during other conditions in both experiments.
Crossmodal reorganization The cortical reorganization leading to specific activation of extrastriate visual areas by tactile stimuli in blind subjects is probably due to cortico-cortical reorganization (Rauschecker et al., 1992). Given our results, we would propose that tactile information is relayed from primary sensorimotor cortex to area 40, from there to area 7 and then to superior area 19 and to the inferior occipito-temporal junction. As shown in primate studies, the number of neurons responding to tactile stimuli in area 7 is increased in blind-raised monkeys (Carlson et al., 1987). This area is strongly connected to extrastriate areas (BA 19) and could reciprocally chain tactile information to extrastriate areas. Reorganization at the thalamic level, as proposed by Hyvarinen et al. (1981b) would be a more parsimonious explanation but, apart from non-specific input from the medial thalamic nuclei (Berman, 1991), a real contribution of thalamo-cortical projections to crossmodal reorganization has not been demonstrated.
Lesion studies There are several reports of lesions in blind Braille readers, aiding the interpretation of our findings. Gloning et al. (1954) studied a patient with a right parietal tumour who developed a Braille alexia, with only slight word-finding difficulties. The location of the lesion coincides with the location of an activation due to Braille reading in the right parietal lobe in the blind subjects in our study. There are two further reports about lesions in blind subjects: Birchmeier (1985) studied a congenitally blind subject who suffered from an ischaemic infarct of the left hemisphere. Unfortunately the extent and location of the lesion were not reported. Given the symptoms of a Wernicke’s aphasia, right hemiparesis and a positive Babinski sign on the right, it is reasonable to assume an ischaemia in the territory of the left middle cerebral artery. Although his hemiparesis resolved completely over a period of 2 months, a profound Braille dyslexia remained. Finally, the case of a blind organist and composer was reported. He suffered from an ischaemic infarct in the territory of the left middle cerebral artery (Signoret et al., 1987) and, besides other neuropsychological deficits, he showed a pronounced Braille alexia and agraphia. In general, the association of Braille alexia with lesions in the right parietal and left temporal lobe suggests a true involvement of these structures in Braille reading as opposed to the notion of non-specific coactivation.
417
higher activations in bilateral temporal regions in the blind. The activation also extended towards the occipito-temporal junction (BA 37; medial temporal gyrus). A posterior extension of activations due to auditory processing has also been reported using ERP (Kujala et al., 1992) and MEG (Kujala et al., 1995). This interaction effect (large difference between Braille reading and auditory processing in blind subjects and no difference between visual reading and auditory processing in sighted subjects in BA 37) could also be due to a similar activation of BA37 by visual and auditory processing in sighted subjects, leading to no significant difference. Therefore, the question of whether increased activation due to auditory processing is an expression of the importance of auditory cues for locomotion of blind subjects (Strelow and Brabyn, 1982) needs further study.
Design The ideal control group for the blind subjects would have been sighted subjects able to read Braille tactually. However, nearly all sighted people, able to read Braille, read it visually. Reading print in sighted subjects as a control condition is a compromise but was only used in the comparison between both blind groups and the sighted group for auditory processing. The main finding of the dependence of primary visual cortex activation on the age of onset of blindness is independent of this control group. The late blind group only consisted of three subjects. This is due to two factors. First, we excluded all subjects with minimal light perception, as this is clearly a confounding factor. Secondly, we were interested in subjects who lost their sight after a defined point in time, pertinent to the development of visual cortex (i.e. puberty). As they started to read Braille late in life, the reading speed of most of these subjects was too slow to match that of the congenitally blind group. Thus, although we recruited subjects from the whole UK over a period of 6 months through the well-established Braille and tape journals of the Royal National Institute for the Blind, we were not able to find more than three late blind subjects who matched the congenitally blind group.
Conclusions Taking earlier data together with our findings, we conclude that differential activation of extrastriate visual areas during Braille reading, relative to auditory processing, supports taskspecific crossmodal responses in extrastriate cortex. However, in late blind subjects with visual experience, where projections between striate and extrastriate cortex have developed normally, it is possible that reciprocal interactions mediate further activation in primary visual cortex. Mental imagery must be considered as a possible mechanism.
Auditory processing Comparing activations due to auditory processing (auditory processing relative to Braille or print reading) between both blind groups and the sighted group showed significantly
Acknowledgements We wish to thank the departmental radiographers A. Brennan, A. Carrol, J. Galliers and G. Lewington for their contribution
418
C. Bu¨chel et al.
to data collection and S. Zeki for an internal review of this manuscript. We are indebted to our subjects, especially the blind subjects who travelled long distances to make this study possible. C.B., C.P., R.S.J.F. and K.F. are supported by the Wellcome Trust. We would also like to thank the Royal National Institute for the Blind in the UK.
References Berman NE. Alterations of visual cortical connections in cats following early removal of retinal input. Dev Brain Res 1991; 63: 163–80. Birchmeier AK. Aphasic dyslexia of Braille in a congenitally blind man. Neuropsychologia 1985; 23: 177–93. Blakemore C. Sensitive and vulnerable periods in the development of the visual system. [Review]. Ciba Found Symp 1991; 156: 129–47. Bourgeois JP, Rakic P. Changes of synaptic density in the primary visual cortex of the macaque monkey from fetal to adult stage. J Neurosci 1993; 13: 2801–20. Bourgeois JP, Rakic P. Synaptogenesis in the occipital cortex of macaque monkey devoid of retinal input from early embryonic stages. Eur J Neurosci 1996; 8: 942–50. Callaway, EM, Katz LC. Effects of binocular deprivation on the development of clustered horizontal connections in cat striate cortex. Proc Natl Acad Sci USA 1991; 88: 745–9. Carlson S, Pertovaara A, Tanila H. Late effects of early binocular visual deprivation on the function of Brodmann’s area 7 of monkeys (Macaca arctoides). Brain Res 1987; 430: 101–11. Chen W, Kato T, Zhu X-H, Ogawa S, Ugurbil K. Primary visual cortex activation during visual imagery in human brain: a fMRI mapping study. Neuroimage 1996; 3: S204. Dehay C, Horsburgh G, Berland M, Killackey H, Kennedy H. Maturation and connectivity of the visual cortex in monkey is altered by prenatal removal of retinal input [see comments]. Nature 1989; 337: 265–7. Comment in: Nature 1989; 340: 180. Deiber M-P, Passingham RE, Colebatch JG, Friston KJ, Nixon PD, Frackowiak RS. Cortical areas and the selection of movement: a study with positron emission tomography. Exp Brain Res 1991; 84: 393–402. Friston KJ, Ashburner J, Frith CD, Poline J-B, Heather JD, Frackowiak RS. Spatial registration and normalization of images. Hum Brain Mapp 1995a; 3: 165–89. Friston KJ, Holmes AP, Worsley KJ, Poline J-B, Frith CD, Frackowiak RS. Statistical parametric maps in functional imaging: a general linear approach. Hum Brain Mapp 1995b; 2: 189–210. Gao JH, Parsons LM, Bower JM, Xiong J, Li J, Fox PT. Cerebellum implicated in sensory acquisition and discrimination rather than motor control [see comments]. Science 1996; 272: 545–7. Comment in: Science 1996; 272: 482–3. Glickstein M. The cerebellum and motor learning. [Review]. Curr Opin Neurobiol 1992; 2: 802–6. Gloning I, Gloning K, Weingarten K, Berner P. Ueber einen Fall mit Alexie der Brailleschrift. Wien Z NervHeilk 1954; 10: 260–73.
Grafton ST, Mazziotta JC, Presty S, Friston KJ, Frackowiak RS, Phelps ME. Functional anatomy of human procedural learning determined with regional cerebral blood flow and PET. J Neurosci 1992; 12: 2542–8. Hubel DH, Wiesel TN. Ferrier lecture. Functional architecture of macaque monkey visual cortex. [Review]. Proc R Soc Lond B Biol Sci 1977; 198: 1–59. Huttenlocher PR, de Courten C, Garey LJ, Van der Loos H. Synaptogenesis in human visual cortex–evidence for synapse elimination during normal development. Neurosci Lett 1982; 33: 247–52. Hyvarinen J, Carlson S, Hyvarinen L. Early visual deprivation alters modality of neuronal responses in area 19 of monkey cortex. Neurosci Lett 1981a ; 26: 239–43. Hyvarinen J, Hyvarinen L, Linnankoski I. Modification of parietal association cortex and functional blindness after binocular deprivation in young monkeys. Exp Brain Res 1981b; 42: 1–8. Kew JJ, Ridding MC, Rothwell JC, Passingham RE, Leigh PN, Sooriakumaran S, et al. Reorganization of cortical blood flow and transcranial magnetic stimulation maps in human subjects after upper limb amputation. J Neurophysiol 1994; 72: 2517–24. Kosslyn SM, Thompson WL, Kim IJ, Alpert NM. Topographical representations of mental images in primary visual cortex. Nature 1995; 378: 496–8. Kujala T, Alho K, Paavilainen P, Summala H, Naatanen R. Neural plasticity in processing of sound location by the early blind: an event-related potential study. Electroencephalogr Clin Neurophysiol 1992; 84: 469–72. Kujala T, Huotilainen M, Sinkkonen J, Ahonen AI, Alho K, Hamalainen MS, et al. Visual cortex activation in blind humans during sound discrimination. Neurosci Lett 1995; 183: 143–6. Neville HJ. Intermodal competition and compensation in development. Evidence from studies of the visual system in congenitally deaf adults. Ann N Y Acad Sci 1990; 608: 71–87. Price DJ, Ferrer JM, Blakemore C, Kato N. Postnatal development and plasticity of corticocortical projections from area 17 to area 18 in the cat’s visual cortex. J Neurosci 1994; 14: 2747–62. Price CJ, Wise RJ, Frackowiak RS. Demonstrating the implicit processing of visually presented words and pseudowords. Cereb Cortex 1996; 6: 62–70. Rakic P, Bourgeois JP, Eckenhoff MF, Zecevic N, Goldman-Rakic PS. Concurrent overproduction of synapses in diverse regions of the primate cerebral cortex. Science 1986; 232: 232–5. Rauschecker JP. Compensatory plasticity and sensory substitution in the cerebral cortex. [Review]. Trends Neurosci 1995; 18: 36–43. Rauschecker JP, Tian B, Korte M, Egert U. Crossmodal changes in the somatosensory vibrissa/barrel system of visually deprived animals. Proc Natl Acad Sci USA 1992; 89: 5063–7. Roland PE, Gulyas B. Visual imagery and visual representation [see comments]. [Review]. Trends Neurosci 1994;17: 281–7. Comment in: Trends Neurosci 1994;17: 287–9, Comment in: Trends Neurosci 1994;17: 290–4, Comment in: Trends Neurosci 1994;17: 514–6. Ro¨sler F, Ro¨der B, Heil M, Hennighausen E. Topographic differences
Activation patterns in blind subjects of slow event-related brain potentials in blind and sighted adult human subjects during haptic mental rotation. Brain Res Cogn Brain Res 1993; 1: 145–59. Sadato N, Pascual-Leone A, Grafman J, Ibanez V, Deiber M-P, Dold G, Hallett M. Activation of the primary visual cortex by Braille reading in blind subjects [see comments]. Nature 1996; 380: 526–8. Comment in: Nature 1996; 380: 479–80. Signoret JL, van Eeckhout P, Poncet M, Castaigne P. Aphasia without amusia in a blind organist. Verbal alexia-agraphia without musical alexia-agraphia in braille [French]. Rev Neurol (Paris) 1987; 143: 172–81. Silbersweig DA, Stern E, Frith C, Cahill C, Holmes A, Grootoonk S, et al. A functional neuroanatomy of hallucinations in schizophrenia. Nature 1995; 378: 176–9. Singer W, Tretter F. Receptive-field properties and neuronal connectivity in striate and parastriate cortex of contour-deprived cats. J Neurophysiol 1976; 39: 613–30. Stephan KM, Fink GR, Passingham RE, Silbersweig D, CeballosBaumann AO, Frith CD, et al. Functional anatomy of the mental representation of upper extremity movements in healthy subjects. J Neurophysiol 1995; 73: 373–86.
419
Strelow ER, Brabyn JA. Locomotion of the blind controlled by natural sound cues. Perception 1982; 11: 635–40. Talairach J, Tournoux P. Coplanar stereotaxic atlas of the human brain. Stuttgart: Thieme; 1988. Veraart C, De Volder AG, Wanet-Defalque MC, Bol A, Michel C, Goffinet AM. Glucose utilization in human visual cortex is abnormally elevated in blindness of early onset but decreased in blindness of late onset. Brain Res 1990; 510: 115–21. Wanet-Defalque MC, Veraart C, De Volder A, Metz R, Michel C, Dooms G, et al. High metabolic activity in the visual cortex of early blind human subjects. Brain Res 1988; 446: 369–73. Wiesel TN, Hubel DH. Comparison of the effects of unilateral and bilateral eye closure on cortical unit responses in kittens. J Neurophysiol 1965; 28: 1029–40. Wright IC, McGuire PK, Poline J-B, Travere JM, Murray RM, Frith CD, et al. A voxel-based method for the statistical analysis of gray and white matter density applied to schizophrenia. Neuroimage 1995; 2: 244–52.
Received September 1, 1997. Accepted November 19, 1997
