Introduction
Blind individuals tend to show better performance than sighted individuals on a variety of auditory and tactile tasks. This is thought to be due both to compensatory hypertrophy – enhanced neuronal processing within auditory and somatosensory cortices, and to cross-modal plasticity – whereby regions of occipital cortex that normally only process visual information begin to process auditory and tactile stimulation. While there is an extensive animal literature examining the long-term effects of early visual deprivation, we still do not have a full understanding of the neurophysiological changes that underlie cross-modal plasticity within the human brain. Here we discuss what is currently known about the behavioral and neurophysiological effects of long-term visual deprivation, and discuss the potential of new imaging techniques to reveal new insights.
Over the last two decades it has been well documented that those who suffer from visual deprivation, especially early in life, show an enhanced ability to use tactile and auditory cues that seems to be accompanied by remarkable changes in the response properties of occipital cortex.
While there is a growing literature demonstrating behavioral differences and differential neural responses to auditory and tactile stimuli in occipital cortex, the neurophysiological changes that occur as a result of blindness in humans are still fairly unclear. There is of course an extensive literature examining the anatomical and neural consequences of visual deprivation using both primate and non-primate animal models. However the literature on the effects of blindness on humans and in animal models has perhaps not been as mutually informative as might be hoped. One reason for this is that it is possible to examine the effects of visual deprivation with high temporal resolution within individual or small groups of cells within animal models, but the tools for assessing the effects of visual deprivation on humans are limited to techniques with much lower spatial or temporal resolution ( Fig. 41.1 ). This limitation in the tools available to study the effects on long-term visual deprivation in humans is of particular concern because it is not clear how easily the animal literature can be generalized to humans. Humans rely more heavily on vision than almost all other animals and have a far larger number of visual sensory areas. Fortunately, as illustrated in Box 41.1 , techniques such as functional magnetic resonance imaging, high-resolution structural imaging, transcranial magnetic stimulation, magnetic resonance spectroscopy, and functional connectivity analyses are now beginning to offer new ways of non-invasively studying the neurophysiological changes that occur as a result of blindness in humans. Table 41.1 summarizes research discussed in this article examining performance on and measuring brain responses to a wide range of auditory and tactile tasks.
Diffusion tensor imaging (DTI). A magnetic resonance (MR) method in which image intensities are proportional to the strength of diffusion along the direction of the magnetic diffusion gradient. Currently, DTI is principally used for the imaging of white matter, since axons in white matter tend to have strong diffusion gradients. More recently tractography (also known as fiber tracking) algorithms have been applied to DTI data in order to estimate the pathways of white matter in the brain.
Electroencephalography (EEG) is the measurement of summed activity of post-synaptic currents produced by the brain as recorded from electrodes placed on the scalp. A surface EEG reading is the summation of the synchronous activity of thousands of neurons that have similar spatial orientation, radial to the scalp. Currents tangential to the scalp are not picked up by EEG, and activity from deep sources is more difficult to detect than currents near the skull. EEG has very high temporal resolution (~1 ms), but poor spatial resolution. Indeed, it is theoretically impossible (the “inverse problem”) to calculate the 3D spatial sources of EEG signals from the 2D electrode array. With assumptions, some estimates of spatial sources can be made but these estimates have spatial resolution on the order of a few centimeters. Recently EEG has been used in conjunction with MEG or fMRI in an attempt to achieve better spatial resolution.
Functional magnetic resonance imaging (fMRI). Hemoglobin is diamagnetic when oxygenated but paramagnetic when deoxygenated. The magnetic resonance signal of blood is therefore slightly different depending on the level of oxygenation. Hemodynamic or blood oxygen level dependent (BOLD) responses are proportional to blood flow and the relative concentrations of deoxygenated to oxygenated blood in the brain area of interest. It should be noted that the precise relationship between neural signals and BOLD remains a matter of active research. With sophisticated research paradigms fMRI can provide spatial resolution on the order of 1 millimeter and temporal resolution of less than a second, but generally has a spatial resolution of a few millimeters and temporal resolution of a few seconds. Since the early 1990s, fMRI has come to dominate the brain mapping field due to its low invasiveness, lack of radiation exposure, and relatively wide availability.
Magnetoencephalography (MEG) is the measurement of summed magnetic fields produced by synchronous electrical activity in the brain as recorded from superconducting quantum interference devices (SQUIDS) placed on the scalp. It is only those magnetic fields generated by bundles of neurons that are perpendicular (as compared to radial in the case of EEG) to the cortical surface that can be measured. Like EEG, MEG has very high temporal resolution, but also suffers from the “inverse” problem. However magnetic fields are less distorted by factors like the skull, allowing somewhat better spatial resolution than EEG.
Magnetic resonance spectroscopy (MRS). An MR method in which levels of metabolites (such as myo-inositol, choline, creatine, N-acetylaspartate, choline, lactate, glutamate, glutamine, and GABA) associated with development and neuronal functioning are measured. Currently, because of the small concentrations of these metabolites, MRS is limited to a spatial resolution on the order of several millimeters and a temporal resolution on the order of minutes.
Positron emission tomography (PET). A short-lived biologically active radioactive tracer isotope is injected into the living subject and local neural activity is inferred based on resulting tissue concentrations within the area of interest. Provides spatial resolution on the order of millimeters, temporal resolution on the order of 10 seconds. Limitations to the use of PET include the high cost and the need to limit radiation exposure.
Transcranial magnetic stimulation (TMS). Weak electric currents are induced in brain tissue by electromagnetic induction due to rapidly changing magnetic fields. These electric currents can be used to trigger neuronal activity over a relatively well-localized region of the cortical surface. TMS is often used to demonstrate causality. If performance for a given task is suppressed by TMS stimulation to a particular brain area, this provides evidence that the region is used in performing that task.
| Study type | Early blind | Late blind | Sighted | ||
|---|---|---|---|---|---|
| Tactile activation | Braille words (Sadato et al. 1996, 1998) | PET | P & E | – | – |
| Braille words (Cohen et al. 1999) | PET | P & E | E | – | |
| Braille words (Burton et al. 2002) | fMRI | P & E | P & E | – | |
| Braille words (Gizewski et al. 2003) | fMRI | P & E | – | N | |
| Angle, width, character discrimination (Sadato et al. 1996, 1998) | PET | P & E | – | N | |
| Sweep (no discrimination) (Sadato et al. 1996, 1998) | PET | N | – | N | |
| Braille letters (Sadato et al. 2002) | fMRI | P & E | E | N | |
| Braille letters (naïve to Braille) (Sadato et al. 2004) | fMRI | – | E | N | |
| Embossed letters (Burton et al. 2006) | fMRI | P & E | P & E | E | |
| Tactile motion (Hagen et al. 2002; Blake et al. 2004) | fMRI/PET | – | – | E | |
| Tactile object recognition (Amedi et al. 2001, 2002; James et al. 2002; Pietrini et al. 2004) | fMRI | – | – | E | |
| Tactile orientation (Sathian et al. 2002) | PET | – | – | E | |
| Tactile spatial frequency (Sathian et al. 2002) | PET | – | – | N | |
| Auditory activation | Frequency discrimination (Alho et al. 1993; Kujala et al. 1995) | ERP/MEG | Y | – | N |
| Frequency discrimination (Kujala et al. 1997) | ERP | Y | Y | N | |
| Frequency discrimination (Kujala et al. 2005) | fMRI | P & E | – | N | |
| Sound localization (Kujala et al. 1992) | ERP | Y | – | N | |
| Sound localization (Leclerc et al. 2000; Gougoux et al. 2005) | PET/ERP | P & E | – | N | |
| Sound localization (De Volder et al. 1999) | PET | P & E | – | N | |
| Sound localization (Weeks et al. 2000) | PET | E – RH only | – | N | |
| Sound localization (Voss et al. 2006) | PET | – | E – RH dom | N | |
| Auditory motion (Poirier et al. 2006) | fMRI | MT+ | – | Y | |
| Auditory motion (Saenz et al. 2008) | fMRI | MT+ | – | N | |
| Auditory language (Burton et al. 2003) | fMRI-PER | P & E | – | N | |
| Auditory language (Amedi et al. 2003) | fMRI | P & E | P & E | N | |
| Auditory language (Burton et al. 2006) | fMRI | – | P & E | N | |
| Tactile functional role | Braille words (Cohen et al. 1999) | TMS | Y | N | – |
| Braille words (Hamilton et al. 2000) | Stroke | Y | – | – | |
| Braille words (Kupers et al. 2007) | TMS | Y | – | – | |
| Braille experience (Liu et al. 2007) | fMRI-PER | Y | – | – | |
| Braille letters (Cohen et al. 1997) | TMS | Y | – | N | |
| Embossed letters (Cohen et al. 1997) | TMS | Y | – | N | |
| Spatial distance (Merabet et al. 2004) | TMS/Stroke | Y | – | Y | |
| Roughness (Merabet et al. 2004) | TMS/Stroke | N | – | N | |
| Tactile sensations (Ptito et al., 2008) | TMS | Y | – | N | |
| Tactile orientation (Zangaladze et al. 1999; Sathian et al. 2002) | TMS | – | – | Y | |
| Auditory functional role | Frequency discrimination (Stevens et al. 2007) | fMRI-PER | Y | – | N |
| Auditory language (Amedi et al. 2003) | fMRI-PER | P | – | N | |
| Sound localization (Leclerc et al. 2000; Gougoux et al. 2005) | PET/ERP | P & E | – | – | |
| Sound localization (Collignon et al. 2007) | TMS | Y | – | N | |
| Auditory language (Amedi et al. 2004) | TMS | Y – RH only | – | N | |
| Pitch (Collignon et al. 2007) | TMS | N | – | N | |
| Intensity (Collignon et al. 2007) | TMS | N | – | N |
The neuronal effects of visual deprivation
Cross-modal processing in visually normal development
During development, while much of sensory cortical architecture and connectivity is determined by genetically controlled factors, further refinement of connections, both within and between brain areas, appears to be dependent on early sensory input during a “critical” period. These experience-dependent effects are believed to be mediated by adjustments in the weighting/gain of pre-existing connectivity, the guidance of new projections to target areas, and the elimination of exuberant projections. A common underlying theme behind all these processes is that competition, based on experience, mediates connectivity; more active projections are strengthened at the expense of those that are less active.
In infant animals, there is substantial physiological evidence for exuberant connectivity within the layers of visual cortex itself (primates, kittens , ); as well as between visual cortex and other cortical areas (primates, cats ) ( Fig 41.2 ). In kittens, there have been reports of projections from auditory/temporal areas to visual cortex, from somatosensory and fronto-parietal cortex to visual cortex. as well as from the subcortical visual thalamus (lateral geniculate nucleus, LGN) to auditory cortex. , These findings suggest that during early development the anatomical connections required for multisensory input to occipital cortex exist, even between primary sensory cortices. A variety of animal models of normal visual development have demonstrated “pruning” (or re-allocation) of a large proportion of these cross-modal connections during the course of maturation. ,
A similar “pruning” is thought to occur in normal human infancy. As a result, it has been suggested that cross-modal connectivity may be more pronounced in infancy than in adulthood. A common technique for measuring brain responses in infants is by measuring event-related potentials (ERP) – the electrical activity of the brain through the skull and scalp. While auditory event-related potential (ERP) responses are usually small or absent in occipital (visual) regions of the human adult brain, large auditory ERP responses are found in occipital regions of 6-month-old infants. These responses decrease as a function of age, implying a reduction in the strength of projections of auditory information to visual cortex.
As shown in Figure 41.2 , and as will be described further below, many animal models suggest that, in the absence of visual input, the normal pruning process fails to occur. According to this model, cross-modal responses in early blind subjects might be mediated through the retention of an “infantile” pattern of anatomical connectivity. However, in contradiction to these results, recent examination of the volume of white and gray matter of early blind human subjects using voxel-based morphometry suggests a reduction in connectivity between cortical areas.
Cross-modal processing in visually normal adults
Over the last three decades, improvements in neuroimaging techniques have made it increasingly easy to study neural processing within humans. Studies before the early 1990s tended to rely on positron emission tomography (PET), where a short-lived biologically active radioactive tracer isotope (such as FDG, an analog of glucose) is injected into the living subject and local neural activity is inferred based on resulting tissue concentrations within the area of interest. Since the early 1990s, functional magnetic resonance imaging (fMRI) has become the technique of choice due to its low invasiveness, lack of radiation exposure, and relatively wide availability ( Fig. 41.3 ). fMRI measures the hemodynamic (blood oxygen level-dependent, BOLD) response, which is proportional to blood flow and the relative concentrations of deoxygenated to oxygenated blood in the brain area of interest.
However, it should be noted that the precise relationship between neuronal signals and the BOLD response is still not clear. The relationship between blood flow, metabolic rate, and neuronal responses remains a matter of active research, and it is even possible that these relationships are altered within the occipital cortex of blind subjects. BOLD responses should not therefore be considered synonymous with neuronal activity.
While primary visual cortex has not been shown to exhibit cross-modal responses to purely tactile or auditory stimuli in visually normal adults, BOLD responses in primary visual cortex can be modulated by information from other senses. Furthermore, many extrastriate visual areas do show cross-modal responses in sighted individuals. Area MT+ shows BOLD responses to tactile motion stimuli, , and the lateral occipital complex (LOC)/inferior temporal (IT) gyrus shows BOLD responses during tactile object recognition tasks. Extrastriate visual areas also show BOLD activation during tactile embossed letter and tactile orientation discrimination tasks.
Recently, transcranial magnetic stimulation (TMS) has been used to examine whether responses to tactile stimuli within occipital cortex play a functional role in sighted subjects. Weak electric currents are induced in brain tissue by rapidly changing magnetic fields (electromagnetic induction), thereby triggering neuronal activity over a relatively well-localized region of the cortical surface ( Fig. 41.4 ). TMS applied over regions of the occipital cortex interferes with the ability of sighted subjects to perform tactile orientation and tactile spatial distance tasks, suggesting functional involvement of these areas in tactile processing. However, it should be noted that other studies have failed to find visual cortical activation for tactile tasks in sighted subjects using fMRI or show interference with tactile tasks using occipital TMS. , It therefore remains unclear how important a role visual cortex plays in the performance of tactile tasks in sighted subjects.
The anatomical connections that underlie these responses in normally sighted individuals are still being determined. Until fairly recently cross-modal modulation within primary sensory areas was assumed to be mediated by feedback from multimodal association areas outside visual cortex, such as intraparietal and superior temporal sulcus cortex, insular cortex, and perhaps even frontal and prefrontal cortex. However, recent anatomical studies have found direct cross-modal projections between sensory cortices in adult primates. Direct projections have been found from auditory to visual cortex, from visual to auditory cortex, , and from visual to somatosensory cortex.
Cross-modal processing in early blind individuals
Measurements of cross-modal responses within early and late blind subjects have tended to either focus on “basic stimuli”, such as simple tactile discrimination or auditory frequency tasks, or have focused on tasks which have strong functional significance, such as Braille reading, auditory localization, or auditory language. While cross-modal plasticity seems to be stronger for functionally relevant tasks, it should be noted that these tasks/stimuli also tend to be more complex, and are therefore more likely to evoke large brain responses.
The accumulation of evidence suggests that, in early blind subjects, primary as well as extrastriate visual areas are extensively recruited for processing of other senses. In contrast, for late blind subjects, there is only strong evidence for activation of extrastriate visual areas – and activation within these areas may not necessarily play a functional role.
Tactile performance
It is often suggested that blind subjects have enhanced tactile abilities. However it is still unclear to what extent differences in tactile performance between sighted and blind subjects should be attributed to blindness per se as compared to practice effects.
For obvious reasons, blind subjects tend to have more extensive experience with Braille than even well-trained sighted subjects (indeed, most sighted teachers of Braille read using vision rather than touch). As a result, non-Braille tactile tasks are typically considered a better measure of the direct effects of blindness (as compared to experience) on tactile abilities. Interestingly, it is not entirely clear whether blind subjects do exhibit enhanced acuity for non-Braille tactile discrimination tasks. In one study, no significant differences in sensory, touch, or two-point discrimination thresholds were found between early blind and sighted subjects. However in several more recent studies, tactile acuity has been found to be significantly higher in early blind subjects than sighted subjects for embossed Roman letters and grating orientation discrimination , tasks. Taken together these data suggest that visual deprivation in early blind subjects may only provide a weak or selective (for some types of task, but not all) advantage for tactile processing.
Most of the benefits in tactile discrimination abilities in blind as compared to sighted subjects are likely to be due to an interaction between blindness and the effects of increased experience and reliance on visual cues. In several studies, although blind subjects initially exhibit superior performance to sighted subjects on a range of tactile tasks, when practice is accounted for between blind and sighted subjects, these differences disappear. , Experiments such as these support the notion that the superiority in tactile tasks found in the blind is driven at least in part by greater tactile experience.
However, in another recent study it has been shown that sighted subjects perform better at Braille letter discrimination when they are blindfolded during a preceding five-day period than when they are allowed normal vision during that five-day period, irrespective of the amount of training in Braille. Indeed, blindfolded subjects who have not been trained at all in the task perform better than non-blindfolded subjects trained in the task over a five-day period. , These findings suggest that visual deprivation does play a role in the enhanced tactile abilities of blind subjects. Furthermore, the fact that sighted Braille teachers read Braille visually (from the shadows of the letters) rather than by touch, suggests that loss of visual input is important for tactile fluency (Pascual-Leone, personal communication).
Given that both visual deprivation and tactile experience play significant roles in subserving the tactile performance benefits observed in blind subjects, an understanding of how deprivation and experience interact may be important in developing effective Braille instruction programs, especially in the partially sighted.
Braille tactile processing
Numerous studies have examined cross-modal responses to Braille reading in the visual cortex of blind subjects. In an early ERP study, early blind subjects were reported to show a more posterior/occipital distribution of response during Braille reading than sighted subjects who read embossed Roman letters. Since then, several neuroimaging studies have demonstrated that early blind subjects consistently show greater responses to Braille words than to non-words (or other appropriate control stimuli) in primary as well as extrastriate visual cortex. In contrast, late blind subjects show greater responses than sighted subjects in extrastriate but not primary visual cortex (see reviews, , though see , ). In early blind subjects the magnitude of BOLD responses to Braille in primary visual cortex is highly correlated on an individual basis with verbal memory ability, suggesting a relationship between the extent of cross-modal responses and Braille abilities, at least in those who are visually deprived at an early age.
Several TMS studies provide clear evidence for the functional role of visual cortex in Braille reading in early blind subjects. TMS delivered to visual cortex while discriminating Braille letters (note that this is a somewhat simpler task than reading Braille text) induces significantly more errors than sham stimulation in early but not late blind subjects ( Fig. 41.5 ). This disruption of Braille performance suggests that the visual cortex does play a functional role in Braille reading at least in early blind subjects , (see reviews ). Additionally, there is the interesting case study of an early blind woman who, following an occipital stroke, lost the ability to read Braille without loss of her ability to detect Braille letters or loss of her other somatosensory abilities.
