Behavioral consequences of blindness
Throughout history, philosophers – and later scientists – were intrigued by blindness and its consequences on human behavior. The blind had inspired reflections on the mental representation of the world. Most famously, the correspondence between John Locke and William Molyneux, the Irish philosopher, which later became known as Molyneux’s problem: Imagine an individual that was born blind and learned the shapes cube and globe using tactile input. If this person were able to see one day, would they be able to distinguish the cube from the globe only based on vision? Locke and Molyneux seemed to agree that the answer is no, implying that representations depend on sensory modality (Degenaar 1996). In a different account some centuries beforehand, Ibn Tufayl, the Arab Andalusian polymath, postulates a contradicting view. Imagine a blind person, “with good capacity […] and solid judgment”, if able to see one day “would find every thing to be exactly agreeable to those notions which he had before; and that colours were such as he had before conceived them to be, by those descriptions he had received”. The only difference, hence, between blindness and sight would be “clearness and extreme delight” (Ockley 1708 in a translation of Ibn Tufayl). Thus, implying a more abstract underlying representation.
More functionally, Diderot, in his “Lettre sur les aveugles” (letter on blindness; 1749), stated confidently that the congenitally blind develop supernormal powers of touch and hearing, which substitute for their loss of vision (Morgan 1999). Over two centuries later, in 2007, the Belgian police recruited blind individuals as detectives because of their skill in segregating speech in recordings with low signal-to-noise ratio (Hötting and Röder 2009).
Notwithstanding the mythical tone of the previous lines, blindness does require a drastic behavioral adaptation. Not able to benefit from visual cues, the blind revert to using tactile, auditory and olfactory strategies to negotiate the world. It is therefore tempting to presume that vision is compensated for by extra computational power to the intact senses. In the next section, I will summarize findings regarding the behavioral differences between the sighted and the early blind.
Comparing the tactile perceptual abilities between the early blind and the sighted showed both the superiority of the blind and the superiority of the sighted but also similarity between the groups. Even when considering the same tactile ability, e.g., grating orientation detection with the fingertips, some results favor the blind (Van Boven et al. 2000; Goldreich and Kanics 2003; Wong et al. 2011) while others find no difference between the blind and the sighted (Grant et al. 2000; Alary et al. 2009). Otherwise, the blind were found similar to the sighted in vibrotactile frequency detection (Grant et al. 2000), surface texture discrimination acuity (Heller 1989; Grant et al. 2000), object shape matching (in children Withagen et al. 2012), dot-pattern discrimination (Stilla et al. 2008), and bar length discrimination (Stevens et al. 1996). Better performance in the blind was demonstrated in real-world texture discrimination (Gurtubay-Antolin and Rodríguez-Fornells 2017), angle size discrimination with the arm out-stretched (but not when the elbow was restrained; Alary et al. 2008), and dot-pattern discrimination (this advantage disappeared after practice; Grant et al. 2000). Inconsistency in the results across tasks motivated looking into the role of perceptual learning, which could potentially underlie some of the observed variability. Wong and colleagues (2011) probed the role of experience in enhanced tactile abilities by comparing the tactile acuity of the blind to that of the sighted on both fingertips and lips. They reasoned that finding a difference on the fingertips but not on the lips favors the hypothesis that hyper-acuity in the blind is due to experience and not merely due to sensory deprivation. Indeed, they showed that while acuity on the lips was similar in both groups, fingers have a higher tactile acuity in the blind. Moreover, the acuity in the fingers used for Braille-reading was correlated with weekly Braille-reading time. Hence, providing evidence of an experience-dependent mechanism, at least for tactile acuity in grating orientation detection (Wong et al. 2011). Notwithstanding, the joint influence of sensory deprivation and experience-dependent mechanisms on tactile perception in the blind is far from being fully understood (Sathian and Stilla 2010).
In general, the blind were found equivalent to the sighted in absolute auditory detection thresholds and reaction time to sounds (Collignon et al. 2006; Cornell Kärnekull et al. 2016). When considering spectral and temporal processing abilities, the early blind show an advantage, when compared to the sighted and the late blind, in pitch discrimination (Gougoux et al. 2004; Rokem and Ahissar 2009; Lerens and Renier 2014) and pitch-timbre categorization (even when controlling for musical training; Wan et al. 2010). Also, absolute pitch was found to be more prevalent in early blind musicians (Hamilton et al 2004). The blind also show a better ability to organize sounds sequentially, supporting an enhanced processing of incoming acoustic information (Boroujeni et al. 2017). In this line, Stevens et al. (2005) used auditory backwards masking to show that an early arrival of a post-stimulus mask impaired performance only in the sighted, providing evidence for an advantage in auditory perceptual consolidation in the blind. When considering verbal input, the blind made less errors during dichotic listening to phonemes (Hugdahl et al. 2004). They also showed an advantage in speech perception with background noise (Niemeyer and Starlinger 1981; Muchnik et al. 1991; Rokem and Ahissar 2009) and in ultra-fast speech comprehension (Gordon-Salant and Friedman 2011; Dietrich et al. 2013).
When considering spatial tasks such as sound-source localization, the blind perform similarly to the sighted when the source is in the horizontal axis (Lessard et al. 1998; Röder et al. 1999; Zwiers et al. 2001a; cf. Macé et al. 2012). They show an advantage, however, when the sources are in the peripheral auditory space (Röder et al. 1999). In addition, a study by Lessard et al. (1998) found that half of the tested blind subjects were able to accurately locate sound-sources monaurally (with one ear blocked) where the sighted presented a systematic bias towards the open ear (Morgan 1999; Van Wanrooij and Van Opstal 2004). This supra-normal ability was later attributed to the more effective use of spectral cues by the blind (Doucet et al. 2005; Voss et al. 2011). In the vertical axis, on the other hand, the blind often perform worse than the sighted (Lewald 2002), especially so under low-SNR conditions (Zwiers et al. 2001b). In a recent study, Voss et al. (2015) showed that the highest performing subjects in monaural localization in the horizontal space are the worst in the vertical space and vice-versa. Thus, providing evidence to support a trade-off in learned perceptual expertise.
When judging distance from sound-source, the blind show an advantage over the sighted for tasks that rely on relative auditory distance information (Voss et al. 2004; Kolarik et al. 2013) but not when relying on absolute auditory distance information (Wanet and Veraart 1985; in adolescents Lai and Chen 2006; in a mixed group of early and late blind Macé et al. 2012; Kolarik et al. 2013). This difference is thought to depend on the fact that the blind possess an internally intact distance representation that is not well-calibrated to the environment due to the absence of vision (Kolarik et al. 2016). Moreover, a major deficit was demonstrated in the blind by using a complex task where subjects had to judge whether the second in a series of three sounds is spatially closer to the first or the third, highlighting further the importance of vision in the calibration of auditory representations (Gori et al. 2014; present also in children Vercillo et al. 2016).
Hence, the emerging picture is that while some auditory functions are enhanced following vision loss, others are impaired. This might reflect the balance between adaptation for blindness and the role of vision in calibrating the auditory space (King 2015).
Olfactory & Gustatory
Higher olfactory performance has been observed in the blind when performing a task of free odor identification (Murphy and Cain 1986; Rosenbluth et al. 2000; Wakefield et al. 2004 in children; cf. Cuevas et al. 2010; Rombaux et al. 2010). This advantage was not present, however, in multiple-choice odor identification (Smith et al. 1993; Rosenbluth et al. 2000; Cuevas et al. 2009, 2010; Beaulieu-Lefebvre et al. 2011; Çomoğlu et al. 2015). An advantage in olfactory sensitivity thresholds was present in some studies (Cuevas et al. 2010; Beaulieu-Lefebvre et al. 2011; Çomoğlu et al. 2015) but not in others (Murphy and Cain 1986; Smith et al. 1993; Rosenbluth et al. 2000; Wakefield et al. 2004 in children). Similarly, some studies showed an advantage for the blind in odor discrimination (Cuevas et al. 2009, 2010; Rombaux et al. 2010; Çomoğlu et al. 2015) while others showed similarity to the sighted (Smith et al. 1993; Beaulieu-Lefebvre et al. 2011). Interestingly, a group of highly trained sighted individuals outperformed both the blind and their sighted controls in odor discrimination (Smith et al. 1993). This suggests that training might be more influential than blindness per se when considering olfactory abilities. Therefore, the overtraining in the blind might confound the interpretation of their superior abilities. Teasing apart the roles played by blindness and training in olfactory abilities would be a useful next step that could help reconcile the reviewed inconsistent findings.
Much less studied is gustatory sensitivity. Recent evidence points to generally worse performance in the blind when compared to the sighted (Gagnon et al. 2013). However, past studies also showed abilities that are equivalent to the sighted (Smith et al. 1993).
Behavioral consequences of blindness
Beyond the traditional senses
Beyond the scope of this dissertation is the impact of blindness on navigation and the development of echolocation skills by some blind individuals. Echo-locating individuals produce mouth clicks, for example, and listen to their echo in order to ‘reveal’ their surroundings. Some individuals are even able to determine information such as the shape and material of objects. Evidence suggests that even when controlling for training, the blind are able to achieve better performance in echolocation tasks (Kolarik et al. 2014). For navigation, the reader is referred to a literature review and discussion by Schinazi et al. (2016).
Perceptual differences between blind and sighted individuals might be expected due to the different ways the two groups experience the world. How would, however, blindness influence cognitive functions?
Numerical skills and language
In numerical cognition, the blind made fewer errors while solving certain basic arithmetic operations, e.g., 9×8 (Dormal, Crollen, et al. 2016) but were not found different from the sighted when solving problems requiring more calculations, e.g., 2 peaches cost 17 cents, how much would a dozen cost? (Rokem and Ahissar 2009). They also performed better than the sighted at numerical estimation (Castronovo and Seron 2007a; Castronovo and Delvenne 2013). Also, the blind demonstrated both the classical numerical distance effect (Szűcs and Csépe 2005) and the SNARC (spatial numerical association of response codes) effect (Castronovo and Seron 2007b) indicating that certain numerical processes do not depend on visual experience.
For language skills, the blind presented higher performance than the sighted in a task of verbal fluency, e.g., produce the largest number of words that start with a certain phoneme, (Occelli et al. 2017).
Classical measures of executive abilities use span tasks where subjects are asked to recall a list of items in order. When measuring recalling span scores, the emerging picture is not very clear. On the one hand, simple digit-, word- and pseudo-word- span scores are mostly found to be higher in the blind (Rokem and Ahissar 2009; Crollen et al. 2011; in children Withagen et al. 2013; Dormal, Crollen, et al. 2016). On the other hand, when adding computational complexity, we find heterogeneous results. In repeating the digits backwards, for example, some authors found that the blind were better than their sighted counterparts (in children Withagen et al. 2013; in adults Occelli et al. 2017) while others did not (Rokem and Ahissar 2009; Castronovo and Delvenne 2013). Similarly, using a task that requires recalling the last word of sentences, some studies showed an advantage to the blind (in children Withagen et al. 2013) while others failed to do so (in children Crollen et al. 2011). In addition, Rokem et al. (2009) found that when controlling for perceptual input the blind lose their advantage in verbal memory span for pseudo-words. Thus suggesting that enhanced stimulus encoding is responsible for the enlarged span as opposed to better executive processing. In line with this suggestion are studies that failed to find an advantage for the blind in tasks of pitch working-memory (Wan et al. 2010), n-back for consonants (Pigeon and Marin-Lamellet 2015) and n-back with Braille letters for the blind and written letters for the sighted (Bliss et al. 2004).
When considering memory capacities, the blind were found better than their sighted counterparts at judging if words were part of previously heard sentences (Röder et al. 2001). They were also found to retain more words from a list of pre-learned abstract words (Amedi et al. 2003; Raz et al. 2005; Occelli et al. 2017). Also, they performed better when recalling words according to their serial position of presentation (Raz et al. 2007) and showed reduced false memory effects (Pasqualotto et al. 2013). Better memory performance was also observed for environmental sounds (e.g. motorcycle) when comparing to the sighted but not to the late blind (Röder and Rösler 2003; Cornell Kärnekull et al. 2016).
The blind showed shorter response times in several attentional tasks (e.g. selective attention) using both auditory and tactile stimuli (Collignon et al. 2006; Pigeon and Marin-Lamellet 2015). However, the latter study showed that performance in selective, sustained and divided attention did not differ from the late blind when using consonants as targets and one-digit numbers as distractors. Also, no advantage to the favor of the blind was observed in task-switching costs (Pigeon and Marin-Lamellet 2015). Using dichotic listening, Hugdahl et al. (2004) provided evidence for an enhanced tuning of attention in the blind. Subjects were able to focus their attention to one ear inhibiting input from the other in order to identify auditory syllables. Furthermore, the blind presented an equally distributed attention across space and, unlike the sighted, did not favor the frontal part of space (Lerens and Renier 2014).
Functional recruitment of the visual cortex
To summarize, blind individuals do show an advantage when compared to the sighted. However, this advantage is neither general nor extensive. It is fairly restricted and it depends on task specificities, sensory modality and expertise. Moreover, under specific circumstances, the blind can also present a deficit in performance. Therefore, the presumption that blind individuals compensate for their lack of vision with more powerful auditory, tactile, olfactory and gustatory abilities should be carefully substantiated and not taken for granted.
Considering studies where a behavioral enhancement was observed in the blind, one cannot but pose the following question: Are the observed effects strictly due to visual deprivation? In this regard, two competing hypotheses are to be considered: 1) The visual deprivation hypothesis: the absence of vision by itself enhances performance; 2) The experience hypothesis: training on a specific task results in better performance. Wong et al. (2011), mentioned earlier, found results that favor hypothesis number 2 because the enhanced acuity was related to the level of expertise on the fingertips and was not at all found on the lips. In a study targeting the effect of visual deprivation in normal adult subjects, Merabet et al. (2008) implemented a 5-day tactile training program while blindfolding half of the subjects for the whole duration of the experiment. They demonstrated that, at the last day of the experiment, tactile acuity measures improved for both the blindfolded and non-blindfolded subjects. However, the blindfolded group made less errors in a Braille letter discrimination task. Hence, demonstrating in normal subjects, that both visual deprivation and training play a role in influencing tactile performance. Therefore, in blind individuals, several mechanisms are possibly at play including deprivation age, training and the specific functionality at hand. An exploration of the interaction between those factors is beyond the scope of this dissertation.
Functional recruitment of the visual cortex
In contrast to the at times inconsistent account of behavioral differences between the blind and the sighted, looking at the brain unveils substantial differences resulting from the lack of vision. At the center of attention is the visual cortex that occupies the occipital lobe extending from the striate and extrastriate cortices both ventrally and dorsally towards the temporal and parietal lobes. Early work by Wanet-Defalque et al. (1988) studying the anatomy of the visual cortex in the blind found that it is qualitatively undistinguishable from the visual cortex of the sighted when imaged using magnetic resonance imaging (MRI). In addition, the authors reported that the regional cerebral metabolic rate for glucose, measured using positron emission tomography (PET), is higher in the blind when compared to blindfolded sighted controls. They concluded that the visual cortex in the blind is indeed functioning and raised questions concerning its functions. The following sections summarize the efforts dedicated to delineating those functions.
Early evidence for a functional role
In one of the first attempts to probe the functions of the visual cortex in the blind, Uhl et al. (1991) compared Braille reading to passing the finger over random dot-patterns. The authors found greater occipital negativity in the blind using scalp-recorded event-related slow negative DC potential shifts. Using a similar task, Uhl et al. (1993) showed higher regional cerebral blood flow (rCBF) indices in the blind when compared to the sighted, but failed to show any differences between the two tasks. It was a few years later that Sadato et al. (1996) confirmed the existence of a task-dependent differential response in the visual cortex of the blind. Braille reading resulted in stronger rCBF than sweeping over a surface homogenously covered with Braille dots. Although not tested for statistically, they also showed that tactile tasks such as judging the width of two grooves also activate the visual cortex of the blind but to a lesser extent. This activation was regarded as a manifestation of crossmodal plasticity in vision-deprived humans (Sadato et al. 1996). Reinforcing the functional relevance of the visual cortex in the blind was a study by Cohen et al. (1997) using transcranial magnetic stimulation (TMS). The authors were able to impair the identification of both Braille and embossed Roman letters in the blind when stimulating the mid-occipital cortex but not when stimulating the somatosensory cortex. Hence, establishing the causal role of the visual cortex in letter identification when performed by early blind individuals. TMS to the occipital cortex was also shown to induce sensations on the fingertips of blind Braille readers, in contrast to the sensations of light produced in the sighted (Ptito, Fumal, et al. 2008). Finally, a case of alexia for Braille following an ischemic occipital stroke in an early blind woman provided even more compelling evidence that the visual cortex is essential to Braille reading (Hamilton et al. 2000).
Specialized functional recruitment
Previously mentioned studies concentrated on the question whether the blind visual cortex is needed for Braille reading. However, Braille reading involves sensory, motor, associative, and executive processes in addition to language functions. Every one of those single functions could, in principle, underlie the Braille-related activation of the visual cortex. Thus, to better understand visual activation in the blind, individual, more circumscribed, processes and functions were studied. What follows is a review of the literature where each section will focus, as much as possible, on specific functions starting from low-level perceptual processing going up the cognitive hierarchy to the execution of high-order functions.
Gizewski et al. (2003) showed that neither median nerve stimulation nor finger tapping activated the visual cortex in the blind suggesting that it is not mere sensory stimulation that is responsible for the visual activation. Supporting this suggestion is the finding that TMS delivered to primary visual cortex (V1) while identifying Braille letters impaired letter identification and not detection in contrast to stimulating primary somatosensory cortex that impaired detection (Hamilton and Pascual-Leone 1998). These results are in line with studies that found no somatosensory processing in the visual cortex when no task was involved (e.g., Sadato et al. 1996; Beisteiner et al. 2015; Pishnamazi et al. 2016). On the other hand, when subjects are asked to perform a task on the perceived stimuli, a different picture emerges. The visual cortex of the blind is recruited during a variety of somatosensory tasks. For example, Rösler et al. (1993) found a pronounced occipital activity in the blind when encoding tactile shapes for a consecutive mental rotation task. Among other examples are the detection of the orientation of gratings applied to the fingertip (Lewis et al. 2010), the orientation of T shapes stimulated on the tongue using an electrode array (Ptito et al. 2005), and the amplitude change of vibro-tactile stimuli (Burton et al. 2010). The visual cortex of the blind is also recruited when performing same/different tasks on the frequency of vibro-tactile stimuli (Burton et al. 2004, 2010) as well as on more elaborate stimuli such as shoes, plastic bottles and masks of faces (Pietrini et al. 2004) and plastic animals that could either be mammals or not (Lewis et al. 2010).
All the above mentioned studies, however, used the non-specific task > rest contrast which resulted in an extensive activation in the visual cortex. This use is not optimal because such contrasts may include many cognitive processes (e.g. executive) that can render the interpretation of activations very difficult. To overcome this problem, Stilla et al. (2008) designed a task of spatial dot-pattern discrimination with a stricter control condition. In the main spatial condition, subjects had to judge if the middle dot in a series of 3 dots is offset to the right or to the left. In the control, temporal, condition, subjects were stimulated for 0.7 and 1.3 s and had to judge if the stimulation duration was short or long (Figure 1.1).
When comparing the main condition to the control condition, they found activations in the left lingual
gyrus and right collateral sulcus in addition to bilateral dorsal activations in the parieto-occipital fissure (Figure 1.2a; Stilla et al. 2008).
Also using a controlled design, Voss et al. (2016) showed a right-lateralized activation in the superior and middle occipital gyri (SOG & MOG) that is modulated by task difficulty of tactile angle-size identification (Figure 1.2b). Additionally, a right-lateralized cluster in the precuneus, sensitive to the symmetry of tactile Braille-like dot patterns, was found when comparing the blind to the sighted (Bauer et al. 2015).
Figure 1.2 – Brain activation during controlled tactile tasks. a | Activation in the spatial > temporal conditions overlaid on a flattened brain. POF=parieto-occipital fissure, IOS=intra-occipital sulcus. Adapted from Stilla et al. (2008). b | Comparing difficult to easy angle-size identification in the blind. Adapted from Voss et al. (2016).
In contrast to somatosensory processing, under no or low attentional demands, certain studies found that auditory stimulation activated the visual cortex in the blind (e.g. Arnaud et al. 2013; Watkins et al. 2013 in anophthalmic subjects; Pishnamazi et al. 2016) while others did not (Kujala et al. 1995; e.g.
Weeks et al. 2000; Stevens and Weaver 2009). Such differences may depend on technical factors such as scanner noise and experimental factors such as the nature of the stimuli and the instructions given to the subjects. Indeed, with the exception of a study in anophthalmic subjects that reported an activation in hMT+/V5 while subjects passively listened to tone trains of 1 s (Watkins et al. 2013), studies not finding an activation for passive auditory stimulation often used functional MRI (fMRI) sampling techniques where the scanner noise does not coincide with the stimuli (e.g. Stevens and Weaver 2009). When performing sound-source localization, early electrophysiological studies generally showed more posterior scalp distributions in the blind compared to the sighted (Kujala et al. 1992; Leclerc et al. 2000). Using imaging, higher rCBF was measured in the right visual cortex of the blind (Weeks et al. 2000). More specifically, higher performance for peripheral sources in the blind was correlated with a more spatially tuned and more posteriorly distributed N1 response (Röder et al. 1999). Moreover, studies have shown that the blind could be split into two groups: 1) individuals who perform better than the sighted on monaural sound-source localization; 2) individuals who perform like the sighted (e.g. Lessard et al. 1998). Only subjects that performed well monaurally were found to activate the right cuneus and lingual gyrus during both monaural (Gougoux et al. 2005) and binaural sound-source localization (Voss et al. 2008). Moreover, in those high performing blind subjects a correlation between performance and change in CBF was found in right striate cortex in addition to the lingual and superior occipital gyri (Gougoux et al. 2005). This finding established a link between superior performance in the blind and activity in their visual cortex. In support of that, TMS delivered to the right dorsal occipital cortex 50msec after stimulus onset successfully disrupted spatial sound processing in the blind. However, it should be noted that a disruption effect was also observed in the sighted, but possibly due to different mechanisms (Collignon et al. 2009).
Table of contents :
1.1. Blindness – definition
1.2. Behavioral consequences of blindness
126.96.36.199. Olfactory & Gustatory
188.8.131.52. Beyond the traditional senses
1.2.2. High-order cognition
184.108.40.206. Numerical skills and language
220.127.116.11. Executive functions
1.3. Functional recruitment of the visual cortex
1.4. Early evidence for a functional role
1.5. Specialized functional recruitment
1.5.1. Somatosensory processing
1.5.2. Auditory processing
1.5.3. Modality-independent perceptual processing
1.5.4. Higher-order processing
1.5.5. Language processing
1.5.6. Other symbolic representations
1.5.7. Long-term memory
1.5.8. Executive functions
1.6. Connectivity of the visual cortex
1.7. Summary and thesis outline
2. Cognitive functions of the visual cortex in the blind
2.1. Distinctive expansion of cognitive networks into the visual cortex in the blind Abstract
2.1.2. Materials and methods
2.1.5. Supplementary information
2.2. Innate connectivity of visual cortex
2.2.1. Hypothesis and analysis
2.3. Limitations and perspectives
3. Electrophysiological investigation of language
3.1. Semantic coding in the visual cortex of the early blind Abstract
3.1.2. Materials and Methods
3.1.5. Supplementary information
4. General discussion
4.1. Thesis overview
4.2. Computation in the blind visual cortex
4.3. Function-connectivity correspondence
4.4. General perspectives