What neuroimaging tells us about sensory substitution

  • Published on

  • View

  • Download

Embed Size (px)


<ul><li><p>Neuroscience and Biobehavioral Revie</p><p>vie</p><p>s</p><p>V</p><p>de</p><p>, 4</p><p>7; a</p><p>Abstract</p><p>itutive modality or is it determined by the nature of the information</p><p>n? This paper reviews the recent neuroimaging studies which have</p><p>investigated the neural bases of sensory substitution. The detailed analysis of available results led us to propose a general scheme of the</p><p>neural mechanisms underlying sensory substitution. Two different main processes may be responsible for the visual area recruitment</p><p>observed in the different studies: cross-modality and mental (visual) imagery. Based on our results analysis, we propose that cross-</p><p>Introduced by Bach-y-Rita in 1969, the sensory sub- ogies (Kaczmarek et al., 1991; Meijer, 1992, Capelle et al.,</p><p>have been reproduced in early (Arno et al., 2001a; Sampaio</p><p>ARTICLE IN PRESS</p><p>Corresponding author. Centre de Medecine Nucleaire Hopital Neuro-</p><p>et al., 2001) and late blind subjects (Cronly-Dillon et al.,1999). Not only these performances were found to beaccessible to blind subjects (Cronly-Dillon et al., 1999;</p><p>0149-7634/$ - see front matter r 2007 Published by Elsevier Ltd.</p><p>doi:10.1016/j.neubiorev.2007.05.010</p><p>Cardiologique, 59 Bd Pinel 69677 BRON Cedex, France.</p><p>Tel.: +33 472 684 961; fax: +33 472 357 345.</p><p>E-mail address: christian.scheiber@chu-lyon.fr (C. Scheiber).stitution concept refers to the use of one sensory modalityto supply information normally gathered from anothersense (Bach-y-Rita et al., 1969). This information isacquired through an articial organ, and then transformedinto a meaningful signal for the substitutive system (Fig. 1).In the case of blindness, visual information can betransmitted through the auditory or tactile channels.Since 1969, several sensory substitution devices have</p><p>been developed, using more and more advanced technol-</p><p>1998). In parallel, several behavioural studies have been ledin order to evaluate the performances allowed by theseprostheses. Tactile- or auditory-for-visual substitutiondevices have been shown to allow blindfolded sightedsubjects to match vibrotactile to visual patterns (Epsteinet al., 1989), to discriminate pattern orientations (Sampaioet al., 2001) and to recognise visual patterns (Arno et al.,1999, 2001a) and graphic representations of objects(Cronly-Dillon et al., 1999). Training is necessary toachieve these performances. Some of these experimentsmodel implies that, with training, sensory substitution mainly induces visual-like perception in sighted subjects and mainly auditory or</p><p>tactile perception in blind subjects. This framework leads us to make some predictions that could easily be tested.</p><p>r 2007 Published by Elsevier Ltd.</p><p>Keywords: Sensory substitution; Blindness; Cross-modality; Mental imagery; Visual perception; Neuroimaging</p><p>Contents</p><p>References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1069modality is the predominant process in early blind subjects whereas mental imagery is predominant in blindfolded sighted subjects. Thisprostheses. Is the perception determined by the nature of the subst</p><p>transmitted by the device? Is it a totally new, amodal, perceptioA major question in the eld of sensory substitution concerns the nature of the perception generated by sensory substitutionRe</p><p>What neuroimaging tells u</p><p>Colline Poiriera, Anne G. DeaLaboratoire de Genie de la Rehabilitation Neurale, Universite catholique</p><p>bInstitut de physique biologique, UMR 7004</p><p>Received 5 March 200ws 31 (2007) 10641070</p><p>w</p><p>about sensory substitution</p><p>oldera, Christian Scheiberb,</p><p>Louvain, Avenue Hippocrate, 54 UCL-54.46, B-1200 Brussels, Belgium</p><p>rue Kirschleger, 67085 Strasbourg, France</p><p>ccepted 19 May 2007</p><p>www.elsevier.com/locate/neubiorev</p></li><li><p>ARTICLE IN PRESSobehFig. 1. (a) Schematic representation of the auditory-for-visual sensory</p><p>substitution device developed by Capelle et al. (1998) and named PSVA,</p><p>adapted to the fMRI environment. (b) The PSVA and its power supply.</p><p>(c and d) Subject using the device in the MRI environment. Normally, the</p><p>PSVA is connected to a tiny head-xed camera. As the subjects cannot</p><p>move their head in the scanner, this camera was replaced by a non-</p><p>magnetic joystick connected to a PC. Using this joystick, the subjects</p><p>could move the patterns they were supposed to recognise. These</p><p>movements made corresponding sounds to change according to the PSVA</p><p>C. Poirier et al. / Neuroscience and BiSampaio et al., 2001), but they also were found moreaccurate as compared to those of blindfolded sightedsubjects (Arno et al., 2001a).More recently, Renier et al. have shown that an</p><p>auditory-for-visual substitution device can mediate visualillusions (Renier et al., 2005a; 2006) and allow depthperception in blindfolded sighted subjects (Renier et al.,2005b). Using a pattern recognition task, it has also beenfound that, as in vision, blindfolded sighted subjects usingan auditory-for-visual substitution device better recognisedvertical bars than horizontal bars, these last ones beingbetter recognised than oblique bars (Poirier et al., 2006a).Subjects were also found to better recognise the size andthe spatial arrangement of the elements constituting thepatterns than the nature of these elements (vertical,horizontal and oblique bars). It is worth noting that theseresults match very well with visual perception rules (e.g.Morrison and Schyns, 2001; Miller and Navon, 2002).All these results raise the question of the nature of the</p><p>perception induced by sensory substitution prostheses. Isthe perception determined by the nature of the substitutivemodality (i.e. tactile or auditory) or is it determined by thenature of the information transmitted by the device (i.e.visual)? Is it a totally new, amodal, perception? Neuroima-ging studies have recently brought partial responses to thisquestion.</p><p>code. These sounds were transmitted via transducers (in the copper box, in</p><p>image (d)) and dedicated plastic conducts that were inserted into the</p><p>subjects ears. Headphones were added for isolation purpose. The plastic</p><p>tube visible in image (d) contained a microphone at its non-visible</p><p>extremity and allowed the experimenter to hear the verbal description</p><p>made by the subject of each pattern.Using Positron Emission Tomography (PET), Arno andcolleagues (2001b) have shown that pattern recognitiontrough an auditory-for-visual device induced the recruit-ment of extra-striate occipital areas (BA 18 and 19) in earlyblind subjects but not in blindfolded sighted controls.Using the same PET technique but another device and adifferent task, Ptito and colleagues (2005) have foundsimilar results: a pattern orientation discrimination task,performed through a tactile-for-visual device stimulatingelectrically the tongue of the subjects, was found to recruitthe extra-striate occipital areas BA 18 and 19 only in blindsubjects but not in sighted controls. Another PET studyhas investigated the neural substrates of a depth perceptiontask through an auditory-for-visual device (Renier et al.,2004, 2005b). Based on three monocular depth cues (therelative target size, the proximity of the target to thehorizon and the linear perspective), this task was found toinvolve the extra-striate area BA 19 in blindfolded sightedsubjects whereas only a slight trend to visual activation wasobserved in early blind subjects. Finally, a FunctionalMagnetic Resonance Imaging (fMRI) study has shownthat pattern recognition through an auditory-for-visualdevice can induce the recruitment of striate (BA 17) andextra-striate (BA 18 and 19) areas in blindfolded sightedsubjects (Poirier et al., 2007) (Fig. 2).The major nding of these studies lies in the recruitment</p><p>of brain areas (BA 17, 18 and 19) usually considered asvisual areas, in addition to auditory or somatosensorycortex activation, in blindfolded sighted (Renier et al.,2005a, b; Poirier et al., 2007) and early blind subjects (Arnoet al., 2001b; Ptito et al., 2005). Two major differentinterpretations of these results can be made.First, visual area activation can reect the use of mental</p><p>(visual) imagery strategies. Visual imagery is known toinduce the recruitment of the striate and extra-striate areasin blindfolded sighted subjects (Kosslyn et al., 1995,Kosslyn and Thompson, 2003). To a lesser extent, earlyblind subjects seem also be able to perform mental imagerytasks (Marmor and Zaback, 1976; Kerr, 1983). The natureof imagery performed by blind subjects, visual or not,remains a source of debate. Nevertheless, this process wasalso found to induce the recruitment of the striate andextra-striate areas in this subject population (De Volder etal., 2001; Vanlierde et al., 2003; Lambert et al., 2004).Second, cross-modality could account for the observed</p><p>results. Cross-modality consists in the recruitment of brainareas normally devoted to processing information fromone sensory modality by the processing of informationcoming from another modality. This phenomenon is wellknown in blind subjects, in whom auditory and tactilestimuli induce visual area recruitment (for a review, seeTheoret et al., 2004). However, recent studies have shownthat this phenomenon also occurs in sighted subjects in amulti-sensory but also in a uni-sensory context. Variousauditory and tactile tasks were found to induce the</p><p>avioral Reviews 31 (2007) 10641070 1065recruitment of the visual areas (e.g. Amedi et al., 2001;Blake et al., 2004). Nevertheless, this process seems to be</p></li><li><p>ARTICLE IN PRESSbehC. Poirier et al. / Neuroscience and Bio1066less important in sighted than in blind subjects (Poirieret al., 2006b).Both hypotheses, mental imagery and cross-modality,</p><p>are not mutually exclusive. It is however difcult todetermine which phenomenon or phenomena occurred ineach study. Indeed, both processes tend to induce similartopographically organised visual activations. Whereas theorganisation of visual information processing in the ventralwhat and dorsal where streams is well known(Ungerleider and Haxby, 1994), visual imagery and cross-modal processing seems to follow the same rules. Whereasimagery of objects or faces mainly induces the recruitmentof the ventral occipitotemporal stream in sighted subjects(Kosslyn et al., 1995), visuo-spatial imagery mainly recruitsthe dorsal occipitoparietal stream (Mellet et al., 1996).Similar results were obtained with imagery tasks performedby blind subjects (De Volder et al., 2001; Vanlierde et al.,2003). Concerning cross-modality, recent studies suggestthat tactile and auditory processes may involve the visualareas normally recruited by the corresponding process inthe visual modality. For instance, auditory and tactile</p><p>Fig. 2. (a) Patterns explored trough an auditory-for-visual device by blindfolde</p><p>elicited by pattern recognition after training (group analysis). The statistical par</p><p>normalised brain MRI (T1-image), allowing the visualisation of brain activat</p><p>corrected for multiple comparisons in the whole brain are displayed. (c) Surface</p><p>rst part and the second part of the exploration, after training (group analysis</p><p>Adapted from Poirier et al. (2006b).avioral Reviews 31 (2007) 10641070motion processing were found to induce the recruitment ofthe visual motion area V5 in blind (Poirier et al., 2006b;Ricciardi et al., 2007) and sighted subjects (Blake et al.,2004; Poirier et al., 2005; Ricciardi et al., 2007) whereastactile processing of shapes or Braille reading in sightedand blind subjects, respectively, induces the recruitment ofthe lateral occipital cortex usually involved in visualprocessing of shapes (Amedi et al., 2001, 2003).However, analysing and comparing the different results</p><p>obtained by the neuroimaging studies having investigatedsensory substitution shed light on the respective contribu-tion of cross-modal plasticity and visual imagery in therecruitment of the visual areas observed in these studies. InPtito et al.s (2005) study, visual occipital areas were foundto be recruited in blind subjects but not in sighted subjects.This recruitment was observed only after training. Theauthors argued that if visual activations observed in blindsubjects were due to mental imagery, such activationswould have been also observed in sighted controls. Theythus interpret their results as a cross-modality consequence.It is however difcult to explain why visual activation was</p><p>d sighted subjects in Poirier et al. (2006b)s study. (b) Brain activation foci</p><p>ametric map is superimposed on a sagittal section (x 10) of an individualion on the striate area V1. Only voxels exceeding a threshold of Po0.01view of the activated brain network during pattern recognition during the</p><p>). A threshold of Po0.01 corrected for multiple comparisons was applied.</p></li><li><p>ARTICLE IN PRESSobehnot found in blind subjects before training. A PET studyabout sound localisation has shown that intensity of cross-modal visual activations was positively correlated tobehavioural accuracy in blind subjects (Gougoux et al.,2005). The very poor accuracy of blind subjects in theorientation task of Ptito et al. (2005) before training couldthus account for the absence of detected activation in visualareas during the rst PET experiment.As in Ptito et als study (2005), Arno et al. (2001b)</p><p>observed occipital area activation in blind subjects but notin blindfolded sighted controls. This result was comfortedby a trans-cranial magnetic stimulation (TMS) study(Collignon et al., 2007). In this study, the authors inducedvirtual lesions of the extra-striate area found activated inArno et al.s (2001b) study in blind and sighted subjectswhile the subjects were trying to recognise pattern with theauditory-for-visual device. These virtual lesions were foundto disrupt the pattern recognition task in blind subjects butnot in sighted controls. These results suggest that therecruitment of visual areas observed in Arno et al.s(2001b) study was due to cross-modality. However, theabsence of occipital activation in sighted subjects must beinterpreted very cautiously. Indeed, absence of result onlymeans that signicant activation was not detected but notthat this activation did not occur. This is particularly truein PET studies, due to the relative poor sensitivity of thistechnique as compared with fMRI. This limitation is alsopresent in the TMS study (Collignon et al., 2007) in whichthe 1-Hz off-line TMS used could have been not enoughdisruptive (as compared with repetitive TMS for instance)and the task not enough demanding (Collignon et al.,2005). These methodological differences could explain whyoccipital areas were found recruited in blindfolded sightedsubjects in Poirier et al.s (2007) fMRI study but not inArno et al.s (2001b), and Ptito et al.s (2005) PET studiesinvolving similar tasks. In both PET studies, the absence ofoccipital area activation in sighted controls should thus notpermit exclusion of the potential use of mental imagerystrategies in sighted subjects, and consequently in blindsubjects: imagery could have occurred during these tasks,inducing the recruitment of...</p></li></ul>


View more >