Cross-Modal Plasticity in the Auditory Cortex of the Congenitally Deaf: an fMRI study using population receptive field analysis

Joana Margarida de Morais Sayal Abreu Campos CROSS-MODAL PLASTICITY IN THE AUDITORY CORTEX OF THE CONGENITALLY DEAF AN fMRI STUDY USING POPULATION RECEPTIVE FIELD ANALYSIS Dissertação no âmbito do Mestrado Interuniversitário em Neuropsicologia Clínica e Experimental, orientada pelo Professor Doutor Jorge Manuel Castelo Branco de Albuquerque Almeida e pela Doutora Zohar Tal e apresentada à Faculdade de Psicologia e Ciências da Educação da Universidade de Coimbra. Janeiro de 2022 Faculty of Psychology and Educational Sciences of University of Coimbra CROSS-MODAL PLASTICITY IN THE AUDITORY CORTEX OF THE CONGENITALLY DEAF An fMRI study using population receptive field analysis Joana Margarida de Morais Sayal Abreu Campos Thesis submitted for the Interuniversity Master in Clinical and Experimental Neuropsychology, supervised by Professor Jorge Almeida and Doctor Zohar Tal and presented to the Faculty of Psychology and Educational Sciences of University of Coimbra. January 2022 CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 2 Resumo Título: Plasticidade Intermodal no Córtex Auditivo de Surdos Congénitos: Análise de Campos Recetivos Populacionais através de um estudo de IRMf Palavras-chave: Neuroplasticidade; Imagem por Ressonância Magnética Funcional; Análise de Campos Recetivos Populacionais; Surdez Congénita; Organização Topográfica A neuroplasticidade é a capacidade manifestada pelo cérebro humano em reorganizar-se e modificar a sua atividade ao longo da vida. Envolve mudanças na estrutura, função, e ligações do cérebro, habitualmente derivadas da exposição a estímulos externos ou internos (como aprender uma nova capacidade), ou como resultado de lesões traumáticas ou privação sensorial. Esta última está presente no caso da perda de visão ou surdez profunda, onde o córtex sensorial privado pode ser recrutado para representar informações sensoriais pertencentes a outras modalidades. Este processo, conhecido como plasticidade intermodal, é o tema central desta tese. Em estudos anteriores, verificou-se que o córtex auditivo de surdos congénitos, mas não de ouvintes, é recrutado durante tarefas visuais. No entanto, não é claro se e até que ponto essas respostas intermodais no córtex auditivo privado representam informações espaciais visuais ou mapeiam o campo visual. Este trabalho aborda essa questão diretamente através de estudos de caso de IRMf, com o objetivo de pesquisar e mapear características de retinotopia no córtex auditivo, similarmente à forma de organização das representações neurais no sistema visual. Dois surdos congénitos e um ouvinte participaram numa experiência de IRMf com estímulos tradicionalmente utilizados para mapear o processamento visual. Foi aplicado um método de retinotopia por ondas progressivas (TWR) e de seguida uma técnica de análise de campos recetivos populacionais (pRF), que tem sido substancialmente usada para mapear gradientes topográficos no cérebro, incluindo retinotopia. Os resultados revelam respostas visuais no córtex auditivo dos surdos congénitos, associadas à apresentação dos estímulos, mas não no participante ouvinte. Estas respostas, predominantemente laterizadas no hemisfério direito, representam o campo visual contralateral e são caracterizadas por grandes campos recetivos, centrados nas proximidades da fóvea. Curiosamente, descobrimos que essas respostas a estímulos visuais refletiam principalmente sinais BOLD negativos no córtex auditivo dos surdos, sugerindo uma representação da informação visual através de sinais de desativação intermodal. Será discutida a interpretação e características dessas representações neuroplásticas e desativações neuronais, assim como os efeitos comportamentais, funcionais e anatómicos da plasticidade intermodal em surdos congénitos. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 3 Summary Title: Cross-Modal Plasticity in the Auditory Cortex of the Congenitally Deaf: an fMRI study using population receptive field analysis Keywords: Neuroplasticity; Functional Magnetic Resonance Imaging; Population Receptive Field Analysis; Congenital Deafness; Topographic Organization Neuroplasticity is the ability of the human brain to reorganize and modify its activity throughout life. It involves changes in the structure, function, and connections within the brain, typically acquired following external or internal stimuli (such as learning a new skill), or as a result of traumatic lesions or sensorial deprivation. The latter is present in the case of profound blindness or deafness, where the sensory-deprived cortex can be recruited to represent sensory information belonging to other modalities. This process, known as cross-modal plasticity, is the core subject of this thesis. Specifically, previous studies indicated that the auditory cortex of congenitally deaf, but not of hearing individuals, is recruited during visual tasks. However, it is not clear if and to what extent these cross-modal responses in the deprived auditory cortex represent visual spatial information or map the visual field. In this work, these questions were addressed directly in an fMRI set of case-studies, aiming to search and map retinotopy features in the auditory cortex, similarly to the well-known organization of neural representations in the visual system. Two congenitally deaf and one hearing participant went through a conventional retinotopy fMRI experiment with visual stimuli designed to map the visual system. We applied both traditional traveling-wave analysis and a biologically inspired population receptive field (pRF) technique, which have been substantially used to map topographic gradients in the brain, including retinotopy. Results reveal retinotopic-related responses in the auditory cortex of the deaf, but not in the hearing, locked to the visual presentation of stimuli. These responses, that were mostly lateralized to the right hemisphere, represented the contralateral visual field and were characterized by large receptive fields, centered to near foveal areas. Interestingly, we found that these responses to visual stimuli predominantly reflected negative BOLD signals in the auditory cortex of the deaf, suggesting that visual information might be represented through cross-modal deactivation signals. The meaning and features of these neuroplastic representations and neuronal deactivations will be discussed, as well the behavioural, functional, and anatomical effects of cross-modal plasticity in the congenitally deaf. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 4 Acknowledgements Completing this master’s and thesis was possible due to the academic and personal contribution of several people during these challenging two and a half years. I would like to express my deepest thanks: To Dr Zohar Tal, for the time we spent meeting and working together, always with a kind-hearted spirit. For her enthusiasm and encouragement. It has been a pleasure having Dr Zohar as my supervisor, and I have learned so much with her. To Professor Jorge Almeida, also my supervisor, for making it so interesting to discuss neuroscience, and for always having clear insights and ideas. For always wanting the best possible and pushing me forward. For the opportunity to be part of Proaction Lab for almost 5 years now. To Professor Óscar Gonçalves, for the support throughout this master on many occasions. And for believing in me. To Proaction Lab: from post-doctoral researchers, PhD students, research assistants, master’s colleagues, and collaborators. I would like to thank their availability, readiness to help, interest in my work, and friendship. To the team that collected this dataset in Beijing, to the participants, and to Dr Alessio Fracasso, that I unfortunately never met in person, for his expertise in pRF models that was so helpful in our analysis. To my internship supervisors, Professor Manuela Vilar, Professor Mário Simões and Dr Diana Duro, as my clinical internship overlapped with my thesis for 9 months. I’d like to thank them for their understanding and motivation during this time. To my friends: in music (OAUC, TAUC, Cithara) and to a very specific group of “always there” friends: Rita, Diana, Mathilde, Simão, Inês and Chico. To my family, my mother and my father, for being my home and unconditional love and support. To my brother, that is not only close emotionally, but is my partner in music and neuroscience. Aos meus avós maternos e paternos, para os quais as palavras são difíceis de escolher. Dedico esta tese à minha avó materna e ao meu avô paterno, que partiram em 2021. Sei que ficariam muito felizes e orgulhosos com a conclusão desta etapa. To Universidade de Coimbra, Universidade de Lisboa and Universidade do Minho. This interuniversity master’s experience was very rewarding, even in these pandemic times demanding unprecedented resilience. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 5 This thesis was supported by a Foundation for Science and Technology (FCT) project grant POCI-01-0145-FEDER-030757 | PTDC/PSI-GER/30757/2017 - “SeeingEars - Vendo com os teus ouvidos: como é que a neuroplasticidade devido a surdez congénita interfere com o processamento neuronal e com estratégias de reabilitação auditiva”, and Programa COMPETE. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 6 Table of Contents 1. Introduction ............................................................................................................................ 9 1.1. The brain and its topographic organization ............................................................................... 9 1.1.1. The visual system and retinotopic organization of V1 ............................................................ 9 1.1.2. The auditory system and its tonotopic organization .............................................................. 11 1.2. Neuroplasticity and cross-modal plasticity ............................................................................. 13 1.2.1. Deafness in humans: behavioural and neuroimaging findings ............................................... 14 1.2.1.1. Deafness in humans: behavioural findings ........................................................................ 14 1.2.1.2. Deafness in humans: anatomical MRI findings ................................................................. 16 1.2.2. Animal models of deafness ................................................................................................ 18 1.3. Functional MRI and its use in neuroscience ........................................................................... 19 1.4. Techniques for measuring visual field maps ........................................................................... 20 1.4.1. Phase encoded retinotopy (travelling wave retinotopy) ......................................................... 21 1.4.2. Population receptive field (pRF) modelling ......................................................................... 23 1.4.2.1. pRF analysis: advantages and applications ....................................................................... 24 1.5. Current study ....................................................................................................................... 25 2. Methods................................................................................................................................ 26 2.1. Participants .......................................................................................................................... 26 2.2. Stimuli and procedure........................................................................................................... 26 2.3. Anatomical and Functional Imaging ...................................................................................... 28 2.4. fMRI pre-processing ............................................................................................................ 28 2.4.1. Anatomical data pre-processing .......................................................................................... 28 2.4.2. Functional data pre-processing ........................................................................................... 28 2.5. Travelling wave retinotopy ................................................................................................... 29 2.6. Population receptive field modelling ..................................................................................... 29 3. Results .................................................................................................................................. 30 4. Discussion and Conclusion .................................................................................................. 36 References ................................................................................................................................ 41 Annexes.................................................................................................................................... 47 CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 7 List of Figures Figure 1 Visual Field Maps in V1, V2 and V3 ...................................................................... 10 Figure 2 Tonotopic Map Layout and Interpretations ............................................................. 12 Figure 3 Visual Field Landmarks .......................................................................................... 21 Figure 4 Travelling Wave Retinotopy Maps ......................................................................... 22 Figure 5 pRF Size Variation With Eccentricity ..................................................................... 24 Figure 6 Experimental Stimuli and Procedure ....................................................................... 27 Figure 7 Retinotopic Organization of the Visual Cortex in Hearing and Deaf Participants .. 31 Figure 8 Explained Variance in Hearing and Deaf Participants – Medial View ................... 32 Figure 9 Explained Variance in Hearing and Deaf Participants – Lateral View ................... 33 Figure 10 Beta Values (Slope) in Hearing and Deaf Participants ........................................... 34 Figure 11 Polar Angle, Eccentricity, Horizontal Location and pRF Width in the Deaf .......... 35 Figure A1 TWR Analysis – Phase maps……………………………………………………..40 CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 8 List of Abbreviations A1 AC BOLD CIs DICOM DMN fMRI GLM GM HG HRF LGN LH MRI NIfTI PAC pRF RF RH STG STS TR TWR V1 WM Area A1; Primary Auditory Cortex Auditory Cortex Blood Oxygen Level Dependent Cochlear Implants Digital Imaging and Communications in Medicine Default Mode Network Functional Magnetic Resonance Imaging General Linear Model Gray Matter Heschl’s Gyrus Hemodynamic Response Function Lateral Geniculate Nuclei Left Hemisphere Magnetic Resonance Imaging Neuroimaging Informatics Technology Initiative Primary Auditory Cortex Population Receptive Field Receptive Field Right Hemisphere Superior Temporal Gyrus Superior Temporal Sulcus Repetition Time Travelling Wave Retinotopy Area V1; Primary Visual Cortex White Matter CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 15 very important evolutionarily. But how does that work in the deaf? They rely more heavily on their remaining senses, since most of their input from the world comes from the binocular visual field (Frasnelli et al., 2011). This lack of auditory-visual convergence is one of the reasons why Bavelier et al. (2006) proposed that hearing loss leads to changes in higher-level attentional processing, with a redistribution of attentional resources to the peripheral visual field. More recent studies like Bottari et al. (2010) have provided complementary knowledge, stating that these differences arise not only from attentional redistribution, but also reorganized sensory processing. Differences between deaf and hearing subjects are seen in detecting motion perception changes: deaf participants are superior in detecting small deviations from horizontal movements (Almeida et al., 2018; Hauthal et al., 2013). However, heightened visual abilities as the latter are not widespread in all the areas of visual cognition (Frasnelli et al., 2011). In fact, deaf individuals can show both better and worse visual skills than hearing controls (Bavelier et al., 2006). Basic sensory thresholds such as contrast sensitivity (Finney et al., 2001), motion velocity (Brozinsky & Bavelier, 2004), motion sensitivity (Bosworth & Dobkins, 1999), brightness discrimination (Bross, 1979), and temporal resolution (Nava et al., 2008; Poizner & Tallal, 1987) do not seem to be enhanced in deaf individuals, which is not the case in more complex tasks, where visual attention and processing of the peripheral visual field are manipulated (Almeida et al., 2018). In the absence of auditory input, in order to monitor extrapersonal space, deaf individuals devote greater processing resources to the monitoring of the peripheral visual field (Bavelier et al., 2000) and enhanced neural responses have been reported under conditions of peripheral compared with central attention in congenitally deaf versus hearing controls (Neville & Lawson, 1987). Also, deaf signers appear to be faster at reorienting their attention compared with hearing controls (Parasnis and Samar, 1985). It has been suggested that sign language usage (and consequent analysis of hand motion) might alter motion processing (Bavelier et al., 2001). Bottari et al. (2010) have inclusively shown that the faster reactivity to visual events in the deaf happens regardless of eccentricity – i.e., both centrally and peripherally on the visual field. In addition to their visual abilities, deaf individuals also appear to outperform controls in tactile sensitivity (Levänen & Hamdorf, 2001), presumably reflecting neuroplasticity and/or attention increase directed to the stimuli (a monotonous sequence of vibratory stimuli). Levänen & Hamdorf (2001) tested for frequency discrimination and detection of random suprathreshold frequency changes, and discovered that congenital deafness enhances the accuracy of suprathreshold tactile change detection. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 16 Overall, but particularly with vision, the stated differences could be easily understood as a compensation for the auditory loss, taking advantage of the auditory-visual convergence typically seen in hearing individuals (but just using vision). The field converges in a common finding: differences in the deaf are focused on abilities that can have a compensatory role regarding early auditory deprivation, such as peripheral visual attention (attention to motion) and orienting mechanisms that have resulted from the reorganization of sensory processing. Deaf individuals show cross modal-plasticity effects particularly on the visual functions that in hearing individuals work in tandem with auditory input. As vision and audition are two important senses needed to navigate through space and time, these compensatory effects can be useful in the perception of daily life danger and spatial orienting. In sum, these behavioural differences are seen mostly in the aspects in which both the deprived and the overtaking sense would provide useful input together (Bell et al., 2019), in order to provide functional advantages. To better understand how these functional and behavioural differences can be represented structurally in the brain, the next section will focus on anatomical studies and how they correlate with functional differences. 1.2.1.2. Deafness in humans: anatomical MRI findings Morphometric studies of the human brain have attempted to investigate anatomical changes between hearing and deaf individuals. Emmorey et al. (2003), using MRI and volumetric analysis, concluded that deaf and hearing individuals do not differ in gray matter (GM) volume of the HG, but deaf individuals show significantly larger gray matter-white matter ratios than hearing subjects in the HG, having less white matter amount in bilateral HG and other areas of the Superior Temporal Gyrus (STG). However, authors such as Tae (2015) and Amaral et al. (2016) contradict this by showing that the deaf have decreased regional GM volume in the left anterior HG, both inferior colliculi, lingual gyri, nuclei accumbens, and left posterior thalamic reticular nucleus in the midbrain compared to hearing, leading to the hypothesis of an underdevelopment of the AC in the deaf. Additionally, hemispheric asymmetries have been found in the subcortical visual and auditory brains of the deaf. Amaral et al. (2016) showed that the right thalamus, the right LGN, and the right inferior colliculus are larger than their left counterparts. As the right AC of the deaf has shown neuroplasticity by representing visual information (Almeida et al., 2015; Finney et al., 2001; Nishimura et al., 1999), these asymmetries suggest that the subcortical areas could be rerouting visual information to the AC. There are also studies with diffusionweighted imaging (Shiell & Zatorre, 2017) and measures of cortical thickness (Shiell et al., CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 17 2016) that correlate the structure of the right planum temporale, a typically auditory region, with visual ability in the deaf. Shiell & Zatorre (2017) found changes in fractional anisotropy, radial diffusivity, and mean diffusivity in the right planum temporale of the deaf, suggesting altered myelination density or crossing fibres. This correlated with the enhanced ability of the deaf to detect visual motion (explored in the next section), evidence of white matter reorganization in favour of visual enhancement. The same region showed increased cortical thickness in the deaf (Shiell et al., 2016). Overall, these differences suggest that the absence of general auditory perception leads to a less developed AC, particularly when it comes to gray matter in the left hemisphere (LH). However, we observe white matter changes in the right AC, as well as differences in cortical thickness and correlations with functional results, which is consistent with seeing more neuroplastic representations in the right hemisphere (Almeida et al., 2015). 1.2.1.3. Deafness in humans: Functional MRI findings Multiple functional MRI studies have indicated that people with sensory deprivation exhibit neuroplasticity and recruit their sensory cortices to process information related to their remaining senses. The neuronal reorganization of the brains from people with congenital deafness might lead to the enhancement of visual perception, and is considered a compensatory mechanism, attempting to provide some substitute for the lost modality (Heimler et al., 2014). Following the findings of behavioural studies, fMRI studies have been showing that the AC of congenitally deaf individuals is recruited for visual tasks. Finney et al. (2001) used a moving dot pattern as the visual stimulus and showed that areas A1 and Brodmann’s areas 42 and 22 (auditory association cortex) were processing that stimuli in the right hemisphere. Bola et al. (2017) used temporally complex sequences of stimuli (rhythms) presented in a visual or auditory modality (flashes for visual and beeps for auditory) in deaf and hearing participants, where both groups performed the visual task and the hearing group also performed the auditory equivalent task. Results suggested that the visual task activates the AC (peaking in the posterior lateral part of the high-level AC) in deaf participants, unlike hearing subjects, and that this activation pattern is similar to the one that the hearing group experiences while in the auditory modality of the task. The authors also disclosed increased functional connectivity in the deaf between the AC and dorsal visual cortex (area V5/MT, associated with processing of dynamic visual stimuli), which was not present for the hearing subjects. Along the same line of research, Almeida et al. (2015) implemented an fMRI experiment typically designed to map visual preferences in the visual cortex and showed that the location of a visual stimulus can be CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 18 decoded from the patterns of neural activity in the AC of deaf, but not hearing individuals, especially in locations within the horizontal plane and the periphery. Studies with visual motion (Retter et al., 2018) have found stronger direction-selective responses in the STS region in deaf participants and less significantly in the PAC. As previously mentioned, vision is not the only altered sense in the deaf population. Congenital deafness also affects how the brain processes somatosensation. For instance, the deaf present greater signal change in the rostrolateral HG than hearing individuals while processing somatosensory and bimodal stimuli (double-flash visual illusion induced by two touches to the face) (Karns et al., 2012). Neuroplasticity might lead to neural reorganization, but it does not necessarily lead to losing its original organizing principles. Literature has shown that despite cross-modal plasticity and sensory deprivation, the AC of the deaf still shows topographic organization: Striem-Amit et al. (2016) used fMRI to look at functional connectivity patterns in the deaf and identified topographic tonotopy-based functional connectivity structures in the AC and extending tonotopic gradients (in auditory core and belt, parabelt, extending to language and speech/voice sensitive regions). This suggests that topographical organization does not need sensory experience to develop and is not affected by brain plasticity. Overall, fMRI studies have shown that properties such as visual motion (Retter et al., 2018), position in the visual field (Almeida et al., 2015; Finney et al., 2001) and rhythm/frequency discrimination (Bola et al., 2017) are represented in early and/or associative AC of deaf humans due to mechanisms of cross-modal plasticity. 1.2.2. Animal models of deafness Animal models enable us to examine behaviour without variables or restrains that are attributed to humans (e.g. language). In humans, cross-modal and neuroplastic effects observed in congenitally deaf humans may result from sensory deprivation or from the altered linguistic experience, such as sign language (Bavelier & Neville, 2002). The use of animal models - e.g., cat studies - show us that cortical reorganization happens independently of language acquisition. Congenitally deaf cats have been considered a suitable model of human congenital deafness (Heid et al., 1998). They show superior performance in visual localization in peripheral visual fields and lower visual movement detection thresholds, compared to hearing cats (Lomber et al., 2010). Lomber and colleagues (2010) used psychophysical tasks and graded cooling, confirming a cross-modal reorganization of the deaf AC. They were able to CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 19 localize individual visual processes in portions of the AC, and particularly demonstrate the role of the superficial layers of the dorsal zone of the AC in enhanced visual motion detection in deaf cats. Moreover, in cats, as in humans, cross-modal plasticity is not restricted to vision. Meredith & Lomber (2011) showed that somatosensory and visual modalities participate in cross-modal reinnervation outside A1, in the anterior auditory field of early-deaf cats. Studies in mice using electrophysiological recording techniques combined with cortical myeloarchitecture also showed that the AC of deaf mice contained neurons that responded to somatosensory and visual information, and also that their primary visual area had an increase in size (Hunt et al., 2006). In sum, animal studies corroborate and share the major findings seen in deaf humans: superiority in localizing stimuli in the peripheral visual field, superiority in motion detection, and neural differences in the processing of both visual and somatosensory information (recruitment of the AC). These findings provide a background for our experiment: the behavioural and functional data clearly show that the AC of the deaf is recruited for visual tasks. More specifically, it processes spatial information regarding the location of visual stimuli. Here, we aim to further explore this cross-modal processing, by investigating if the AC of deaf individuals represents low-level spatial features of visual stimuli, and if these representations are organized in a retinotopic format. We will use fMRI as our neuroimaging method and pRF analysis as our modelling technique. These methods will be described below. 1.3. Functional MRI and its use in neuroscience Magnetic resonance imaging (MRI) is a technique that uses a strong magnetic field to create images of biological tissue - in our case, the brain. The MR scanner uses a series of changing magnetic gradients and oscillating electromagnetic fields, known as the pulse sequence, causing energy to be absorbed and emitted by the atom’s nuclei. Adjustments in parameters of the radio frequency excitation pulses and the magnetic field gradients make it possible to acquire information about structure (anatomical imaging), flow (perfusion imaging), or neural activity (functional imaging) (Logothetis & Wandell, 2004). In the case of functional MRI (fMRI), it allows for the identification of where in the brain particular mental processes occur, and to characterize those patterns of brain activation. This is possible by measuring changes in blood oxygenation over time, as its levels change rapidly following the activity of neurons in specific brain regions (Huettel et al., 2008). Neural activity leads to the consumption of oxygen, ATP, and glucose, followed by an increase in cerebral blood flow. There is a mismatch between cerebral blood flow and oxygen consumption CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 20 (leading to an oversupply of oxygen), which is one of the bases of the blood-oxygen-leveldependent (BOLD) signal. This BOLD signal is sensitive to alterations in deoxygenated haemoglobin levels and cerebral blood volume, and relies on the phenomenon of neurovascular coupling: the link between changes in neuronal activity and the constriction or dilation of micro vessels in the brain (for more on how to interpret the BOLD signal to make inferences about the neural signal see Logothetis & Wandell, 2004). Besides MRI, several other techniques are used in neuroimaging, such as electroencephalography (EEG), single-unit recording, positron emission tomography (PET), and magnetoencephalography (MEG). All of these differ in their spatial and temporal resolution, how they acquire information, whether their measurements are direct or indirect, and invasiveness. fMRI is a non-invasive technique that has a high spatial resolution, and because of that it is increasingly used in basic and applied neuroscience. The fMRI BOLD signal that is measured from each voxel (the basic functional unit, usually a 3x3x3 millimetre volumetric unit) represents the average activity of all the neurons within that voxel. Based on the topographic organization of different cortical and subcortical areas, neighbouring cells are characterized by similar receptive fields properties, and thus, their averaged response to stimulation of their receptive field will result in an increase of the BOLD signal. This, as well as the ability of fMRI to obtain signal from the whole brain, make fMRI imaging a useful and efficient method to map the topographic organization of the brain. Today it is one of the most important methods to measure spatial organization, which is what we aim to do. Engel et al. (1994, 1997) used fMRI 27 years ago to measure the retinotopic organization within the human visual cortex, using visual stimuli to map eccentricity and polar angle preferences: specifically, they used expanding rings and rotating wedges respectively. These stimuli were used to create travelling waves of neural activity in the retinotopically organized cortex (see stimuli in Figure 4). 1.4. Techniques for measuring visual field maps Detailed measurements of visual field maps in individual subjects with fMRI have been performed with several techniques, including the travelling wave retinotopy (TWR) technique and more recently population receptive field (pRF) modelling. TWR - or phase-encoded retinotopy - has been the gold standard paradigm for the last two decades. pRF is an innovative approach that allows for more accurate and complete mapping, and solves some concerns presented by conventional mapping methods. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 21 1.4.1. Phase encoded retinotopy (travelling wave retinotopy) TWR is a method used in fMRI experiments to measure retinotopic organization in the human cortex and is based on a traveling wave of neural activity within retinotopically organized visual areas, created by the presentation of a visual stimulus. Because of this retinotopic organization, the presentation of stimuli as the rotating wedges and expanding rings (Engel et al., 1994) can elicit a continuous traveling wave of neural activity in the visual cortex (Engel et al., 1997). The continuous and periodic variation of the visual stimuli location generates fMRI responses that vary systematically in their delay, which is measured as the phase of the sinusoid that best fits the data. While measuring eccentricity preferences (ring expanding from the fovea to periphery), we see that voxels that respond later in time represent more peripheral parts of the visual field, producing a traveling wave of activity moving from the posterior to the anterior part of the calcarine sulcus. This means that the visual field location representations at each voxel are estimated from the relative timing of that voxel's fMRI response. This timing is quantified as the phase of the fMRI signal, calculated by taking the Fourier transform of the voxel's time series (Engel et al., 1994; Engel, 2012). The two orthogonal dimensions needed to identify a unique location in the visual space (x, y) are polar angle (stimulus is rotating wedge) and eccentricity (stimulus is expanding ring), represented by two polar coordinates: an angle (theta, direction from the centre to the point), and a radius (distance from the centre to the point). Taken together, these two measurements specify the most effective visual field position (Engel et al., 1997; Engel, 2012; Sereno et al., 1995; Wandell et al., 2007) (see Figure 3). Figure 3 Visual Field Landmarks Note. Visual field landmarks and representation of eccentricity (radius) and polar angle (theta) meaning. F – fixation point; HM – horizontal meridian; UVM – upper vertical meridian; LVM – lower vertical meridian. From “Retinotopic Organization in Human Visual Cortex and the CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 22 Spatial Precision of Functional MRI”, by S. Engel, G. Glover and B. Wandell, 1997, Cerebral Cortex, 7(2):181-92. Adapted with permission from Oxford University Press. These stimuli were designed to maximally stimulate the primary visual cortex and elicit an fMRI signal modulation on the order of 1%-3%, or 15-20 standard deviations above background noise (Brewer & Barton, 2012). An example of retinotopic maps obtained by the TWR method is shown in Figure 4, where we see a selection of voxels that have a powerful response above a defined threshold of coherence, that is the correlation between the response of that voxel and the sinusoidal model, at the frequency of stimulus presentation (Brewer & Barton, 2012). Figure 4 Travelling Wave Retinotopy Maps – Medial view of the Visual Cortex Note. Color map shows the response phase at each location for polar angle (A) and eccentricity experiments (B). The solid white lines indicate the boundaries between visual areas. Only voxels with a powerful response at a coherence ≥ 0.25 are coloured. Regarding polar angle preferences, we can see that the lower vertical meridian is represented along the upper bank of the sulcus (A). We also see that foveal preferences fall along the posterior end of the sulcus, and periphery preferences are represented in more anterior locations in the calcarine (B). POS, parietal-occipital sulcus; Cal-S, calcarine sulcus. From “Visual Field Map Organization in Human Visual Cortex”, by A. Brewer and B. Barton, in Visual CortexCurrent Status and Perspectives, 2012, IntechOpen. Reprinted with permission. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 23 These phase-encoded designs using the TWM are informative about neural tuning properties and allow for an estimation of the entire visual field layout, measuring cortical field maps. The application of phase-encoded methods was particularly effective in defining field maps in early visual areas, which are characterized by neurons with small receptive fields that are mostly confined to one hemifield. However, this method is less efficient for mapping areas with large receptive fields that include the fovea (Brewer & Barton, 2012; Wandell et al., 2007). As we are not only analysing early visual areas but also the AC (that might have several receptive field sizes and visual preferences), pRF analysis seemed to be a more advantageous method. Below is a description of this technique and why it is preferable in our study. 1.4.2. Population receptive field (pRF) modelling pRF modelling is a computational imaging approach that builds models predicting the neural responses to a stimulus, and can be used in a wide range of conditions (in our case, reconstructing cortical visual field maps). It couples fMRI signals (at millimetre scale) with receptive field properties of visual neurons (at micron scale). pRF estimates not only the visual field map, but properties such as receptive field size, laterality, and surround suppression (Dumoulin & Wandell, 2008), and is estimated for individual subjects. These RF properties change systematically across eccentricity and between visual field maps (Fracasso et al., 2016). The key pRF parameters are receptive field position and receptive field size, units that can be compared over different instruments and different subjects. The shape of the receptive field is a two-dimensional circularly symmetric (isotropic) Gaussian in the visual field, described by field position (x,y) and spread (s), both in visual degrees. For each voxel, the pRF parameters are adjusted to match predicted and measured fMRI time series (Wandell & Winawer, 2015). pRF analysis can be used in experiments where the traditional mapping stimuli are used, but also with a series of bar patterns that sweep through the visual field in different directions, or a series of stimuli placed at different visual field positions. Figure 5 demonstrates that pRF size increases along the visual hierarchy (from V1 to V3) and, simultaneously, within a given visual area there is a positive correlation between pRF eccentricity and size. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 24 Figure 5 pRF Size Variation With Eccentricity Note. pRF size increases with eccentricity and along the maps in visual areas, with smaller pRF sizes in V1 and larger in ventral and lateral visual areas (A). The increasing radius of each circle, representing pRF size, in each eccentricity position and visual area (B). From “Computational neuroimaging and population receptive fields”, by B. Wandell and J. Winawer, 2015, Trends in Cognitive Sciences, 19 (6), 349-357. Reprinted with permission from Elsevier. 1.4.2.1. pRF analysis: advantages and applications Although the pRF method was developed for retinotopic mapping in the visual cortex (Dumoulin & Wandell, 2008), it has had other applications such as measuring tonotopic maps and estimating bandwidth for voxels in the human AC (Thomas et al., 2015). Applications on the study of visual cortical responses using pRF analysis include research on clinical conditions and cognitive task demands (Wandell & Winawer, 2015). Examples are studies on attention (Sprague & Serences, 2013), plasticity of the adult visual cortex (Papanikolaou et al., 2014), developmental plasticity (Haak et al., 2014), psychiatric and neurological disorders such as dementia (Brewer & Barton, 2012) and autism (Hadjikhani et al., 2004; Schwarzkopf et al., 2014). The advantages of pRF over TWR include the fact that pRF analysis is more precise in the case of visual field maps with large receptive fields. Moreover, pRF analysis provides not only the preferred center for each voxel’s pRF but also pRF size/spread. Furthermore, it also provides information on laterality. These are the reasons why pRF was chosen for this study, as it reveals topographical organization more clearly than conventional methods and gives us access to additional information (Brewer & Barton, 2012). CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 31 Figure 7 Retinotopic Organization in the Visual Cortex of Hearing and Deaf Participants Note. Eccentricity and polar angle representations along the early visual cortex. A. eccentricity. The colour blue represents the centre of the visual field (0 visual degrees), up to red (periphery, 6 or more visual degrees) B. polar angle (ranging from -π (red) to π). RHright hemisphere. LHleft hemisphere. Threshold is explained variance ≥ 0.20. Dashed lines delineate the calcarine sulcus and possible borders between V1 and V2. pRF scaled model including only positive beta values. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 32 In this analysis, we expected to see strong visual preferences and organization in early visual areas. We clearly see eccentricity preferences ranging from the fovea to the periphery along the axis of the calcarine sulcus in the early visual cortex. Regarding polar angle, we see a clear contralateral representation of the visual field (RH represents left visual field), and the expected mapping of the stimulus regarding the y/vertical axis. In fact, polar angle is one of the clearer measures to define the borders between V1, V2, and V3. This is particularly discernible for the participants deaf 01 and hearing. Next, we turned to explore the negative pRF model, by comparing the extracted parameters between the negative and positive models. First, we looked at the variance explained by each one of the models, focusing on early visual areas (see Figure 8). The maps show that both models perform similarly and provide good fitting of the data in the early visual cortex. The negative pRF model covers additional areas compared to the positive model, mainly in medial parietal areas (see Szinte and Knapen (2020) for a similar result). Figure 9 shows the same variance explained maps in a lateral view including the AC. Figure 8 Variance Explained in Hearing and Deaf Participants – Medial View of the Visual Cortex Note. Variance explained in a medial view of early visual areas of the 3 subjects. Threshold is r 2 ≥ 0.08. pRF scaled model including positive and negative beta values (A) and only positive beta values (B). RH – right hemisphere; LHleft hemisphere. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 33 Figure 9 Variance Explained in Hearing and Deaf Participants – Lateral view of the Brain Note. Variance explained in a lateral view of the brain, including the AC. Threshold is r2 ≥ 0.08. pRF scaled model including positive and negative beta values (A) and only positive beta values (B). The hearing subject does not show differences between models, but the deaf subjects show more clusters of significant explained variance in the model that includes negative beta values (A). RH – right hemisphere; LHleft hemisphere. Overall, we see that the hearing subject does not appear to show clear differences between the two models, but the deaf show some more variance explained in the model that includes negative beta values. Thus, it seems that potential retinotopic responses in the AC are better captured by the negative pRF model, which led us to focus on the model that includes both positive and negative pRFs in the following analysis steps. To further characterize the responses to the visual stimuli we plotted the beta values of the best fitted predictor at each cortical location, colour coded by its sign (Figure 10A and 10B). This figure shows that in early visual areas the beta values are predominately positive across the participants, with negative values at the medial parietal cortex (in areas of the DMN, similar to Szinte and Knapen, 2020). On a lateral view (Figure 10A), deaf (but not the hearing) individuals show more representations related to negative betas values, mainly in the prefrontal and temporal lobes. The difference between the hearing and deaf participants is more evident when plotting the negative and positive values in separate maps (Figure 10C and 10D). This is particularly observable in deaf 01, which shows pertinent clusters in the STS, STG, and HG, but also in deaf 02. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 34 Figure 10 Beta Values (Slope) in Deaf and Hearing Participants Note. Slope (beta values from ≤ -4 to ≥4, warm colours code positive values and cold colours code negative values), in the lateral (A) and medial (B) view in the model with positive and negative beta values, and lateral views of the masked model with just positive beta values (C) and with just negative beta values (D). Threshold is r2 ≥ 0.08. The findings of Figure 10 lead us to explore which visual features are represented in the auditory areas delineated by the negative model. Since visual responses were seen only in the AC of the deaf and only for negative pRF tuning, we focused on these subjects in the last analysis, aiming to explore which visual features are represented in the AC of the deaf. For that, we extracted the parameters representing polar angle, eccentricity, x location, and pRF size (see Figure 11) only in voxels with negative beta values, as it is the predominant representation in the deaf (but not in the hearing). CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 35 Figure 11 Polar Angle, Eccentricity, Horizontal Location and pRF Width in the Deaf Note. Visual representations in the AC of the deaf participants. These parameters were tested with the masked model with just negative beta values. A. Polar angle ranges from -π (red) to π. B. In eccentricity preferences, the colour blue represents the centre of the visual field (0 visual degrees), extending until red (periphery, 6 or more visual degrees). C. The horizontal visual location was measured on a scale to ≤ -6 to ≥ 6. D. pRF size values are between 0 and ≥ 5. Threshold is r2 ≥0.08. Dotted lines delineate the STS and HG. We can draw several conclusions regarding these maps: overall, explained variance values are not high in the AC, and although the RH shows stronger responses than the left, there are no clear retinotopic maps. Then, in polar angle, we see representations of the left visual field in the RH, which is contralateral and in accordance with what happens in the visual cortex. The LH does not show clusters large enough to draw any conclusion. Regarding eccentricity, the maps suggest that the preferred area represented in the AC is mostly centre of the visual field, which is not what we expected given the trend towards periphery CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 36 representations seen in the literature. X position shows more clearly what we had already seen in polar angle: there is a notable preference for representing the left visual field, particularly in the RH. Width does not seem to show a consistent gradient, but it shows a tendency for high values, meaning a large receptive field size. It also does not seem to be positively correlated with eccentricity preferences. These clusters in the deaf appear in the RH along the STS, STG, the inferior circular sulcus of the insula and there is also a cluster in the Transverse Temporal Gyrus (TTG), which corresponds to the Heschl's gyrus. This means that they are seen both in early and associative auditory areas, but more significantly in the associative. 4. Discussion and Conclusion This thesis aimed to investigate cross-modal plasticity in congenitally deaf participants: firstly, by replicating previous results indicating that there are representations of visual information in the AC of the deaf but not the hearing; secondly, by attempting to map these representations, and more specifically, uncover retinotopic features of the AC (if any) that are similar to the visual cortex. This set of single cases included an fMRI experiment with stimuli typically used in classic retinotopic experiments designed to map the visual cortex. We analysed the data with a traditional travelling-wave method, and then focused on our elected technique: the pRF analysis. The results have demonstrated that there are visual-related responses in the AC of the deaf following a cyclic visual presentation, but not in the hearing. Concerning the retinotopy features, several aspects are of relevance: there are notably more representations in the RH/right AC of the deaf than in the left, and these clusters represent information from the contralateral visual field, therefore corresponding to the left visual field and in accordance with cortical visual processing. This anatomical laterality (predominancy of the RH) in the deaf is coherent with literature on hemispheric asymmetries, where anatomical studies have shown increased white matter volume in the RH compared to the LH (Amaral et al., 2016; Tae, 2015). It also supports fMRI studies stating the supremacy of the right AC of the deaf in representing visual information (Almeida et al., 2015; Finney et al., 2001), finding also present in connectivity and cortical thickness studies: Bola et al. (2017) revealed functional coupling between the AC and area V5 in the deaf (an area responsible for the processing of dynamic visual stimuli), and Shiell et al. (2016) have suggested a behaviour-structure correlation in deaf participants where the ones with better performances at visual motion detection have increased cortical thickness in the right planum temporale. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 37 Regarding the maps representing polar angle and eccentricity, although we do not see clear retinotopic gradients fully covering the visual field, the responses in the AC share some retinotopic features: pRFs in the AC represent information from large parts of the contralateral visual field, and the pRF centres are mostly localized to the foveal or near central part of the visual field. Functional laterality is reinforced by the map of horizontal position, supporting this contralateral representation of the visual field in the RH. pRF size (width) is overall large in the auditory areas but there is no gradient or positive correlation with eccentricity, as it happens with RFs in the visual system (Dumoulin & Wandell, 2008). In fact, if there is a correlation, it tends towards a negative one: we see large width pRFs with centres localized in the fovea, meaning that they represent large parts of the visual field. Altogether, the values of explained variance are low in the AC of the deaf, compared to the values in the visual system, and deaf 01 clearly shows more clusters in the maps than deaf 02. However, the explained variance values are still higher than the values found in the hearing participant, and higher than the average values obtained at non-visual areas (white matter). Moreover, these clusters appear in anatomical auditory regions that have previously been associated with neuroplasticity following sensorial deprivation: the HG and the STS (Almeida et al., 2015; Daphne Bavelier et al., 2001; Finney et al., 2001; Retter et al., 2018). To give an example, Retter et al. (2018) used visual motion stimuli (random-dot kinematograms) in an fMRI experiment and reported that the PAC of deaf (and not the hearing) shows directionselective visual motion responses, and that the right STS is extensively recruited for that processing. Bavelier et al. (2001) corroborate this recruitment of the posterior STS in a similar task. These areas belong to the early and associative AC: while the HG is the first cortical structure to process auditory information and is part of the PAC, the STS has been considered the main region for audio-visual integration, although it has also been associated with biological motion perception, speech processing, processing of faces and theory of mind (Hein & Knight, 2008). A critical finding in our results is that the visual-related responses in the deaf were revealed through the application of the “negative” pRF model - i.e., with the negative beta values. We can discuss the meaning of these negative values, that are associated with negative BOLD signal and therefore would potentially represent neural deactivation or suppression. Negative cross-modal modulation has been reported previously in the literature with hearing population, i.e., visual activity modulating auditory activity in the form of deactivation in the AC (Laurienti et al., 2002; Mozolic et al., 2008). Although some areas of the cortex (polysensory areas) can be involved in a variety of higher-order multisensory perceptions, sensoryspecific cortices (such as visual and auditory) do not function as independently as it was considered in the past. Laurienti et al. (2002) used fMRI in a passive stimulation paradigm CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 38 with 3 conditions (a visual stimulus alone, an auditory stimulus alone, and a combined visualauditory stimulus) and concluded that visual stimulation resulted in negative BOLD signals in the AC, namely along the superior and middle temporal gyri with a peak in Brodmann area 22 (located in the PAC). A similar effect occurred with the auditory condition, as negative BOLD was observed in voxels of extrastriate visual areas, but not for the combined stimulus condition. Additionally, it has been proposed that the difficulty of the task determines the degree of deactivation (Hairston et al., 2008). This is also true for intra-modal deactivations (Wilson et al., 2019), where negative BOLD responses are elicited within the stimulated cortex, as unilateral stimulus can deactivate the ipsilateral sensory cortex (e.g., unilateral hand movements activate the contralateral sensorimotor cortex and deactivate the ipsilateral cortex (Allison et al., 2000); early visual areas that are tuned to foveal representation respond with positive BOLD signal to foveal visual stimuli and with negative BOLD signal when the stimuli are presented in the periphery, and vice versa (Shmuel et al., 2006). This task-specific deactivation however is controversial and is not reported in all brain imaging studies using single sensory stimuli (Laurienti et al., 2002). This can be due to different experimental conditions in distinct contexts and with different tasks, and because most studies focus on brain activation and positive modulation results instead of negative BOLD signal and deactivation. We see this contrast in our dataset, among the deaf and hearing participants. Given that all participants were presented with the same stimuli in an equal context, differences between deaf-hearing could arise due to their own functional and anatomical brain properties. It would be erroneous to generalize this hypothesis since a group analysis has not been performed (which would require more participants), but it is an assumption to further explore. Besides, if deactivation/suppressing nonrelevant information contributes to the maximal perceptual integrity and attention to sensory information (Hairston et al., 2008), one could argue that this deactivation is involved in the superior performance of the deaf in some visual tasks, as previously mentioned. They would direct more attention to the task than the hearing participants and therefore exhibit superior abilities. As discussed in the introduction of this thesis, the answer is perhaps found in the middle: enhancements of the congenitally deaf in non-deprived senses result from attention reorienting and neuronal reorganization. Also, we could hypothesize that given that the pRF centres are located near the fovea in the AC of the deaf (and the peak of deactivation happens with foveal representations), this deactivation might sharpen their ability to distinguish between central and peripheral visual representations. Nevertheless, intra and cross-modal deactivations between sensory cortices are not the only occasions where negative BOLD is present. Various studies regarding the DMN mention deactivation, negative BOLD signal, and inclusively use pRF as an analysis technique. As a matter of fact, Szinte & Knapen (2020) coined the term “negative pRFs” to refer to signals CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 39 better predicted by deactivation. They used fMRI and pRF analysis to prove that the DMN selectivity deactivates as a function of the position of a visual stimulus, and that it acts as a negatively modulated high level visual network. This shares some similarity with our results (both in deaf and hearing but mostly in deaf) in medial parietal areas, belonging to the DMN (see figure 10B) - that englobes regions in the lateral prefrontal, posteromedial, inferior parietal cortices, as well as regions in the lateral and medial temporal cortex. They also stated, as we did, that the signal resulting from visual areas (both high and low level) is best explained with positive pRF models, so models predicting a positive amplitude modulation (positive slope). Furthermore, they also perceive that these voxels preferably represent the contralateral visual field (as the AC of our deaf subjects, RH) and their RF sizes increased with eccentricity (retinotopy feature that we could not observe). DMN areas typically show activation during rest (Raichle 2001) and self-related high-level cognitive tasks such as episodic and semantic memory or mind wandering and deactivation during attention-demanding and externally oriented tasks (Alves et al., 2019). For instance, motor movements such as hand and foot movements deactivate the DMN (Nakata et al., 2019). Another aspect we could reflect on is the matter of controlling attention maintenance, by turning an apparently passive task into an active one. In this experiment, the participants viewed the stimuli on the screen and solely had to fixate on the central fixation point. Other identical experiments have used behavioural tasks such as clicking on a response button to report motion direction changes (see Retter et al., 2018), or monitor the stimuli to report the changes in the end of the block (see Bavelier et al., 2001) or even click on a button when the colour of the fixation dot changed (see Szinte & Knapen, 2020). It remains undisclosed whether adding such tasks that modulate spatial attention would increase attention recruitment or attention maintenance throughout the runs of our dataset, and their effects in the amplitude of the BOLD results, either positive or negative in regard to the baseline. We could hypothesise that a larger attention recruitment would provide clearer results. Various steps can be performed to further explore how the visual cortex of the deaf use additional resources from the brain, namely the AC. In future research, collecting a larger sample might strengthen the results and potentiate a group analysis, contrasting the hearing and the deaf, and different parameters. Even if the pRF model was applied at an individual level, more participants would permit comparisons and generalizations. This would be particularly informative given that deaf 01 and deaf 02 do not behave exactly the same, and it would allow to potentially reinforce the results of deaf 01. Then, a Region of Interest (ROI) analysis with selected areas in the visual and auditory cortices would provide a more detailed statistical analysis of the data. Additional experiments could include repeating this study with other CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 40 sensory modalities, such as searching for negative BOLD in the visual cortex of the blind and looking for tonotopic organization in this recruitment or deactivation. Research on neuroplasticity is relevant to neuroscience and neuropsychology both for its contribution to basic science, i.e., for the state of the art on the brain’s topographical organization and reorganization, but also for applied neurorehabilitation purposes: devices such as cochlear implants (CIs) can be developed to compensate for the sensory loss in the deaf population. These implants convert auditory signals into electric impulses delivered to the acoustic nerve, replacing normal cochlear functions, and require a deep comprehension of the deprived brain’s ability to reverse changes. Different areas and functions have their own windows of sensitivity and susceptibility to plasticity. Some mechanisms may be available throughout life (cortico-cortical plasticity), but others are hardly maintained in adults (subcortical plasticity) (Bavelier & Neville, 2002). This raises the question of how cross-modal plasticity will interact with the presence of auditory input. If the AC has reorganized to process vision or touch, will neuroplasticity be detrimental for CIs? Human and animal data suggest that it interferes with the resettlement of the regained auditory inputs (Kral et al., 2006; Kral & Sharma, 2012; Sharma et al., 2014). Thus, it can be hypothesised that cross-modal plasticity is adaptive for sensory deprivation but maladaptive for sensory recovery. Heimler et al. (2014) have suggested a more balanced framework for this impact, stating that cross-modal plasticity might also have maladaptive outcomes in sensory deprivation - being related with impaired functions in the remaining senses, namely temporal tactile processing (Bolognini et al., 2012) - and potential adaptive outcomes in sensory restoration – since functionally selective plasticity might facilitate the recovery of specific cognitive functions (Hassanzadeh, 2012). Sensory rehabilitation programmes should consider multisensory substitutions trainings for the deaf, as they more successfully lead to sensory recovery across the lifespan (Heimler & Amedi, 2020). This thesis was a set of case-studies that aimed at investigating cross-modal plasticity in the AC of congenitally deaf participants through an fMRI experiment using pRF analysis. Results showed retinotopic-related responses predominantly with negative BOLD signal in the AC of the deaf but not in the hearing. Understanding how the brain is organized and reorganized through neuroplasticity contributes for the state of the art in neuroscience, and for a better and more independent experience of the world by people affected by sensory loss. CROSS-MODAL PLASTICITY IN THE CONGENITALLY DEAF 47 Annexes Figure A1 TWR analysis – phase maps Note. Phase maps. Parameters were the correlation coefficient and the phase value. This analysis was performed separately for (A) eccentricity (ring) and (B) polar angle (wedge) representations. The phase of the response is represented by a colour code, overlayed on the anatomical data. The colour scale in A has blue representing the fovea and red the periphery of the visual filed. The colour scale in B ranges from 0 to 2π. The threshold is coherence ≥ 0.5. RH – right hemisphere. LH – left hemisphere.