Basic Science

Nature vs. Nurture Factors in Shaping Brain Specializations


Our basic science work focuses on the principles that drive specializations in the human brain and their dependence on specific experiences during development (i.e. critical/sensitive periods) versus learning in the adult brain. I will specifically focus on studying nature vs. nurture factors in shaping category selectivity in the human brain (ERC project BrainVisionRehab). A key part of the project involves various technologies (see BrainTech technologies), including:


  1. The use of visual-to-auditory Sensory Substitution Devices (SSDs)
  2. A speech-to-touch sensory substitution approach which improves the ability of the hearing-impaired to perceive sound in noisy environments.
  3. I will discuss the technology, behavior and neural correlates
    of these novel sensory


Our basic science work strongly encourages a paradigm shift in the conceptualization of our sensory brain by suggesting that visual experience during critical periods is not necessary in order to develop anatomically consistent specializations in higher-order ‘visual’ or ‘auditory’ regions. This also has implications for rehabilitation by suggesting that multisensory, rather than unisensory, training might be more effective. Our new ERC grant, “How experience shapes the brain” – the NovelExperieSense project – studies nature vs. nurture factors in shaping topographical maps. Initial results from the grant’s various projects suggest that the same conclusions are also applicable to early sensory areas like retinotopic and tonotopic areas. (Key references: Hiemler & Amedi, “Emerging Notions on Brain Specializations and Critical Periods Resulting from the Studying of the Blind Brain with Sensory Substitution.” Neuroscience Biobehavioral Reviews 2020; Amedi et al, “Task Selectivity as a Comprehensive Principle for Brain Organization.” Trends in Cognitive Sciences, 2017).


(“The best technologies make the invisible visible.” – Beau Lotto). The second ERC project also explores novel ways of transmitting non-invasive invisible topographical information to individuals with normal senses by using similar training and SSD protocols to couple it with input from ‘invisible’ sensors, like infrared or ultrasound images, and testing whether novel topographical representations can emerge in the brain to input that was never experienced during development or evolution and whether this type of putative specialization will be consistent across the board among different individuals who become experts in using such devices.


The Neural Recycling Theory and the Brain as a Sensory Independent Task Machine


Highlights of our Work:


Our work aims to unravel the properties driving the sensory brain organization and uncover the extent to which specific unisensory experiences during critical periods are essential (or not essential) for the development of the natural sensory specializations. Our work focused on two fundamental discoveries:


  1. Using the congenitally blind adult brain as a working model of a brain developing without any visual experience, we documented that essentially most, if not all, higher-order ‘visual’ cortices can maintain their anatomically consistent category-selectivity (e.g., for body shapes, letters, numbers or faces) even if the input is provided by an atypical sensory modality learned in adulthood, and that such task-specific sensory-independent specializations emerge after few hours of specific training (e.g. Abboud et al., Nat Comm 2015; Amedi et al Trends Cog Sci 2017). Our work strongly encourages a paradigm shift in the conceptualization of our sensory brain by suggesting that visual experience during critical periods is not necessary in order to develop anatomically consistent specializations in higher-order ‘visual’ regions. I will integrate this theory with a prominent theory in cognitive neuroscience, ‘neural recycling’, by the Dehaene lab, and will propose an integrated framework supporting the notion of the brain as a task-machine rather than as a sensory machine. as classically conceived. Within this framework, we also suggested the potential mechanisms underlying the emergence of sensory brain specializations:

a) Pre-programmed sensory-independent task-specific computations that each specialized area/network processes (e.g., 3D reconstruction of geometry in object related area independently of the sensory modality input);

b) Partly innate network connectivity biases linking each specific cortical area to `the rest of the brain (Heimler et al., 2015 Curr Opin Neurobiol; Hannagan et al., 2015 TICS). Our emphasis on the task-selective and sensory independent brain organization also led to a paradigm shift in rehabilitation by suggesting that multisensory rather than unisensory training might be more effective (e.g. Reich et al Curr Opion in Neuorl 2012; Heimler et al., 2015 Curr Opin Neurobiol).


The BrainVisionRehab project – ERC StG: Nature vs. Nurture in the Development of Category Selectivity in the Adult Brain


What is the main driving force shaping the emergence of category selectivity in the human brain? It is well established that higher-order sensory cortices are divided into regions that are highly anatomically consistent across individuals, and that respond to specific stimuli categories (e.g., in vision: broad division of labor between ventral (“what”) and dorsal (“where”) streams; within these streams, category selectivity to specific visual objects such as letters (Visual Word Form Area - VWFA), faces (e.g. Fusiform Face Area -FFA), or body parts (Visual Extra-Striate Area - EBA) has been repeatedly observed; in audition: category-selectivity to human voices; rhythm sequences perception, language and timbre selectivity, etc.).


To what extent do such specializations arise solely from specific unisensory experiences during critical periods? Our work on brain development without visual experience (i.e. the congenitally blind adult brain as a working model), has documented several key findings. For instance, we showed that higher-order ‘visual’ cortices do develop anatomically consistent category-selectivity (e.g., for dorsal/ventral division of labor as well as for specific stimuli categories such as body shapes, letters, numbers or faces) without any visual experience even if the input is provided by an atypical sensory modality learned in few hours of dedicated training during adulthood (Striem-Amit et al., Plos One 2011; Reich et al Curr Biol 2011; Striem-Amit et al., Neuron 2012; Curr Biol 2014; Abboud et al., Nat Comm 2015).


Our work during our ERC Starting Grant (ERC-2012-Stg 310809) promotes a paradigm shift in how we conceptualize our sensory brain by suggesting that visual experience during critical periods is not essential to develop anatomically consistent specializations in the studied higher-order ‘visual’ regions so far (see Amedi et al, TICS 2017, which also includes supportive results in the auditory system). This work has been published in over 30 papers in leading neuroscience and neuroimaging journals, and has yielded a theoretical framework that was presented in several recent reviews. We posit that two non-mutual exclusive principles underlie the emergence of task-selective sensory-independent brain organization in higher-order sensory cortices:

(a) Pre-programmed sensory-independent task-specific computations that each specialized area/network process (e.g., shape analysis for letter symbols independent of the sensory modality);

(b) Partly innate network connectivity biases linking each specific cortical area to the rest of the brain (Heimler et al., Curr Opin Neurobiol 2015; Hannagan et al., Trends Cog Sci 2015; Amedi et al., Trends Cog Sci 2017).


We also translated this theoretical basic science work to novel ways to perform rehab and translational work in sensory (mainly visual and auditory) impairments. This also resulted in the creation of devices and in >20 translational related papers. Our research suggests (contrary to the current prevalent approach), that multisensory, rather than unisensory, technologies and training might be more effective in cases of sensory restoration and rehab (e.g. Reich et al Curr Opion Neurol 2012; Heimler et al., Curr Opin Neurobiol 2015).


  • In collaboration with Porf. Stanislas Deheane & Prof. Laurent Cohen

Critical Periods theory vs. Reversed Plasticity Gradient theory (RVP theory)


In a recent review paper, we suggested a new way to look at the issue of critical periods. We revised studies with visual and auditory deprived/recovery populations to argue for the need to redefine the crucial role of unisensory-specific experiences during critical periods (CPs) in the emergence of sensory specializations. Specifically, we highlight that these studies, with an emphasis on results with congenitally blind adults using visual sensory-substitution devices, consistently document that typical specializations (e.g., in visual cortex) could arise in adulthood as well via other sensory modalities (e.g., audition), even after relatively short (tailored) trainings. Altogether, these studies suggest that:


  1. Brain specializations are driven by sensory-independent computations rather than by unisensory-specific inputs
  2. Specific computation-oriented trainings, even if executed during adulthood, can guide the sensory brain to display/recover core properties of brain specializations.

We thus introduce here the concept of a reversible plasticity gradient, namely that brain plasticity spontaneously decreases with age in line with CPs theory, but it nonetheless can be reignited across the lifespan, even without any exposure to unisensory (e.g., visual) experiences during childhood, thus diverging dramatically from CPs assumptions. (Key references: Hiemler & Amedi, “Emerging Notions on Brain Specializations and Critical Periods Resulting from the Studying of the Blind Brain with Sensory Substitution.” Neuroscience Biobehavioral Reviews 2020; Amedi et al, “Task Selectivity as a Comprehensive Principle for Brain Organization.” Trends in Cognitive Sciences, 2017).


The TopoBrain Project: Widespread Multiple, Mirror-Symmetric Topographical Maps as a Key Principle of Human Brain Organization

In recent years, we also investigated the representations of topographical maps for different sensory and high-level cognitive functions like mental imagery in the adult brain, ultimately strengthening the notion that topographical maps are another key principle of brain organization (together with category-selective organization). Specifically, following in the footsteps of data and methods from the visual system, using fMRI, we were the first to map a whole-body sensory-motor Penfield homunculus in primary somatosensory (Tal et al., Cereb Cortex 2017) and motor cortices (Zeharia et al., PNAS 2012; J Neurosci 2015). Using uni- frequency (developed by others) and multi-frequency spectral Fourier analysis (developed in our lab; Hertz & Amedi, Neuroimage, 2010), we also reported that topographic maps in the auditory system (Striem-Amit et al., 2011) and in the sensory-motor system (Zeharia et al., PNAS 2012) extend well beyond primary and secondary sensory cortices. For instance, using spectral analyses together with multi-voxel pattern analyses (MVPA) and a support vector machine (SVM), we showed that sensory-motor homunculi are much more widespread than classically conceived, and whole-body maps are also present in subcortical structures and even in the occipito-parietal cortex, of a total count of more than 12 sensory-motor homunculi to date (some already identified in animal studies and some novel and discovered by my team: Zeharia et al. PNAS 2012; J Neurosci 2015; Zeharia et al, submitted; Tal et al, submitted). In many cases, these cortical maps were organized in mirror-symmetric topographical organization similarly to the well-known multiple retinotopic mirror-symmetric preferences in the visual system, and all of them (cortical and subcortical) were further supported by corresponding topographic organization observed through resting state functional connectivity (rs-fMRI) analysis. Finally, we have preliminary results suggesting that topographical representations can also characterize higher-level cognitive functions. Specifically, we found a whole-body novel homunculi map in the occipito-parietal regions for entirely imagined body experiences (“The Imaginary Homunculus,” Tal et al., in prep).


  • In Collaboration with Prof. Tamar Flash & Dr. Nava Levit-Binnun


The ExperieSense Project – ERC CON: Nature vs. Nurture in the Development of Sensory Topographical Maps in the Adult Brain


Until now, we worked towards the unraveling of the key organizational principles of the human sensory brain, focusing on the stability and flexibility of the organization of our natural sensory systems at early and later stages of sensory processing. Our new project (ERC Consolidator Grant ERC-2017-COG 773121), which was recently approved for funding, concerns a more extreme case – namely, the extent to which the healthy human brain can develop brain specializations which will show anatomical consistency across participants, for entirely novel sensory experiences (NSEs) to which the brain was never exposed during evolution. One example is infrared (IR) vision conveyed through sound/touch or ultrasound perception.


Our central hypothesis is that specific computations (and/or specific connectivity profiles), rather than specific unisensory inputs, better explain the principles underlying brain organization. We thus propose that after intensive training on NSEs during adulthood, the brain would develop dedicated and anatomically consistent novel specializations in higher-order sensory cortices, and more specifically, in the cortical region that shares some basic computation-related properties and/or shares unique connectivity profiles with the novel task, regardless of the input modality (i.e., similarly to what it has been shown in the brain of blind and deaf adults for visual/auditory stimuli conveyed through atypical sensory modalities (Amedi et al., 2017). Even more intriguingly, will NSEs develop dedicated topographical biases? This question might appear far-fetched, as topographical maps are classically conceptualized as intrinsically unisensory and their development as strictly constrained by unisensory critical periods. However, a series of preliminary results from my lab with sensory-deprived populations suggest that even the arising of topographical organizations might be less constrained by unisensory critical periods than generally assumed. Recently, using rs-fMRI data, we successfully mapped large-scale retinotopy in the brain of sighted individuals and then used the resulting retinotopic activations as seeds to test whether a similar organization was present in the blind brain. Intriguingly, we found that similar resting state functional connections are preserved even without any visual experience in congenitally blind adults and even in microphthalmic patients (Striem-Amit et al, Brain 2015). Similarly, preserved topographical mapping was also shown for tonotopy in congenitally deaf adults (Striem-Amit et al., Sci Rep 2017). The classic theory would predict that such maps are merely epiphenomena and do not systematically impact functional brain organization. Crucially, though, another series of preliminary unpublished results from my lab with congenitally blind adults suggest that even the emergence of topographical biases (and perhaps also full topographical maps) might be driven more by specific computations than by specific sensory experiences. And furthermore, that the (sensory) brain might be topographical in nature and topography might be a self-emerging property even for perceptual experiences acquired in adulthood.


Specifically, our data show an intriguing double dissociation. Two tasks, namely Braille reading (Reich et al., in prep) and face recognition using visual-to-auditory Sensory Substitution Devices (SSD) learned in adulthood (Arbel et al., in prep), elicit the selective recruitment of posterior V1 (foveal region) while anterior V1 (peripheral region) is showing negative BOLD activations (reduction in the hemodynamic fMRI signal below the average baseline). On the contrary, we found anterior (low resolution peripheral) V1 activations in both congenitally blind and blindfolded sighted adults during navigation through virtual mazes using visual-to-auditory SSDs. This selective recruitment was observed only once participants managed to learn to successfully navigate via sounds (Maidenbaum et al., in prep). We propose that such double dissociation can be explained by describing the so-called foveal vs. peripheral bias in retinotopic organization as high vs. low shape resolution analyses bias (i.e., reading and face perception both require high-resolution shape analyses, similar to foveal tasks, while navigation requires low shape-resolution analyses, similar to peripheral tasks).


This suggests that large-scale ‘retinotopic’ biases may not necessarily be constrained by the input sensory modality and strongly supports the prediction that novel topographic biases will emerge in the sensory cortex best suited to process a given specific computation (e.g., sensory information carrying spectral or periodic features would elicit functional tonotopy-like activations in auditory cortex). However, these results do not demonstrate the emergence of full sensory-independent ‘retinotopic’ maps, only strong biases. If topography is a self-emerging property of brain organization, though, we should be able to observe the emergence of full novel topographical maps if the right technology and training is provided. Using the population receptive field (pRF) method (Dumoulin, NeuroImage, 2008) which models fMRI responses to capture the preferred location in space and tuning width of the underlying neuronal population for a given visual stimulus, we obtained preliminary results suggesting that well-trained congenitally blind SSD users do develop full novel topographical maps in occipital and parietal regions in response to visual-to-auditory SSD stimuli mimicking the visual stimuli commonly used for retinotopic mapping (Hofstetter et al., in prep).


These results, in turn, albeit preliminary, strongly suggest that contrary to classic predictions, it is indeed possible to develop full novel topographical maps in adulthood and, ultimately, that topographical organization might be a self-emerging and sensory-modality flexible principle of brain organization. We plan to further investigate the principles underlying the emergence of
topographical organization by extending our approach:

  1. Beyond visual cortices to the deprived primary auditory cortices;
  2. To the sighted brain, with participants learning to interpret visual/auditory information, and moreover, NSEs, through other sensory modalities. We predict, for instance, that NSEs carrying spectral information such as IR vision conveyed through sound/vibration would elicit topographical biases in the auditory cortex, independently of the sensory modality used. And if we convey IR vision through bidimensional sounds (i.e., conveying also shape or spatial information), novel topographies would emerge in the occipital (‘visual’) cortex. We also aim to test the stability and flexibility of topographical maps by expanding the perceived spectrum of existing senses: if we convey backward visual information through sound or ultrasound information through touch, would the typical topographical representation expand in order to represent the additional information as well? In addition, how will these putative novel (topographical) specializations interact/clash with the existing and typical sensory specializations?


  • In collaboration with Prof. Serge Dumoulin

Bridging the Gap Between Body, Mind and Emotion: Elucidating the Representations of Body Maps in the Brain's Self-Referential Network


In Collaboration with The Sagol Center for Brain and Mind, directed by Dr. Nava Levit-Binnun.

Mind body interventions (MBIs) are becoming increasingly prevalent in both daily and clinical settings. In these practices, there is extensive use of body-based techniques for achieving beneficial mental and emotional goals. In certain MBIs, the practitioner aims to reduce the unconstrained train of thought, termed mind-wandering or rumination, which has been associated with reduced emotional wellbeing. From a neurofunctional point of view, it has been linked to the neural system known as the default mode network (DMN), a well-established large-scale neural network. Indeed, MBIs have been shown to reduce both mind-wandering and DMN activity. This project aims to investigate the link between the DMN, body processing and emotion. The first goal of this project is to investigate the anatomical overlap between the default mode network and neural representations of the body. The second goal is to elucidate the functional connectivity of body representations found in self-related processing brain regions. The third goal is to look for body representations specific to mental imagery of touch and mental body scan, without motor or tactile stimulation. The fourth goal is to investigate the effect of mental body scan training on neural body representations and their functional connectivity patterns. The fifth goal is to draw conclusions that may aid in devising specific MBIs and body scan techniques to assist in the promotion of the mental and emotional benefit of the intervention. Meantime in the project we had re-discoved a new Homunculus predicted by Penfield. A full body somato-motor gradient (homunculus) was found in the precuneus, which is a node of the DMN. This homunculus is functionally connected to other body representations, including one in the insula which is associated both with body representation and affective regulation. Key references: Zeharia N., Hofstetter S., Flash T. and Amedi A.“A Whole-Body Sensory-Motor Gradient is Revealed in the Medial Wall of the Parietal Lobe,” Journal of Neuroscience 2 October 2019, 39 (40) 7882-7892; "Topographic Neural Specialization for Internally-Driven Whole-Body Tactile Stimulation” & "Unique Modulation of the Default-Mode Network by Facial Movements," both under revision by Neuroimage.