Periodic Reporting for period 1 - SOMATOGPS (The Body Positioning System: A GPS for somatosensory space)
Período documentado: 2023-04-01 hasta 2025-09-30
When a fly lands on your right hand, it is natural to swipe it off with your left. This behavior is so commonplace, and feels so effortless, that we hardly give it a second thought. However, this “simple” behavior (tactile localization) is actually an incredibly complicated problem the brain must solve on a daily basis. The body’s posture changes from moment-to-moment. The fly may can be at the same location on the skin but radically different locations in the three-dimensional space around you. Knowledge of body posture (proprioception) is needed to solve this problem, yet the brain initially processes proprioceptive and tactile signals independently. Furthermore, the relationship between these signals and the physical body is ambiguous. Despite this, the brain can somehow accurately pinpoint the fly in three-dimensional space and reach towards it.
Over a century of research has made progress at a conceptual understanding of the process, highlighting the importance of stored body representations and spatial coordinate systems. Yet we still lack a formal theory of somatosensory space, a crucial step in the scientific process. Understanding tactile localization on a computational level is important for making concrete progress on societal issues, such as designing prosthetics that can be embodied. SOMATOGPS introduces the first neurocomputational framework aimed at solving this mystery.
My novel approach leverages an analogy with computations used by manmade positioning technology in order to gain insight into potential solutions. Global Positioning Systems (GPS) ingeniously turn localization into a problem of geometry, pinpointing an object on Earth by calculating its distance from multiple satellites. I propose that the brain uses a somatosensory version of GPS—a Body Positioning System (BPS)—that reduces localization to its geometry. By keeping track of the distances between each body part and their distances from the fly, the brain could compute a reach to it. The main aim of SOMATOGPS is to characterize how the BPS is implemented by neural computations. I will develop novel behavioral and neuroimaging paradigms to measure localization in three dimensions, model the underlying geometric computations, and manipulate the geometry of the body in order to perturb these neural computations. This innovative proposal provides the first computational model of tactile localization, representing a new state-of-the-art in our understanding of somatosensory space.
Over a century of research has made progress at a conceptual understanding of the process, highlighting the importance of stored body representations and spatial coordinate systems. Yet we still lack a formal theory of somatosensory space, a crucial step in the scientific process. Understanding tactile localization on a computational level is important for making concrete progress on societal issues, such as designing prosthetics that can be embodied. SOMATOGPS introduces the first neurocomputational framework aimed at solving this mystery.
My novel approach leverages an analogy with computations used by manmade positioning technology in order to gain insight into potential solutions. Global Positioning Systems (GPS) ingeniously turn localization into a problem of geometry, pinpointing an object on Earth by calculating its distance from multiple satellites. I propose that the brain uses a somatosensory version of GPS—a Body Positioning System (BPS)—that reduces localization to its geometry. By keeping track of the distances between each body part and their distances from the fly, the brain could compute a reach to it. The main aim of SOMATOGPS is to characterize how the BPS is implemented by neural computations. I will develop novel behavioral and neuroimaging paradigms to measure localization in three dimensions, model the underlying geometric computations, and manipulate the geometry of the body in order to perturb these neural computations. This innovative proposal provides the first computational model of tactile localization, representing a new state-of-the-art in our understanding of somatosensory space.
The main goal of SOMATOGPS is to develop a novel framework to discover and characterize the neural computations underlying the Body Positioning System (BPS). This means going beyond a conceptual understanding of tactile localization and body representation, and aiming to understand these phenomena from a neurocomputational perspective. Three key objectives are helping guide our work in the direction of this goal.
Our first key objective is to develop novel paradigms to precisely measure localization in three dimensional space. We have made several acheivements for this objective. First, we designed a novel paradigm that combined VR and motion-tracking, which allowed us 3D body representation for the first time. We then extended this paradigm to allow for measuring all aspects of the BPS simultaneously. Finally, we adapted this approach to the EEG, which allowed us to make first steps towards understanding the BPS at the neural level . In all cases, these paradigms involved high-density mapping of space, a necessary step to characterize the properties of somatosensory space.
Our second key objective is to model the neural computations underlying the BPS. We have made several acheivements for this objective. All experiments, such as those mentioned above, involve using data to model the underlying computational components of the BPS. This has led to the first models of somaotsensory body representation, tactile localization, and the coordinate systems used by the brain. These acheivements were made possible by the unique high-density mapping approach we take in our paradigms.
Our final key objective is to manipulate these computations by changing how the brain represents the size of the body. We do so by training participants to use finger-extending exoskeletons. We have found that these exoskeletons stretch the user’s somatosensory finger representations, leading to a modification in the computations that localize touch on their fingers. Speifically, because distances within the finger representation are now larger, the BPS changes how it localizes touch in a manner predicted by our computational model.
In total, we are already making good progress on unraveling the BPS and how it takes sensory signals and creates a sense of self.
Our first key objective is to develop novel paradigms to precisely measure localization in three dimensional space. We have made several acheivements for this objective. First, we designed a novel paradigm that combined VR and motion-tracking, which allowed us 3D body representation for the first time. We then extended this paradigm to allow for measuring all aspects of the BPS simultaneously. Finally, we adapted this approach to the EEG, which allowed us to make first steps towards understanding the BPS at the neural level . In all cases, these paradigms involved high-density mapping of space, a necessary step to characterize the properties of somatosensory space.
Our second key objective is to model the neural computations underlying the BPS. We have made several acheivements for this objective. All experiments, such as those mentioned above, involve using data to model the underlying computational components of the BPS. This has led to the first models of somaotsensory body representation, tactile localization, and the coordinate systems used by the brain. These acheivements were made possible by the unique high-density mapping approach we take in our paradigms.
Our final key objective is to manipulate these computations by changing how the brain represents the size of the body. We do so by training participants to use finger-extending exoskeletons. We have found that these exoskeletons stretch the user’s somatosensory finger representations, leading to a modification in the computations that localize touch on their fingers. Speifically, because distances within the finger representation are now larger, the BPS changes how it localizes touch in a manner predicted by our computational model.
In total, we are already making good progress on unraveling the BPS and how it takes sensory signals and creates a sense of self.
SOMATOGPS has made methodological advances beyond the current state-of-the-art. All the tasks we use involve high-density mapping of somatosensory space, a practice that is not common in the field. However, without high-density mapping, accurately characterizing the BPS would not be possible. This is especially true of neuroimaging experiments, where somatosensory space is typically measured using two postures (e.g. uncrossed or crossed across the midline). By measuring several (between eight and ten) different spatial locations, we are able to precisely characterize somatosensory space at the behavioral and neural level. This approach is also necessary to leverage computational modelling when analysing our data, which is not commonly part of methodological toolkit of most somatosensory and body representation research. The explicit focus on modelling taken in SOMATOGPS reflects an important methodological advance. Furthermore, the models we build to characterize the body in space are more geometrically complex than traditionally. Our models of body representation and tactile localization are therefore the new state-of-the-art in the field.
SOMATOGPS has also made theoretical advances beyond the current state-of-the-art. This is thanks in large part to our emphasis on understanding body representation from a computational perspective. Our framework therefore goes above and beyond the emphasis on conceptual models, which has dominated the field for the last hundred years. We are making good progress on actually formalizing components of these models as well as demonstrating that certain assumptions made by conceptual models was unnecessary. For example, we have used models to challenge the assumption that body representations are spatially distorted and that the default reference frame of somatosensory space is centred on the trunk. We are currently forming the theoretical bedrock of a new state-of-the-art understanding of the body in the brain.
SOMATOGPS has also made theoretical advances beyond the current state-of-the-art. This is thanks in large part to our emphasis on understanding body representation from a computational perspective. Our framework therefore goes above and beyond the emphasis on conceptual models, which has dominated the field for the last hundred years. We are making good progress on actually formalizing components of these models as well as demonstrating that certain assumptions made by conceptual models was unnecessary. For example, we have used models to challenge the assumption that body representations are spatially distorted and that the default reference frame of somatosensory space is centred on the trunk. We are currently forming the theoretical bedrock of a new state-of-the-art understanding of the body in the brain.