Periodic Reporting for period 2 - IntuitiveBCI (Next-generation Brain-Computer Interfacing to enable intuitive, skilled control)
Période du rapport: 2023-06-01 au 2024-11-30
Brain-computer interfaces (BCIs) that “decode” the user’s intent from the activity of hundreds of neurons are emerging as potential solutions to restore loss function to people suffering from neurological conditions such as paralysis. By mapping this decoded intent into command signals, BCIs have enable participants to control a computer cursor, a robot, or even a stimulator that reanimates their paralysed muscles restoring some ability to walk or use their hands. However, BCIs still face several challenges that need to overcome if they are to become clinically available solutions.
In the IntuitiveBCI project, we aim to help address one of these problems: building a decoder that allows us to accurately and robustly infer the user’s intent across many conditions, a main limitation of current BCI decoders that usually only work well during the very specific task they were built for. Since becoming a more proficient BCI user may not be too dissimilar to becoming a better guitar player, we are studying the contributions of key brain structures to learning, executing, and adapting a skill, with the end goal of building what we learn into future BCI decoders. Ultimately, we hope that our findings will advance our understanding of the brain and also lead to BCI decoders that afford better control even across different tasks.
In the IntuitiveBCI project, we aim to help address one of these problems: building a decoder that allows us to accurately and robustly infer the user’s intent across many conditions, a main limitation of current BCI decoders that usually only work well during the very specific task they were built for. Since becoming a more proficient BCI user may not be too dissimilar to becoming a better guitar player, we are studying the contributions of key brain structures to learning, executing, and adapting a skill, with the end goal of building what we learn into future BCI decoders. Ultimately, we hope that our findings will advance our understanding of the brain and also lead to BCI decoders that afford better control even across different tasks.
Our work so far has fallen in two broad categories: basic research into how the brain learns, controls, and adapts movement, and engineering work to build better BCI decoders, by developing new experimental paradigms that we are piloting.
Understanding how the brain controls movement: we have continued our research on a new approach to describe how the brain works not based on the activity of the individual neurons that make it up, but based on the collective activity of these populations of neurons. With this view, we have been able to show that even if each is unique, different animals produce the same collective activity when performing the same actions. This is an important finding both for understanding the brain and for BCIs, since it will allow building BCI decoders that work across individuals. In addition to this work, we have also shown that two key parts of the motor network in the brain work more close together to produce movement than traditionally thought, which we are further exploring in a new experimental paradigm we have built in to identify the contributions of many brain areas during naturalistic behaviour in mice.
Understanding how skills are learnt and adapted: here, we have complemented past experiments by our group and others by building computational models of how skills are learnt and later adapted, with the ultimate goal of testing them in BCI experiments we are piloting. Most importantly, we have shown that how someone learns a skill, not only how good they are at that skill, influences their ability to learn future skills. We have also learned that the motor system can adapt a learnt skill to a surprising degree without the need of changing the connectivity of its neurons, contrary to long-held assumptions in the field.
Engineering next-generation BCIs decoders: we have developed a new setup to perform BCI learning and control experiments in mice. This setup allows us to take information from hundreds of neurons from different parts of the brain and integrate them in order for mice to control a visual or an auditory cursor —not unlike how you move a computer cursor but directly using their “thoughts”. With this new setup, we are comparing BCI control from different brain structures involved in the generation of skills.
In summary, our research has shed light into how the brain learns, performs and adapts skilled movements, including uncovering fundamental commonalities across individuals. Our ongoing BCI work is setting the foundation for these principles to be applied for the development of novel BCI decoders that we hope will provide more accurate and robust control.
Understanding how the brain controls movement: we have continued our research on a new approach to describe how the brain works not based on the activity of the individual neurons that make it up, but based on the collective activity of these populations of neurons. With this view, we have been able to show that even if each is unique, different animals produce the same collective activity when performing the same actions. This is an important finding both for understanding the brain and for BCIs, since it will allow building BCI decoders that work across individuals. In addition to this work, we have also shown that two key parts of the motor network in the brain work more close together to produce movement than traditionally thought, which we are further exploring in a new experimental paradigm we have built in to identify the contributions of many brain areas during naturalistic behaviour in mice.
Understanding how skills are learnt and adapted: here, we have complemented past experiments by our group and others by building computational models of how skills are learnt and later adapted, with the ultimate goal of testing them in BCI experiments we are piloting. Most importantly, we have shown that how someone learns a skill, not only how good they are at that skill, influences their ability to learn future skills. We have also learned that the motor system can adapt a learnt skill to a surprising degree without the need of changing the connectivity of its neurons, contrary to long-held assumptions in the field.
Engineering next-generation BCIs decoders: we have developed a new setup to perform BCI learning and control experiments in mice. This setup allows us to take information from hundreds of neurons from different parts of the brain and integrate them in order for mice to control a visual or an auditory cursor —not unlike how you move a computer cursor but directly using their “thoughts”. With this new setup, we are comparing BCI control from different brain structures involved in the generation of skills.
In summary, our research has shed light into how the brain learns, performs and adapts skilled movements, including uncovering fundamental commonalities across individuals. Our ongoing BCI work is setting the foundation for these principles to be applied for the development of novel BCI decoders that we hope will provide more accurate and robust control.
By answering the long-standing question of “what is similar in the activity of the brains of two different people when they do the same thing,” we have set the foundation for using a BCI or any other neurotechnology that has needs a model that relates brain activity to movement, sensation, disease state, etc across individuals. This could accelerate the deployment of these emergent technologies as clinical products.
Our basic neuroscience work in skill execution, acquisition and learning also invites rethinking some of our assumptions about how the brain works. In addition to adding to the general effort of understanding the most complicated system we are aware of –the brain–, it may inspire new ways of thinking about the various neurological diseases that impair movement, perhaps inspiring new approaches to understand and even treat them.
Our basic neuroscience work in skill execution, acquisition and learning also invites rethinking some of our assumptions about how the brain works. In addition to adding to the general effort of understanding the most complicated system we are aware of –the brain–, it may inspire new ways of thinking about the various neurological diseases that impair movement, perhaps inspiring new approaches to understand and even treat them.