Neural interfaces are essentially engineered devices that directly link the nervous system to the outside world. Therefore, neural interfaces bypass the body, which is the normal intermediary connecting the nervous system and the world; for example, to speak your nerves must first activate your mouth and throat, and to hear sound waves must first enter your ear before being transduced by nerves into impulses that the brain interprets as sound. In the first example signals go from nerves to the body to effect the world (speech), and in the second the world goes to the body to affect the nervous system (hearing). Similarly, neural interfaces can operate in both directions. Our research involves both working with the nervous system’s outputs (through computer algorithms that try to decode neural outputs for brain-computer interfaces) and neural inputs (electrically stimulating nerves to improve bladder function). Some of our projects are described below:
Brain-Computer Interface Design and Validation
One of the challenges limiting the effectiveness of current brain-computer interfaces is the difficulty in designing an algorithm that can both decode the user’s intent from abstract neural signals and constantly adapt to the user’s changing strategies for operating the device. Our work uses a novel paradigm, based on artificial neural networks and computational neuroscience, that allows us to test many such algorithms in closed-loop with many users to determine which ones improve users performance the most, and why. Our studies have yielded insights into the limitations of focusing algorithm design on task error, and the potential problems associated with computer algorithms that learn faster than the human user.
Neural Control of the Lower Urinary Tract
All clinical strategies to treat lower urinary tract dysfunction, be they neuromodulation or pharmacotherapy, rely on a basic theory of how the system functions. By probing the neurophysiology of the urinary tract with animal models, our lab is refining our theory of urinary tract control and function, which will lead to new therapies and a improvements of existing treatments. We have discovered a number of features about how sensory information is represented and processed in the lower urinary tract, and how this could affect reflex control of the bladder and urethra.
Targeted Electrical Stimulation of the Lower Urinary Tract
As we age or are afflicted with neurological injury, our bladders no longer function effectively, prevents hundreds of thousands of people from properly emptying their bladder. We are using electrophysiology and behavioral studies in animal models to examine why the lower urinary tract loses its functionality as we age or suffer a spinal cord injury. We are delivering electrical stimulation of nerves that control the lower urinary tract to restore some lost urinary tract functionality. These novel stimulation strategies span different stimulation sites, such as the urethra and spinal cord, and different stimulation modalities, such as closed-loop stimulation triggered off bladder events or low-amplitude stimulation designed to boost neural sensitivity.
Co-Adaptation of Human and Artificial Learners
More often people have to learn how to perform complex tasks with the help of technology that is also adapting. We study how the dual-learning environment affects user control strategies and proficiency, and how we can better design artificial systems to optimize human motor learning skill. Our results have shown that simply changing visual cues, even if they are irrelevant to the task, can influence how people learn. This finding is especially important for brain-computer interface development because users must learn to operate a complex new system, and it is not known what feedback is most helpful for them during learning.