Towards neurally guided physical therapy
MetadataShow full item record
Fine motor skills such as individual finger movements are impaired after neurological injury (e.g. stroke). Conventional therapy, operating at the limb, has limited success in rehabilitating fine motor skills after stroke. This work lays the foundation for guiding physical therapy of the hand from the brain, rather than from the limb. This work aims to answer three fundamental questions: (1) how do people learn to control their own neural activity? (2) how can we best decode patterns of neural activity related to individual fingers? (3) can people learn to shift the patterns of neural activity associated with each of their fingers? We first investigated how people learn to control their own neural activity. Neurofeedback experiments in the MRI scanner are expensive, time-consuming, and rely on human participants to learn to control their own brain activity. Minor errors in data processing, feedback delivery, or instructions to participants can ruin an fMRI neurofeedback experiment, without any indication as to what was the problem. Here, we investigate how the properties of the fMRI signal, feedback timing, and self-regulation strategies affect this learning, using a simulated neurofeedback environment to compare how participants' strategies interact with feedback timing. In an experiment with human participants playing a simple neurofeedback game with a simulated brain, continuous feedback led to faster learning than an intermittent feedback signal. However, in a computer model of automatic reward-based learning, intermittent feedback was more reliable. These results provide critical guidelines to the design of fMRI neurofeedback experiments. Next, we developed techniques to most accurately decode individual finger presses in real-time from fMRI recordings. The neural correlates of individual finger movements can be revealed using multivoxel pattern analysis (MVPA) of fMRI data. Neurofeedback of MVPA, known as decoded neurofeedback, has manipulated behaviors such as visual perception and confidence judgements. However, this technique has yet to be applied to sensorimotor behaviors associated with individual fingers. Here we investigated how best to decode patterns of neural activity from sensorimotor cortex in real-time. To set key parameters for the experiment, we used offline simulations of decoded neurofeedback using previously recorded fMRI data to predict neurofeedback performance. We show that these predictions align with real neurofeedback performance at the group level and can also explain individual differences in neurofeedback success. Finally, we investigated if people could learn to shift the neural patterns related to their own finger movements, and how this might affect fine motor skills. Deficits in individual finger movements after stroke are associated with weakened, overlapping neural activity patterns. Here we investigated whether neural activity patterns of fingers in sensorimotor cortex could be shifted using decoded neurofeedback in healthy individuals. This is meant to provide the groundwork to using neurofeedback on stroke patients, except in the opposite direction, by decreasing confusion between finger pairs instead of increasing confusion between them. We discovered that participants could learn to shift the pattern associated with their ring finger but not that of their middle finger. We also found that participants' finger movement behaviors changed in the ring and little fingers, but not in the index or middle fingers. Our results show that neural activity and behaviors associated with the ring finger are more readily modulated than those associated with the middle finger. These results have broader implications for rehabilitation of individual finger movements, which may be limited or enhanced by individual finger plasticity after neurological injury.