A powerful idea in neuroscience links motor control with action observation. Recent work has demonstrated that when we observe the actions of others we activate the same neural circuitry responsible for planning and executing our own actions. While much of the work in this area has been centered around cognitive and social mechanisms such as action understanding, empathy and theory of mind, we have been pursuing a wholly different idea, namely that neural mechanisms linking observation and action also facilitate neuroplasticity in somatosensory and motor regions of the brain, and can result in behavioural benefits for motor skill learning.
We use computational models of neuromuscular systems such as the upper limb to test hypotheses about how the brain controls voluntary movement, and how motor learning is achieved. An emphasis is placed on including realistic physiological, mechanical and neural properties of the neuromuscular system. We use models to study the form of time-varying control signals to muscles that the central nervous system must generate to produce voluntary movement. We combine model predictions with empirical measurements of limb kinematics and patterns of muscle activation measured using electromyography. Recently we have begun to develop artificial neural network models that can be trained to control physiologically realistic neuromuscular models of upper limb movement (MotorNet). We use them to test hypotheses about how information for skilled movement and for motor learning is represented across a distributed network of computational units.
Whereas there has been extensive work on the neural mechanisms that subserve voluntary limb movement, comparatively little is known about how the motor system modulates the mechanical properties of the limb through the neural control of limb stiffness, e.g. by modulating reflex gains, and/or by co-contracting antagonist muscles around a joint. Our long-term goal is to understand how stiffness control is integrated into the ongoing control of movement and how it is used in an adaptive fashion during interactions with the environment. We view stiffness control as playing an active part in producing movements that differ in rate, trajectory requirements and accuracy and in maintaining stability when manipulating objects and interacting with the environment. We are using recurrent neural network models (MotorNet) to study how neural control signals for movement and stiffness are represented in a distributed network of computational units.
A large body of research has explored how adaptation in sensory systems (e.g. vision and proprioception) affects motor performance, however we know little about how motor learning affects the function of sensory systems. The goal of this project is to explore changes in visual and somatosensory systems as a consequence of motor action and in particular, motor learning. Experiments are designed to test the hypothesis that visual and somatosensory processing are modulated in specific ways with recent motor behaviour and are further modified after motor learning. The idea is that motor learning changes not only motor behaviour, but also changes how we sense and perceive our own actions, and the world around us.
We are interested in motor learning in speech production, in particular using experimental models of auditory perturbations such as formant shifts. This is a relatively new research thread in the lab. Aims include establishing links between sensory/motor adaptation in the upper limb and in speech production, and also studying the ways in which these two sensorimotor systems differ, in particular in the context of sensory/motor learning.
Our research is (or has in the past been) funded by grants from NSERC (Canada), CIHR (Canada) and NIH (USA).