Seminars
The basal ganglia are often thought to select discrete actions. This talk presents evidence for a broader role: the basal ganglia help control continuous changes in the body and coordinate the sequence of smaller action elements that make up complex behavior. Recordings from mice show that different parts of the basal ganglia represent variables such as movement velocity, body position, and head orientation. These signals may serve as instructions for brainstem systems that maintain posture and guide movement. In this view, the basal ganglia operate within layered feedback loops, continually adjusting the body’s configuration as behavior unfolds. Experiments also show that separate basal ganglia circuits control different components of an action, including aiming, reaching, and licking. Activating one circuit can initiate or prolong a particular component while suppressing competing components, whereas another circuit can terminate the current component and help restart the sequence. These findings suggest that the basal ganglia function less like a simple “action gate” and more like a dynamic coordinator that organizes continuous movements and their transitions.
As we move through the world, the brain must distinguish sensory feedback caused by our own actions from motion imposed by the environment. This distinction supports stable perception, adaptive behavior, and a sense of agency. I will use self-motion to examine how the nervous system makes this distinction during natural behavior.Vestibular afferents provide complementary representations of motion: regular afferents convey information through a robust rate code, while irregular afferents provide temporally precise signals suited to rapid, dynamic behaviors. But these signals alone cannot indicate whether motion is self-generated or external. That distinction emerges centrally: vestibular nuclei neurons preferentially encode unexpected motion, and thalamic signals carry this information to cortical regions involved in perceiving self-motion.How does the brain identify which sensory feedback to expect? My recent work points to a cerebellar computation in which weighted combinations of Purkinje-cell activity transform motor, vestibular, and proprioceptive signals into a prediction of the sensory consequences of active movement. Comparing this prediction with incoming feedback attenuates expected, self-generated signals while preserving sensitivity to unexpected motion. This computation also adapts: when the relationship between action and sensory feedback changes, neural responses recalibrate alongside behavior.These findings link sensory coding and prediction to learning and agency. A match between predicted and actual sensory consequences may contribute to the sense that “I caused this.” The same principle matters for vestibular prostheses: the brain must learn how artificial input relates to self-generated movement and incorporate it into behavior. Together, this work shows how the brain uses predictions to distinguish the sensory consequences of our own actions from changes in the world around us.
