Error detection by the olivocerebellar system and its role in motor adaptation

The main aim of this study is to understand how the Inferior Olivary nucleus (IO) contributes to cerebellar dependent motor learning.

Motor learning is the process by which we acquire and store memories that encode movements 1. This is largely done by the cerebellum 2. By integrating information coming from the sensory systems and the spinal cord, it is able to assess the execution of each movement 2. When an actual motor output doesn't match the intended movement, changes are implemented 3. Cerebellar disorders are commonly associated with ataxias, that is a lack of control and coordination upon faster movements 4. The treatments available for cerebellar ataxias are limited with physical therapy being the most common form of care 5. With a deeper understanding on the neural computations underlying cerebellar learning, we can aid future efforts to understand and correct these conditions.

The IO is postulated to be the site that detects the mismatch between the predicted and actual movement, by receiving an internal copy of a motor command as well as sensory input and consequently re-afferent information on the actual motor output 6. Upon mismatch, it generates a sensory error signal thought to be used for the online control of movement by the cerebellum, to enable motor adaptation to the changing environment 6,7. Despite its central role in motor learning, how the IO encodes this sensory error signal is unknown 7.

Recording neural activity in the mammalian IO during a behavioural task, presents a series of complications and challenges. To overcome these challenges, this study will utilise the larval zebrafish as a model system 8. We can exploit its genetic amenability and optical transparency to use the latest imaging tools, one being Voltron 9. Voltron, a 'chemigenetic' voltage sensor will be expressed specifically in the IO neurons. A calcium reporter GCaMP 10, will be expressed brain-wide to report on neural activity across the entire brain. We will use the well-established Opto Motor Response (OMR), motor adaptation paradigm in a closed loop virtual reality setup 11,12. Since the stimulus will be in the experimenters' control, we can induce sensory prediction errors as the zebrafish is engaging in behaviour and have a comprehensive overview of IO and whole brain activity during this. The advantage to using Voltron enables us to not only see individual spike activity, but also subthreshold membrane fluctuations 9. We can relate both these aspects in IO activity to stimulus, behaviour and whole brain activity. Under the hypothesis that the IO encodes the sensory error signal used to guide cerebellar learning, we can use our techniques to test these key predictions: Firstly, we should record representations of the sensory environment and motor output in IO activity: The sign and amplitude of deviation of the actual movement from the predicted movement is seen in IO activity. Features of IO activity should predict the subsequent motor adaptation. Finally, by selective activation of the IO adhering to these features, we should 'trigger' the coinciding behaviour.