A cerebellar internal model calibrates a feedback controller involved in sensorimotor control

Check out our most recent work from Daniil A. Markov, Luigi Petrucco and, Andreas M. Kist, who investigated the role of the cerebellum in long-term motor adaptation using behavioral experiments and light-sheet imaging in the whole brain and the cerebellum of behaving fish larvae.

When animals move, they receive visual feedback from their locomotion. In this video, a larva swims in a closed-loop experiment in response to a moving grating. When the larva swims, the grating moves backward. If the feedback is slower or delayed, the swimming bouts become longer.

The cerebellum is a structure that is regarded as important in using sensory feedback to predict the consequences of our own actions. With some transgenic lines created by Andreas M. Kist, we can destroy all Purkinje cells and see what happens to the behavior.

Surprisingly, even when messing so badly with the fish cerebellum, the ability to adapt to unexpected perturbations of the feedback does not appear to change!

But… If the feedback is perturbed consistently over a long time, larval zebrafish adapt their behavior in a long-term manner. This time, the long-term adaptation turns out to be cerebellum-dependent!

We conclude that optomotor swimming and its acute adaptation can be implemented by a feedback controller. We modeled such a feedback controller and showed that it can indeed implement acute adaptation data.

What about the cerebellum then? In long-term adaptation, the cerebellum could lead to a recalibration of the feedback controller parameters, in particular sensory integration time constants.

To prove this hypothesis, we performed imaging in the whole brain and in the cerebellum of fish while they were performing long-term adaptation. During the task, we can find sensory and motor responses in different parts of the brain.

In the long-term adaptation, Purkinje cells underwent slow changes in their responses during the experiment in the group of learner fish, but not in non-learners and controls.

Finally, the time constants of neurons integrating directional motion over time were changing in the long-term adaptation, confirming the hypothesis that the implementation of long-term motor adaptation could come from the recalibration of sensory integration time constants.

That’s it! If you want to know more, check out the paper and share it on Twitter!

A cerebellar internal model calibrates a feedback controller involved in sensorimotor control
Scroll to top