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

Abstract

Highlights• Larval zebrafish acutely adapt their ongoing behavior to changed visual reafference• This acute motor adaptation is likely to be implemented by a feedback controller• Persistent change in reafference modifies the parameters of this controller• The cerebellum is required for this recalibration SummaryAnimals must adapt their behavior to survive in a changing environment. Behavioral adaptations can be evoked by two mechanisms: feedback control and internal-model-based control. Feedback controllers compare current with desired sensory state and generate motor output to minimize this difference. In contrast, internal models use previous sensory-motor history to update parameters required for predictive motor control. In this study we use multiple perturbations in visual feedback to show that larval zebrafish acutely adapt their swimming behavior depending on the feedback received while swimming. These short-term changes do not affect an initial stereotyped ballistic portion of the swimming event and are unaffected by the ablation of cerebellar Purkinje cells. Contrarily, consistent long-term changes induce larvae to gradually change their behavior, including the ballistic period kinematics, in a cerebellar-dependent process. We conclude that swimming in larval zebrafish can be understood as a feedback control system that is recalibrated by cerebellar output.

Structural and functional organization of visual responses in the inferior olive of larval zebrafish

“…These visual responses were direction-selective and, for the most part, not tightly associated with swimming events (Figure 2C), suggesting a strong sensory component in their activity. Consistent with previous reports describing complex spikes in Purkinje cells, (Knogler et al, 2019;Markov et al, 2021), IO activity was generally phase-locked to motion onset (Figure 2C), suggesting a role of these neurons as 'sensors', rather than 'integrators', of sensory evidence (Bahl and Engert, 2020;Dragomir et al, 2020;Markov et al, 2021).…”

Rapid acquisition of internal models of stimulus patterns by Purkinje cells enables faster behavioral responses

“…1b), to look for indications that fish are predicting the onset of optic flow. Latency to respond to optic flow would be a function of pre-motor processing, unlike swim bout kinematics which are modulated within a bout using sensory feedback 32, 40 .…”

PyZebraScope: an open-source platform for brain-wide neural activity imaging in zebrafish

“…The modular architecture of PyZebrascope further enables advanced microscope control and image processing algorithms. For example, a common issue during whole-brain imaging in zebrafish is the sample’s slow drift resulting from gravity force, tail motions of unparalyzed fish 14 , or the pressure of pipette attachment during fictive recording in paralyzed fish 11,13,30 . A small amount of drift, especially in the axial direction along which the volumetric scan is undersampled, can result in the loss of neurons during imaging because the neuronal diameter in the zebrafish brain is usually less than 5 μm.…”

“…These advantages of DSLM are best exploited in the optically transparent brain of larval zebrafish. DSLM enabled studies of whole-brain neural dynamics during visually-evoked swimmming 11,12 , motor learning 13,14 , learned helplessness 15 , threat escape 16 , and body posture change 17 . Usage of the DSLM also revealed spontaneous noise dynamics across the brain 18 and how those dynamics change during neural perturbations 19,20 or administration of psychoactive reagents 21 .…”