A Cerebellar Learning Model of Vestibulo-Ocular Reflex Adaptation in Wild-Type and Mutant Mice

Claudia Clopath, Aleksandra Badura, Chris I. De Zeeuw, Nicolas BrunelJournal of Neuroscience 2014 · DOI

Abstract

Mechanisms of cerebellar motor learning are still poorly understood. The standard Marr-Albus-Ito theory posits that learning involves plasticity at the parallel fiber to Purkinje cell synapses under control of the climbing fiber input, which provides an error signal as in classical supervised learning paradigms. However, a growing body of evidence challenges this theory, in that additional sites of plasticity appear to contribute to motor adaptation. Here, we consider phase-reversal training of the vestibulo-ocular reflex (VOR), a simple form of motor learning for which a large body of experimental data is available in wild-type and mutant mice, in which the excitability of granule cells or inhibition of Purkinje cells was affected in a cell-specific fashion. We present novel electrophysiological recordings of Purkinje cell activity measured in naive wild-type mice subjected to this VOR adaptation task. We then introduce a minimal model that consists of learning at the parallel fibers to Purkinje cells with the help of the climbing fibers. Although the minimal model reproduces the behavior of the wild-type animals and is analytically tractable, it fails at reproducing the behavior of mutant mice and the electrophysiology data. Therefore, we build a detailed model involving plasticity at the parallel fibers to Purkinje cells' synapse guided by climbing fibers, feedforward inhibition of Purkinje cells, and plasticity at the mossy fiber to vestibular nuclei neuron synapse. The detailed model reproduces both the behavioral and electrophysiological data of both the wild-type and mutant mice and allows for experimentally testable predictions.

Challenging the Marr–Albus–Ito Dogma

For decades, the "Marr–Albus–Ito" theory has been the bedrock of cerebellar neuroscience. It posits that motor learning happens at a single location: the synapses between parallel fibers (from granule cells) and Purkinje cells. In this view, climbing fibers act as "teachers," providing an error signal that weakens these synapses to correct movements. However, this elegant theory has hit a wall. Recent experiments with mutant mice—specifically those with altered granule cell excitability or missing inhibition—showed behavioral deficits that the standard model couldn't explain. Furthermore, electrophysiological recordings showed that while animals learned a task, the activity of their Purkinje cells didn't change as much as the theory predicted. There was a clear gap between the "minimalist" theory and biological reality.

The Detailed Circuit Model

To bridge this gap, the authors developed a biologically representative circuit model that expands upon the traditional framework. This "Detailed Model" incorporates several critical components missing from the minimal model:

  • Molecular Layer Interneurons (INs): These provide feedforward inhibition to Purkinje cells, a crucial regulatory mechanism.
  • A Second Plasticity Site: Specifically at the Mossy Fiber (MF) to Medial Vestibular Nuclei (MVN) synapse, allowing for distributed learning.
  • Synaptic Bounds: Realistic limits on how much a synapse can strengthen or weaken, reflecting physiological constraints.
The detailed cerebellar circuit model (A) and the experimental setup for VOR adaptation (B).

Experimental Validation: VOR Phase-Reversal

The models were tested using a classic motor task: Vestibulo-Ocular Reflex (VOR) phase-reversal. This reflex normally keeps eyes steady on a target while the head moves. In "phase-reversal" training, the visual world is manipulated to force the mouse to eventually move its eyes in the same direction as its head—a complete reversal of the natural reflex. The study utilized new electrophysiology from wild-type (WT) mice and existing behavioral data from two mutant strains: PC-$Δγ2$ (lacking PC inhibition) and GC-KCC2 (hyper-excitable granule cells).

Key Results and Neural Signatures

The detailed model successfully replicated the 4-day learning curve of WT mice, showing a decrease in gain on Day 1 followed by a gradual phase reversal on Days 2-4. Crucially, it also correctly predicted the failure of mutant mice to achieve phase reversal, matching experimental observations where mutants could decrease reflex gain but were unable to flip the phase.

Behavioral adaptation (gain and phase) over 4 days and corresponding electrophysiological changes in Purkinje cell activity.

Unlike the minimal model, the detailed model correctly showed that Purkinje cell firing patterns remain relatively stable after learning. This is because the "heavy lifting" of the memory is moved to the vestibular nuclei, supporting the "Transfer Hypothesis" of memory consolidation.

Insights into Memory and the "Headroom" Problem

The model suggests that the cerebellar cortex (GC-PC synapses) is the "fast learner," reacting quickly to error signals. Over time, this memory is transferred to the vestibular nuclei (MF-MVN synapses) for long-term storage. This explains why Purkinje cell activity doesn't need to change drastically to maintain a learned behavior.

Furthermore, the model reveals why mutant mice fail at phase reversal through the "Headroom Problem." To maintain normal baseline function, these mutants must compensate by lowering their initial synaptic weights. This leaves the synapses sitting near their lower bounds (saturated), providing no "headroom" to weaken further during the difficult phase-reversal task. Finally, the smooth transition of eye-movement phase seen in mice is best explained by a ~100ms delay in the error signal provided by climbing fibers, a detail that proved essential for model accuracy.

Limitations and Future Work

While the model is robust, it does not account for the initial 30–60 degree phase shift seen in mice before training, likely due to the poor visual acuity of mice which the model doesn't fully simulate. Future work could involve testing the model against other mutant strains to see if this multi-site framework holds up across all forms of cerebellar-dependent motor learning.

Summary updated

Abstract

Mechanisms of cerebellar motor learning are still poorly understood. The standard Marr-Albus-Ito theory posits that learning involves plasticity at the parallel fiber to Purkinje cell synapses under control of the climbing fiber input, which provides an error signal as in classical supervised learning paradigms. However, a growing body of evidence challenges this theory, in that additional sites of plasticity appear to contribute to motor adaptation. Here, we consider phase-reversal training of the vestibulo-ocular reflex (VOR), a simple form of motor learning for which a large body of experimental data is available in wild-type and mutant mice, in which the excitability of granule cells or inhibition of Purkinje cells was affected in a cell-specific fashion. We present novel electrophysiological recordings of Purkinje cell activity measured in naive wild-type mice subjected to this VOR adaptation task. We then introduce a minimal model that consists of learning at the parallel fibers to Purkinje cells with the help of the climbing fibers. Although the minimal model reproduces the behavior of the wild-type animals and is analytically tractable, it fails at reproducing the behavior of mutant mice and the electrophysiology data. Therefore, we build a detailed model involving plasticity at the parallel fibers to Purkinje cells' synapse guided by climbing fibers, feedforward inhibition of Purkinje cells, and plasticity at the mossy fiber to vestibular nuclei neuron synapse. The detailed model reproduces both the behavioral and electrophysiological data of both the wild-type and mutant mice and allows for experimentally testable predictions.
BibTeX
@article{clopath2014cerebellar,
  author = {Claudia Clopath and Aleksandra Badura and Chris I. De Zeeuw and Nicolas Brunel},
  title = {{A Cerebellar Learning Model of Vestibulo-Ocular Reflex Adaptation in Wild-Type and Mutant Mice}},
  journal = {Journal of Neuroscience},
  year = {2014},
  doi = {10.1523/jneurosci.2791-13.2014},
}
Figure