Cerebellar granule cell axons support high-dimensional representations

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Classical cerebellar theories propose high-dimensional sensorimotor representations for motor learning.
• Recent studies suggested low-dimensional activity in cerebellar granule cells (GrCs).

Purpose of the Study:

• To investigate the dimensionality of GrC axon population activity.
• To understand sensorimotor representations from the perspective of Purkinje cell decoders.

Main Methods:

• Utilized three-dimensional acousto-optic lens two-photon microscopy.
• Recorded activity from hundreds of cerebellar granule cell axons during spontaneous behaviors.

Main Results:

• GrC axon population activity is high-dimensional and distributed.
• Little fine-scale spatial structure was observed in GrC activity.
• Distinct behavioral states were encoded along orthogonal dimensions in the neuronal activity space.

Conclusions:

• Cerebellar cortex supports high-dimensional representations, contrary to some recent suggestions.
• The cerebellum segregates behavioral state-dependent computations into orthogonal subspaces, similar to the neocortex.
• Findings align with cerebellar pattern separation theories and suggest common population coding features between cerebellum and neocortex.