Neural Noise- Cerebellar population coding and behavior were framed in an electrophysiology-friendly framework (Journal of Neurophysiology).1
- Relevant to motor control and potential neuroprosthetics.1 1
- Full citation: Reza Shadmehr and Mohammad Amin Fakharian, “A neural language for the cerebellum: control of behavior via competing populations of Purkinje cells,” J Neurophysiol 135(3):557–578 (March 2026); DOI: 10.1152/jn.00253.2025; PMCID: PMC12940533. Both authors from the Laboratory for Computational Motor Control, Dept. of Biomedical Engineering, Johns Hopkins School of Medicine.2
- Core framework: individual Purkinje cells (P-cells) fire for all movement directions and their spike rates resist straightforward interpretation as motor commands; meaning emerges only at the population level via a “potent vector” assigned to each P-cell, estimated from climbing-fiber input and spike-triggered averaging.2
- The “tug-of-war” model proposes that neighboring P-cells with opposing potent vectors cancel each other’s downstream effects, ensuring movement accuracy; reward-relevant saccades showed better balance and cancellation (perpendicular component near zero) than reward-irrelevant ones, implicating the cerebellum in value-based motor optimization.2
- In saccade experiments with n=154 P-cells, summing population activity produced a behaviorally meaningful burst-pause structure: an initial push toward the target (acceleration) followed by a braking signal near deceleration onset, consistent with fastigial nucleus inactivation studies producing hypermetric ipsilateral / hypometric contralateral saccades.2
- The framework generalizes across behaviors: reaching in mice and monkeys, tongue movements in mice (lobule VI/VII P-cells), and saccades all showed similar population-level burst-pause computation; optogenetic suppression of P-cells displaced movements in the direction predicted by each cell’s potent vector, with simultaneous suppression of two cells producing superadditive displacement (amplitude more than doubled).2
- Anatomically, neighboring P-cells likely form local “cliques” within ~100 µm, and their coordinated activity could be read out by downstream fastigial nucleus and burst-generator circuits; the paper explicitly discusses future holographic optogenetic stimulation to test causal predictions of the framework.2
- Translational relevance: the potent-vector dictionary could serve as a “Rosetta Stone” for decoding cerebellar output across behaviors, potentially informing closed-loop cerebellar stimulation strategies and improving neural decoding in cerebellar-targeting neuroprosthetics (e.g., for ataxia, tremor, or rehabilitation after cerebellar stroke).2
- Key quantitative details: complex-spike-induced pause in simple spikes lasts 10–20 ms; direction spacing of “imagined sister P-cells” used to correct sampling bias is π/4 increments (8 cells per recorded cell); one example group of cells had potent-vector separation of ~70 degrees, illustrating near-opposing but not fully canceling pairs.2