50 Years Since the Marr, Ito, and Albus Models of the Cerebellum.
Level 5 - mechanism / opinion, no new human data
Narrative review and theoretical proposal without empirical data.
PubMed 32599123 · doi:10.1016/j.neuroscience.2020.06.019
What was done
The authors conducted a narrative review evaluating the 50-year-old cerebellar function models of David Marr, Masao Ito, and James Albus against recent computational and experimental findings. They analyzed shared concepts and differences among the models regarding synaptic plasticity, climbing fiber signals, granule cell coding, and internal models, and proposed a revised computational framework.
What was found
The abstract reports no quantitative data or empirical numbers. It outlines the common ground across all three models in parallel-fiber-Purkinje-cell plasticity guided by climbing-fiber activities during sensorimotor learning. It identifies divergences in potentiation versus depression, teaching versus error signals, pattern recognition versus control objectives, and internal model structures, and proposes hierarchical reinforcement learning with multiple internal models as a new direction.
Why it matters
The paper synthesizes 50 years of foundational cerebellar motor control theories and suggests modern reinforcement learning frameworks to advance computational neuroscience models of the cerebellum.
Limits
This is a theoretical narrative review without new experimental, human, or animal data. The abstract provides no systematic literature search strategy or quantitative synthesis.
Cited by
- supports During learning, making errors sends signals relayed through the cerebellum that prompt neural circuitry to reshape.