What the cerebellum computes

https://doi.org/10.1016/S0166-2236(03)00054-7Get rights and content

Abstract

The brain is an organ that processes information. Brain systems such as the cerebellum receive inputs from other systems and generate outputs according to their internal rules of information processing. Thus, our understanding of the cerebellum is ultimately best expressed in terms of the information processing it accomplishes and how cerebellar neurons and synapses produce this processing. We review evidence that indicates how Pavlovian eyelid conditioning reveals cerebellar processing to be an example of feedforward control. Eyelid conditioning demonstrates a capacity for learning in the cerebellum that is error driven, associative and temporally specific – as is required for feedforward control. This computation-centered view is consistent with a variety of proposed functions of the cerebellum, including sensory–motor integration, motor coordination, motor learning and timing. Moreover, feedforward processing could be the common link between motor and non-motor functions of the cerebellum.

Section snippets

From a model system to modeling a system

Although initially viewed as a model of associative learning, it is increasingly clear that Pavlovian eyelid conditioning is an especially useful means of studying cerebellar computation and its underlying mechanisms. This transition has occurred with the steady accumulation of evidence to indicate that eyelid conditioning engages the cerebellum in a straightforward and conceptually useful manner. In a typical experiment, the paired presentation of a cue stimulus such as a tone and a

Feedforward control and learning

Making accurate movements requires sensory input. In principle, this input can be applied to motor commands in two general ways: feedback and feedforward. This distinction can be illustrated using control of room temperature. A standard thermostat controls room temperature via feedback. It activates the heating–cooling system when the thermometer (sensory input) signals that room temperature differs from the target setting. Thus, feedback about current performance is used to generate output

Temporally specific learning

The demands on this learning are made more complex by the likelihood that not all sensory signals will predict errors with the same time delay. The time-dependent properties of eyelid conditioning show that cerebellar learning is well suited to solve this problem.

In the thermostat analogy, some sensors might be activated when a small window is opened, and others activated by opening a larger window. On a cold day, these signals would predict the same error (room too cold), but with different

Mechanisms of temporally specific learning

Analysis of cerebellum-dependent behaviors such as eyelid conditioning and adaptation of the vestibulo–ocular reflex (VOR) has revealed a great deal about how cerebellar neurons and synapses accomplish this computation 8, 14. Evidence from such studies indicates that plasticity in both the cerebellar cortex and cerebellar nuclei is involved 15, 16. For learning that increases cerebellar output, climbing fiber inputs drive the induction of plasticity in the cortex. Recent evidence indicates that

Concluding remarks

Pavlovian eyelid conditioning has emerged as a valuable tool for studying cerebellar computation. Eyelid conditioning engages the cerebellum in a relatively direct way: mossy and climbing fibers, respectively, convey information about the tone and the puff to the cerebellum, and cerebellar output drives the expression of the conditioned response. Because of this straightforward mapping, the behavioral properties of eyelid conditioning are a reasonable first approximation of the input–output

References (39)

  • N.E. Berthier et al.

    Activity of deep cerebellar nuclear cells during classical conditioning of nictitating membrane extension in rabbits

    Exp. Brain Res.

    (1990)
  • J.F. Medina et al.

    Computer simulation of cerebellar information processing

    Nat. Neurosci.

    (2000)
  • M.D. Mauk et al.

    A model of Pavlovian eyelid conditioning based on the synaptic organization of the cerebellum

    Learn. Mem.

    (1997)
  • I. Gormezano
  • N. Schneiderman et al.

    Conditioning of the nictitating membrane of the rabbit as a function of CS–US interval

    J. Comp. Physiol. Psychol.

    (1964)
  • M.D. Mauk et al.

    Learning-dependent timing of Pavlovian eyelid responses: differential conditioning using multiple interstimulus intervals

    Behav. Neurosci.

    (1992)
  • M. Ito

    Neurophysiological aspects of the cerebellar motor control system

    Int. J. Neurol.

    (1970)
  • J.L. Raymond

    The cerebellum: a neuronal learning machine?

    Science

    (1996)
  • S.P. Perrett et al.

    Extinction of conditioned eyelid responses requires the anterior lobe of cerebellar cortex

    J. Neurosci.

    (1995)
  • Cited by (201)

    • The role of cerebellum in the child neuropsychological functioning

      2020, Handbook of Clinical Neurology
      Citation Excerpt :

      For Honda et al. (2018), a tandem configuration of forward and inverse internal models enables efficient motor learning in the cerebellum. Many studies provided evidence supporting the forward model hypothesis (Mauk et al., 2000; Ohyama et al., 2003; Ebner and Pasalar, 2008; Cerminara et al., 2009; Imamizu and Kawato, 2009, 2012; Brandauer et al., 2010; Fautrelle et al., 2011; Brooks et al., 2015; Giovannucci et al., 2017; Popa and Ebner, 2019; Tanaka et al., 2019). During the acquisition of motor skills, movements become increasingly controlled by cerebellar internal models and are eventually executed automatically in a feedforward manner (Ramnani, 2006).

    • Corollary Discharge Signals in the Cerebellum

      2019, Biological Psychiatry: Cognitive Neuroscience and Neuroimaging
    View all citing articles on Scopus
    View full text