Learning to predict the future: the cerebellum adapts feedforward movement control
Introduction
The cerebellum has a strikingly homogeneous cortical architecture, and for that reason cerebellar function has been considered one of the more tractable problems in neuroscience [1]. Yet, after decades of research there is no consensus on how the cerebellum operates, what information it processes and stores, or how it learns. This review focuses on new findings suggesting that the cerebellum functions as a sensorimotor predictor. The following questions are addressed: does the cerebellum play a primary role in predictive motor control? If so, what predictive information does the cerebellum compute? What types of error information are used to learn and calibrate cerebellar prediction mechanisms? These questions ultimately require answers spanning the molecular, cellular, circuit and behavioral levels. The focus of this review is on recent behavioral- and circuit-level analyses of motor function.
Section snippets
Cerebellar damage and predictive movement control
Cerebellar damage does not cause loss of movement, but instead leads to clear and consistent abnormalities in movement that can include a lack of coordination, increased variability, tremor and poor accuracy. These abnormalities can manifest in eye and limb movements, walking and balance. Many theories about the normal role of the cerebellum have been formed on the basis of deficits in one or a few movement types. As discussed below, theories involving predictive control can explain the effects
Cerebellar computation: what is predicted?
In the movement studies of people with cerebellar damage reviewed above, loss of feedforward control indicates that the cerebellum is necessary for behaviors requiring prediction. However, these studies do not tell us what specific predictive information is computed by the cerebellum. That issue is more directly addressed by neurophysiological, neuroimaging and computational studies. Computational theories of cerebellar function are not discussed exhaustively here, because there are several
Learning to predict: is one error equivalent to another?
Cerebellar learning is thought to depend largely on error feedback. Activation in the cerebellum is increased during more difficult movements, which consequently results in more error feedback [30]. For error feedback to be useful in predictive control, errors from past movements must be used to update subsequent movements. In a recent neuroimaging experiment, Diedrichsen et al. [31••] studied reaching movements in which random errors were induced in one of three ways: by mechanically
Conclusions
Recent work has highlighted the importance of cerebellar control over predictive motor behaviors. The cerebellum seems most crucial for adapting these movements to novel situations through trial-by-trial learning mechanisms. This result holds across many different movement types — walking, reaching, pinching, catching and balance. However, different cerebellar regions are likely to be involved in the control of these movement types. Medial cerebellar structures are probably more important for
References and recommended reading
Papers of particular interest, published within the annual period of review, have been highlighted as:
• of special interest
•• of outstanding interest
Acknowledgements
Thanks to J Diedrichsen, J Bastian, and E Connor for helpful discussion and comments. Supported by National Institutes of Health grants HD040289 and HD04741.
References (36)
- et al.
An internal model for sensorimotor integration
Science
(1995) - et al.
Inverse-dynamics model eye movement control by Purkinje cells in the cerebellum
Nature
(1993) - et al.
Consolidation of motor memory
Trends Neurosci
(2006) - et al.
Adaptation to gradual as compared with sudden visuo-motor distortions
Exp Brain Res
(1997) A theory of cerebellar cortex
J Physiol
(1969)The cerebellum, predictive control and motor coordination
Novartis Found Symp
(1998)- et al.
Cerebellar contributions to locomotor adaptations during splitbelt treadmill walking
J Neurosci
(2006) - et al.
Hereditary cerebellar ataxia progressively impairs force adaptation during goal-directed arm movements
J Neurophysiol
(2004) - et al.
Intact ability to learn internal models of arm dynamics in Huntington's disease but not cerebellar degeneration
J Neurophysiol
(2005) - et al.
Motor disorder in Huntington's disease begins as a dysfunction in error feedback control
Nature
(2000)
Influence of vision on upper limb reaching movements in patients with cerebellar ataxia
Brain
Contralateral cerebellar damage impairs imperative planning but not updating of aimed arm movements in humans
Exp Brain Res
Cerebellar subjects show impaired adaptation of anticipatory EMG during catching
J Neurophysiol
Predictive and reactive finger force control during catching in cerebellar degeneration
Cerebellum
Cerebellar control of postural scaling and central set in stance
J Neurophysiol
Predictive saccades to a regularly alternating target in cerebellar patients
Ann N Y Acad Sci
Cerebellar involvement in anticipating the consequences of self-produced actions during bimanual movements
J Neurophysiol
Analysis of cerebellar motor disorders by visually guided elbow tracking movement
Brain
Cited by (490)
Disrupting somatosensory processing impairs motor execution but not motor imagery
2023, Human Movement ScienceSusceptibility to walking balance perturbations in young adults is largely unaffected by anticipation
2023, Human Movement ScienceNeurophysiology of cerebellar ataxias and gait disorders
2023, Clinical Neurophysiology PracticeOut of control: computational dynamic control dysfunction in stress- and anxiety-related disorders
2024, Discover Mental HealthSELECTIVE INFERENCE FOR SPARSE MULTITASK REGRESSION WITH APPLICATIONS IN NEUROIMAGING
2024, Annals of Applied Statistics