David Marr (1969) believed that all cerebellar learning is initiated by the neocortex such that a particular context (conveyed through the neocortex) triggers an automated response:

“The learning of a context will arise if the combination of olivary cell firing [through the cerebellum] and that particular context is a frequent one, as it would be, for example, while the child (under cerebral control) was 'learning' to stand. Once the context is learned [as transmitted by the mossy fibres], the reflex automatically becomes operative when it is required.” (Marr 1969, p. 467)

Furthermore, Marr (1969) asserted that the cerebellum, which has a uniform structure across all vertebrates, performs two types of learning: learning to maintain posture and balance through the reflexes (which is part of unconscious control but only once shaped by declarative learning, Marr 1969; Tehovnik, Hasanbegović, Chen 2024) and learning movements to act (which is part of volitional, skeletal control as guided by the neocortex, Tehovnik, Hasanbegović, Chen 2024) (the idea that the cerebellum mediates learning was proposed by Brinkley in 1964 and followed up in detail by Marr 1969). Under normal circumstances these two forms of learning co-occur. As already understood by Marr, parallel fibre-Purkinje synapses are modified by one climbing fibre from the inferior olive, a fibre that wraps itself around the initial segment and the dendritic arbors of a single Purkinje neuron (Eccles 1967; Ito 2008; Loyola, De Zeeuw et al. 2019). Each climbing fibre responds to a neocortical command for the induction of a standing reflex, for example, which involves aligning the vestibular axis of the head with respect to the gravitational axis as the legs and arms support the body in a standing posture by a child, an ability acquired between the ages of 12 and 18 months postnatally. In short, to learn to stand a cue is sent via the neocortex through the inferior olive signaling to the cerebellar Purkinje circuits that a ‘standing context’ must be stored. A Purkinje cell can initiate an elemental posture/movement if the context is optimal; the context is set by the mossy fibres as when one’s head is elevated and supported by an upright body. Such a context is established not only by the vestibular and proprioceptive senses, but also by the visual and tactile senses to discriminate between an upright posture as compared to a sitting or crawling posture as it would apply to an infant. Once learning is complete (which can take up to one year for an infant learning to stand and walk), centi-second exposure to a context (e.g., a call by a mother) is sufficient to generate a standing reflex following by walking by way of the Purkinje cells and cerebellar nuclei, as the climbing fibres remain relatively quiescent (Loyola, De Zeeuw 2019). The Purkinje cells, which number in the millions [i.e., 15 million for humans, Andersen et al. 1992; Braitenberg and Atwood 1958; Nairm et al. 1989], learn different contexts as transmitted via the mossy fibres.

For cerebellar circuits to learn various postures and movement sequences, the olive must be privy to command signals (e.g., retinal slip, an air puff, a mother’s voice, a reward or punishment signal). Marr (1969) suggested that the cerebellum can manage ongoing postural shift and movement without the direct participation of the neocortex by storing the symbolic sensory representations normally housed in the neocortex: the storage of the symbolic sensory representation after learning is done by the granular cells, the most numerous neurons in the mammalian central nervous system (Herculano-Houzel 2009). But as suspected by Marr (1969, p. 467), without the neocortex the postures and movements of the body cannot be optimized by the high-fidelity declarative information stored in and provided by the neocortex (Schiller and Tehovnik 2015; Tehovnik, Hasanbegović, Chen 2024; Tehovnik, Patel, Tolias et al. 2021).

More Edward J Tehovnik's questions See All
Similar questions and discussions