In 1980, Fred Miles proposed (contrary to the views of Masao Ito 1972) that the purpose of the cerebellum is to house efference-copy representations of all movements being commanded by the neocortex (Miles et al. 1980b). This means that the cerebellum’s role in oculo- and skeletomotor behavior is maximal while new routines are being learned, as borne out by many experimental studies (e.g., Ito et al. 1974a; Kassardjian et al. 2005; Miles and Lisberger 1981; Robinson 1976; Sendhilnathan, Goldberg 2020b; Swain, Thompson et al. 2011; Takahara et al. 2003; Takemori and Cohen 1974). Nevertheless, what this means is that the cerebellum must always be in the neural loop participating in the execution of any behavior to ascertain that all the efference-copy representations remain updated by continuously comparing the representation on record to the sensory feedback during task execution (for not to remain in the loop means the automated behavior begins to degrade, particularly as related to the timing of the response, see Fig. S5,QR of Wagner et al. 2019). This process is realized by having the latency of task execution oscillate across trials such that the long-latency trials are meant to send a command signal (in efference copy form) via the Purkinje neurons in real time (Lisberger 1984) so that comparisons can be made against the sensory feedback to tweak the efference-copy code. It has been suggested that at any one time the human cerebellum can update as many as 50,000 independent efference-copy representations (Heck and Sultan 2002; Sultan and Heck 2003). And we know that during task execution that the entire cerebellar cortex is engaged including circuits not necessary for task execution (Hasanbegović 2024). This global reach assures that all aspects of a behavior are perfected through continuous sensory feedback.
Consciousness, as mediated by the neocortex (Tononi et al. 2008ab), selects what is to be learned (Hebb 1949, 1961, 1968). The learning process here is meant to add new declarative conscious information to neocortical stores. The energy consumed by neocortical neurons is 20 times more expensive than that consumed by cerebellar fibres across all mammals including humans (Herculano-Houzel 2011). Also, neocortex (or consciousness) never sleeps during waking periods even when we are immobile (Chomsky 2023; Herculano-Houzel 2011). This concurs with the observation that to drive a brain-machine interface implanted in the neocortex requires tremendous concentration and effort on the part of the patient (Bublitz et al. 2018). It is this effort (in the absence of automaticity via the cerebellum) that will prevent brain-machine interfaces by way of the neocortex from becoming usable technology. To overcome the shortcoming, methods will need to be developed whereby the efference-copy capabilities of the cerebellum are integrated with the neocortical brain-machine interface as the interface records stimulus-response associations. To date, all neocortically-based brain-machine interfaces (e.g., Lorach et al. 2023; Martin et al. 2014; Metzger, Chang et al. 2023; Nicolelis 2019; Tehovnik and Chen 2015; Tehovnik et al. 2013; Willett, Shenoy et al. 2021, 2023; also see: https://en.wikipedia.org/wiki/Neuralink) have no clinical utility if this problem is not solved, since current interfaces do not automate the learned response: extreme conscious effort will always be required to control a device.
- Badges
- Science method