The polysynaptic connections between the neocortex and the cerebellum (as verified by fMRI resting-state functional connectivity) are such that the anterior lobe of the cerebellum mediates skeletomotor processing due to its connections with M1 and S1, that the mediolateral lobe mediates object processing due to its connections with the orbital and temporal cortices, and that the posterior lobe mediates spatial processing due to its connections with MT/MST (middle temporal and middle superior temporal cortices), the retrosplenial and parietal cortices, and the medial frontal lobes which house the ocular eye and head fields in primates (Chen and Tehovnik 2007; Tehovnik, Patel, Tolias et al. 2021). Van Overwalle et al. (2023) re-investigated the connections between the neocortex and the cerebellum as it pertains to human cognition using a database of 44,500 participants; it was hypothesized over two decades ago that the cerebellum is centrally involved in cognitive control (Schmahmann 1997). However, unlike the neocortical output which encodes information according to the senses, the cerebellar output operates according to a firing-rate code that is used in the contraction of muscles by circuits in the brain stem and spinal cord, converting all information entering the cerebellum into a muscle code (Herzfeld, Lisberger et al. 2023; Schiller and Tehovnik 2015; Tehovnik, Patel, Tolias et al. 2021). Furthermore, when the efference-copy signal is interrupted by electrical stimulation delivered either to the cerebellum or the saccade generator in the brain stem, primates never correct their memories following such interruption (Tehovnik, Patel, Tolias et al. 2021). A similar result occurs when the ocular proprioceptors are activated (Chen 2019; Roll and Roll 1987; Roll et al. 1991; Valey et al. 1994, 1995, 1997), which are known to send short-latency signals (within 4 ms) to the cerebellum for processing (Fuchs and Kornhuber 1969) which displaces the actual target location preventing a correction through vision: it takes over 30 ms for a visual signal to arrive at the cerebellum and brain stem (Miles and Lisberger 1981), thus being much too long to counter the effect of the proprioceptive perturbation. When perturbations occur by stimulation of the midbrain or neocortex the correct target location through memory is always acquired after the displacement; these regions are outside the efference-copy loop (Loyola et al. 2019; Shadmehr 2020; Tehovnik, Patel, Tolias et al. 2021).

Figure 1 summarizes the data of Van Overwalle et al. (2023, Fig. 4/King et al. 2019, re-analyzed) according to cognitive variables. Notice that in both anterior and posterior cerebellum there is a large representation from S1 and M1, which is consistent with the finding that 49% of the neocortex is dedicated to proprioception and movement (Sarubbo et al. 2020). Moreover, attentional processes have been attributed to the midline region of lobule V1 which contains neurons that respond during the execution of saccadic eye movements medially and head movements laterally (Fig. 8 of Tehovnik, Patel, Tolias et al. 2021). Much like the neocortex for which language/object encoding represents over 47% of the hemispheric real estate (Sarubbo et al. 2020), the cerebellar mediolateral lobe—which represents over half of the cerebellum—is also activated during language and object processing, even though a range of cognitive descriptors have been used to label the cerebellar functions such as ‘executive control’ and ‘mentalizing’ (i.e., watching movies), which in the neocortex includes the temporal and orbital cortices that store object information (Brecht and Freiwald 2012; Bruce et al. 1981; Schwarzlose et al. 2005; Schwiedrzik, Freiwald et al. 2015; Freiwald and Tsao 2010). The mentalizing/default label is known to include the cingulate cortex which is a fibre bundle linking various regions of the neocortex to the limbic system (Tehovnik, Hasanbegović, Chen 2024). It is noteworthy that the cerebellar real estate dedicated to limbic processes is minimal; the hippocampus (unlike the neocortex and cerebellum) is involved in transferring information rather than in the storage of information (Corkin 2002; Knecht 2004; Morrison and Hof 1997; Munoz-Lopez et al. 2010; Roux et al. 2021; Scoville and Milner 1957; Squire et al. 2001; Xu et al. 2016); it has been estimated that the storage capacity of the human cerebellum is 2.8 x 10^14 bits of information or 2^(2.8 x 10^14) possibilities, and the storage capacity of the human neocortex is 1.6 x 10^14 bits or 2^(1.6 x 10^14) possibilities (Huang 2008; Tang et al. 2001; Tehovnik, Hasanbegović, Chen 2024).

Figure 2 (modified from Fig. 4/King et al. 2019 of Van Overwalle et al. 2023) is used to simplify the representation for the cerebellum. As before, both the anterior and posterior lobes are dedicated to skeletomotor control with the posterior lobes also participating in spatial processing (a characteristic of MT/MST, the retrosplenial and parietal lobes, and medial frontal lobes). The oculomotor region is confined to lobule VI, and the mediolateral lobe subserves language and object processing. The cognitive labels used in the study of Van Overwalle et al. (2023) are indicated in parentheses.

Given that cognition depends on synaptic connectivity (‘for anesthesia [which disables the synapses] eliminates all sensation’, Hebb 1968) it is no surprise that this process has been ascribed to the cerebellum (Schmahmann 1997), which is polysynaptically connected to the neocortex with a comparable and proportionate representation of all neocortical functions (Buckner et al. 2011; King et al. 2019; Tehovnik, Patel, Tolias et al. 2021; Van Overwalle et al. 2023). The cerebellum, however, is not necessary for cognition even though severely damaged (or missing) it produces the retardation of one’s movements/ expressions (Yu et al. 2014). Cerebellar patients are still aware of the outside world since their vital senses are intact; yet they have great difficulty moving about. But if one cannot express their cognition/consciousness through dance, drawing, speaking, reading, and writing, for example, then the quality of life is severely compromised. The late Stephen Hawking, who suffered from ALS (amyotrophic lateral sclerosis), is a case in point: toward the end of his life his movements were reduced to the transfer of 0.1 bits per second, which was based on the output of a cheek muscle and information of which was transferred to operate a communication device (Tehovnik, Patel, Tolias et al. 2021). Importantly, neocortical neurons consume 20 times more energy per neuron than do cerebellar neurons during immobility (Herculano-Houzel 2011). This has been attributed to the neocortex requiring high energy consumption to support cognition while immobile, i.e., while thinking, whereas the cerebellum is engaged when movements are being generated, which is always required to update an efference-copy code as it pertains to a behavior being updated (Tehovnik, Hasanbegović, Chen 2024). In closing, the functionality of the cerebellum and the neocortex can be described by Kahneman’s (2011) ‘thinking fast’ and ‘thinking slow’. ‘Thinking fast’ is dependent on rapid motor responses with little thinking, which has been associated with the cerebellum (Tehovnik, Hasanbegović, Chen 2024; Tehovnik, Patel, Tolias et al. 2021). ‘Thinking slow’ refers to the slow process of learning something new, which has been associated with the neocortex (Chen and Wise 1995ab; Hebb 1949, 1968; Kimura 1993; Ojemann 1991; Ito, Maldonado et al. 2022; Schwarzlose et al. 2005). But to be clear, both ‘thinking fast’ and ‘thinking slow’ require the cerebellum and the neocortex, but the difference is in the number of synapses recruited for information storage and behavioral execution: ‘thinking fast‘ necessitates fewer synapses than ‘thinking slow‘, since the latter is involved in the storage of new information through declarative and procedural learning, and in the creation of an efference-copy representation at the Purkinje neurons for the task being learned.