“Theoretically, the computation of passive motion requires a comparison between an internal estimate of the sensory consequences of active self-motion (i.e., forward model) and the actual sensory feedback (reviewed in Cullen 2011). Cerebellar output neurons dynamically encode this difference during self-motion; fastigial neurons are insensitive to active motion and encode an explicit representation of passively applied self-motion [Brooks and Cullen 2013]. Specifically, the two distinct fastigial nucleus populations … selectively and dynamically encode passive head and body motion relative to space. Moreover, our evidence to date suggests that this cerebellar-dependent mechanism uses an internal model of the expected sensory consequences of active head motion to selectively cancel responses to active motion.” [Cullen 2015, pp. 206-207, in: Baumann, Sokolov et al. 2015]. See the figure depicting a circuit diagram vis-à-vis the cerebellum which is central to the internal representation of self-motion or movement commands in general (Fig. 1).
Accordingly, in the experiments of Noda et al. (1991) when electrical stimulation deviates the eyes by fastigial activation before a monkey generates a memory-guided saccadic eye movement, the animal fails to acquire the target location since the brain assumes that the deviation originated from an external source (Cullen 2015), much like what happens by pushing on an eye to make a viewed visual target jump [see: Roll and Roll 1987; Roll et al. 1991; Valey et al. 1994, 1995, 1997]. This is exactly what occurs when the proprioceptors of the eyes are activated electrically in a monkey rewarded for generating a saccade to a viewed target (Chen 2019): once the proprioceptors are stimulated, the monkey never evokes a saccade to the correct target location since the brain (by way of the proprioceptors) understands that the deviation originated from an external source (Cullen 2015). Furthermore, when electrical stimulation is used to deviate the eyes by activation of the neocortex or superior colliculus of a monkey performing a memory-guided saccade task, a monkey readily elicits a corrective saccade to acquire the remember target position (Schiller and Sandell 1983; Sparks and Mays 1983; Tehovnik and Sommer 1996). This indicates that such stimulation of the neocortex and superior colliculus is interpreted by the cerebellum as a command generated by the monkey to move the eyes and therefore the stimulation is not disruptive to the ocular performance of the animal (i.e., the animal always acquires the target).
Figure 1: The cerebellum receives a plethora of inputs with the purpose of dissociating self-motion (as signaled by a motor efference copy representation) and body motion triggered by external sources as conveyed by the vestibular, visual, proprioceptive/somatosensory systems. Some neural structures contributing to this process include the superior colliculus, the hippocampus, and the thalamus via, respectively, the primary motor cortex, the premotor cortex, and the posterior parietal cortex. The illustration is from figure 5 of Cullen (2015; in: Baumann, Sokolov et al. 2015).