The neurophysiology of primate express, oculomotor and skeletomotor behavior has received renewed interest in systems neuroscience (e.g., Cecala et al. 2023; Ito et al. 2022; Mekhaiel et al. 2023). This interest is driven by the need to determine the brain circuits involved in establishing behavioral automaticity. The notion that the subcortical structure, the superior colliculus, is necessary for this process and that it relays sensory-cortical information from posterior regions of the neocortex to the brain stem is well accepted (Schiller and Tehovnik 2015; for a recent review see Tehovnik 2024). Disclosing the details of the subcortical circuits that finalize automatic responses is a work in progress, however.

As discussed, the cerebellar Purkinje neurons establish an efference-copy representation of automated motor responses for such behaviors as movement of the head to trigger a vestibuloocular reflex or presentation of a visual stimulus that triggers express saccades (Lisberger 1984; Miles and Lisberger 1981; Schiller and Tehovnik 2015; Tehovnik 2024; Tehovnik et al. 2021). The Purkinje channel subserves the creation of automatic motor responses. The best evidence for this, electrophysiologically, comes from overtraining a cerebellar circuit using a classical conditioning paradigm, whereby electrical stimulation is delivered to the input fibres of the cerebellar cortex (the mossy or parallel fibres) to induce a fictive conditioned response, and electrical stimulation is delivered to the inferior olive (or to the climbing fibres) to induce a fictive unconditioned response (Figure 33). Once conditioning is secured between a conditioned stimulus (CS) and an unconditioned stimulus (US), there is an immediate drop in discharge by the Purkinje neurons (which is well-documented for all forms of cerebellar conditioning, Swain et al. 2011). Figure 34 shows the pattern of Purkinje firing once conditioning has been programmed, such that the conditioned stimulus evokes a drop in discharge. Across the 20 trials illustrated, notice that the latency to evoke a drop in firing oscillates from a low value of 30 ms to a high value of about 100 ms. The figure shows at least three cycles of oscillation, such that every 8 trials a full range of latencies is exhibited by the neuron.

The reason this oscillatory pattern is important (a pattern exhibited by overtrained, conditioned Purkinje neurons) is that it mirrors the latency of the behavioral response (Gallistel et al. 2022). As well, when subjects are overtrained on either VOR or on express saccades it is noteworthy that a similar oscillation of behavioral latencies occurs across trials, which can be grouped in blocks of 8 trials or so (Lisberger 1984; Schiller and Tehovnik 2015). For example, in the case of express saccades (with latencies < 125 ms) versus regular saccades (with latencies > 125 ms), the two saccade types oscillate across trials (Schiller and Tehovnik 2015).[1]

We propose that the oscillatory pattern reflects the utilization of a short and long latency circuit through the cerebellum to continuously update all automatic responses. For the short-latency circuit (i.e., short-latency Purkinje discharge), there is a coincidence of transmission through the Purkinje-nuclear and nuclear-only pathway to induce a combined EPSP at the nuclear neurons (see circuit, Figure 33). For the long-latency circuit, there is no coincidence between the two signals: the Purkinje-nuclear EPSP is far slower than the nuclear-only EPSP. The timing of the latter, the nuclear-only EPSP should remain invariant (and short) once an animal is overtrained. You might ask: why should a system have such oscillations in transmission latency across trials? This assures that the Purkinje ‘efference-copy’ circuit is always up to date, based on the latest sensory feedback to act as a check and balance during the longest trials once automaticity is finalized. The worst thing that can happen is to execute an automated act that has become obsolete: e.g., when Alec Baldwin discharged a gun that he believed was not loaded, this led to the death of his colleague. Clearly, Baldwin’s efference-copy signal was not up to date.

That a short-latency cerebellar circuit [i.e., a circuit that bypasses the cerebellar cortex and utilizes a direct nuclear pathway] is sufficient for the execution of automated acts is supported empirically. Disruption of just the Purkinje neurons (while sparing the cerebellar nuclei) affects the learning of new vestibuloocular-reflex gain changes or new visual-object-response associations, while sparing the execution of old gain changes or old associations (Kassardjian et al. 2005; Sendhilnathan and Goldberg 2020b). However, damage of both the Purkinje neurons and the cerebellar nuclei abolishes all learned behavior, new and old, using an eyeblink conditioning paradigm (Takahara et al. 2003).

The foregoing needs to be verified by simultaneously recording from both the Purkinje neurons and the cerebellar nuclear neurons, as an animal performs under-learned and overlearned routines. And the effect of paired Purkinje and nuclear inactivation must be compared to the effect of individual inactivation of each area using an assortment of behavioral paradigms. This should help us deduce the parameters that govern the automaticity process by the cerebellum. As suggested by Kahneman (2011) over 50% of behavior produced by humans occurs automatically, which he refers to as ‘thinking fast’.

Footnote:

[1] The behavioral parameters that affect the oscillations between express and regular saccades have yet to be deduced, but task difficulty is one obvious parameter. Nonetheless, during complex tasks such as match to sample or visual discrimination express saccades are abolished in primates (Schiller and Tehovnik 2015).

Figure 33. The cerebellar circuit for classical conditioning. Pathways in green indicate the input fibres to the cerebellar cortex: the mossy fibres (mf), the granular cells (Grc), and the parallel fibres from the granular cells (pf) that synapse onto the Purkinje neurons (Pc) shown in grey. Pathways in red indicate the climbing fibres (cf) from the inferior olive. The output of the cerebellar cortex starting at the Purkinje neurons and terminating on the cerebellar nuclei (the anterior interpositus nucleus, AIN) is illustrated. The unconditioned stimulus (US) pathway is activated by electrically stimulating the inferior olive or the climbing fibres. The conditioned stimulus (CS) pathway is activated by stimulating the mossy fibres (mf CS) or the parallel fibres (pf CS). When the conditioned stimulus coincides with the unconditioned stimulus (CS for 300 ms followed by a US), the conditioned stimulus begins to induce a decrease in the discharge of the Purkinje neurons (top-right inset). A golgi circuit (Gc) is also indicated. Derived from Fig. 1B of Gallistel et al. (2022).

Figure 34. Raster data are shown for one Purkinje neuron whose discharge is abruptly terminated after a conditioned stimulus (electrical activation of the mossy fibres) is delivered after the establishment of conditioning. (A) A total of 20 trials is illustrated. The first blue bar indicates the onset of the conditioning train of pulses, and the second blue bar indicates the offset of the train. The green marker shows the termination of Purkinje discharge, and the red marker shows the resumption of Purkinje discharge. (B) The cumulative histogram of Purkinje discharge for the 20 trials. From Fig. 2AB of Gallistel et al. (2022).