Nadia L Cerminara
Department of Physiology and Pharmacology, School of Medical Sciences, University of Bristol, Bristol, UK

https://doi.org/10.36866/pn.78.16

For activities as diverse as catching a ball, or reaching out to pick up a cup of coffee, the cerebellum is thought to play a prominent role in the accurate execution of visually guided movements. But how exactly the cerebellum controls the interactions between our limbs and the external environment is a matter of debate. Behavioural experiments have shown that the minimum interval needed for visual information to influence an ongoing movement is approximately 80–100 ms. This neural delay time in sensory processing and motor execution is far too long to permit effective feedback control; therefore, one must be able to predict and anticipate the position and trajectory of the target. One structure that seems particularly critical in this prediction process is the cerebellum.

Predictive behaviour can be achieved through internal models (Miall et al. 1993; Wolpert et al. 1995). Broadly speaking, internal models are defined as neural representations of our bodies and objects in the external environment and can be of two types: forward and inverse. Forward models make predictions about the behaviour of the motor system and external objects whereas inverse models transform a desired goal into the appropriate plan of action. The cerebellum is thought to form internal models through a learning process in which a simulation of the desired movement is constructed and modified by repeated practice of the movement. This allows us to rapidly and precisely execute a desired movement without depending exclusively on on-line sensory feedback from the moving body part to guide the movement.

In the context of moving our limb towards an object, say a cup of coffee, a forward model would predict the position or velocity of the limb, whereas an inverse model would transform the desired trajectory of the limb into the appropriate joint forces/torques. However, there is disagreement in the literature regarding the existence of inverse versus forward models in the cerebellum. On the one hand, Purkinje cell activity in regions of the cerebellum concerned with eye movements have been interpreted as representing inverse models (Gomi et al. 1998). On the other hand, in a recent study of non-human primates performing a circular manual tracking task under various force loads, Purkinje cell activity was not altered with the change in hand force and arm muscle activity that occurred as a result of varying the loads (Pasalar et al. 2006), leading to the conclusion that rather than an inverse dynamics model of the arm, Purkinje cell activity represents the kinematic output of arm movements. This kinematic representation may in fact correspond to the output of a forward model that predicts the consequences of limb movements. Therefore, in relation to control of parts of the body, such as the eyes or limbs, the debate between forward and inverse models remains unresolved.

One way to inform the debate is to seek evidence for the existence of internal models in the cerebellum of the movement of objects in the external world. The situation here is simpler; if an internal model of an external moving object exists, it can only be of the forward type, making predictions about the object’s future position and velocity. When pursuing a moving target, it is necessary to overcome the delays in visual feedback by predicting the future location of the target based on its motion. It has been suggested that on-line visual information is combined with a representation of target kinematics to make an internal model and hence predict future target location (Barborica & Ferrera, 2003). Indirect evidence for the operation of internal models in the lateral cerebellum of objects and tools in the external world has been obtained in human imaging studies (Imamizu et al. 2000). However, to obtain direct evidence recordings from individual cerebellar neurones are required.

Previous work from our laboratory (Miles et al. 2006) in cats trained to perform a visually guided reach–retrieval task has shown that Purkinje cells in the lateral cerebellum, a region known to be involved in the visual guidance of movement (Stein & Glickstein, 1992), were responsive to the on-going motion of the visual target, displaying tonically altered rates of simple spike discharge for as long as the target was moving (Fig. 1A). The altered tonic discharge rate was found to encode the speed of the target, as individual Purkinje cells displayed a ‘preferred’ target velocity when tested against a range of speeds (Fig. 1B). Since the cats were familiar with the motion of the external target which moved in a predictable fashion, could this pattern of neural activity reflect the operation of a forward model? Alternatively, was the neuronal activity simply coding the on-line motion of the target driven by the visual stimulus?

In order to distinguish between these two possibilities, temporary visual denial of the target can be used. Whilst invisible, the tonic increase in discharge rate usually seen whilst the visible target is moving would be expected to disappear if the cells were being passively driven by the sensory stimulus whereas such activity would be expected to survive if a forward model of target motion had been formed (Fig. 1C).

Again we used single unit recordings in cats trained to perform a predictable visually guided reaching task. Cats were trained to reach (after receipt of a ‘go’ signal) into a moving visual target travelling horizontally at a constant velocity. The target for reach consisted of a hollow Perspex tube dimly lit by a ring of LEDs. The tube was initially stationary 7 cm to the left of centre (as viewed by the cat) at a comfortable height for reaching (Fig. 2A). The tube then moved at a constant velocity rightwards across the cat’s visual field (Fig. 2B). Experiments were conducted without ambient illumination in a light-proof room. Thus, the only source of visual information available to the cat was from the target LEDs. At various stages of the target’s motion, illumination of the ring of LEDs around the tube was temporarily extinguished during which time the animal was in total darkness.

For target-related Purkinje cells that displayed tonically altered simple spike activity during on-going movement of the visual target, there was a similar pattern of tonic activity when the cat’s view of the target was occluded (Fig. 2C and D). This result therefore shows that visual feedback is not necessary to maintain the pattern of discharge. Instead, the finding is consistent with the hypothesis that a forward model of target movement has been constructed which predicts the target’s velocity and position and thereby maintains neural activity in the absence of sensory information. Such a mechanism is likely to be important for movement planning and control for the interception of a moving object – just the sort of skilled movement that is severely affected when the cerebellum is damaged. Further experiments are required to address the learning process involved in acquiring such  models.

References

Barborica A & Ferrera VP (2003). Estimating invisible target speed from neuronal activity in monkey frontal eye field. Nat Neurosci 6, 66–74.

Cerminara NL, Apps R & Marple-Horvat DE (2009). An internal model of a moving visual target in the lateral cerebellum. J Physiol 587, 429–442. http://jp.physoc.org/content/587/2/429.long

Gomi H, Shidara M, Takemura A, Inoue Y, Kawano K & Kawato M (1998). Temporal firing patterns of Purkinje cells in the cerebellar ventral paraflocculus during ocular following responses in monkeys I. Simple spikes. J Neurophysiol 80, 818–831.

Imamizu H, Miyauchi S, Tamada T, Sasaki Y, Takino R, Putz B, Yoshioka T & Kawato M (2000). Human cerebellar activity reflecting an acquired internal model of a new tool. Nature 403, 192–195.

Miall RC, Weir DJ, Wolpert DM & Stein JF (1993). Is the cerebellum a Smith predictor? J Mot Behav 25, 203–216.

Miles OB, Cerminara NL & Marple-Horvat DE (2006). Purkinje cells in the lateral cerebellum of the cat encode visual events and target motion during visually guided reaching. J Physiol 571, 619–637.

Pasalar S, Roitman AV, Durfee WK & Ebner TJ (2006). Force field effects on cerebellar Purkinje cell discharge with implications for internal models. Nat Neurosci 9, 1404–1411.

Stein JF & Glickstein M (1992). Role of the cerebellum in visual guidance of movement. Physiol Rev 72, 967–1017.

Wolpert DM, Ghahramani Z & Jordan MI (1995). An internal model for sensorimotor integration. Science 269, 1880–1882.

Acknowledgements

The research was supported by the BBSRC and The Wellcome Trust.