Neuronal Substrates of Spatial Transformations in Vestibuloocular and Vestibulocollic Reflexes a

This paper will examine the neuronal mechanisms that allow vestibular reflex systems to produce eye or head rotations that are in the appropriate direction to compensate for perturbations that disturb the angular orientation of the head. After a decade of exploring the dynamic transformations that occur in vestibular and optokinetic reflexes, a number of investigators became interested in analysis of the three-dimensional spatial properties of these reflexes. Robinson' and Pellionisz and GrafZ proposed models of spatial transformations in the vestibuloocular reflex (VOR). Graf,'.' M~Crea, ' .~ Uchino,"" Wilson,"." and their colleagues used electrophysiological and morphophysiological techniques to reveal circuitry involved in these transformations. Simpson and colleagues studied spatial transformations in the rabbit's optokinetic r e f l e ~ , ' ~ while we"-'7 and others'"~'' undertook parallel studies of the cat's VOR and vestibulocollic reflex (VCR). The latter studies have provided information on vestibular reflexes elicited by both semicircular canal and otolith receptors of the vestibular labyrinth. We are closer to understanding the neuronal substrates of the canal-ocular and canal-neck reflexes, which will therefore be the focus of this review. The spatial properties of the input to the canal-ocular and canal-neck reflexes are simple since all afferent fibers of a given canal have the same spatial tuning, which is determined by the geometry of the canal. The directional sensitivity of each afferent of a given canal can thus be represented by a unit vector perpendicular to the plane of that canal extending in the direction corresponding to the rotation that activates that afferent according to the right-hand rule. Throughout this paper we will refer to such a vector as the maximal activation direction (MAD) vector of an afferent, neuron, or muscle. The response of a canal afferent to rotation about other axes is simply the product of the maximal response it exhibits for rotation around its MAD vector multiplied by the cosine of the angle between the axis and its MAD vector. The MAD vectors of the three canals in the cat are well characterized by the work of Curthoys and colleagues.z" The remainder of this review will examine first what spatial transformations take place between the incoming canal signal and muscle output in the VOR and VCR and second how those transformations may be implemented by vestibuloocular and vestibulospinal neurons.