The superior colliculus: A window for viewing issues in integrative neuroscience

The superior colliculus (SC) serves as a fruitful site for investigating a variety of interesting problems in integrative neuroscience. Situated at the interface between sensory and motor processing, the SC contains computational maps of cells responsive to visual, auditory and somatosensory stimuli as well as cells which fire before saccadic eye and head movements. In this chapter the computational issues of the coordination of different motor programs, population coding of saccades, target selection, derivation of eye position signals from corollary discharge, and coordinate transformations necessary for sensorimotor integration will be discussed with respect to the SC. The intermediate and deeper layers of the superior colliculus (SC) are sites where auditory, visual, and somatosensory signals converge and an area where motor commands for orienting movements of the eyes and head are generated (see Sparks, 1986; Sparks and Mays, 1990 for recent reviews). Studies of the SC concerned with the translation of sensory signals into motor commands have been forced to address some of the most important and perplexing problems of integrative neuroscience. In the process, significant advances have been made in understanding: a) the role of computational maps in processing sensory and motor signals; b) the coordination of different motor programs; c) the extraction of information from the spatial and temporal pattern of activity in populations of neurons; d) the role of corollary discharge in guiding movements; and e) mechanisms for selecting a single response for execution from a large repertoire of potential responses. The experimental strategies developed for studying the SC have also been fruitful when applied to studies of other brain areas. In this chapter, we review the advances listed above, emphasizing studies related to the question of sensorimotor integration-the translation of sensory signals into motor commands. Computational maps Many neural computations are performed by computational maps in the brain (see Knudsen, du Lac, and Esterly, 1987, for a review). For example, sensory neurons may be organized in two-or three-dimensional arrays with systematic variations in the value of a computed parameter (e.g., the angle of orientation of a line segment, the direction or velocity of stimulus movement) across each dimension of the array. Neurons, as elements of the map, may be thought of as an array of preset processors or filters, each tuned slightly differently, and operating in parallel on incoming signals. Consequently, signals are transformed, almost instantaneously, into a distribution of neural activity within the computational map. The …