Anatomical organization in the SC

The superior and inferior colliculi form the roof of the midbrain. In mammals the SC is composed of seven alternating fibrous and cellular layers. On the basis of anatomical and behavioral data, these layers are grouped into two functional units: (1) superficial and (2) deep compartments. The superficial layers (stratum zonale, stratum griseum superficiale, and stratum opticum) receive inputs devoted almost exclusively to vision. Cells in the superficial layers of each colliculus are activated by stimuli appearing in the contralateral visual field and are topographically organized according to receptive field location (Cynader and Berman, 1972). Neurons with receptive fields near the center of the visual field are located anteriorally; those with receptive fields in the periphery are located posteriorly. Cells with receptive fields in the upper visual field are located medially; those with receptive fields in the lower visual field are located laterally. The perifoveal representation is enlarged, with over one-third of the collicular surface devoted to the central 10 degrees of the visual field. The representation of the horizontal meridian runs from anterolateral to posteromedial. The visual signals observed are in retinal coordinates; cells respond to visual stimuli if, and only if, particular regions of the retina are activated. The outputs of the superficial layers are primarily ascending and terminate, for the most part, in various regions of the thalamus, including the pulvinar (see Sparks and Hartwich-Young, 1989, for a review).

In contrast, the intermediate (stratum griesum intermedium, stratum album intermedium) and deeper (stratum griseum profundum and stratum album profundum) layers—collectively, the deep layers—receive sensory inputs of several modalities (e.g., visual, auditory, and somatosensory) and contain neurons with motor properties. In their early description of SC neurons discharging before saccadic eye movements, Wurtz and Goldberg (1972) noted that the neurons have movement fields, that is, each neuron discharges before or during saccades having a particular range of directions and amplitudes. The size of the movement field is a function of the amplitude of the optimal movement. Some neurons that discharge prior to saccades also have visual receptive fields, while other neurons have only movement fields. Neurons discharging in response to visual stimuli and prior to eye movements have overlapping, but not necessarily coextensive, movement and receptive fields (Anderson et al., 1998; Wurtz and Goldberg, 1972).

Based on both neural recording and microstimulation experiments in head-restrained animals, it has been established that saccade direction and amplitude are topographically organized in the deep layers of the SC (Robinson, 1972; Schiller and Stryker, 1972). Neurons discharging prior to small saccades are located anteriorly, and those firing before large saccades are found posteriorly. Cells near the midline discharge prior to movements with up components, and those on the lateral side discharge maximally before movements with down components. Microstimulation of the deep layers produces a saccadic eye movement with an amplitude and a direction similar to those of the optimal vector encoded by the neurons near the tip of the electrode.

Despite the general correspondence between the motor and overlying sensory maps, there is no essential functional linkage between retinotopically coded visual activity in the superficial layers and saccade-related premotor activity in the deep layers of the SC. Vigorous activity may occur in the superficial layers and may not be translated into saccade-related discharge in underlying cells in the deep layers. Conversely, saccade-related activity recorded from neurons in the intermediate and deeper layers may not be triggered by activity of the overlying visual neurons coding retinal error, that is, the distance and direction of the target image from the fovea. Thus, the activity of visual neurons in the superficial layers is neither necessary nor sufficient to produce activation of saccade-related neurons in the deep layers of the underlying colliculus (Mays and Sparks, 1980).

Yet, chemical inactivation in hamster SC (Mooney et al., 1992) and in vitro experiments in SC slices using whole-cell patch-clamp methods (Isa et al., 1998; Lee et al., 1997) have demonstrated synaptic transmission from superficial to intermediate layers. Excitatory postsynaptic potentials, evoked with mono- and polysynaptic latencies, were recorded from neurons in the intermediate layers when the overlying superficial layer was stimulated. In the presence of bicuculline, neurons in the intermediate layers even exhibited a burst upon stimulation of the superficial layers, suggesting that the signal transmission is suppressed by GABAergic inhibition. The burst property of intermediate layer neurons was also facilitated by activation of nicotinic acetylcholine receptors (Isa et al., 1998). Behaviorally, the occurrence of short-latency express saccades increased after microinjections of the acetylcholinergic agonist nicotine in the SC of awake, behaving monkeys (see Kobayahsi et al., 2001, for a review). Thus, this interlaminar circuitry may play a critical role in reducing saccadic reaction time and triggering express saccades, which are absent following ablation of the SC (Schiller et al., 1987). A description of SC participation underlying saccadic initiation has been reviewed recently (Munoz et al., 2000; see Sparks et al., 2000, for a slightly different perspective).

Unlike the interlaminar organization, the neurons within the deep layers likely use local excitation and distant inhibition mechanisms to shape the evolution of the population activity that leads to the generation of each saccade. The electrophysiological (McIlwain, 1982; Munoz and Istvan, 1998) and pharmacophysiological (Meredith and Ramoa, 1998; Pettit et al., 1999) experiments that provide credence for such connectivity monitored extracellular activity while stimulation pulses were delivered to different parts of the collicular map. Thus, intracollicular connections likely shape the spatial and temporal profiles of activity, although a potential confound may be introduced by stimulation of fibers of passage. Before and during each saccade, neurons in approximately 25 to 30% of the collicular map discharge, and the size of the active area remains relatively invariant across saccades of all amplitudes and directions (Anderson et al., 1998; McIlwain, 1975; Munoz and Wurtz, 1995b). The SC neurons project predominantly to brainstem structures that process the collicular commands to produce an appropriate movement.