Figure 10.

A schematic illustration of the transformation of signals from SC to motor command for visually guided or microstimulation-evoked saccades. (A,B) Visually guided saccades to retinally identical targets from two different eye positions (i). The site of activity in the SC is the same (ii, colored spots) but the level of activity is lower for the more contralateral initial fixation position [cooler colors in (B) than in (A)]. A site-weighted average of SC activity is calculated, thus removing the eye position signal and producing an output that is the same for different eye positions [iii, same number of spikes in both (A) and (B)] and specifies desired eye displacement. As in common eye displacement models of the oculomotor pulse-step generator, an eye position signal (iv) is combined with this signal, producing a motor command that depends on both eye position and saccade amplitude (v) and generating an accurate saccade to the target (vi). (C,D). How this circuit would have to operate in order to account for the eye position dependence of stimulation-evoked saccades.Stimulation evokes activity at the same site in the SC regardless of initial fixation, but the stimulation combines with endogenous eye position signals to produce a hill of activity that varies in level. The stimulation-evoked activity is too weak to trigger normalization, so the output is lower and is not invariant across eye positions [different numbers of spikes in (Ciii) and (Diii); and fewer spikes than the corresponding (Aiii) and (Biii)]. The motor command that is generated does not have the correct number of spikes to bring the eyes all the way to the (non-existent) target (Cv and Dv), and the saccades fall short by amounts that depend on eye position (Cvi and Dvi).