Figure 4.

The linear model accurately predicts threshold shifts resulting from stimulation with electrode triplet patterns. A, Depiction of activation threshold as a linear function of currents passed through three electrodes (gray plane, equation), for an ON parasol cell. Vectors show currents associated with a tested triplet pattern (black) and the two pair patterns with the same secondary/primary current ratios as was used in the triplet pattern (orange and teal). Neighboring schematics depict current ratios (represented as diameters of circles) corresponding to each pattern. Threshold predictions lie at the intersection of each vector and the threshold plane. Measured threshold is shown in red. Inset scale bars, 0.1 μA. B–F, Comparison of observed and predicted thresholds to electrode triplet patterns for each of the five tested RGCs. Triplet pattern depicted in A is marked with an arrow and a small black outline in B. Thresholds are plotted as fractional shifts from the model value for primary-alone threshold (f⁻¹(0.5) in Eq. 3). Dashed black lines indicate 1:1 correspondence between predicted and measured values. Coefficients of determination (R²), indicating the proportion of measured variance in the data that is explained by the linear model, are shown in the top left of each panel. Two data points outside the range of the horizontal axis in B are shown as triangles at the axis limits. These two data points correspond to patterns containing high secondary/primary current ratios, which frequently require a piecewise linear model to describe (see below), and are therefore excluded from calculation of R². B, C, ON parasol cells. D, ON midget cell. E, OFF parasol cell. F, ON parasol cell.