Figure 3.

Summary of the Microstimulation Effect

(A) This figure illustrates the shifts when we fitted the standardized psychophysical data pooled across microstimulation sites in monkey FLE. The range of cylinder disparities was standardized for all 20 microstimulation sites, such that the largest disparities at each site were set equal to −1 and 1. Data were averaged across sites. The proportion of choices in the preferred direction was plotted against the standardized cylinder disparity, and pairs of cumulative Gaussians were fitted. Data from all sites are shown in green; data for sites with a significant shift are shown in black and gray. For the significant stimulation sites, one pair of fits, for microstimulated (in gray) and nonmicrostimulated trials (in black), was only allowed to differ in their horizontal offset (solid line); the other pair in offset and slope (1/SD) of the fitted curve (dashed line). For FLE, the pooled shift measured +0.42 and the SD (which is a measure of psychometric threshold) was 0.64. This yields a ratio of shift/SD of 0.66, equivalent to almost two thirds of threshold. The pooled data for all microstimulation sites (green) shows an almost identical pattern (shift/SD = 0.62). The difference is small because most sites showed a significant shift and there were two significant shifts in the nonpreferred direction. Error bars depict the SEM.

(B) For monkey ICA, response data were standardized and pooled for microstimulation sites with significant shifts (in black and gray) and across all sites (in green), as in (A). The pooled shift for significant microstimulation sites measured +0.39 and the SD 0.66. The shift is equivalent to about two thirds of threshold (SD), with a ratio of shift/SD 0.59 (for all sites shift/SD = 0.36).

(C) The distribution of normalized microstimulation shifts across all 48 sites from both monkeys is shown. The majority of significant horizontal shifts with electrical microstimulation (black bars; χ², p < 0.05) are in the direction predicted by the preference at the stimulated site (shift/threshold > 0), with only three significant shifts in the null direction. Some shifts of the psychometric curve are larger than discrimination threshold (SD) (shift/threshold > 1).

(D) Normalized shifts plotted separately for sites with a preference for far and near disparities. While there were overall fewer sites with a preference for far disparity, there seemed to be little difference in the distribution.

(E) Normalized shifts were plotted separately for sites with a preference for CW or CCW response preference. No significant differences between the distributions of normalized shifts were observed for all comparisons (χ², p > 0.05).

See also Figure S3.