Figure 5. Focal inactivation of SC impairs decision termination.

(A) Experimental setup. Muscimol was injected unilaterally in the intermediate and deep layers of SC. On half of the experiments, Neuropixels recordings were obtained from ipsilateral area LIP.

(B) Saccadic latencies measured in an instructed delayed saccade task were slowed by ~50 ms relative to pre-injection in a region of the contralateral visual field (heatmap). The left-choice target was placed in the center of this region.

(C) An impairment in the mechanism for detecting the threshold might result in a requirement for stronger signals to achieve termination for ${\mathrm{T}}_{\text{in}}$ choices. This is equivalent to application of higher decision threshold for ${\mathrm{T}}_{\text{in}}$ choices (left panel) while leaving the threshold for ${\mathrm{T}}_{\text{out}}$ choices unchanged (right panel).

(D) Simulated choice proportions (bottom) and RT (top), generated by the model in (C), before (black) and after (green) a 70% increase in the ${\mathrm{T}}_{\text{in}}$ decision threshold. Curves show the fit of a bounded evidence-accumulation model. Inset depicts the predicted effect of SC inactivation on the slope of the choice function.

(E) Choice-RT data before (black) and after (green) unilateral SC inactivation. Inset depicts the model-derived ${\mathrm{T}}_{\text{in}}$ decision threshold before and after SC inactivation. SC inactivation increased the ${\mathrm{T}}_{\text{in}}$ decision threshold by 31.9%.

(F) Same as (E) but for saline/sham injection experiments.

(G) Effects of muscimol (green) and saline/sham (gray) injection on choice bias (top) and the slope of the choice function (bottom). Positive values for choice bias indicate a bias toward ipsilateral (rightward) choices. Asterisk denotes a statistically significant difference (p<.01, likelihood ratio test).

(H) Muscimol-induced slowing of left-choice RTs (top) is substantially larger than the slowing of saccades in (B) (dotted line). The same analysis for the saline controls is shown on the bottom.