Figure 2. Time course of neural activity in the variable duration task.

(A) Average activity from 90 leader neurons aligned to the onset of motion. A leader neuron contains one of the choice targets, T⁺, in its response field (shading in diagrams). Colors indicate motion strength and direction. Both correct and error trials are included. Inset shows the effect of signed motion strength on buildup rate during the first 200 ms of putative integration (gray scale bar on the abscissa). Buildup rates are plotted as a function of signed motion strength (line, least-squares regression). Error bars are SE. Leader neurons reflect the sensory evidence bearing on the choice target in the response field, although the next eye movement is to the green, choice-neutral target (T⁰). See also Figure S2.

(B) Average activity of the leader neurons aligned to task events following the motion-viewing epoch. The coherence dependence gives way to a discrete binary representation of the decision outcome before the saccade to T⁰ (arrow 1). The representation disappears after the saccade, and it is recovered as the pursuit eye movement places the gaze at the original FP (arrow 3). There is a perisaccadic response associated with T⁺ choices. Only correct trials are included.

(C and D) Average activity from 177 supporter neurons aligned to the same events as in (A) and (B). The neurons first represent the decision outcome after the saccade to T⁰ (arrow 2). They retain this representation until reacquisition of the original FP at the end of the smooth-pursuit eye movement. Then, they show only a nonselective post-saccadic response, as both T⁺ and T⁻are outside the response field.

(E) Choice selectivity (d’) of individual neurons (rows). The neurons are ordered by the time of maximum d’ up to the 1^st saccade (leader neurons; top) or time of maximum d’ throughout the trial (supporter neurons; bottom). Some supporter neurons represented the choice just before the 1^st saccade and not after. Note that as a population, LIP represents the decision at all times: from motion viewing, through the IEM, to the final saccade to T⁺ or T⁻.