Fig. 6. Cross-trial representations of task variables.

a, b The relevant signal variance (RSV) for the present-trial sRAs of benefit (green), choice (orange), and expected reward (blue) was computed with respect to the previous-trial (left panels) or present-trial (right panels) variables and plotted as a function of time from fixation (left panels) or offer onset (right panels) on the present trial for monkeys N (a) and K (b). Peak RSV for benefit (a, right panel) is not shown so as to conserve the vertical scale across axes. Horizontal gray bars and solid gray squares indicate the 2.5-to-97.5 percentile range and median, respectively, of previous-trial events, as labeled. RSV during the previous trial (left panels) was smaller than during the present trial (right panels) likely due to the wide temporal variability between trials—reward delivery of 0–3 s depending on reward number and ITI of 1–2 s depending on when animal initiated fixation—leading to temporal smearing of otherwise robust, temporally aligned responses. c, d Log₁₀ probability (one-sided, uncorrected) of data in a and b, respectively, was derived empirically in comparison to random dimensions (see “Methods”). Note: Offer-aligned axes (right panels) of a–d recapitulate Fig. 4c–f and are provided for reference. e, f The RSV for the previous-trial sRAs of previous benefit (green), previous choice (orange), and experienced reward (blue) was computed with respect to the previous-trial variables for both temporal alignments (unlike for a, b, where offer-aligned RSV was with respect to present-trial variables). Other conventions as in a and b. g, h Log₁₀ probability (one-sided, uncorrected) of data in e and f, respectively, was derived empirically in comparison to random dimensions (see “Methods”). Horizontal dashed lines in c, d, g, and h correspond to p = 0.05. As for the present-trial variables, individual units were selective for multiple previous-trial variables (i.e., mixed), and previous-trial sRAs were specific to their targeted variable and largely separable (Supplementary Figs. 26–29 and Supplementary Table 4).