Fig. 6: Predictive relationship between choice and confidence signals in LIP.
From: Neural basis of concurrent deliberation toward a choice and confidence judgment
a, Neural DV from the choice decoder as a function of time, aligned to saccade onset for leftward (contralateral) choices and 0% coherence trials only. Traces are separated by whether the wager decoder predicted a high (purple) or low (gold) bet. Shaded regions are s.e.m. Black bar at the top indicates a statistically significant difference between the traces (one-tailed Wilcoxon signed-rank test with Šidák correction). b, Same as a but for rightward (ipsilateral) choice trials. c, log odds from the choice decoder (leftward choices only) as a function of the probability of a high bet predicted by the wager decoder, based on the time window from −0.2 s to 0.1 s relative to the saccade. Each dot is an individual trial and the black line is a linear regression. Unlike a, purple and gold represent the behavioral wager outcome and not the decoder prediction. d, Same as c but for rightward choice trials. e, Corrected R2 values from a linear regression relating trial-by-trial choice decoding strength and wager decoder probability (P(high)), as a function of time lag between them. Values are computed using the time window 0 to 0.4 s from motion onset (MO). The s.e. is indicated by the shaded regions (barely visible). f, Same as e but using the window −0.4 to 0 s relative to saccade onset (SO). g, Same as e but using the window −0.2 to 0.2 s from saccade onset.
