Temporal properties of prior estimation.

<p>(a) Relationship between the average underlying probe priors and average estimated priors for one subject using an window of <b><i>w</i></b> = 21 trials. (b) Results of aggregating the correlation shown in (a) across different subjects and different window sizes (or α values). The argmax values refer to the window sizes and α values that resulted in maximum correlation coefficients for the probe priors estimated from correct trials. (c) Data (circles) and logistic regression fit (solid line) between detection accuracies and prior probabilities estimated for one subject using a window size of <b><i>w</i></b> = 19 trials. Priors were put in discrete bins for illustration, but were treated as a continuous variable in all analyses. (d) Results of aggregating the fits (RSS, residual sums of squares) like the one shown in (c) across different subjects and different window sizes (or α values). The argmin values refer to the window sizes and α values that resulted in the minimum fit errors for the probe priors estimated from correct trials.