File:Developing-Bayesian-adaptive-methods-for-estimating-sensitivity-thresholds-(d′)-in-Yes-No-and-Video1.ogv

English: The qYN method applied to estimate the sensitivity threshold in simple detection. The true YN psychometric function (black line; upper left panel) and the qYN's updating estimate of the psychometric function (red dotted line) are presented. The trial history (upper middle panel) presents the signal contrast presented on each trial and the subsequent response. Each dot represents the contrast of the selected stimulus, as a function of trial number, and the dot's color represents whether the observer responded Yes (blue), or No (black). Shaded regions that correspond to confidence intervals for the sensitivity threshold estimates (65% in dark red and 95% in light red) as a function of trial number, are overlaid on the trial history. In the adjacent panel (upper right), calculation of the stimulus selection algorithm is presented and the true threshold value is presented for reference (dotted line); expected information gain is normalized by the maximum of the expected information gain estimate. Therefore, the selected stimuli correspond to peaked regions of the information gain function. The Bayesian update of the pdf defined over three qYN parameters (threshold and steepness of the sensitivity function and decision criterion) is presented. The one-dimensional marginals for two sensitivity parameters (threshold and steepness) and one decision parameter are presented (lower panels); the ground truth for each parameter is presented for reference (dotted lines).