Fig. 3: Modelling discrimination decisions.
From: Prior information differentially affects discrimination decisions and subjective confidence reports
a Model simulations of decision accuracy. Target decisions were simulated from the Flexible model across different posterior levels, between the two conditions, and at three different values of wchoice: 1 (optimal weighting of prior information), 0.33 (overweighting of prior), and 3 (underweighting of prior), shown from left to right respectively. These values are representative of the range found in the data, and capture either over- or underestimated variance by a factor of 3. The resulting decision accuracies are shown here. The model predicts that optimally using prior information will lead to no difference in accuracy between the two conditions, whereas overweighting prior information will lead to higher accuracy when the lead is stronger, and underweighting prior information will lead to higher accuracy when the target is stronger. b Data and predictions of the individually fit model. The left panel shows the mean observed accuracies per posterior level and condition. The right panel shows the predicted accuracies generated from sampling each individual participant’s fit posterior parameter distributions 10 times and simulating 720 trials for each of those sampled parameters, also using that participant’s staircased coherences, internal noise and decision bias. Error bars capture standard deviation (SD) of accuracies across participants (N = 20). c Hierarchical model posterior distribution for wchoice. The posterior distribution for the group mean parameter of the weighting of prior information in the decision, wchoice. The blue shaded region shows the 89% credible interval and the vertical black dashed line reflects optimal weighting of the prior in the decision (wchoice = 1). A weight above 1 captures overestimation of the variance and hence underweighting of the prior. Source data are provided as a Source Data file.
