Fig 1.
Go-RT slowing in Experiment 1.
A participant’s Go-RTs for one condition. The line indicates the increasing number of Change-trials in that condition, the circles are Go-RTs.
Fig 2.
Likelihood estimates during Experiment 1.
The PSI marginal method’s posterior distributions for the example participant. Probabilities are averaged over the error rate dimension, colors indicate the probability of a parameter combination, the point of maximum probability is indicated in black. Colors range from minimum to maximum probability per distribution. Around Change-trial 9, the most likely threshold parameter reaches the boundaries of the method’s parameter space.
Fig 3.
Accuracy throughout Experiment 1.
The proportion of successful Change-trials (averaged over bins of 8 trials) of the PSI marginal and Staircase methods. Participants were split into two groups based on whether they showed substantial Go-response slowing.
Fig 4.
Different methods’ performance with stationary GO-RTs.
Performance of the PSI marginal method, PSI adjusted method, and the staircase method with stationary Go-RTs. (A) Correlations of estimated SSRTs with simulated SSRTs. (B) Mean absolute deviation between estimated and simulated SSRTs. (C) Linear slope between estimated and simulated SSRTs. Filled areas indicate the standard error of the mean over the 150 simulated experiments.
Fig 5.
Different methods’ performance with slowing GO-RTs.
Performance of the PSI marginal method, PSI adjusted method, and the staircase method with increasing Go-RTs. (A) Correlations (B) Mean absolute deviation (C) Linear slope. Filled areas indicate the standard error of the mean.
Fig 6.
PSI marginal method’s performance for different degrees of slowing inside parameter space.
Panel A shows correlation coefficients between simulated and estimated SSRTs, panel B shows mean absolute deviation between simulated and estimated SSRTs, panel C shows mean linear slope relating simulated and estimated SSRTs. Colors indicate different degrees of response slowing. Until trial 100 (red line), all simulated participants are inside the method’s respective parameter space.
Fig 7.
PSI adjusted method’s performance for different degrees of slowing inside parameter space.
As in Fig 6, but now the performance of the PSI adjusted method is depicted.
Fig 8.
Marginal entropy of the PSI marginal method over the course of the task.
The marginal entropies of the threshold, slope and error rate parameters are shown, over the course of the task, for the PSI marginal method. Line pattern indicates which parameter’s entropy is shown. The upper panel shows the results for 5 ms of slowing per Stop-trial, the middle panel for 10 ms, and the lower panel for 15 ms.
Fig 9.
Go-RT slowing in Experiment 2.
As Fig 1, a participant’s Go-RTs for one condition in the PSI adjusted method’s block. The line indicates the increasing number of Change-trials, the circles are Go-RTs.
Fig 10.
Likelihood estimates during Experiment 2.
As Fig 2, the PSI adjusted method’s posterior distribution for the example participant. Colors indicate probability of the parameter combinations, the error rate dimension was averaged over. Despite considerable Go-response slowing, the parameters receiving the highest probability do not reach the upper boundaries of the parameter space.
Fig 11.
Parameter estimates for Experiment 1 and 2.
Highest probability parameters for the PSI marginal method (Experiment 1) and PSI adjusted method (Experiment 2) over the course of the experiment. The participants were split intro groups depending on whether the showed go-response slowing. Every line represents one of the stop-change task’s conditions of one participant; thus, there are four lines per participant. Black lines indicate a parameter reaching the upper end of the parameter space, red lines indicate parameters reaching the lower end. Parameter estimates were slightly smoothed using a cubic spline to facilitate visual inspection. Note that while maximum probability parameters are shown, SSRTs are eventually computed based on the expected parameters.
Fig 12.
Accuracy throughout Experiment 2.
As Fig 3, the proportion of successful Change-trials for the PSI adjusted and Staircase methods, separately for slowed and non-slowed participants.