Problem 2
Using simulated annealing I fit 5 different shifted Wald models to the experimental data that
you visualized in Homework 2. The shifted Wald approximates a diffusion to one boundary, and
has 3 parameters – gamma (drift rate), alpha (boundary separation or caution)
and theta (non decision time or shift).
The results of these analysis are provided in 5 separate mat files.
described in more detail below.
The analysis produced different number of parameter estimates in each model depending
on the details of the model.
All of the analysis only used the correct trials, i.e.,rt(correct == 1). You need to figure out the
number of samples used in the analysis to do the homework.
Each of the files has the same set of variables, but there are only two important ones
fun – the value of the negative log likelihood at the minimum.
x – the parameters of the model.
Model 1: simple.mat
This is the simplest model, where a single estimate of gamma, alpha, and theta were obtained for
all of the data.
x = [gamma, alpha,theta]
Model 2: cond3_gamma.mat
In this model, gamma was allowed to vary between conditions, but alpha and theta took one value
for all 3 conditions
x = [gamma[1],gamma[2],gamma[3],alpha,theta]
Model 3: cond3_gamma_alpha.mat
In this model, gamma and alpha were alllowed to vary between conditions, but theta took one value
x = [gamma[1],gamma[2],gamma[3],alpha[1],alpha[2],alpha[3],theta]
Model 4: simple_ind.mat
In this model, gamma, alpha, and theta were allowes to take different values for each subject.
x has 102 parameters:
34 values of gamma,
34 values of alpha,
34 values of theta.
Model 5: cond3_gamma_ind.mat
In this model, single subject estimates are obtained for gamma, alpha, and theta. In addition
gamma is allowed to vary between conditions.
x has 170 parameters:
34 values of gamma in condition 1,
34 values of gamma in condition 2,
34 values of gamma in condition 3,
34 values of alpha,
34 values of theta
The post different shifted Wald models to the experimental data appeared first on My Assignment Online.