Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
This link contains the FORTRAN source codes used for simulating the baseline logicalrule models, the EBRW model, and the free stimulus drift-rate model from: Fific, M., Little, D.R., & Nosofsky, R.M. (accepted pending revision). Logical-rule models of classification response times: A synthesis of mental-architecture, randomwalk, and decision-bound approaches. Psychological Review. serst.for = serial self-terminating model serexh.for = serial exhaustive model parst.for = parallel self-terminating model parexh.for = parallel exhaustive model coact.for = coactive model ebrw.for = EBRW model freestm.for = free stimulus drift-rate model ************************************************************* Each of the simulation subroutines receives the following inputs and produces the following outputs: intsep = an indicator variable equal to 1 for separable-dimension stimuli and equal to 0 for integral-dimension stimuli (relevant only for EBRW model). abase = parameter vector (described in next section) x = stimulus coordinates matrix [x(i,m) is the value of stimulus i on dimension m] cat = category-membership vector [cat(i) is the category assignment for stimulus i] nstim = number of stimuli [9 in the experiments from Fific et al.] nsim = number of simulations per stimulus [10000 in Fific et al.] is1 = a five digit integer-valued random seed nbin = number of RT quantiles into which the correct RTs are divided quant = quantile cutoff matrix for the correct RTs associated with each stimulus [quant(i,j) is the jth quantile cutoff for stimulus i] theory = theoretical predictions output matrix [theory(i,k) gives the predicted probability that a response for stimulus i is correct and falls in the k'th RT quantile (k=1,nbin) or that the response is incorrect (k=nbin+1)] mrt = predicted mean correct RT output vector [mrt(i) is the predicted mean correct RT for stimulus i] (Continued) ******************************************************************* For the logical-rule models and the EBRW model, the input parameters (abase) are as follows (see Fific et al. for an explanation of the free parameters): abase(1) = px abase(2) = Dx abase(3) = Dy abase(4) = σx abase(5) = σy abase(6) = +A abase(7) = -B abase(8) = mean of normal distribution that is exponentiated abase(9) = standard deviation of normal distribution that is exponentiated Note: μR = exp[abase(8)+.5*abase(9)] σR2 = {exp[abase(9)]-1}*{exp[2*abase(8)+abase(9)]} abase(10) = k ******************************************************************* (Continued) ******************************************************************* For the free stimulus-drift-rate model, the input parameters are as follows: abase(1) = +A abase(2) = -B abase(3) = mean of normal distribution that is exponentiated abase(4) = standard deviation of normal distribution that is exponentiated abase(5) = k abase(6) through abase(14) = p(1) through p(9) ******************************************************************* ******************************************************************* Within each simulation, the following other standard subroutines are called: rand = assigns a (uniformly distributed) random value between 0 and 1 to pran zscor = provides the z value below which a given pran of the cases fall in a standardized normal distribution. pzscor = provides the proportion of cases p that fall below a given z value in a standardized normal distribution. **********************************************************************