feat: edit prl_fictitious_rewmag model

Author: bugoverdose

R/R/prl_fictitious_rewmag.R

@@ -14,8 +14,8 @@
 #' @templateVar LENGTH_DATA_COLUMNS 5
 #' @templateVar DETAILS_DATA_1 \item{subjID}{A unique identifier for each subject in the data-set.}
 #' @templateVar DETAILS_DATA_2 \item{choice}{Integer value representing the option chosen on that trial: 1 or 2.}
-#' @templateVar DETAILS_DATA_3 \item{rewardMagnitude1}{The reward amount associated with option 1 on a given trial, which is paid out if option 1 is chosen and is the correct choice (1~99).}
-#' @templateVar DETAILS_DATA_4 \item{rewardMagnitude2}{The reward amount associated with option 2 on a given trial, which is paid out if option 2 is chosen and is the correct choice (1~99).}
+#' @templateVar DETAILS_DATA_3 \item{rewardMagnitude1}{The amount of reward paid out if option 1 is chosen and is the correct choice (0~99).}
+#' @templateVar DETAILS_DATA_4 \item{rewardMagnitude2}{The amount of reward paid out if option 2 is chosen and is the correct choice (0~99).}
 #' @templateVar DETAILS_DATA_5 \item{outcome}{Integer value representing the outcome of that trial (where reward == 1, and loss == -1).}
 #' @templateVar LENGTH_ADDITIONAL_ARGS 0
 #' 
@@ -39,7 +39,7 @@ prl_fictitious_rewmag <- hBayesDM_model(
     "eta" = c(0, 0.5, 1),
     "alpha" = c(-Inf, 0, Inf),
     "beta" = c(0, 1, 10),
-    "gamma" = c(0, 0.5, 2)
+    "gamma" = c(0, 0.5, 1)
   ),
   additional_args = NULL,
   regressors      = list(

commons/models/prl_fictitious_rewmag.yml

@@ -21,11 +21,9 @@ data_columns:
   subjID: A unique identifier for each subject in the data-set.
   choice: "Integer value representing the option chosen on that trial: 1 or 2."
   rewardMagnitude1:
-    The reward amount associated with option 1 on a given trial,
-    which is paid out if option 1 is chosen and is the correct choice (1~99). 
+    The amount of reward paid out if option 1 is chosen and is the correct choice (0~99).
   rewardMagnitude2:
-    The reward amount associated with option 2 on a given trial,
-    which is paid out if option 2 is chosen and is the correct choice (1~99). 
+    The amount of reward paid out if option 2 is chosen and is the correct choice (0~99).
   outcome:
     Integer value representing the outcome of that trial (where reward == 1,
     and loss == -1).
@@ -41,7 +39,7 @@ parameters:
     info: [0, 1, 10]
   gamma:
     desc: reward magnitude sensitivity
-    info: [0, 0.5, 2]
+    info: [0, 0.5, 1]
 regressors:
   ev_c: 2
   ev_nc: 2

commons/stan_files/prl_fictitious_rewmag.stan

@@ -7,12 +7,12 @@
  */
 
 data {
-  int<lower=1> N;                       // Number of subjects
-  int<lower=1> T;                       // Maximum number of trials across subjects
-  int<lower=1, upper=T> Tsubj[N];       // Number of trials/blocks for each subject
-  real rewardMagnitude[N, T, 2];        // Expected reward of each option for each trial for each subject (1~99)
-  int<lower=-1, upper=2> choice[N, T];  // The choices subjects made
-  real outcome[N, T];                   // The outcome
+  int<lower=1> N;                                   // Number of subjects
+  int<lower=1> T;                                   // Maximum number of trials across subjects
+  int<lower=1, upper=T> Tsubj[N];                   // Number of trials/blocks for each subject
+  real<lower=0, upper=99> rewardMagnitude[N, T, 2]; // Expected reward of each option for each trial for each subject
+  int<lower=-1, upper=2> choice[N, T];              // The choices subjects made
+  real<lower=-1, upper=1> outcome[N, T];            // The outcome
 }
 
 transformed data {
@@ -39,12 +39,12 @@ transformed parameters {
   vector<lower=0, upper=1>[N] eta;
   vector[N] alpha;
   vector<lower=0, upper=10>[N] beta;
-  vector<lower=0, upper=2>[N] gamma;
+  vector<lower=0, upper=1>[N] gamma;
 
   for (i in 1:N) {
     eta[i]   = Phi_approx(mu_pr[1] + sigma[1] * eta_pr[i]);
     beta[i]  = Phi_approx(mu_pr[3] + sigma[3] * beta_pr[i]) * 10;
-    gamma[i] = Phi_approx(mu_pr[4] + sigma[4] * gamma_pr[i]) * 2;
+    gamma[i] = Phi_approx(mu_pr[4] + sigma[4] * gamma_pr[i]);
   }
   alpha = mu_pr[2] + sigma[2] * alpha_pr;
 }
@@ -79,15 +79,19 @@ model {
 
     for (t in 1:(Tsubj[i])) {
       // Compute action probabilities with reward magnitude
-
       ev_diff = (ev[1] - ev[2]) + (gamma[i] * (rewardMagnitude[i,t,1] - rewardMagnitude[i,t,2]) / 99.0);
       prob_logits[1] = beta[i] * (alpha[i] - ev_diff);
       prob_logits[2] = 0;
       choice[i, t] ~ categorical_logit(prob_logits);
 
       // Prediction error
-      reward             = outcome[i,t] * (rewardMagnitude[i,t,choice[i, t]] / 99.0);
-      reward_of_unchosen = (-outcome[i,t]) * (rewardMagnitude[i,t,3-choice[i, t]] / 99.0);
+      if (outcome[i,t] > 0) {
+        reward             = rewardMagnitude[i, t, choice[i, t]] / 99.0;
+        reward_of_unchosen = 0;
+      } else {
+        reward             = 0;
+        reward_of_unchosen = rewardMagnitude[i, t, 3-choice[i, t]] / 99.0;
+      }
       PE   = reward             - ev[choice[i, t]];
       PEnc = reward_of_unchosen - ev[3-choice[i, t]];
 
@@ -103,7 +107,7 @@ generated quantities {
   real<lower=0, upper=1> mu_eta;
   real mu_alpha;
   real<lower=0, upper=10> mu_beta;
-  real<lower=0, upper=2> mu_gamma;
+  real<lower=0, upper=1> mu_gamma;
 
   // For log likelihood calculation
   real log_lik[N];
@@ -136,7 +140,7 @@ generated quantities {
   mu_eta    = Phi_approx(mu_pr[1]);
   mu_alpha  = mu_pr[2];
   mu_beta   = Phi_approx(mu_pr[3]) * 10;
-  mu_gamma  = Phi_approx(mu_pr[4]) * 2;
+  mu_gamma  = Phi_approx(mu_pr[4]);
 
   { // local section, this saves time and space
     for (i in 1:N) {
@@ -166,8 +170,13 @@ generated quantities {
         y_pred[i, t] = categorical_logit_rng(prob_logits);
 
         // prediction error
-        reward             = outcome[i,t] * (rewardMagnitude[i,t,choice[i, t]] / 99.0);
-        reward_of_unchosen = (-outcome[i,t]) * (rewardMagnitude[i,t,3-choice[i, t]] / 99.0);
+        if (outcome[i,t] > 0) {
+          reward             = rewardMagnitude[i, t, choice[i, t]] / 99.0;
+          reward_of_unchosen = 0;
+        } else {
+          reward             = 0;
+          reward_of_unchosen = rewardMagnitude[i, t, 3-choice[i, t]] / 99.0;
+        }
         PE   = reward             - ev[choice[i, t]];
         PEnc = reward_of_unchosen - ev[3-choice[i, t]];