• RWiener package provides R functions for the Wiener diffusion model.
• For further Information check:
  • https://cran.r-project.org/web/packages/RWiener/index.html
  • https://github.com/cran/RWiener/tree/master

The tutorial begins by setting up the R environment and providing a brief overview of the functions and parameters associated with the diffusion model. The functions introduced include dwiener, pwiener, qwiener, rwiener, nlm, and optim. The parameters of the diffusion model discussed are α (boundary separation), 𝜏 (non-decision time), β (initial bias), and δ (drift rate parameter). There's also a note on the Maximum likelihood estimation method using rWiener.

A dataset is generated using the rwiener function with 100 observations and parameters like boundary separation, non-decision time, bias, and drift rate.

This section demonstrates how to obtain the density of a specific quantile using the dwiener function.

The tutorial explains how to calculate the cumulative distribution using the pwiener function.

The inverse CDF function, qwiener, is introduced to find the appropriate quantile for a given probability.

The wiener_plot function is used to visualize the observed lower and upper responses.

The tutorial demonstrates how to fit the diffusion model to the data using optimization algorithms like Nelder-Mead and BFGS. Comparing Model vs. Real Data:

The tutorial provides code to generate modeled data and compare it to real data using density plots.

This section introduces various criteria functions like wiener_likelihood, wiener_deviance, wiener_aic, and wiener_bic to evaluate the fit of the model.

The tutorial provides a comprehensive pipeline for analyzing the data:

• Pipeline Part 1: Parameter Estimation for each participant.
• Pipeline Part 2: Estimation parameters are extracted.
• Pipeline Part 3: Modeled data is generated.
• Pipeline Part 4: Modeled and real data are compared.
• Mean Parameters of Model vs. Real Data: The tutorial concludes by plotting the mean parameters of the model against the real data.

This tutorial, titled "ADVANCED_Tutorial", provides a comprehensive guide on advanced data analysis using the Drift Diffusion Model (DDM) in R. The tutorial is structured as follows:

• Simulated data is generated for two groups (A and B) for multiple participants.
• Helper functions are defined to convert milliseconds to seconds and compute deleted trials.
• Results for deleted trials are printed.

• Reaction time distributions are plotted for different conditions and groups.

• The data is preprocessed.
• The DDM is fit to the data for each participant.
• Results, including model coefficients and information criteria, are saved and printed.

• The tutorial calculates the differences in BIC and AIC between the simple and complex models for each participant.
• Based on these differences, a majority vote is taken to decide the preferred model.

• The tutorial plots the beta and delta parameters for each reward type.

• Paired t-tests are conducted for beta (bias) and delta (drift rate) between group levels.

• Sanity Check: Ensure that optimization routines meet a predefined criterion for convergence. Check that certain parameters (like Non-decision time and boundary separation) adhere to logical constraints.
• Predictive Check: Validates the model by using parameter estimates to generate simulated datasets. The simulated data should resemble key features of the empirical data.
• Parameter Recovery Study: After simulating data, the DDM is applied to see if the original parameters can be recovered.