binaryRL (version 0.9.0)

rpl_e: Step 4: Replaying the experiment with optimal parameters

Description

After completing Step 3 using `fit_p` to obtain the optimal parameters for each subject and saving the resulting CSV locally, this function allows you to load that result dataset. It then applies these optimal parameters back into the reinforcement learning model, effectively simulating how the "robot" (the model) would make its choices.

Based on this generated dataset, you can then analyze the robot's data in the same manner as you would analyze human behavioral data. If a particular model's fitted data can successfully reproduce the experimental effects observed in human subjects, it strongly suggests that this model is a good and valid representation of the process.

Usage

rpl_e(
  data,
  id = NULL,
  result,
  model,
  model_name,
  param_prefix,
  n_trials = NULL
)

Value

A list, where each element is a data.frame representing one subject's results. Each data.frame includes the value update history for each option, the learning rate (eta), utility function (gamma), and other relevant information used in each update.

Arguments

data

[data.frame] This data should include the following mandatory columns:

• "sub"

• "time_line" (e.g., "Block", "Trial")

• "L_choice"

• "R_choice"

• "L_reward"

• "R_reward"

• "sub_choose"

id

[vector] Specifies the subject ID(s) for whom optimal parameters are to be fitted. If you intend to fit all subjects within your dataset, it is strongly recommended to use id = unique(data$Subject). This approach accounts for cases where subject IDs in the dataset may not be simple numeric sequences (e.g., "1", "2", "3", "4") or may contain string entries (e.g., "1", "2", "3", "004"). Using id = 1:4 could lead to errors if IDs are not sequentially numbered integers.

default: id = NULL

result

[data.frame] Output data generated by the `fit_p()` function. Each row represents model fit results for a subject.

model

[function] A model function to be applied in evaluating the experimental effect.

model_name

[character] A character string specifying the name of the model to extract from the result.

param_prefix

[character] A prefix string used to identify parameter columns in the `result` data

default: param_prefix = "param_"

n_trials

[integer] Represents the total number of trials a single subject experienced in the experiment. If this parameter is kept at its default value of `NULL`, the program will automatically detect how many trials a subject experienced from the provided data. This information is primarily used for calculating model fit statistics such as AIC (Akaike Information Criterion) and BIC (Bayesian Information Criterion).

default: n_trials = NULL

Examples

if (FALSE) {
list <- list()

list[[1]] <- dplyr::bind_rows(
  binaryRL::rpl_e(
    data = binaryRL::Mason_2024_Exp2,
    result = read.csv("../OUTPUT/result_comparison.csv"),
    model = binaryRL::TD,
    model_name = "TD",
    param_prefix = "param_",
  )
)

list[[2]] <- dplyr::bind_rows(
  binaryRL::rpl_e(
    data = binaryRL::Mason_2024_Exp2,
    result = read.csv("../OUTPUT/result_comparison.csv"),
    model = binaryRL::RSTD,
    model_name = "RSTD",
    param_prefix = "param_",
  )
)

list[[3]] <- dplyr::bind_rows(
  binaryRL::rpl_e(
    data = binaryRL::Mason_2024_Exp2,
    result = read.csv("../OUTPUT/result_comparison.csv"),
    model = binaryRL::Utility,
    param_prefix = "param_",
    model_name = "Utility",
  )
)
}