Utility: Model: Utility

In binaryRL: Reinforcement Learning Tools for Two-Alternative Forced Choice Tasks

Model: Utility

Description

U(R) = {R}^{\gamma}

V_{new} = V_{old} + \eta \cdot (U(R) - V_{old})

Usage

Utility(params)

Arguments

params

[vector] algorithm packages accept only one argument

Value

loss [numeric]

algorithm packages accept only one return

Examples

## Not run: 
Utility <- function(params) {
  res <- binaryRL::run_m(
    data = data,
    id = id,
    eta = c(params[1]),
    gamma = c(params[2]),
    tau = c(params[3]),
    priors = priors,
    n_params = n_params,
    n_trials = n_trials,
    mode = mode,
    policy = policy
  )

  assign(x = "binaryRL.res", value = res, envir = binaryRL.env)
  loss <- switch(EXPR = estimate, "MLE" = -res$ll, "MAP" = -res$lpo)
  switch(EXPR = mode, "fit" = loss, "simulate" = res, "replay" = res)
}

## End(Not run)

binaryRL documentation built on Aug. 21, 2025, 6:01 p.m.