Skip to contents

Supported Distributions

Source: vignettes/distributions.Rmd
distributions.Rmd

This vignette lists the accumulator distributions currently available in AccumulatR and the parameter names they use.

## 
## Attaching package: 'AccumulatR'
## The following object is masked from 'package:stats':
## 
##     simulate

Naming convention

Parameter names are built as accumulator.parameter. For example, go.m, stop.shape, and choice.v refer to parameters for the accumulators go, stop, and choice.

Available distributions 

data.frame(
  distribution = c("lognormal", "gamma", "exgauss", "LBA", "RDM"),
  parameters = c(
    "m, s",
    "shape, rate",
    "mu, sigma, tau",
    "v, B, A, sv",
    "v, B, A, s"
  ),
  stringsAsFactors = FALSE
)
##   distribution     parameters
## 1    lognormal           m, s
## 2        gamma    shape, rate
## 3      exgauss mu, sigma, tau
## 4          LBA    v, B, A, sv
## 5          RDM     v, B, A, s
  • lognormal: m and s, the usual log-scale location and spread parameters.
  • gamma: shape and rate.
  • exgauss: mu, sigma, and tau.
  • LBA: v, B, A, and sv.
  • RDM: v, B, A, and s.

Outside of these all accumulators can have a non-decision time t0 and a trigger probability q. For the exgauss t0 and mu are redundant (they both constitute a shift in rt) and one must be fixed.

Example

The same model can combine different accumulator distributions. The only thing that changes is the parameter suffix used in the parameter vector.

model <- race_spec() |>
  add_accumulator("go", "lognormal") |>
  add_accumulator("stop", "exgauss") |>
  add_outcome("go", "go") |>
  add_outcome("stop", "stop") |>
  finalize_model()

params <- c(
  go.m = log(0.30),
  go.s = 0.18,
  stop.mu = 0.10,
  stop.sigma = 0.04,
  stop.tau = 0.08
)

build_param_matrix(model, params, n_trials = 2)
##      q w t0        p1   p2   p3
## [1,] 0 1  0 -1.203973 0.18 0.00
## [2,] 0 1  0  0.100000 0.04 0.08
## [3,] 0 1  0 -1.203973 0.18 0.00
## [4,] 0 1  0  0.100000 0.04 0.08

Choose the distribution that matches the accumulator you want to specify, then use its parameter names consistently in build_param_matrix(), simulate(), and likelihood evaluation.