Simulate from a generalized drift-diffusion model

Draws choices and response times by solving the model's own Fokker-Planck equation and sampling the resulting defective densities, so the draws come from the density gddm() fits rather than from a discretized forward simulation, whose first passages are late by however much the step misses excursions between monitoring times.

Usage

gddm_simulate(
  n,
  ...,
  coh = 0,
  drift = gddm_drift_constant(),
  bound = gddm_bound_constant(),
  start = gddm_start_point(),
  lapse = c("none", "uniform"),
  control = gddm_control()
)

Arguments

n

Number of trials.

...

Parameter values by name: mu, bs, ndt and whatever else the chosen components need (alpha, leak, tau, kappa, bias, sz, lapse). Anything not given takes the component's own starting value. Each is recycled to length n and must be constant within every distinct value of coh, because one solve serves each of those and reads its parameters at its first trial. A parameter that varies inside one is refused by name; give coh a distinct value per parameter setting, or call this once per setting.

coh

Coherence covariate, recycled to length n. It is also what separates parameter settings, whether or not a drift term reads it. Only the drift value is meaningful with gddm_drift_coherence().

drift, bound, start, lapse, control

As in gddm().

Value

A data frame with rt, upper, cond and, when a coherence drift is used, coh.

See also

gddm()

Examples

set.seed(2)
head(gddm_simulate(20, mu = 1.5, bs = 2, ndt = 0.2,
                   control = gddm_control(t_max = 2, dt = 0.02,
                                          ny = 101)))
#>          rt upper cond
#> 1 0.8319035     1    1
#> 2 0.5872643     1    1
#> 3 1.1134061     1    1
#> 4 0.4327229     1    1
#> 5 0.5594095     1    1
#> 6 0.6637092     1    1