Simulate from a linear ballistic accumulator

Draws from the generative process directly: a uniform start point and a normal drift rate per accumulator, and whichever reaches the threshold first. This is the joint draw of choice and time, which is what the model produces and what a fit needs; simulate() on a fitted object instead redraws only the time, holding each row's observed choice.

Usage

lba_simulate(n, v, A = 0.5, k = 0.4, ndt = 0.2, sd_v = 1, posdrift = TRUE)

Arguments

n

Number of trials.

v

Drift means, one per accumulator. Its length sets the number of accumulators. May be a matrix with one row per trial, for a drift that varies with a covariate.

A

Upper end of the start-point range.

k

Distance from the top of the start-point range to the threshold, so the threshold is A + k.

ndt

Non-decision time.

sd_v

Drift standard deviation, fixed. One number or one per accumulator.

posdrift

Truncate the drift distribution at zero? Matches the lba() argument of the same name.

Value

A data frame with choice (the accumulator that responded, from 1) and rt.

Examples

set.seed(1)
dat <- lba_simulate(500, v = c(2.5, 1.5, 1.0), A = 0.5, k = 0.4,
                    ndt = 0.2)
table(dat$choice)
#> 
#>   1   2   3 
#> 321 126  53 
tapply(dat$rt, dat$choice, mean)
#>         1         2         3 
#> 0.4374168 0.4691844 0.4590923