lbaModel provides fast and flexible tools for working with the Linear Ballistic Accumulator (LBA) model, a widely used framework in cognitive psychology and neuroscience for simulating and analysing choice and response time (RT) data.
Key features:
- ⚡ Likelihood computation for parameter estimation
- 🎯 Data simulation at the subject or population level
- 📊 Probability density (PDF) & cumulative distribution (CDF) evaluation
- 🔗 Design-based parameter mapping for factorial experiments
- 💻 C++ backend for high-performance computation
While designed to be part of the ggdmc ecosystem, lbaModel is also fully functional as a standalone package.
Here’s a schematic of the LBA model showing accumulators racing to threshold (similar to diffusion models, but with linear deterministic growth.)
Figure: The LBA model assumes evidence accumulates linearly and independently across choices, with start points, drawing from a uniform distribution and drift rates. The first accumulator to hit its threshold determines the response and response time.
library(ggdmcModel)
library(ggdmcPrior)
model <- BuildModel(
p_map = list(A = "1", B = "1", t0 = "1", mean_v = "M", sd_v = "1", st0 = "1"),
match_map = list(M = list(s1 = "r1", s2 = "r2")),
factors = list(S = c("s1", "s2")),
constants = c(st0 = 0, sd_v = 1),
accumulators = c("r1", "r2"),
type = "lba"
)
# Set population-level prior
pop_mean <- c(A = 0.4, B = 0.5, mean_v.false = 0.15, mean_v.true = 2.5, t0 = 0.3)
pop_scale <- c(A = 0.1, B = 0.1, mean_v.false = 0.2, mean_v.true = 0.2, t0 = 0.05)
pop_dist <- BuildPrior(
p0 = pop_mean,
p1 = pop_scale,
lower = c(0, 0, 0, 0, 0),
upper = rep(NA, length(pop_mean)),
dists = rep("tnorm", length(pop_mean)),
log_p = rep(FALSE, length(pop_mean))
)
plot_prior(pop_dist)sub_model <- setLBA(model)
pop_model <- setLBA(model, population_distribution = pop_dist)
# One subject
p_vector <- c(A = 0.75, B = 1.25, mean_v.false = 1.5, mean_v.true = 2.5, t0 = 0.15)
dat <- simulate(sub_model, nsim = 256, parameter_vector = p_vector, n_subject = 1)
# Multiple subjects
hdat <- simulate(pop_model, nsim = 128, n_subject = 32)# Parameters
params_tmp <- list(
A = c(0.5, 0.5),
b = c(1.0, 1.0),
mean_v = c(2.0, 1.0),
sd_v = c(1.0, 1.0),
st0 = c(0.0, 0.0),
t0 = c(0.2, 0.2)
)
# Convert to matrix
param_list2mat <- function(param_list) {
n_row <- length(param_list[[1]])
n_col <- length(param_list)
out <- matrix(NA, nrow = n_row, ncol = n_col)
for (i in seq_len(n_col)) out[, i] <- param_list[[i]]
t(out)
}
params <- param_list2mat(params_tmp)
time_params <- c(0, 5, 0.01)
nv <- ncol(params)
is_pos <- rep(TRUE, nv)
pdfs <- theoretical_dlba(params, is_pos, time_params)
cdfs <- theoretical_plba(params, is_pos, time_params)install.packages("lbaModel")
⚠️ Requires development tools and extra dependencies.
# install.packages("devtools")
devtools::install_github("yxlin/lbaModel")- R (≥ 3.3.0)
- ggdmcPrior
- ggdmcModel
- Rcpp (≥ 1.0.7)
- RcppArmadillo (≥ 0.10.7.5.0)
- ggdmcHeaders
If you use lbaModel, please cite:
- Brown & Heathcote (2008). The simplest complete model of choice response time: Linear ballistic accumulation. Cognitive Psychology, 57(3), 153–178.: https://doi.org/10.1016/j.cogpsych.2007.12.002
- Lin & Strickland (2020). Evidence accumulation models with R: A practical guide to hierarchical Bayesian methods. The Quantitative Methods for Psychology, 16(2), 133–149. https://doi.org/10.20982/tqmp.16.2.p133
glba: Generalised LBA model fitting via MLErtdists: Density/distribution for LBA and diffusion models.
- Tight integration with
ggdmcfor hierarchical Bayesian inference - Full support for design-based parameter mapping in factorial designs
- Optimised C++ backend with extensions (
ggdmcLikelihood,pPDA) for high-performance parallelised LBA
Contributions are welcome! Please feel free to submit issues, fork the repo, or open pull requests.
- Maintainer: Yi-Shin Lin
- 📧 yishinlin001@gmail.com