Created
December 21, 2021 03:31
-
-
Save jvcasillas/628232a0c759e97b9911ab3eeff26241 to your computer and use it in GitHub Desktop.
ddm_sims
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
library(tidyverse) | |
# DDM simulation function | |
sim_ddm <- function(drift_rate, boundary_separation, bias, ndt, n_sims, seed = NULL) { | |
set.seed(seed) | |
sim_df = NULL | |
for (sim in 1:n_sims){ | |
step = 1 | |
value = bias | |
n = 1 | |
while (abs(value[n]) < boundary_separation) { | |
n = n + 1 | |
value[n] = (value[n - 1] + rnorm(1, 0, abs(drift_rate))) | |
step[n] = n | |
} | |
dd_df <- tibble(sim_n = sim, step, value) %>% | |
mutate( | |
sim_n = as.factor(sim_n), | |
value = case_when( | |
value > boundary_separation ~ boundary_separation, | |
value < -boundary_separation ~ -boundary_separation, | |
TRUE ~ .$value) | |
) | |
sim_df = rbind(sim_df, dd_df) | |
} | |
return(sim_df) | |
} | |
# Get α, β, δ, and τ from DDM | |
# Plot 20 random simulations after fitting model | |
sim_ddm( | |
drift_rate = 0.8, | |
boundary_separation = 1, | |
bias = 0, | |
ndt = 0.2, | |
n_sims = 20, | |
seed = 12345) %>% | |
ggplot(., aes(x = step, y = value, color = sim_n)) + | |
geom_line(show.legend = F) | |
# You can increase the number of sims and plot the average trajectories too |
I haven't done much DDM. I am planning to collect some data online next semester that I want to look at with DDM so I am exploring options.
@jvcasillas Can you share the code you used to make the figure as well? I think it would be a good way to depict predictions one has about the different parameters.
Ah the plot you are referring to is just a schematic to explain the model (it's in the method section of the paper Im working on). I use it when I explain what each parameter represents. Here is the code:
# DDM explanation -------------------------------------------------------------
# Generate skewed RT-like data
my_skewed_x <- fGarch::rsnorm(100, mean = 2.5, sd = 1.5, xi = 3.5)
# Correct response dist
inset_top <- tibble(x = my_skewed_x) %>%
ggplot(., aes(x = x)) +
geom_density(fill = viridis::viridis_pal(option = "B", begin = 0.3)(1)) +
theme_void()
# Incorrect response dist
inset_bottom <- tibble(x = my_skewed_x) %>%
ggplot(., aes(x = x)) +
geom_density(fill = viridis::viridis_pal(option = "B", begin = 0.75)(1)) +
scale_y_reverse() +
theme_void()
threshold_labs <- tibble(
step = 80, value = c(1.4, -1.4),
labs = c("Correct response", "Incorrect response")
)
# Main plot
main_plot <- sim_ddm(q_type = "yn", eq = "low", lt = "low", seed = 20180119,
drift_rate = 0.11, boundary_separation = 1.25, bias = 0, ndt = 0, n_sims = 2) %>%
ggplot(., aes(x = step, y = value, color = sim_n)) +
geom_vline(xintercept = 1) +
geom_hline(yintercept = c(-1.25, 1.25), lty = 3) +
geom_line(show.legend = F, size = 1.25) +
geom_segment(inherit.aes = F, aes(x = 1, xend = 98, y = 0, yend = 0),
lty = 2, size = 0.25) +
coord_cartesian(ylim = c(-2, 2), xlim = c(-5, 100)) +
scale_x_continuous(breaks = c(1, seq(10, 100, 10)),
labels = c("0", seq(10, 100, 10))) +
scale_y_continuous(breaks = NULL) +
labs(y = expression(alpha), x = "Time step") +
annotate("text", x = -4.5, y = 0.15, size = 5, label = expression(tau)) +
geom_segment(inherit.aes = F, aes(x = -5, y = 0, xend = 0, yend = 0),
arrow = arrow(length = unit(0.2, "cm")), size = 0.5) +
geom_segment(inherit.aes = F, aes(x = -5, y = 0, xend = -9, yend = 0),
arrow = arrow(length = unit(0.2, "cm"))) +
annotate("text", x = 101, y = 0, size = 5, label = expression(beta)) +
geom_segment(inherit.aes = F, aes(x = 101, y = 0.2, xend = 101, yend = 0.5),
arrow = arrow(length = unit(0.2, "cm")), size = 0.5) +
geom_segment(inherit.aes = F, aes(x = 101, y = -0.2, xend = 101, yend = -0.5),
arrow = arrow(length = unit(0.2, "cm"))) +
annotate("text", x = 32, y = 0.25, size = 5, label = expression(delta)) +
geom_segment(inherit.aes = F, aes(x = 30.5, y = 0.35, xend = 25, yend = 0.7),
arrow = arrow(length = unit(0.2, "cm"))) +
geom_text(inherit.aes = F, data = threshold_labs, hjust = 0,
aes(x = step, y = value, label = labs), family = "Times") +
scale_color_viridis_d(option = "B", begin = 0.3, end = 0.75) +
theme_minimal(base_family = 'Times', base_size = 13) +
theme(panel.grid.major = element_line(size = 0.2),
panel.grid.minor = element_blank(),
axis.title.y = element_text(size = rel(.9), hjust = 0.5))
# Combine plots
ddm_explanation <- main_plot + annotation_custom(
grob = ggplotGrob(inset_top),
ymin = 1.25, ymax = 2, xmin = -2.5, xmax = 75) +
annotation_custom(
grob = ggplotGrob(inset_bottom),
ymin = -1.25, ymax = -2, xmin = -2.5, xmax = 75)
# -----------------------------------------------------------------------------
This is how Im using it:
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Yep. I looked into a few other options (there is an interesting one in python but I couldnt get it to work: http://ski.clps.brown.edu/hddm_docs/index.html), but brms is pretty much home for me now. How do you usually fit yours?