Strong mnemonic prediction errors increase cognitive control, attention, and arousal.

Code ↔ Paper

3 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 3 matches

[1] § Results › Effects of mnemonic prediction errors › Distinct pupil features reflect MPEs and attentional orienting ↔ scripts/pupil/pupil_temporal_pca.Rmd, lines 1339 to 1389 · score 0.64 · cognitive effort, temporal PCA, orienting responses, PC6, PC2, PC5 · Report a mismatch
[2] § Methods › EEG and pupillometry analyses › Pupillary temporal PCA ↔ scripts/pupil/pupil_temporal_pca.Rmd, lines 161 to 281 · score 0.51 · Temporal PCA, matrix, rotation, varimax, RT, components · Report a mismatch
[3] § Methods › EEG and pupillometry analyses › Pupillary temporal PCA ↔ scripts/pupil/pupil_temporal_pca_include_incorrect.Rmd, lines 186 to 309 · score 0.51 · Temporal PCA, matrix, rotation, varimax, RT, components · Report a mismatch

Paper

Loaded from Europe PMC by your browser; OSCR does not store it: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The paper is loaded by your browser, with JavaScript: read it at doi.org or on Europe PMC.

The authors' code

R Markdown · 2,507 lines · 88 KB · GPL-3.0 · 2 matches

---
title: "Temporal PCA on pupil data during associative retrieval"
author: "Alice M. Xue ([email hidden])"
date: "`r format(Sys.time(), '%B %d, %Y')`"
params:
exp_id: "MPE"
run_models: yes
output:
html_document:
toc: yes
toc_depth: 3
toc_float:
collapsed: no
self_contained: true
pdf_document:
toc: yes
toc_depth: '3'
---
```{r setup, include=FALSE}
library(plyr)
library(Hmisc)
library(ggplot2)
library(knitr)
library(kableExtra)
library(data.table)
library(psycho)
library(lme4)
library(sjPlot)
library(afex)
library(viridis)
library(showtext)
library(tidyverse)
library(mgcv)
library(itsadug)
library(RColorBrewer)
library(ggcorrplot)
library(devtools)
library(factoextra)
library(tidyjson)
library(purrr)
library(broom)
library(zoo)
library(psych)
library(patchwork)
library(emmeans)
library(ggpubr)
library(stats)
library(broom.mixed)
library(pracma)
library(afex)
library(stringr)
library(ggeffects)
font_add(family = "Avenir Next",
regular = "../utils/AvenirNext-Regular.ttf",
bold = "../utils/AvenirNext-DemiBold.ttf",
italic = "../utils/AvenirNext-Italic.ttf")
showtext.auto()
op <- showtext_opts()
fontfamily <- "Avenir Next"
set_theme(
base=theme_classic(base_size = 12,
base_family="Avenir Next")
)
options(dplyr.summarise.inform = FALSE)
#knitr::opts_chunk$set(dev = 'pdf') # set output device to pdf
knitr::opts_chunk$set(echo = TRUE, warning=FALSE, message = FALSE)
```
```{r, echo=F}
source("../utils/summary_se_utils.R")
```
```{r get_behavior, echo=F}
set.seed(478)
exp_id <- params$exp_id
run_models <- params$run_models
data_ret <- fread("../../data/behavior/MPE_asso_retrieval_behavior.csv") %>%
mutate(key = match_response.keys,
rt = match_response.rt * 1000)
data_recog <- fread("../../data/behavior/MPE_mismatch_recog_behavior.csv") %>%
mutate(rt = recognition_response.rt * 1000)
recognition_behavior <- data_recog %>%
mutate(trialN = trials.thisTrialN) %>%
select(trialN,
recognition_response.keys,
recognition_response.rt,
is_old,
img,
subjectId,
expId) %>%
mutate(hit = is_old & (recognition_response.keys==1)) %>%
mutate(false_alarm = !is_old & (recognition_response.keys==1)) %>%
mutate(miss = is_old & (recognition_response.keys==2)) %>%
mutate(correct_rejection = !is_old & (recognition_response.keys==2)) %>%
mutate(trial_outcome = case_when(hit ~ "Hit",
false_alarm ~ "FA",
miss ~ "Miss",
correct_rejection ~ "CR")) %>%
group_by(subjectId,
trialN,
trial_outcome,
is_old,
img,
recognition_response.keys,
expId) %>%
summarize(recognition_response.rt = mean(recognition_response.rt)) %>%
ungroup()
```
```{r load_models, echo=F, eval=!run_models}
load("tpca_models_correct_only.RData")
```
```{r save_or_load_data, echo=F}
retrieval_pupil <- fread("../../data/pupil/MPE_imm_retrieval_pupil.csv")
retrieval_pupil.del <- fread("../../data/pupil/MPE_del_retrieval_pupil.csv")
retrieval_pupil <- bind_rows(retrieval_pupil, retrieval_pupil.del) %>%
mutate(expId = case_when(grepl("MPE1", subjectId) ~ "imm",
grepl("MPE3", subjectId) ~ "del",
grepl("MPE5", subjectId) ~ "del")) %>%
mutate(across(expId,
factor,
levels=c("imm", "del"))) %>%
mutate(room_setup = case_when(grepl("MPE1", subjectId) ~ "1", # eeg + pupil, shared-eeg lab
grepl("MPE3", subjectId) ~ "2", # pupil only, shared-eeg lab
grepl("MPE5", subjectId) ~ "3")) # pupil only, 4th floor lab
```
```{r labels, echo=F}
palette <- c("#008080", "#9fdfbf")
hit_miss_palette <- c("#6E9075", "#FC7753")
ret_labels_strong <- tibble(event = c("cue", "delay", "probe", "ITI"),
time = c(0, 1, 2, 2.75))
ret_labels_weak <- tibble(event = c("cue", "delay", "probe", "ITI"),
time = c(0, 1, 2, 2.75))
ret_labels <- rbind(ret_labels_strong, ret_labels_weak)
ret_labels$position <- 0.12
```
```{r tempPCA_trial, echo=F}
subj_timeseries_trial <- retrieval_pupil %>%
dplyr::filter(trial_outcome %in% c("Hit", "CR")) %>%
mutate(pupil_diameter = pupil_diameter_z)
subj_timeseries_trial %>%
pivot_wider(id_cols = c("subjectId",
"expId",
"room_setup",
"trial_outcome",
"match_response.rt",
"association_strength",
"trialN",
"img"),
names_from = c("time"),
values_from = c("pupil_diameter")) %>%
mutate(across(trial_outcome,
factor,
levels=c("Hit", "CR", "Miss", "FA"))) %>%
group_by(trial_outcome, association_strength) %>%
summarize(count = n()) %>%
ungroup() %>%
mutate(total_ntrials = sum(count)) %>%
mutate(percent_data = round(count / total_ntrials * 100, 2)) %>%
kable() %>%
kable_styling(bootstrap_options = "striped",
full_width = F,
position = "left")
subj.data <- subj_timeseries_trial %>%
pivot_wider(id_cols = c("trial_outcome",
"association_strength",
"subjectId",
"expId",
"room_setup",
"trialN",
"img",
"trialN_block",
"match_response.rt",
"block"),
names_from = c("time"),
values_from = c("pupil_diameter"))
pca_input <- subj.data %>%
select(-trial_outcome,
-association_strength,
-subjectId,
-expId,
-room_setup,
-match_response.rt,
-trialN,
-img,
-trialN_block,
-block)
ncomp <- 6
pc_names <- paste("PC", 1:ncomp, sep = "")
data.pca <- prcomp(pca_input, center=T, scale=T)
# scree plot
plot(data.pca)
# change the sign of PC4 columns to flip the sign of the loadings
rot <- data.pca$rotation
rot[, c(4)] <- rot[, c(4)] * -1
data.pca$rotation <- rot
raw.loadings <- data.pca$rotation[,1:ncomp] %*% diag(data.pca$sdev, ncomp, ncomp)
rotated.loadings <- varimax(raw.loadings)$loadings
inv.loadings <- t(pracma::pinv(rotated.loadings))
loadings <- rotated.loadings
scores <- scale(pca_input) %*% inv.loadings
subj.pred <- as_tibble(scores) %>%
rename_with(~ gsub("V", "PC", .))
fviz_eig(data.pca, addlabels = TRUE)
pca.loadings <- as_tibble(matrix(as.numeric(loadings),
attributes(loadings)$dim,
dimnames=attributes(loadings)$dimnames),
rownames="time") %>%
mutate(time = as.numeric(time))
pca.loadings <- pca.loadings %>%
rename_with(~ gsub("V", "PC", .))
ss_loadings <- colSums(loadings^2)
var_explained <- as_tibble(ss_loadings/nrow(loadings)) %>%
mutate(percent_var_explained = value * 100) %>%
add_column(component = pc_names)
var_explained <- var_explained %>%
arrange(desc(percent_var_explained)) %>%
mutate(component_renamed = paste("PC", row_number(), sep=""))
pca.loadings <- pca.loadings %>%
pivot_longer(cols = pc_names,
names_to = "component",
values_to = "loading") %>%
left_join(var_explained, by=c("component")) %>%
rename(component_orig = component,
component = component_renamed)
# ordered by "peak time"
ordered_pcs <- c("PC2",
"PC5",
"PC1",
"PC6",
"PC3",
"PC4")
component_order_with_var_explained <- c("PC2: 25.89%",
"PC5: 5.66%",
"PC1: 28.81%",
"PC6: 1.83%",
"PC3: 19.29%",
"PC4: 14.68%")
```
```{r plot_pc_loadings, echo=F}
rainbow <- c("#E82C42",
"#F46D43",
"#FFCC00",
"#29A343",
"#1E8DD2",
"#59169C")
pca.loadings %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
mutate(across(component,
factor,
levels=component_order_with_var_explained)) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(linewidth=1.2) +
geom_text(aes(label = event,
x = time,
y = 0.9),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 6,
colour = "gray37",
family = fontfamily) +
labs(x = "Time (s)",
y = "PC loading",
title = "Components extracted by temporal PCA") +
scale_color_manual(values=rainbow, name="") +
theme(text = element_text(size=15))
pca.loadings %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color="black", linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 3,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component) +
labs(x = "Time (s)",
y = "PC loading",
title = "Temporal PCA") +
scale_color_brewer(palette="Set2") +
theme(legend.position = "none")
subj.data.scores <- subj.data %>%
add_column(subj.pred) %>%
pivot_longer(cols = pc_names,
names_to = "component",
values_to = "component_score") %>%
left_join(var_explained, by=c("component")) %>%
rename(component_orig = component,
component = component_renamed)
scores <- subj.data.scores %>%
select(subjectId,
trialN,
img,
trialN_block,
block,
trial_outcome,
association_strength,
component,
component_score)
save(loadings,
inv.loadings,
subj.data.scores,
file="pupil_tPCA_correct_only.RData")
write.csv(scores, "pupil_tPCA_scores_correct_trials_only.csv")
write.csv(pca.loadings, "pupil_tPCA_loadings_correct_trials_only.csv")
```
```{r loadings_plts, echo=F}
loadings.plt <- pca.loadings %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color="black", linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1,
size = 3,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component, ncol=1) +
labs(x = "Time (s)",
y = "PC loading") +
theme(legend.position = "none")
loadings.vrt.plt <- pca.loadings %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
mutate(across(component,
factor,
levels=component_order_with_var_explained)) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color="black", linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 3,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component, ncol=1) +
labs(x = "Time (s)",
y = "PC loading") +
theme(legend.position = "none")
```
## Memory outcome models
Match v. mismatch, correct v. incorrect
```{r match_mismatch_acc_models, echo=F, eval=run_models, fig.width=6.5, fig.height=3, results="asis"}
models <- c()
models.compact.no.mismatch <- c()
for (i in 1:ncomp) {
mod.data <- subj.data.scores %>%
dplyr::filter(component == paste("PC", i, sep="")) %>%
mutate(is_mismatch = if_else(trial_outcome %in% c("Hit", "Miss"),
0,
1)) %>%
mutate(is_incorrect = if_else(trial_outcome %in% c("Hit", "CR"),
0,
1))
mod <- mod.data %>%
lmer(component_score ~ is_mismatch * association_strength +
(is_mismatch * association_strength | subjectId) +
(1 | expId),
data = .)
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- mod.data %>%
lmer(component_score ~ is_mismatch * association_strength +
(is_mismatch * association_strength | subjectId),
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- mod.data %>%
lmer(component_score ~ is_mismatch * association_strength +
(is_mismatch + association_strength | subjectId) +
(1 | expId),
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- mod.data %>%
lmer(component_score ~ is_mismatch * association_strength +
(is_mismatch + association_strength | subjectId),
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- mod.data %>%
lmer(component_score ~ is_mismatch * association_strength +
(1 | subjectId) +
(1 | expId),
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- mod.data %>%
lmer(component_score ~ is_mismatch * association_strength +
(1 | subjectId),
data = .)
}
final.mod.formula <- formula(mod) [3]
final.mod.formula <- paste("component_score ~", final.mod.formula)
no.mismatch.formula <- gsub("is_mismatch \\* ", "", final.mod.formula)
no.mismatch.formula <- gsub("is_mismatch \\+ ", "", no.mismatch.formula)
mod.no.mismatch <- mod.data %>%
lmer(as.formula(no.mismatch.formula),
data = .)
if (isSingular(mod.no.mismatch) | any(grepl("failed to converge", mod.no.mismatch@optinfo$conv$lme4$messages))) {
mod.no.mismatch <- mod.data %>%
lmer(component_score ~ association_strength +
(association_strength | subjectId),
data = .)
}
if (isSingular(mod.no.mismatch) | any(grepl("failed to converge", mod.no.mismatch@optinfo$conv$lme4$messages))) {
mod.no.mismatch <- mod.data %>%
lmer(component_score ~ association_strength +
(1 | subjectId),
data = .)
}
models <- c(models, mod)
models.compact.no.mismatch <- c(models.compact.no.mismatch, mod.no.mismatch)
}
```
```{r show_match_mismatch_acc_models, echo=F, eval=T, fig.width=6.5, fig.height=3, results="asis"}
for (i in 1:ncomp) {
fig1 <- pca.loadings %>%
dplyr::filter(component == paste("PC", i, sep="")) %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color="black", linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 3,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component) +
labs(x = "Time (s)",
y = "PC loading") +
scale_color_brewer(palette="Set2") +
scale_fill_brewer(palette="Set2") +
theme(legend.position = "none")
fig2 <- subj.data.scores %>%
dplyr::filter(component == paste("PC", i, sep="")) %>%
group_by(subjectId, association_strength, trial_outcome, component) %>%
summarize(component_score = mean(component_score)) %>%
summarySEwithin(.,
measurevar="component_score",
withinvars=c("association_strength",
"trial_outcome",
"component"),
idvar="subjectId") %>%
mutate(across(trial_outcome,
factor,
levels=c("Hit", "CR", "Miss", "FA"))) %>%
ggplot(aes(x = association_strength,
y = component_score,
fill = trial_outcome)) +
stat_summary(fun = "mean",
geom = "bar",
color = "black",
position=position_dodge()) +
geom_errorbar(aes(ymin=component_score-se, ymax=component_score+se),
position=position_dodge(.9),
width=.2) +
facet_wrap(~component, scales = "free") +
scale_color_manual(values=c("#1a75ff", "#ff751a", "#80b3ff", "#ffb380"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a", "#80b3ff", "#ffb380"), name="") +
labs(x = "", y = "PC score")
print(fig1 + fig2)
mod <- models[[i]]
mod.no.mismatch <- models.compact.no.mismatch[[i]]
title <- formula(mod) [3]
title <- gsub("is_mismatch", "Mismatch", title)
title <- gsub("is_incorrect", "Accuracy", title)
title <- gsub("association_strength", "Strength", title)
title <- gsub(":", " x ", title)
tidy(mod, effects = "fixed", conf.int = T, conf.level = 0.95) %>%
mutate(CI = paste0("[",
sprintf("%.03f", conf.low),
", ",
sprintf("%.03f", conf.high),
"]",
sep="")) %>%
select(term, estimate, CI, p.value, std.error, df) %>%
rename("Estimate" = estimate,
"p-value" = p.value,
"SE" = std.error,
"DF" = df
) %>%
mutate(term = gsub("is_mismatch", "Mismatch", term),
term = gsub("is_incorrect", "Accuracy", term),
term = gsub("association_strengthWeak \$1x\$", "Strength", term),
term = gsub(":", " x ", term)) %>%
kable(caption=paste("PC", i, " score ~ ", title, sep=""),
digits=3) %>%
kable_styling() %>%
cat()
tidy(mod, effects = "ran_pars", scales = "vcov") %>%
select(term, estimate) %>%
mutate(term = gsub("var__", "Variance ", term),
term = gsub("cov__", "Covariance ", term),
term = gsub("\$Intercept\$", "intercept", term),
term = gsub("is_mismatch", "Mismatch", term),
term = gsub("is_incorrect", "Accuracy", term),
term = gsub("association_strengthWeak \$1x\$", "Strength", term),
term = gsub("\\.", ", ", term)) %>%
kable(digits=3, caption="Random effects") %>%
kable_styling() %>%
cat()
anova(mod.no.mismatch, mod, test="LRT") %>%
tidy() %>%
kable(caption=paste("mod.no.mismatch:",
formula(mod.no.mismatch) [3])) %>%
kable_styling() %>%
cat()
print(paste("No mismatch compact model failed to converge: ",
isSingular(mod.no.mismatch) | any(grepl("failed to converge", mod.no.mismatch@optinfo$conv$lme4$messages))))
emm <- emmeans(mod, ~ is_mismatch | association_strength, lmerTest.limit = 10346)
print(emm)
tidy(emm, effects = "fixed", conf.int = T, conf.level = 0.95) %>%
kable(digits=3, caption=paste("PC", i, "; Emmeans", sep="")) %>%
kable_styling() %>%
cat()
emm.pairs <- pairs(emm)
print(emm.pairs)
tidy(emm.pairs, effects = "fixed", conf.int = T, conf.level = 0.95) %>%
kable(digits=3, caption=paste("PC", i, "; Pairwise comparisons", sep="")) %>%
kable_styling() %>%
cat()
}
```
```{r full_model_all_pc, echo=F, eval=F}
mod <- subj.data.scores %>%
mutate(is_mismatch = if_else(trial_outcome %in% c("Hit", "Miss"),
0,
1)) %>%
mutate(is_incorrect = if_else(trial_outcome %in% c("Hit", "CR"),
0,
1)) %>%
lmer(component_score ~ component * is_mismatch * association_strength +
(component + is_mismatch + association_strength | subjectId),
data = .)
mod %>%
summary()
tidy(mod, effects = "fixed", conf.int = T, conf.level = 0.95) %>%
select(term, estimate, conf.low, conf.high, p.value, std.error, df) %>%
kable(digits=3) %>%
kable_styling()
tidy(mod, effects = "ran_pars", scales = "vcov") %>%
select(term, estimate) %>%
kable(digits=3, caption="Random effects") %>%
kable_styling()
```
## RT models
```{r get_rt_data_obj, echo=F}
### joint plot
rt_data <- subj.data.scores %>%
mutate_at(c('trial_outcome'), as.character) %>%
mutate(trial_outcome=replace(trial_outcome, trial_outcome=="Hit", "Match")) %>%
mutate(trial_outcome=replace(trial_outcome, trial_outcome=="CR", "Mismatch")) %>%
dplyr::filter(trial_outcome %in% c("Match", "Mismatch"))
```
```{r match_mismatch_rt, echo=F, eval=run_models, fig.width=7, fig.height=10}
models.rt <- c()
for (pc in 1:ncomp) {
mod <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(match_response.rt = log(match_response.rt)) %>%
lmer(match_response.rt ~ component_score * association_strength * trial_outcome +
(component_score * association_strength * trial_outcome | subjectId) +
(1 | expId),
data = .)
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(match_response.rt = log(match_response.rt)) %>%
lmer(match_response.rt ~ component_score * association_strength * trial_outcome +
(component_score * trial_outcome * association_strength | subjectId),
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(match_response.rt = log(match_response.rt)) %>%
lmer(match_response.rt ~ component_score * association_strength * trial_outcome +
(component_score + trial_outcome + association_strength | subjectId) +
(1 | expId),
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(match_response.rt = log(match_response.rt)) %>%
lmer(match_response.rt ~ component_score * association_strength * trial_outcome +
(component_score + trial_outcome + association_strength | subjectId),
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(match_response.rt = log(match_response.rt)) %>%
lmer(match_response.rt ~ component_score * association_strength * trial_outcome +
(component_score | subjectId) +
(1 | expId),
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(match_response.rt = log(match_response.rt)) %>%
lmer(match_response.rt ~ component_score * association_strength * trial_outcome +
(1 | subjectId),
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
print(paste("MODEL DID NOT CONVERGE FOR PC", pc))
}
models.rt <- c(models.rt, mod)
}
```
```{r binned_rt, echo=F, fig.width=7, fig.height=10}
nbins <- 4
rt.binned.plt <- rt_data %>%
group_by(subjectId, component, association_strength, trial_outcome) %>%
mutate(trial_count = n()) %>%
ungroup() %>%
dplyr::filter(trial_count > nbins-1) %>%
group_by(subjectId, component, association_strength, trial_outcome) %>%
mutate(component_score_bin = cut_interval(component_score,
n = nbins,
labels = F)) %>%
ungroup() %>%
group_by(subjectId, trial_outcome, association_strength,
component, component_score_bin) %>%
summarize(rt = median(match_response.rt)) %>%
summarySEwithin(.,
measurevar="rt",
withinvars=c("association_strength",
"trial_outcome",
"component",
"component_score_bin"),
idvar="subjectId") %>%
ggplot(aes(x = as.numeric(as.character(component_score_bin)),
y = rt,
color = trial_outcome,
group = trial_outcome)) +
geom_line(linewidth=0.8) +
geom_point(size=1.5) +
geom_errorbar(aes(ymin=rt-se, ymax=rt+se),
width=0.2) +
facet_grid(rows=vars(component),
cols=vars(association_strength),
axes="all_x",
scales = "free") +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "PC score quartile", y = "RT (s)") +
theme(strip.text.y = element_blank())
loadings.plt + rt.binned.plt +
plot_layout(widths=c(1,2))
```
```{r show_mismatch_rt_models, echo=F, fig.width=6.5, fig.height=3, results="asis"}
for (pc in 1:ncomp) {
fig1 <- pca.loadings %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color="black", linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 3,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component) +
labs(x = "Time (s)",
y = "PC loading") +
scale_color_brewer(palette="Set2") +
scale_fill_brewer(palette="Set2") +
theme(legend.position = "none")
mod <- models.rt[[pc]]
fig2 <- mod %>%
plot_model(type="pred",
terms=c("component_score[all]",
"trial_outcome[all]",
"association_strength[Strong (4x), Weak (1x)]")) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(y = "RT (s)") +
theme(plot.title=element_blank(),
axis.title.x=element_blank())
print(fig1 + fig2)
title <- formula(mod)[3]
title <- gsub("trial_outcome", "Mismatch", title)
title <- gsub("component_score",
paste("PC", pc, " score", sep=""),
title)
title <- gsub("association_strength", "Strength", title)
title <- gsub(":", " x ", title)
tidy(mod, effects = "fixed", conf.int = T, conf.level = 0.95) %>%
mutate(CI = paste0("[",
sprintf("%.03f", conf.low),
", ",
sprintf("%.03f", conf.high),
"]",
sep="")) %>%
select(term, estimate, CI, p.value, std.error, df) %>%
rename("Estimate" = estimate,
"p-value" = p.value,
"SE" = std.error,
"DF" = df
) %>%
mutate(term = gsub("trial_outcomeMismatch", "Mismatch", term),
term = gsub("component_score",
paste("PC", pc, " score", sep=""),
term),
term = gsub("association_strengthWeak \$1x\$", "Strength", term),
term = gsub(":", " x ", term)) %>%
kable(caption=paste("RT ~ ", title, sep=""),
digits=3) %>%
kable_styling() %>%
cat()
tidy(mod, effects = "ran_pars", scales = "vcov") %>%
select(term, estimate) %>%
mutate(term = gsub("var__", "Variance ", term),
term = gsub("cov__", "Covariance ", term),
term = gsub("\$Intercept\$", "intercept", term),
term = gsub("is_mismatch", "Mismatch", term),
term = gsub("is_incorrect", "Accuracy", term),
term = gsub("association_strengthWeak \$1x\$", "Strength", term),
term = gsub("\\.", ", ", term)) %>%
kable(digits=3, caption="Random effects") %>%
kable_styling() %>%
cat()
slopes <- emtrends(mod, pairwise ~ association_strength * trial_outcome, var = "component_score", lmerTest.limit = 10346)
print(slopes)
tidy(slopes$emtrends, conf.int = T, conf.level = 0.95) %>%
kable(caption=paste("slopes: RT ~ ", title, sep=""),
digits=3) %>%
kable_styling() %>%
cat()
tidy(slopes$contrasts, conf.int = T, conf.level = 0.95) %>%
select(contrast, estimate, std.error, df, conf.low, conf.high, statistic, adj.p.value) %>%
kable(caption=paste("slope contrasts: RT ~ ", title, sep=""),
digits=3) %>%
kable_styling() %>%
cat()
}
```
## Individual PC plots
```{r indivi_pc_plots, fig.width=6, fig.height=2, echo=F}
showtext_opts(dpi = 300)
for (pc in 1:ncomp) {
fig1 <- pca.loadings %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color="black", linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 3,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component) +
ylim(-0.25, 1) +
labs(x = "Time (s)",
y = "PC loading") +
scale_color_brewer(palette="Set2") +
scale_fill_brewer(palette="Set2") +
theme(legend.position = "none")
fig2 <- subj.data.scores %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(trial_outcome = ifelse(as.character(trial_outcome) %in% c("Hit"),
"Match", "Mismatch")) %>%
group_by(subjectId, association_strength, trial_outcome, component) %>%
summarize(component_score = mean(component_score)) %>%
summarySEwithin(.,
measurevar="component_score",
withinvars=c("association_strength",
"trial_outcome",
"component"),
idvar="subjectId") %>%
ggplot(aes(x = exp_id,
y = component_score,
fill = trial_outcome)) +
stat_summary(fun = "mean",
geom = "bar",
color = "black",
position=position_dodge()) +
geom_errorbar(aes(ymin=component_score-se, ymax=component_score+se),
position=position_dodge(.9),
width=.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "", y = "PC score") +
theme(axis.title.x=element_blank(),
axis.text.x=element_blank(),
axis.ticks.x=element_blank())
fig3 <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
group_by(subjectId, component, association_strength, trial_outcome) %>%
# exclude any subjects with # trials < # bins
mutate(trial_count = n()) %>%
ungroup() %>%
dplyr::filter(trial_count > nbins-1) %>%
group_by(subjectId, component, association_strength, trial_outcome) %>%
mutate(component_score_bin = cut_interval(component_score,
n = nbins,
labels = F)) %>%
ungroup() %>%
group_by(subjectId, trial_outcome, association_strength,
component, component_score_bin) %>%
summarize(rt = median(match_response.rt)) %>%
summarySEwithin(.,
measurevar="rt",
withinvars=c("association_strength",
"trial_outcome",
"component",
"component_score_bin"),
idvar="subjectId") %>%
mutate(across(component,
factor,
levels=ordered_pcs)) %>%
ggplot(aes(x = as.numeric(as.character(component_score_bin)),
y = rt,
color = trial_outcome,
group = trial_outcome)) +
geom_line(linewidth=0.8) +
geom_point(size=1.5) +
geom_errorbar(aes(ymin=rt-se, ymax=rt+se),
width=0.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "PC score quartile", y = "RT (s)") +
guides(color = "none")
joint.fig <- (fig1 + fig2 + fig3) +
plot_layout(widths=c(1,1,1),
guides="collect", axis_titles = "collect") &
theme(strip.text.x = element_text(size=12),
strip.text.y = element_blank(),
strip.background = element_blank(),
text = element_text(size = 10),
)
ggsave(paste("tPCA_figures_correct_only/PC", pc, "_scores_and_RT.pdf", sep=""),
plot=joint.fig,
width=22,
height=7,
dpi=300,
create.dir=T,
units="cm")
}
showtext_opts(op)
```
```{r other_version_of_pc_plots, fig.width=6, fig.height=2, echo=F}
showtext_opts(dpi = 300)
for (pc in 1:ncomp) {
fig1 <- pca.loadings %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color="black", linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 3,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component) +
ylim(-0.25, 1) +
labs(x = "Time (s)",
y = "PC loading") +
scale_color_brewer(palette="Set2") +
scale_fill_brewer(palette="Set2") +
theme(legend.position = "none",
strip.text.x = element_text(size=12),
strip.background = element_blank())
fig2 <- subj.data.scores %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(trial_outcome = ifelse(as.character(trial_outcome) %in% c("Hit"),
"Match", "Mismatch")) %>%
group_by(subjectId, association_strength, trial_outcome, component) %>%
summarize(component_score = mean(component_score)) %>%
summarySEwithin(.,
measurevar="component_score",
withinvars=c("association_strength",
"trial_outcome",
"component"),
idvar="subjectId") %>%
ggplot(aes(x = exp_id,
y = component_score,
fill = trial_outcome)) +
stat_summary(fun = "mean",
geom = "bar",
color = "black",
position=position_dodge()) +
geom_errorbar(aes(ymin=component_score-se, ymax=component_score+se),
position=position_dodge(.9),
width=.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "", y = "PC score") +
theme(axis.title.x=element_blank(),
axis.text.x=element_blank(),
axis.ticks.x=element_blank())
fig3 <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
group_by(subjectId, component, association_strength, trial_outcome) %>%
# exclude any subjects with # trials < # bins
mutate(trial_count = n()) %>%
ungroup() %>%
dplyr::filter(trial_count > nbins-1) %>%
group_by(subjectId, component, association_strength, trial_outcome) %>%
mutate(component_score_bin = cut_interval(component_score,
n = nbins,
labels = F)) %>%
ungroup() %>%
group_by(subjectId, trial_outcome, association_strength,
component, component_score_bin) %>%
summarize(rt = median(match_response.rt)) %>%
summarySEwithin(.,
measurevar="rt",
withinvars=c("association_strength",
"trial_outcome",
"component",
"component_score_bin"),
idvar="subjectId") %>%
mutate(across(component,
factor,
levels=ordered_pcs)) %>%
ggplot(aes(x = as.numeric(as.character(component_score_bin)),
y = rt,
color = trial_outcome,
group = trial_outcome)) +
geom_line(linewidth=0.8) +
geom_point(size=1.5) +
geom_errorbar(aes(ymin=rt-se, ymax=rt+se),
width=0.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "PC score quartile", y = "RT (s)") +
guides(color = "none")
joint.fig <- (fig1 + fig2 + fig3) +
plot_layout(widths=c(1,1,1),
guides="collect", axis_titles = "collect") &
theme(strip.text.y = element_blank()
)
ggsave(paste("tPCA_figures_correct_only/PC", pc, "_scores_and_RT_framed.pdf", sep=""),
plot=joint.fig,
width=23,
height=7,
dpi=300,
create.dir=T,
units="cm")
}
showtext_opts(op)
```
```{r rainbow_indivi_pc_plots, fig.width=2, fig.height=6, echo=F}
showtext_opts(dpi = 300)
full.rainbow.fig <- list()
for (pc in 1:ncomp) {
ordered.rainbow <- c("#FFCC00", "#E82C42", "#59169C", "#1E8DD2", "#F46D43", "#29A343")
pc.color <- ordered.rainbow[[pc]]
fig1 <- pca.loadings %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color=pc.color, linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 3,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component) +
ylim(-0.25, 1) +
labs(x = "Time (s)",
y = "PC loading") +
scale_color_brewer(palette="Set2") +
scale_fill_brewer(palette="Set2") +
theme(legend.position = "none")
fig2 <- subj.data.scores %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(trial_outcome = ifelse(trial_outcome %in% c("Hit"),
"Match", "Mismatch")) %>%
group_by(subjectId, association_strength, trial_outcome, component) %>%
summarize(component_score = mean(component_score)) %>%
summarySEwithin(.,
measurevar="component_score",
withinvars=c("association_strength",
"trial_outcome",
"component"),
idvar="subjectId") %>%
ggplot(aes(x = exp_id,
y = component_score,
fill = trial_outcome)) +
stat_summary(fun = "mean",
geom = "bar",
color = "black",
position=position_dodge()) +
geom_errorbar(aes(ymin=component_score-se, ymax=component_score+se),
position=position_dodge(.9),
width=.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "Memory outcome", y = "PC score") +
theme(axis.text.x=element_blank())
fig3 <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(trial_outcome = ifelse(trial_outcome %in% c("Hit"),
"Match", "Mismatch")) %>%
group_by(subjectId, component, association_strength, trial_outcome) %>%
# exclude any subjects with # trials < # bins
mutate(trial_count = n()) %>%
ungroup() %>%
dplyr::filter(trial_count > nbins-1) %>%
group_by(subjectId, component, association_strength, trial_outcome) %>%
mutate(component_score_bin = cut_interval(component_score,
n = nbins,
labels = F)) %>%
ungroup() %>%
group_by(subjectId, trial_outcome, association_strength,
component, component_score_bin) %>%
summarize(rt = median(match_response.rt)) %>%
summarySEwithin(.,
measurevar="rt",
withinvars=c("association_strength",
"trial_outcome",
"component",
"component_score_bin"),
idvar="subjectId") %>%
mutate(across(component,
factor,
levels=ordered_pcs)) %>%
ggplot(aes(x = as.numeric(as.character(component_score_bin)),
y = rt,
color = trial_outcome,
group = trial_outcome)) +
geom_line(linewidth=0.8) +
geom_point(size=1.5) +
geom_errorbar(aes(ymin=rt-se, ymax=rt+se),
width=0.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "PC score quartile", y = "RT (s)") +
guides(color = "none")
full.rainbow.fig[[pc]] <- fig1 / fig2 / fig3
}
rainbow.fig <- (full.rainbow.fig[[2]] |
full.rainbow.fig[[5]] |
full.rainbow.fig[[1]] |
full.rainbow.fig[[6]] |
full.rainbow.fig[[4]] |
full.rainbow.fig[[3]]) +
plot_layout(
guides = "collect", axes = "collect", axis_titles = "collect") &
theme(strip.text.x = element_text(size=12),
strip.text.y = element_blank(),
strip.background = element_blank(),
text = element_text(size = 10),
legend.position = "left"
)
ggsave(paste("tPCA_figures_correct_only/rainbow_all_PC_scores_and_RT.pdf", sep=""),
plot=rainbow.fig,
width=40,
height=18,
dpi=300,
create.dir=T,
units="cm")
showtext_opts(op)
```
```{r rainbow_with_titles_indivi_pc_plots, fig.width=2, fig.height=6, echo=F}
showtext_opts(dpi = 300)
titles <- list("PC1"="'Memory reinstatement strength'",
"PC2"="'Preparatory attention'",
"PC3"="'Prediction error x strength'",
"PC4"="'Orienting response to errors'",
"PC5"="'Cognitive effort'",
"PC6"="'Mismatch detection'")
full.rainbow.fig <- list()
for (pc in 1:ncomp) {
ordered.rainbow <- c("#FFCC00", "#E82C42", "#59169C", "#1E8DD2", "#F46D43", "#29A343")
pc.color <- ordered.rainbow[[pc]]
fig1 <- pca.loadings %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color=pc.color, linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 3,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component) +
ylim(-0.25, 1) +
labs(x = "Time (s)",
y = "PC loading",
title = titles[[pc]]) +
theme(legend.position = "none",
plot.title=element_text(hjust=0),
plot.subtitle=element_text(size=12))
fig2 <- subj.data.scores %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(across(trial_outcome,
factor,
levels=c("Hit", "Miss", "CR", "FA"))) %>%
group_by(subjectId, association_strength, trial_outcome, component) %>%
summarize(component_score = mean(component_score)) %>%
summarySEwithin(.,
measurevar="component_score",
withinvars=c("association_strength",
"trial_outcome",
"component"),
idvar="subjectId") %>%
ggplot(aes(x = exp_id,
y = component_score,
fill = trial_outcome)) +
stat_summary(fun = "mean",
geom = "bar",
color = "black",
position=position_dodge()) +
geom_errorbar(aes(ymin=component_score-se, ymax=component_score+se),
position=position_dodge(.9),
width=.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "Memory outcome", y = "PC score") +
theme(axis.text.x=element_blank())
fig3 <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
group_by(subjectId, component, association_strength, trial_outcome) %>%
# exclude any subjects with # trials < # bins
mutate(trial_count = n()) %>%
ungroup() %>%
dplyr::filter(trial_count > nbins-1) %>%
group_by(subjectId, component, association_strength, trial_outcome) %>%
mutate(component_score_bin = cut_interval(component_score,
n = nbins,
labels = F)) %>%
ungroup() %>%
group_by(subjectId, trial_outcome, association_strength,
component, component_score_bin) %>%
summarize(rt = median(match_response.rt)) %>%
summarySEwithin(.,
measurevar="rt",
withinvars=c("association_strength",
"trial_outcome",
"component",
"component_score_bin"),
idvar="subjectId") %>%
mutate(across(component,
factor,
levels=ordered_pcs)) %>%
ggplot(aes(x = as.numeric(as.character(component_score_bin)),
y = rt,
color = trial_outcome,
group = trial_outcome)) +
geom_line(linewidth=0.8) +
geom_point(size=1.5) +
geom_errorbar(aes(ymin=rt-se, ymax=rt+se),
width=0.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "PC score quartile", y = "RT (s)") +
guides(color = "none")
full.rainbow.fig[[pc]] <- fig1 / fig2 / fig3
}
rainbow.fig <- (full.rainbow.fig[[2]] |
full.rainbow.fig[[5]] |
full.rainbow.fig[[1]] |
full.rainbow.fig[[6]] |
full.rainbow.fig[[3]] |
full.rainbow.fig[[4]]) +
plot_layout(guides="collect",
tag_level = 'new') +
plot_annotation(tag_levels = list(c("A", "", "",
"B", "", "",
"C", "", "",
"D", "", "",
"E", "", "",
"F", "", ""))) &
theme(strip.text.x = element_text(size=12),
strip.text.y = element_blank(),
strip.background = element_blank(),
text = element_text(size = 10),
plot.margin = margin(0, 10, 0, 10),
plot.title.position = "plot",
plot.subtitle=element_text(hjust=0, margin=margin(t=5)),
plot.tag = element_text(face="bold", margin=margin(t=5, b=5)),
axis.title.y = element_text(margin = margin(r=0)),
axis.text.y = element_text(margin = margin(l=3, r=3)))
ggsave(paste("tPCA_figures_correct_only/rainbow_all_PC_scores_and_RT_titled.pdf", sep=""),
plot=rainbow.fig,
width=46,
height=16,
dpi=300,
create.dir=T,
units="cm")
showtext_opts(op)
```
```{r custom_indivi_pc_plots, fig.width=6, fig.height=2, echo=F}
indiv_pc_plots <- list()
showtext_opts(dpi = 300)
for (pc in 1:ncomp) {
fig1 <- pca.loadings %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color="black", linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 4,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component) +
ylim(-0.25, 1) +
labs(x = "Time (s)",
y = "PC loading",
subtitle = titles[[paste("PC", pc, sep="")]]) +
scale_color_brewer(palette="Set2") +
scale_fill_brewer(palette="Set2") +
theme(legend.position = "none",
plot.title=element_text(hjust=-0.1),
plot.subtitle=element_text(size=12))
fig2 <- subj.data.scores %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(across(trial_outcome,
factor,
levels=c("Hit", "Miss", "CR", "FA"))) %>%
group_by(subjectId, association_strength, trial_outcome, component) %>%
summarize(component_score = mean(component_score)) %>%
summarySEwithin(.,
measurevar="component_score",
withinvars=c("association_strength",
"trial_outcome",
"component"),
idvar="subjectId") %>%
ggplot(aes(x = exp_id,
y = component_score,
fill = trial_outcome)) +
stat_summary(fun = "mean",
geom = "bar",
color = "black",
position=position_dodge()) +
geom_errorbar(aes(ymin=component_score-se, ymax=component_score+se),
position=position_dodge(.9),
width=.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "", y = "PC score") +
theme(axis.title.x=element_blank(),
axis.text.x=element_blank(),
axis.ticks.x=element_blank())
fig3 <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
group_by(subjectId, component, association_strength, trial_outcome) %>%
# exclude any subjects with # trials < # bins
mutate(trial_count = n()) %>%
ungroup() %>%
dplyr::filter(trial_count > nbins-1) %>%
group_by(subjectId, component, association_strength, trial_outcome) %>%
mutate(component_score_bin = cut_interval(component_score,
n = nbins,
labels = F)) %>%
ungroup() %>%
group_by(subjectId, trial_outcome, association_strength,
component, component_score_bin) %>%
summarize(rt = median(match_response.rt)) %>%
summarySEwithin(.,
measurevar="rt",
withinvars=c("association_strength",
"trial_outcome",
"component",
"component_score_bin"),
idvar="subjectId") %>%
mutate(across(component,
factor,
levels=ordered_pcs)) %>%
ggplot(aes(x = as.numeric(as.character(component_score_bin)),
y = rt,
color = trial_outcome,
group = trial_outcome)) +
geom_line(linewidth=0.8) +
geom_point(size=1.5) +
geom_errorbar(aes(ymin=rt-se, ymax=rt+se),
width=0.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "PC score quartile", y = "RT (s)") +
guides(color = "none")
joint.fig <- (fig1 + fig2 + fig3) +
plot_layout(widths=c(1,1,1),
guides="collect", axes = "collect", axis_titles = "collect") &
theme(strip.text.x = element_text(size=12),
strip.text.y = element_blank(),
strip.background = element_blank(),
text = element_text(size = 10),
plot.title=element_text(hjust=0.5)
)
indiv_pc_plots[[paste("PC", pc, sep="")]] <- fig1 + fig2 + fig3
# ggsave(paste("tPCA_figures_correct_only/PC", pc, "_scores_and_RT.pdf", sep=""),
# plot=joint.fig,
# width=22,
# height=7,
# dpi=300,
# units="cm")
}
plot.full <- (indiv_pc_plots[["PC2"]] /
indiv_pc_plots[["PC5"]] /
indiv_pc_plots[["PC1"]] /
indiv_pc_plots[["PC6"]] /
indiv_pc_plots[["PC3"]] /
indiv_pc_plots[["PC4"]]) +
plot_layout(guides="collect",
tag_level = 'new') +
plot_annotation(tag_levels = list(c("A", "", "",
"B", "", "",
"C", "", "",
"D", "", "",
"E", "", "",
"F", "", ""))) &
theme(strip.text.x = element_text(size=12),
strip.text.y = element_blank(),
strip.background = element_blank(),
text = element_text(size = 10),
plot.margin = margin(0, 0, 0, 5),
plot.title.position = "plot",
plot.subtitle=element_text(hjust=0, margin=margin(t=5)),
plot.tag = element_text(face="bold", margin=margin(t=0)),
axis.title.y = element_text(margin = margin(r=0)),
axis.text.y = element_text(margin = margin(l=3, r=3)))
ggsave(paste("tPCA_figures_correct_only/full_PC_plot.pdf", sep=""),
plot=plot.full,
width=9,
height=16,
dpi=300,
create.dir=T,
units="in")
showtext_opts(op)
```
## Individual PC plots: score by RT
```{r rainbow_indivi_pc_plots_yx, fig.width=2, fig.height=6, echo=F}
showtext_opts(dpi = 300)
full.rainbow.fig <- list()
for (pc in 1:ncomp) {
ordered.rainbow <- c("#FFCC00", "#E82C42", "#59169C", "#1E8DD2", "#F46D43", "#29A343")
pc.color <- ordered.rainbow[[pc]]
fig1 <- pca.loadings %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color=pc.color, linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 3,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component) +
ylim(-0.25, 1) +
labs(x = "Time (s)",
y = "PC loading") +
scale_color_brewer(palette="Set2") +
scale_fill_brewer(palette="Set2") +
theme(legend.position = "none")
fig2 <- subj.data.scores %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(across(trial_outcome,
factor,
levels=c("Hit", "Miss", "CR", "FA"))) %>%
group_by(subjectId, association_strength, trial_outcome, component) %>%
summarize(component_score = mean(component_score)) %>%
summarySEwithin(.,
measurevar="component_score",
withinvars=c("association_strength",
"trial_outcome",
"component"),
idvar="subjectId") %>%
ggplot(aes(x = exp_id,
y = component_score,
fill = trial_outcome)) +
stat_summary(fun = "mean",
geom = "bar",
color = "black",
position=position_dodge()) +
geom_errorbar(aes(ymin=component_score-se, ymax=component_score+se),
position=position_dodge(.9),
width=.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "Memory outcome", y = "PC score") +
theme(axis.text.x=element_blank())
fig3 <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(rt = round(match_response.rt, 1)) %>%
group_by(association_strength) %>%
mutate(rt_bin = cut_interval(rt,
n = nbins)) %>%
ungroup() %>%
group_by(subjectId, trial_outcome, association_strength,
component, rt_bin) %>%
summarize(component_score = mean(component_score)) %>%
summarySEwithin(.,
measurevar="component_score",
withinvars=c("association_strength",
"trial_outcome",
"component",
"rt_bin"),
idvar="subjectId") %>%
mutate(across(component,
factor,
levels=ordered_pcs)) %>%
ggplot(aes(x = rt_bin,
y = component_score,
color = trial_outcome,
group = trial_outcome)) +
geom_line(linewidth=0.8) +
geom_point(size=1.5) +
geom_errorbar(aes(ymin=component_score-se, ymax=component_score+se),
width=0.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(y = "PC score", x = "RT bin") +
guides(color = "none") +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
full.rainbow.fig[[pc]] <- fig1 / fig2 / fig3
}
rainbow.fig <- (full.rainbow.fig[[2]] |
full.rainbow.fig[[5]] |
full.rainbow.fig[[1]] |
full.rainbow.fig[[6]] |
full.rainbow.fig[[3]] |
full.rainbow.fig[[4]]) +
plot_layout(
guides = "collect", axes = "collect", axis_titles = "collect") &
theme(strip.text.x = element_text(size=12),
strip.text.y = element_blank(),
strip.background = element_blank(),
text = element_text(size = 10),
legend.position = "left"
)
ggsave(paste("tPCA_figures_correct_only/rainbow_all_PC_scores_and_RT_yx.pdf", sep=""),
plot=rainbow.fig,
width=40,
height=18,
dpi=300,
create.dir=T,
units="cm")
showtext_opts(op)
```
```{r rainbow_with_titles_indivi_pc_plots_yx, fig.width=2, fig.height=6, echo=F}
showtext_opts(dpi = 300)
titles <- list("PC1"="'Memory reinstatement strength'",
"PC2"="'Preparatory attention'",
"PC3"="'Prediction error x strength'",
"PC4"="'Orienting response to errors'",
"PC5"="'Cognitive effort'",
"PC6"="'Mismatch detection'")
full.rainbow.fig <- list()
for (pc in 1:ncomp) {
ordered.rainbow <- c("#FFCC00", "#E82C42", "#59169C", "#1E8DD2", "#F46D43", "#29A343")
pc.color <- ordered.rainbow[[pc]]
fig1 <- pca.loadings %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color=pc.color, linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 3,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component) +
ylim(-0.25, 1) +
labs(x = "Time (s)",
y = "PC loading",
title = titles[[pc]]) +
theme(legend.position = "none",
plot.title=element_text(hjust=0),
plot.subtitle=element_text(size=12))
fig2 <- subj.data.scores %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(across(trial_outcome,
factor,
levels=c("Hit", "Miss", "CR", "FA"))) %>%
group_by(subjectId, association_strength, trial_outcome, component) %>%
summarize(component_score = mean(component_score)) %>%
summarySEwithin(.,
measurevar="component_score",
withinvars=c("association_strength",
"trial_outcome",
"component"),
idvar="subjectId") %>%
ggplot(aes(x = exp_id,
y = component_score,
fill = trial_outcome)) +
stat_summary(fun = "mean",
geom = "bar",
color = "black",
position=position_dodge()) +
geom_errorbar(aes(ymin=component_score-se, ymax=component_score+se),
position=position_dodge(.9),
width=.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "Memory outcome", y = "PC score") +
theme(axis.text.x=element_blank())
fig3 <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(rt = round(match_response.rt, 1)) %>%
group_by(association_strength) %>%
mutate(rt_bin = cut_interval(rt,
n = nbins)) %>%
ungroup() %>%
group_by(subjectId, trial_outcome, association_strength,
component, rt_bin) %>%
summarize(component_score = mean(component_score)) %>%
summarySEwithin(.,
measurevar="component_score",
withinvars=c("association_strength",
"trial_outcome",
"component",
"rt_bin"),
idvar="subjectId") %>%
mutate(across(component,
factor,
levels=ordered_pcs)) %>%
ggplot(aes(x = rt_bin,
y = component_score,
color = trial_outcome,
group = trial_outcome)) +
geom_line(linewidth=0.8) +
geom_point(size=1.5) +
geom_errorbar(aes(ymin=component_score-se, ymax=component_score+se),
width=0.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(y = "PC score", x = "RT bin") +
guides(color = "none") +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
full.rainbow.fig[[pc]] <- fig1 / fig2 / fig3
}
rainbow.fig <- (full.rainbow.fig[[2]] |
full.rainbow.fig[[5]] |
full.rainbow.fig[[1]] |
full.rainbow.fig[[6]] |
full.rainbow.fig[[3]] |
full.rainbow.fig[[4]]) +
plot_layout(guides="collect",
tag_level = 'new') +
plot_annotation(tag_levels = list(c("A", "", "",
"B", "", "",
"C", "", "",
"D", "", "",
"E", "", "",
"F", "", ""))) &
theme(strip.text.x = element_text(size=12),
strip.text.y = element_blank(),
strip.background = element_blank(),
text = element_text(size = 10),
plot.margin = margin(0, 10, 0, 10),
plot.title.position = "plot",
plot.subtitle=element_text(hjust=0, margin=margin(t=5)),
plot.tag = element_text(face="bold", margin=margin(t=5, b=5)),
axis.title.y = element_text(margin = margin(r=0)),
axis.text.y = element_text(margin = margin(l=3, r=3)))
ggsave(paste("tPCA_figures_correct_only/rainbow_all_PC_scores_and_RT_titled_yx.pdf", sep=""),
plot=rainbow.fig,
width=46,
height=16,
dpi=300,
create.dir=T,
units="cm")
showtext_opts(op)
```
```{r custom_indivi_pc_plots_yx, fig.width=6, fig.height=2, echo=F}
indiv_pc_plots <- list()
showtext_opts(dpi = 300)
for (pc in 1:ncomp) {
fig1 <- pca.loadings %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(component = paste(component, ": ", round(percent_var_explained, 2),
"%",
sep = "")) %>%
ggplot(aes(x = time,
y = loading,
color = component,
fill = component)) +
geom_vline(xintercept = c(0, 1, 2, 2.75),
linetype = "dashed",
color = "gray37") +
geom_line(color="black", linewidth=1) +
geom_text(aes(label = event,
x = time,
y = 0.75),
data = ret_labels,
inherit.aes = FALSE,
angle = 90,
hjust = 1,
vjust = 1.2,
size = 4,
colour = "gray37",
family = fontfamily) +
facet_wrap(~component) +
ylim(-0.25, 1) +
labs(x = "Time (s)",
y = "PC loading",
subtitle = titles[[paste("PC", pc, sep="")]]) +
scale_color_brewer(palette="Set2") +
scale_fill_brewer(palette="Set2") +
theme(legend.position = "none",
plot.title=element_text(hjust=-0.1),
plot.subtitle=element_text(size=12))
fig2 <- subj.data.scores %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(across(trial_outcome,
factor,
levels=c("Hit", "Miss", "CR", "FA"))) %>%
group_by(subjectId, association_strength, trial_outcome, component) %>%
summarize(component_score = mean(component_score)) %>%
summarySEwithin(.,
measurevar="component_score",
withinvars=c("association_strength",
"trial_outcome",
"component"),
idvar="subjectId") %>%
ggplot(aes(x = exp_id,
y = component_score,
fill = trial_outcome)) +
stat_summary(fun = "mean",
geom = "bar",
color = "black",
position=position_dodge()) +
geom_errorbar(aes(ymin=component_score-se, ymax=component_score+se),
position=position_dodge(.9),
width=.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(x = "", y = "PC score") +
theme(axis.title.x=element_blank(),
axis.text.x=element_blank(),
axis.ticks.x=element_blank())
fig3 <- rt_data %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(rt = round(match_response.rt, 1)) %>%
group_by(association_strength) %>%
mutate(rt_bin = cut_interval(rt,
n = nbins)) %>%
ungroup() %>%
group_by(subjectId, trial_outcome, association_strength,
component, rt_bin) %>%
summarize(component_score = mean(component_score)) %>%
summarySEwithin(.,
measurevar="component_score",
withinvars=c("association_strength",
"trial_outcome",
"component",
"rt_bin"),
idvar="subjectId") %>%
mutate(across(component,
factor,
levels=ordered_pcs)) %>%
ggplot(aes(x = rt_bin,
y = component_score,
color = trial_outcome,
group = trial_outcome)) +
geom_line(linewidth=0.8) +
geom_point(size=1.5) +
geom_errorbar(aes(ymin=component_score-se, ymax=component_score+se),
width=0.2) +
facet_grid(cols=vars(association_strength),
rows=vars(component)) +
scale_color_manual(values=c("#1a75ff", "#ff751a"), name="") +
scale_fill_manual(values=c("#1a75ff", "#ff751a"), name="") +
labs(y = "PC score", x = "RT bin") +
guides(color = "none") +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
joint.fig <- (fig1 + fig2 + fig3) +
plot_layout(widths=c(1,1,1),
guides="collect", axes = "collect", axis_titles = "collect") &
theme(strip.text.x = element_text(size=12),
strip.text.y = element_blank(),
strip.background = element_blank(),
text = element_text(size = 10),
plot.title=element_text(hjust=0.5)
)
indiv_pc_plots[[paste("PC", pc, sep="")]] <- fig1 + fig2 + fig3
# ggsave(paste("tPCA_figures_correct_only/PC", pc, "_scores_and_RT.pdf", sep=""),
# plot=joint.fig,
# width=22,
# height=7,
# dpi=300,
# units="cm")
}
plot.full <- (indiv_pc_plots[["PC2"]] /
indiv_pc_plots[["PC5"]] /
indiv_pc_plots[["PC1"]] /
indiv_pc_plots[["PC6"]] /
indiv_pc_plots[["PC3"]] /
indiv_pc_plots[["PC4"]]) +
plot_layout(guides="collect",
tag_level = 'new') +
plot_annotation(tag_levels = list(c("A", "", "",
"B", "", "",
"C", "", "",
"D", "", "",
"E", "", "",
"F", "", ""))) &
theme(strip.text.x = element_text(size=12),
strip.text.y = element_blank(),
strip.background = element_blank(),
text = element_text(size = 10),
plot.margin = margin(0, 0, 0, 5),
plot.title.position = "plot",
plot.subtitle=element_text(hjust=0, margin=margin(t=5)),
plot.tag = element_text(face="bold", margin=margin(t=0)),
axis.title.y = element_text(margin = margin(r=0)),
axis.text.y = element_text(margin = margin(l=3, r=3)))
ggsave(paste("tPCA_figures_correct_only/full_PC_plot_yx.pdf", sep=""),
plot=plot.full,
width=9,
height=16,
dpi=300,
create.dir=T,
units="in")
showtext_opts(op)
```
## Subsequent memory - CR trials
```{r submem_hist, echo=F, eval=T}
palette <- c("#59735e", "#b7c8ba")
palette <- c("#008080", "#9fdfbf")
subs.exclude <- c("MPE514", "MPE527") # insufficient number of trials
subj.data.submem <- subj.data.scores %>%
dplyr::filter(!subjectId %in% subs.exclude) %>%
left_join(recognition_behavior,
by = c("subjectId", "expId", "img"),
suffix = c(".retrieval", ".recognition")) %>%
dplyr::filter(!is.na(trial_outcome.recognition)) %>%
dplyr::filter(trial_outcome.retrieval == "CR") %>%
mutate(across(trial_outcome.recognition,
factor,
levels=c("Hit", "Miss"))) %>%
mutate(trial_outcome.recognition = recode(trial_outcome.recognition,
"Hit" = 1,
"Miss" = 0)) %>%
group_by(subjectId, component) %>%
mutate(component_score = scale(component_score)) %>%
ungroup() %>%
mutate(expId = recode(expId,
"imm" = "Experiment 1 (same day)",
"del" = "Experiment 2 (48 hrs later)"))
```
### Models
```{r recog_models, echo=F, results="asis", eval=run_models}
recog.glmer.models <- list()
for (i in 1:ncomp) {
mod.data <- subj.data.submem %>%
dplyr::filter(component == paste("PC", i, sep=""))
mod <- mod.data %>%
glmer(trial_outcome.recognition ~ component_score * association_strength * expId +
(component_score * association_strength | subjectId),
data = .)
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- mod.data %>%
glmer(trial_outcome.recognition ~ component_score * association_strength * expId +
(component_score + association_strength | subjectId),
family = "binomial",
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- mod.data %>%
glmer(trial_outcome.recognition ~ component_score * association_strength * expId +
(component_score | subjectId),
family = "binomial",
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- mod.data %>%
glmer(trial_outcome.recognition ~ component_score * association_strength * expId +
(1 | subjectId),
family = "binomial",
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
print("MODEL DID NOT CONVERGE")
}
recog.glmer.models[[i]] <- mod
}
```
```{r show_recog_mods, echo=F, results="asis", fig.height=3, fig.width=6}
for (i in 1:ncomp) {
mod <- recog.glmer.models[[i]]
tidy(mod, effects = "fixed", conf.int = T, conf.level = 0.95, exponentiate=T) %>%
# select(term, estimate, conf.low, conf.high, p.value, std.error, df) %>%
kable(caption=paste("PC", i, ": ", formula(mod)[3], sep=""),
digits=3) %>%
kable_styling() %>%
cat()
tidy(mod, effects = "ran_pars", scales = "vcov") %>%
select(term, estimate) %>%
kable(digits=3, caption="Random effects") %>%
kable_styling() %>%
cat()
if (i == 1) {
slopes <- emtrends(mod, pairwise ~ association_strength * expId, var = "component_score")
print(slopes)
title <- formula(mod)[3]
tidy(slopes$emtrends, conf.int = T, conf.level = 0.95) %>%
kable(caption=paste("slopes: remember ~ ", title, sep=""),
digits=3) %>%
kable_styling() %>%
cat()
tidy(slopes$contrasts, conf.int = T, conf.level = 0.95) %>%
select(contrast, estimate, std.error, df, conf.low, conf.high, statistic, adj.p.value) %>%
kable(caption=paste("slope contrasts: remember ~ ", title, sep=""),
digits=3) %>%
kable_styling() %>%
cat()
}
plt <- mod %>%
plot_model(type="pred",
terms=c("component_score[all]",
"association_strength[Strong (4x), Weak (1x)]",
"expId"),
show.data=F) +
scale_color_manual(values=palette, name="") +
scale_fill_manual(values=palette, name="") +
scale_y_continuous(limits = c(0, 1)) +
labs(y = "P(remember probe)",
x = paste("PC", i, " score (z)", sep=""),
title = paste("PC", i, sep="")) +
theme(plot.subtitle = element_text(size = 12))
print(plt)
nbins <- 4
submem.summary <- subj.data.submem %>%
dplyr::filter(component == paste("PC", i, sep="")) %>%
group_by(subjectId, association_strength, expId) %>%
mutate(component_score_bin = cut(component_score,
breaks = nbins,
labels = F)) %>%
ungroup() %>%
group_by(subjectId, component_score_bin, association_strength, expId) %>%
summarize(component_score = mean(component_score),
prob.remember = mean(trial_outcome.recognition)) %>%
ungroup() %>%
group_by(component_score_bin, association_strength, expId) %>%
mutate(component_score_mean = mean(component_score)) %>%
ungroup() %>%
mutate(component_score_mean = as.numeric(component_score_mean)) %>%
summarySEwithin(.,
measurevar="prob.remember",
withinvars=c("association_strength",
"component_score_mean"),
betweenvars=c("expId"),
idvar="subjectId") %>%
mutate(component_score_mean = as.numeric(as.character(component_score_mean)))
pred <- data.frame(ggpredict(mod, terms = c("component_score[all]",
"association_strength[Strong (4x), Weak (1x)]",
"expId"))) %>%
rename(component_score_mean = x,
association_strength = group,
expId = facet)
plt <- ggplot() +
geom_ribbon(data = as.data.frame(pred),
aes(x = component_score_mean,
ymin = conf.low,
ymax = conf.high,
fill = association_strength),
alpha = 0.2, color = NA) +
geom_line(data = as.data.frame(pred),
aes(x = component_score_mean,
y = predicted,
color = association_strength),
linewidth = 1) +
geom_point(data = submem.summary,
aes(x = component_score_mean,
y = prob.remember,
color = association_strength),
size = 1.5) +
geom_errorbar(data = submem.summary,
aes(x = component_score_mean,
ymin = prob.remember - se,
ymax = prob.remember + se,
color = association_strength),
width = 0.2) +
facet_wrap(~ expId) +
scale_color_manual(values = palette, name = "") +
scale_fill_manual(values = palette, name = "") +
scale_y_continuous(limits = c(0, 1)) +
labs(y = "P(remember probe)",
x = paste("PC", i, " score (z)", sep=""),
title = paste("PC", i, sep = "")) +
theme(strip.text.x = element_text(size=12))
print(plt)
showtext_opts(dpi = 300)
ggsave(paste("submem_figures/submem_PC", i, ".pdf", sep=""),
plot=plt,
width=17,
height=8,
dpi=300,
create.dir=T,
units="cm")
showtext_opts(op)
# print(mod %>%
# plot_model(type="pred",
# pred.type = "re",
# terms=c("component_score",
# "subjectId",
# "expId",
# "association_strength"
# ),
# ci.lvl = NA,
# show.data=F,
# color = "gray",
# line.size = 0.2) +
# scale_y_continuous(limits = c(0, 1)) +
# theme(legend.position = "none"))
}
```
## Subsequent memory RT
```{r rt_data_submem, echo=F}
### joint plot
rt_data.submem <- subj.data.submem %>%
dplyr::filter(trial_outcome.recognition == 1) %>%
mutate(across(association_strength,
factor,
levels=c("Strong (4x)", "Weak (1x)")))
```
```{r submem_rt_joint_plot, echo=F, fig.width=10, fig.height=8, results="asis", eval=run_models}
models.recog.rt <- list()
for (pc in 1:ncomp) {
mod <- rt_data.submem %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(recognition_response.rt = log(recognition_response.rt)) %>%
lmer(recognition_response.rt ~ component_score * association_strength * expId +
(component_score * association_strength * expId | subjectId),
data = .)
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- rt_data.submem %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(recognition_response.rt = log(recognition_response.rt)) %>%
lmer(recognition_response.rt ~ component_score * association_strength * expId +
(component_score + association_strength + expId | subjectId),
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- rt_data.submem %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(recognition_response.rt = log(recognition_response.rt)) %>%
lmer(recognition_response.rt ~ component_score * association_strength * expId +
(component_score | subjectId),
data = .)
}
if (isSingular(mod) | any(grepl("failed to converge", mod@optinfo$conv$lme4$messages))) {
mod <- rt_data.submem %>%
dplyr::filter(component == paste("PC", pc, sep="")) %>%
mutate(recognition_response.rt = log(recognition_response.rt)) %>%
lmer(recognition_response.rt ~ component_score * association_strength * expId +
(1 | subjectId),
data = .)
}
models.recog.rt[[pc]] <- mod
}
```
```{r submem.binned.rt, echo=F, fig.width=7, fig.height=10, eval=T}
nbins <- 4
rt.binned.plt <- rt_data.submem %>%
mutate(trial_outcome = trial_outcome.recognition) %>%
mutate(across(association_strength,
factor,
levels=c("Strong (4x)", "Weak (1x)"))) %>%
group_by(subjectId, component, association_strength, trial_outcome, expId) %>%
mutate(trial_count = n()) %>%
ungroup() %>%
dplyr::filter(trial_count > nbins-1) %>%
group_by(subjectId, component, association_strength, trial_outcome, expId) %>%
mutate(component_score_bin = cut_interval(component_score,
n = nbins,
labels = F)) %>%
ungroup() %>%
group_by(subjectId, trial_outcome, association_strength,
component, component_score_bin, expId) %>%
summarize(rt = median(recognition_response.rt)) %>%
summarySEwithin(.,
measurevar="rt",
withinvars=c("association_strength",
"trial_outcome",
"component",
"component_score_bin"),
betweenvars=c("expId"),
idvar="subjectId") %>%
mutate(across(component,
factor,
levels=ordered_pcs)) %>%
ggplot(aes(x = as.numeric(as.character(component_score_bin)),
y = rt,
color = association_strength,
group = association_strength)) +
geom_line(linewidth=0.8) +
geom_point(size=1.5) +
geom_errorbar(aes(ymin=rt-se, ymax=rt+se),
width=0.2) +
facet_grid(rows=vars(component),
axes="all",
scales = "free") +
scale_color_manual(values=palette, name="") +
scale_fill_manual(values=palette, name="") +
labs(x = "PC score", y = "RT (s)") +
theme(strip.text.y = element_blank()) +
facet_grid(rows = vars(component),
cols = vars(expId))
loadings.vrt.plt + rt.binned.plt +
plot_layout(widths=c(1,1.5))
```
### Models
```{r print_submem_rt_models, echo=F, results="asis", eval=T}
for (pc in 1:ncomp) {
tidy(models.recog.rt[[pc]], effects = "fixed", conf.int = T, conf.level = 0.95) %>%
select(term, estimate, conf.low, conf.high, p.value, std.error, df) %>%
kable(caption=paste("PC", pc, ": ", formula(mod)[3], sep=""),
digits=3) %>%
kable_styling() %>%
cat()
tidy(models.recog.rt[[pc]], effects = "ran_pars", scales = "vcov") %>%
select(term, estimate) %>%
kable(digits=3, caption="Random effects") %>%
kable_styling() %>%
cat()
}
```
## Save
```{r save_models, echo=F, eval=run_models}
save(models,
models.compact.no.mismatch,
models.rt,
recog.glmer.models,
models.recog.rt,
file="tpca_models_correct_only.RData")
```

pupil_temporal_pca.Rmd at commit d144397, under GPL-3.0 · at the source

Overview

Authors: Alice M Xue^1,2, Jenna Jokhani^1,2, Anthony M Norcia^1,2, Anthony D Wagner^1,2

ORCID iDs: Alice M Xue, Anthony D Wagner

Department of Psychology, Stanford University Stanford United States
Wu Tsai Neurosciences Institute, Stanford University Stanford United States

Institutions: Stanford University (United States)

Journal: eLife, volume 15, article RP109678

Dates: published online 21 October 2026

Type: Research article · Language: English

License: CC BY

Identifiers: DOI 10.7554/elife.109678 · PMID 42905440 · PMCID PMC13686618 · OpenAlex W7130661913

Open access: gold, a free copy (OpenAlex)

Status: code verified (how the code is found and verified)

Categories: EEG (modality), human (organism), cognitive (subfield)

Methods: Spectral & time-frequency, Connectivity, Statistics, Smoothing, state filtering, decompositions, Machine learning, Preprocessing, Single-unit activity, calcium imaging, Physiology & signal measures

Keywords: arousal, attention, cognitive control, episodic memory, learning, prediction error, Human

MeSH: Arousal*, Attention*, Cognition*, Memory*, Adult, Brain, Electroencephalography, Female, Humans, Male, Pupil, Young Adult (* major topic)

Journal subjects: Neuroscience

Topic: Memory Processes and Influences (Cognitive Neuroscience, Neuroscience), according to OpenAlex

Funding: National Institute on Aging (R01AG065255); National Science Foundation (Graduate Research Fellowship Program DGE-2146755); School of Medicine, Stanford University (Stanford Interdisciplinary Graduate Fellowship)

Citations: not cited yet (Europe PMC); 96 references in the paper

Abstract

Ongoing experience is continuously processed in the context of past events. Divergence between current experience and memory-based predictions (i.e., a mnemonic prediction error; MPE) is theorized to be a signal that encoding, as opposed to retrieval, should be prioritized. We asked how MPEs place demands on control and attention-related processes, and whether these demands differ as a function of prediction strength. We investigated these questions across two experiments, wherein we recorded scalp electroencephalography and/or pupillometry as 101 young human adults performed an associative memory task. Strong MPEs, more so than weak MPEs, increased physiological indices of cognitive control (frontal theta), attention (posterior alpha and pupil size), and arousal (pupil size). Greater cognitive control (frontal theta) during strong MPEs promoted better learning of prediction-violating (i.e., unexpected) stimuli. Collectively, these findings reveal multiple pathways through which the mind and brain respond adaptively to violations of strong mnemonic predictions.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repository

Its files are read in the Code ↔ Paper reader above, with 3 matches between paragraphs and lines of code.

alicexue/mpe-attention-cog-ctrl

License: GPL-3.0

State: the link answers, verified on 25 October 2026

Evidence: files inventoried

Commit: d1443971f1b402b0abcfa740f67bedc99daf88db, 20 October 2026

Languages: R (20)

Size: 34 files, 20 scripts

Software Heritage: archived

Found in: “Data availability”

Holds: README, license file, 16 notebooks

Not found: CITATION.cff, environment file, tests, continuous integration, documentation

Tools: broom (17 files), ggplot2 (17 files), patchwork (17 files), tidyverse (17 files), afex (16 files), data.table (16 files), emmeans (16 files), lme4 (16 files), ggpubr (15 files), mgcv (15 files), psych (15 files), brms (1 file), easystats (1 file)

Availability: 1 check, the latest on 25 October 2026: the link answers

25 October 2026: the link answers

22 files

scripts/behavior/behavioral_analysis.Rmd, R, 864 lines
scripts/eeg/compile_eeg_and_pupil_betas.R, R, 213 lines
scripts/eeg/generate_power_reports.R, R, 75 lines
scripts/eeg/mediation_models_brms.Rmd, R, 597 lines
scripts/eeg/power_and_pupil_cross_corr.Rmd, R, 991 lines
scripts/eeg/spectral_analyses_and_pupil.Rmd, R, 3,646 lines
scripts/eeg/spectral_analyses_and_sub_mem.Rmd, R, 1,236 lines
scripts/eeg/spectral_analyses_include_incorrect.Rmd, R, 1,917 lines
scripts/eeg/spectral_timeseries.Rmd, R, 1,155 lines
scripts/eeg/theta_alpha_corr.Rmd, R, 1,350 lines
scripts/pupil/compare_experiment_tPCA_loadings.Rmd, R, 148 lines
scripts/pupil/generate_pupil_reports.R, R, 45 lines
scripts/pupil/pupil_retrieval_timeseries.Rmd, R, 2,008 lines
scripts/pupil/pupil_retrieval_timeseries_by_experiment.Rmd, R, 1,896 lines
scripts/pupil/pupil_retrieval_timeseries_include_incorrect.Rmd, R, 1,190 lines
scripts/pupil/pupil_temporal_pca.Rmd, R, 2,507 lines, 2 matches
scripts/pupil/pupil_temporal_pca_bootstrap.Rmd, R, 1,115 lines
scripts/pupil/pupil_temporal_pca_include_incorrect.Rmd, R, 2,370 lines, 1 match
scripts/pupil/pupil_tpca_next_trial_effects.Rmd, R, 612 lines
scripts/utils/summary_se_utils.R, R, 136 lines
LICENSE, License, 674 lines
README.md, Text, 3 lines

At the source: github.com/alicexue/mpe-attention-cog-ctrl

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
20 scripts, each with its path and the digest of its content;
3 matches between paragraphs of the paper and lines of the code (method lexical-v1);
neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

No dataset and no data link were found in the paper.

Data availability

Data and code for this manuscript are accessible at https://github.com/alicexue/mpe-attention-cog-ctrl (copy archived at Xue, 2026).

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 25 October 2026: the first record

Recorded: type, language, journal, volume, pages, dates, 4 authors, 7 keywords, 12 MeSH terms, 3 funders, 94 references.

Cite

This paper

Xue, A. M., Jokhani, J., Norcia, A. M., & Wagner, A. D. (2026). Strong mnemonic prediction errors increase cognitive control, attention, and arousal. eLife, 15, RP109678. https://doi.org/10.7554/elife.109678

BibTeX

@article{xue2026strong, author = {Xue, Alice M and Jokhani, Jenna and Norcia, Anthony M and Wagner, Anthony D}, title = {{Strong mnemonic prediction errors increase cognitive control, attention, and arousal}}, journal = {eLife}, year = {2026}, month = oct, volume = {15}, pages = {RP109678}, publisher = {eLife Sciences Publications, Ltd}, issn = {2050-084X}, doi = {10.7554/elife.109678}, url = {https://doi.org/10.7554/elife.109678}, pmid = {42905440}, pmcid = {PMC13686618} }

RIS

TY - JOUR AU - Xue, Alice M AU - Jokhani, Jenna AU - Norcia, Anthony M AU - Wagner, Anthony D TI - Strong mnemonic prediction errors increase cognitive control, attention, and arousal T2 - eLife J2 - eLife PY - 2026 DA - 2026/10/21 VL - 15 SP - RP109678 SN - 2050-084X PB - eLife Sciences Publications, Ltd DO - 10.7554/elife.109678 UR - https://doi.org/10.7554/elife.109678 LA - en ER -

CSL-JSON

{ "id": "10.7554/elife.109678", "type": "article-journal", "title": "Strong mnemonic prediction errors increase cognitive control, attention, and arousal", "container-title": "eLife", "author": [ { "family": "Xue", "given": "Alice M" }, { "family": "Jokhani", "given": "Jenna" }, { "family": "Norcia", "given": "Anthony M" }, { "family": "Wagner", "given": "Anthony D" } ], "container-title-short": "eLife", "volume": "15", "page": "RP109678", "DOI": "10.7554/elife.109678", "PMID": "42905440", "PMCID": "PMC13686618", "ISSN": "2050-084X", "publisher": "eLife Sciences Publications, Ltd", "URL": "https://doi.org/10.7554/elife.109678", "language": "en", "issued": { "date-parts": [ [ 2026, 10, 21 ] ] } }

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1186/s13229-025-00678-w [code]: Locus coeruleus tonic upregulation increases selectivity to inconspicuous auditory information in autistic compared to non-autistic individuals: a combined pupillometry and electroencephalography study.
Journal: Molecular autism
In common: brms, easystats, psych, 5 other tools, EEG, cognitive, 6 references
[2] doi:10.1371/journal.pone.0355165 [code]: Pupillary dynamics during hands-off L2 driving and transitions of control under high cognitive load.
Journal: PloS one
In common: brms, mgcv, easystats, 6 other tools, cognitive, 4 references
[3] doi:10.1073/pnas.2503535122 [code]: Hippocampal mismatch signals are based on episodic memories and not schematic knowledge.
Journal: Proceedings of the National Academy of Sciences of the United States of America
In common: cognitive, 11 references
[4] doi:10.3758/s13423-022-02124-x [code]: Explicitly predicting outcomes enhances learning of expectancy-violating information.
Journal: Psychonomic bulletin & review
In common: afex, easystats, psych, 6 other tools, cognitive, 3 references
[5] doi:10.1186/s11689-025-09651-3 [code]: The association between infant EEG aperiodic exponent and the trajectory of restricted and repetitive behaviors for toddlers with and without autism.
Journal: Journal of neurodevelopmental disorders
In common: brms, afex, mgcv, 8 other tools, EEG
[6] doi:10.1162/imag_a_00312 [code]: Alpha-band fluctuations represent behaviorally relevant excitability changes as a consequence of top-down guided spatial attention in a probabilistic spatial cueing design.
Journal: Imaging neuroscience (Cambridge, Mass.)
In common: brms, afex, broom, 6 other tools, EEG, cognitive, 2 references
[7] doi:10.1038/s41593-024-01672-w [code]: Cognitive control training with domain-general response inhibition does not change children's brains or behavior.
Journal: Nature neuroscience
In common: brms, afex, broom, 8 other tools, cognitive
[8] doi:10.1038/s41398-026-03948-0 [code]: Locus coeruleus modulation of neurophysiological sensory selectivity differs in autism and other mental health conditions.
Journal: Translational psychiatry
In common: easystats, psych, emmeans, 5 other tools, EEG, 4 references
[9] doi:10.1016/j.neuroimage.2026.122115 [code]: Midfrontal theta power relates to response speeding following frustrative nonreward.
Journal: NeuroImage
In common: easystats, psych, emmeans, 5 other tools, EEG, cognitive, 4 references
[10] doi:10.1038/s41467-022-31040-w [code]: Hippocampal representations switch from errors to predictions during acquisition of predictive associations.
Journal: Nature communications
In common: cognitive, 10 references

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Add the badge to the code's README

If you are an author or a maintainer of this paper's code, you can add this badge. It says what OSCR knows of the paper (its code, its validated tracing map) and links the code to this page. Paste one of these snippets under the README's title. Keep the two marks around it, so that the badge can be replaced in place and is never added twice. OSCR changes nothing in your repository.

Markdown

<!-- oscr-badge:start -->
[![OSCR](https://openscicode.org/badge/doi/10.7554/elife.109678.svg)](https://openscicode.org/badge/doi/10.7554/elife.109678)
<!-- oscr-badge:end -->

reStructuredText

.. oscr-badge:start

.. image:: https://openscicode.org/badge/doi/10.7554/elife.109678.svg
   :target: https://openscicode.org/badge/doi/10.7554/elife.109678
   :alt: OSCR

.. oscr-badge:end

HTML

<!-- oscr-badge:start -->
<a href="https://openscicode.org/badge/doi/10.7554/elife.109678"><img src="https://openscicode.org/badge/doi/10.7554/elife.109678.svg" alt="OSCR"></a>
<!-- oscr-badge:end -->

On GitHub, open the README, choose Edit (the pencil), paste the snippet under the title, then "Commit changes".

alicexue/mpe-attention-cog-ctrl: README.md on GitHub

If you choose to, you can have the badge kept up to date for you, with oscr badge --write from the command line or with the GitHub Action in your repository's workflows. Every other form and option (a card, AsciiDoc, LaTeX, shields.io) is on the badge page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: code errors and mismatches between the code and the paper about this paper and its code, from readers and its authors, beside the paper and its tracing map.

Report a mismatch between the code and the paper · Report a code error (signed in)

Reproductions: reports from readers who ran the authors' code and did not get the paper's result, what they ran, with which environment, commit and data.

Report a reproduction failure (signed in)

Versions of its code: the releases of its code tied to a version of this paper (the preprint, the accepted manuscript, the version of record…), each with the tracing map versioned with it.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.