The human hippocampus can pattern separate memories by meaning.

Code ↔ Paper

9 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 9 matches

[1] § Behavioral Methods › Materials. › Covariate assessments. ↔ scripts/smst2_bids_participants_maker.py, lines 191–215 · score 0.70 · WAIS IV, Digit Symbol, subtests, Vocabulary, covariate, words
[2] § Behavioral Methods › MRI Methods. › Hippocampal subfield segmentation. ↔ scripts/smst_mr1_HCmask_snapshot.sh, lines 1–50 · score 0.64 · coronal slices, template T1, brain, masks, SI
[3] § Behavioral Methods › Materials. › Main task (sMST). › Stimuli. ↔ scripts/smst2_bids_tsv_maker.py, lines 303–386 · score 0.63 · Hungarian National Corpus, imageability, metric, arousal, concreteness, fillers
[4] § Behavioral Methods › MRI Methods. › Analysis. ↔ scripts/smst_mr1_analysis.Rmd, lines 156–247 · score 0.61 · close lure, distant lure, lure correct, incorrect, hits, foil
[5] § Results › RS Sensitive Clusters in the Hippocampal Head Differentially Activate for Exact Repeats and Modified Repeats During Encoding. ↔ scripts/smst_mr1_analysis.Rmd, lines 2190–2332 · score 0.61 · DG CA3, right hemisphere, left hemisphere, CA12, ROI, Facets
[6] § Results › Neural Response Is Increased for Semantically Close Modified Phrases Compared to Exact Repetitions, with No Further Difference between Semantic Similarity Bins. ↔ scripts/smst_mr1_analysis.Rmd, lines 2190–2332 · score 0.54 · DG CA3, distant repeats, CA1, anatomically, ANOVA, head
[7] § Behavioral Methods › Analysis. ↔ scripts/smst_mr1_analysis.Rmd, lines 398–440 · score 0.54 · Wald, confint, intercept, arousal, concreteness, GLMEMs
[8] § Results › RS Sensitive Clusters in the Hippocampal Head Differentially Activate for Exact Repeats and Modified Repeats During Encoding. ↔ scripts/smst_mr1_analysis.Rmd, lines 442–540 · score 0.53 · Post hoc, covariate, clusters, encoding
[9] § Results › Clusters Sensitive to Repetition at Encoding Do Not Track Mnemonic Discrimination Success within Lures but Differentiate Correctly Rejected Lures from Correctly Rejected Foils During Recogni ↔ scripts/smst_mr1_analysis.Rmd, lines 3552–3596 · score 0.50 · lure correct rejections, target hits, modified repeat, baseline, clusters

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: 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 · 5,188 lines · 155 KB · no license · 6 matches

---
title: "Analysis pipeline for the behavioral and functional MRI data from the semantic Mnemonic Similarity Task"
output: html_document
editor_options:
chunk_output_type: console
---
# Glossary & packages
ENC = Encoding task phase
REC = Recognition task phase
RS = repetition suppression (first presentation - repeat)
PS = pattern separation (hippocampal operation)
MD = mnemonic discrimination (behavioural correlate)
LDI = lure discrimination index (proxy of MD)
```{r}
knitr::opts_chunk$set(echo = TRUE)
library(ppcor)
library(tidyverse)
library(tidytuesdayR)
library(readxl)
library(tidyr)
library(scales)
library(ggplot2)
library(psycho)
library(ggpubr)
library(rstatix)
library(afex)
library(MuMIn)
library(lme4)
library(lmerTest)
library(ggstatsplot)
library(sjPlot)
library(smplot2)
library(openxlsx)
library(stats)
library(performance)
library(emmeans)
library(purrr)
library(broom)
library(ggpattern)
library(ggtext)
library(dplyr)
library(ppcor)
library(readr)
library(stringr)
library(grid)
library(gridExtra)
library(gtable)
library(patchwork)
theme_set(theme_light())
custom_theme = theme(
plot.title = element_text(color="black", size=30, face="bold", hjust = 0, vjust = 3), #30 for covariate plots
axis.text.x = element_text(vjust = 0.5, hjust=0.5, size = 20, face="bold"), #was 16 for other plots besides rec resp rate
axis.text.y = element_text(vjust = 0.5, hjust=0.5, size = 20, face="bold"),
axis.title.x = element_text(vjust = -2, hjust=0.5, size = 26, face="bold"),
axis.title.y = element_text(vjust = 3, hjust=0.5, size = 26, face="bold"),
legend.text = element_text(vjust = 0.5, hjust=0.5, size = 22, face="bold"),
legend.title = element_text(vjust = 0.5, hjust=0.5, size = 24, face="bold"),
strip.text = element_text(vjust = 0.5, hjust=0.5, size = 22, face="bold"),
plot.margin = margin(30, 30, 30, 30),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.line = element_line(colour = "black")
)
```
# Data handling
## Read-in MR
```{r}
semmst_mr_data = read.xlsx("./data/fmri/ses-01/subfield_data_featquery_2025.xlsx")
```
## Preprocess MR
```{r}
semmst_mr_data_filtered = semmst_mr_data %>%
dplyr::select(-c(cope_num, query_type, groupanalysis_type)) %>%
dplyr::rename(ID = subject) %>%
mutate(
ID = as.factor(ID),
space_type = as.factor(space_type),
task = as.factor(task),
cope_name = as.factor(cope_name),
hemisphere = as.factor(hemisphere),
roi = as.factor(roi),
mask_type = as.factor(mask_type),
contrast = as.factor(contrast),
cluster = as.factor(cluster),
estimate = as.numeric(estimate),
)
```
## Read-in behavior
```{r}
semmst_behav_data <- read_xlsx("./data/behavioural/smst/smst_mr_behvaioural_grand.xlsx")
```
## Preprocess behavior
```{r}
# Clean and factorize
semmst_behav_data_filtered <- semmst_behav_data %>%
dplyr::rename(ID = participant_id) %>%
mutate(ID = factor(str_c("sub-", ID)),
task = factor(task, levels = c("enc", "rec")),
run = factor(run),
trial_type = factor(trial_type),
stim_file = factor(stim_file),
itemno = factor(itemno),
order = factor(order),
response = tolower(response),
sex = factor(sex, levels = c("male", "female")),
handedness = factor(handedness, levels = "R"),
response_time = as.numeric(response_time),
onset = as.numeric(onset),
correctness = factor(correctness)) %>%
filter(ID %in% semmst_mr_data_filtered$ID)
semmst_mr_behav_data_enc <- semmst_behav_data_filtered %>%
filter(task == "enc",
trial_type != "FILLER") %>%
mutate(former_type = ifelse(trial_type=="REPEAT", "TARGET", "LURE")) %>%
dplyr::select(ID, noun, former_type) %>%
distinct()
semmst_mr_behav_data_rec <- semmst_behav_data_filtered %>%
filter(task == "rec",
!is.na(response)) %>%
left_join(semmst_mr_behav_data_enc, by = c("ID", "noun") ) %>%
mutate(response = factor(response, levels=c("old", "new"))) %>%
mutate(trial_type = factor(trial_type)) %>%
mutate(former_type = factor(former_type)) %>%
mutate(correctness = factor(correctness, levels=c(0,1)))
```
### LDI
```{r}
semmst_behav_data_rec_filtered = semmst_mr_behav_data_rec %>%
filter(task == "rec") %>%
filter(response != "None") %>%
mutate(response = as.factor(response))
semmst_behav_data_rec_wide_dprime = semmst_behav_data_rec_filtered %>%
dplyr::count(ID, trial_type, response) %>%
unite("trial_resp", c("trial_type", "response"), sep = "_") %>%
spread(trial_resp, n) %>%
mutate_at(vars(matches("_")), ~ifelse(is.na(.), 0, .)) %>%
mutate(targ_n = TARGET_new + TARGET_old) %>%
mutate(close_n = CLOSE_new + CLOSE_old) %>%
mutate(distant_n = DISTANT_new + DISTANT_old) %>%
mutate(foil_n = FOIL_new + FOIL_old) %>%
mutate(LURE_old = CLOSE_old + DISTANT_old) %>%
mutate(LURE_new = CLOSE_new + DISTANT_new) %>%
mutate(lure_n = LURE_new + LURE_old)
# Simple recognition - TARG vs FOIL
rec_dprime <- psycho::dprime(
n_hit = semmst_behav_data_rec_wide_dprime$TARGET_old,
n_fa = semmst_behav_data_rec_wide_dprime$FOIL_old,
n_miss = semmst_behav_data_rec_wide_dprime$TARGET_new,
n_cr = semmst_behav_data_rec_wide_dprime$FOIL_new,
n_targets = semmst_behav_data_rec_wide_dprime$targ_n,
n_distractors = semmst_behav_data_rec_wide_dprime$foil_n,
adjusted = TRUE
)
# Close lure discrimination - TARG vs CLOSE
close_dprime <- psycho::dprime(
n_hit = semmst_behav_data_rec_wide_dprime$TARGET_old,
n_fa = semmst_behav_data_rec_wide_dprime$CLOSE_old,
n_miss = semmst_behav_data_rec_wide_dprime$TARGET_new,
n_cr = semmst_behav_data_rec_wide_dprime$CLOSE_new,
n_targets = semmst_behav_data_rec_wide_dprime$targ_n,
n_distractors = semmst_behav_data_rec_wide_dprime$close_n,
adjusted = TRUE
)
# Distant lure discrimination - TARG vs DISTANT
dist_dprime <- psycho::dprime(
n_hit = semmst_behav_data_rec_wide_dprime$TARGET_old,
n_fa = semmst_behav_data_rec_wide_dprime$DISTANT_old,
n_miss = semmst_behav_data_rec_wide_dprime$TARGET_new,
n_cr = semmst_behav_data_rec_wide_dprime$DISTANT_new,
n_targets = semmst_behav_data_rec_wide_dprime$targ_n,
n_distractors = semmst_behav_data_rec_wide_dprime$distant_n,
adjusted = TRUE
)
# All lure discrimination - TARG vs LURE
lure_dprime <- psycho::dprime(
n_hit = semmst_behav_data_rec_wide_dprime$TARGET_old,
n_fa = semmst_behav_data_rec_wide_dprime$LURE_old,
n_miss = semmst_behav_data_rec_wide_dprime$TARGET_new,
n_cr = semmst_behav_data_rec_wide_dprime$LURE_new,
n_targets = semmst_behav_data_rec_wide_dprime$targ_n,
n_distractors = semmst_behav_data_rec_wide_dprime$lure_n,
adjusted = TRUE
)
semmst_behav_data_rec_summary <- semmst_behav_data_rec_wide_dprime %>%
transmute(
ID,
# Recognition
rec_dprime = rec_dprime$dprime,
rec_correct = (TARGET_old) / (targ_n),
rec_incorrect = (TARGET_new) / (targ_n),
# Close LDI
close_dprime = close_dprime$dprime,
close_correct = (CLOSE_new) / (close_n),
close_incorrect= (CLOSE_old) / (close_n),
# Distant LDI
distant_dprime = dist_dprime$dprime,
distant_correct = (DISTANT_new) / (distant_n),
distant_incorrect= (DISTANT_old) / (distant_n),
# Lure LDI
lure_dprime = lure_dprime$dprime,
lure_correct = (LURE_new) / (lure_n),
lure_incorrect= (LURE_old) / (lure_n),
)
```
## Merge
```{r}
semmst_merged_data <- semmst_mr_data_filtered %>%
left_join(
semmst_behav_data_filtered %>%
dplyr::select(ID, sex:GRO_score_SES02) %>%
distinct(ID, .keep_all = TRUE),
by = "ID"
) %>%
left_join(semmst_behav_data_rec_summary, by = "ID") %>%
droplevels()
```
## FILTER for HYPOTHESIS TESTING
```{r}
# REPETITION SENSTIVITY
semmst_merged_data_filtered = semmst_merged_data %>%
filter(contrast %in% c("ENC_ERP_ALLRS")) %>%
filter(space_type == "standard") %>%
filter(voxels > 50) %>%
filter(as.numeric(as.character(cluster)) < 140) %>%
mutate(roi_cluster = paste(hemisphere, roi, cluster, sep = "_")) %>%
mutate(roi_hemisphere = paste(hemisphere, roi, sep = "_")) %>%
mutate(Vocabulary_scale = scale(Vocabulary)) %>%
mutate(Digit_Symbol_scale = scale(Digit_Symbol)) %>%
mutate(NonWord_sum_scale = scale(NonWord_sum))
# ANATOMY CONTROL
semmst_merged_data_filtered_anat = semmst_merged_data %>%
filter(contrast %in% c("anatHC")) %>%
mutate(roi_cluster = paste(hemisphere, roi, cluster, sep = "_")) %>%
mutate(roi_hemisphere = paste(hemisphere, roi, sep = "_")) %>%
mutate(Vocabulary_scale = scale(Vocabulary)) %>%
mutate(Digit_Symbol_scale = scale(Digit_Symbol)) %>%
mutate(NonWord_sum_scale = scale(NonWord_sum))
```
# Revision data handling - read-in, filter and merge
```{r}
# Read in
semmst_mr_data_supplement = read.xlsx("./data/fmri/ses-01/subfield_data_featquery_supplement.xlsx")
# Filter
semmst_mr_data_supplement_filtered = semmst_mr_data_supplement %>%
dplyr::select(-any_of(c("cope_num", "query_type", "groupanalysis_type"))) %>%
dplyr::rename(ID = subject) %>%
mutate(
ID = as.factor(ID),
space_type = as.factor(space_type),
task = as.factor(task),
cope_name = as.factor(cope_name),
hemisphere = as.factor(hemisphere),
roi = as.factor(roi),
mask_type = as.factor(mask_type),
contrast = as.factor(contrast),
cluster = as.factor(cluster),
estimate = as.numeric(estimate)
)
# Merge
semmst_merged_data_supplement <- semmst_mr_data_supplement_filtered %>%
left_join(
semmst_behav_data_filtered %>%
dplyr::select(ID, sex:GRO_score_SES02) %>%
distinct(ID, .keep_all = TRUE),
by = "ID"
) %>%
left_join(semmst_behav_data_rec_summary, by = "ID") %>%
droplevels()
```
## CLEAN
```{r}
semmst_merged_data_supplement_filtered = semmst_merged_data_supplement %>%
mutate(roi_cluster = paste(hemisphere, roi, cluster, sep = "_")) %>%
mutate(roi_hemisphere = paste(hemisphere, roi, sep = "_")) %>%
mutate(Vocabulary_scale = scale(Vocabulary)) %>%
mutate(Digit_Symbol_scale = scale(Digit_Symbol)) %>%
mutate(NonWord_sum_scale = scale(NonWord_sum))
```
# Results
## Analysis 1 - Behavioural
### ANOVA on LDI
```{r}
semmst_behav_data_rec_LDI_ANCOVA = semmst_behav_data_rec_LDI_ANOVA %>%
left_join(semmst_behav_data_rec_filtered %>% dplyr::select(ID, age, sex, education, Vocabulary, Digit_Symbol, NonWord_sum) %>% distinct(), by="ID") %>%
mutate(Vocabulary_scale = scale(Vocabulary)) %>%
mutate(Digit_Symbol_scale = scale(Digit_Symbol)) %>%
mutate(NonWord_sum_scale = scale(NonWord_sum)) %>% drop_na()
# ASSUMPTION CHECKS
## OUTLIER
semmst_behav_data_rec_LDI_ANCOVA %>%
group_by(dprime_type) %>%
identify_outliers(dprime_value)
##NORMALITY
semmst_behav_data_rec_LDI_ANCOVA %>%
group_by(dprime_type) %>%
shapiro_test(dprime_value) %>% print(n=40)
ggqqplot(semmst_behav_data_rec_LDI_ANCOVA, "dprime_value", ggtheme = theme_bw()) +
facet_grid(. ~ dprime_type, labeller = "label_both")
# ANCOVA
ANCOVA_behav_data_rec_LDI = semmst_behav_data_rec_LDI_ANCOVA %>%
anova_test(dv = dprime_value, wid = ID, within = c(dprime_type), covariate = c(sex, age, Vocabulary_scale, Digit_Symbol_scale, NonWord_sum_scale))
get_anova_table(ANCOVA_behav_data_rec_LDI)
# POST-HOC
semmst_behav_data_rec_LDI_ANCOVA_PWC = semmst_behav_data_rec_LDI_ANCOVA %>%
pairwise_t_test(
dprime_value ~ dprime_type,
paired = TRUE,
detailed = TRUE
) %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
ungroup() %>%
mutate(label = paste0("p=", signif(p.adj, 3)))
semmst_behav_data_rec_LDI_ANCOVA_PWC %>%
dplyr::select(-c(alternative, method, .y.)) %>%
print(n = 50)
# REPORT
## dprime_type 174.597 <0.001 *** 0.582000
## NonWord_sum_scale 8.346 0.008 ** 0.229000
## Digit_Symbol_scale:dprime_type 3.593 0.035 * 0.028000
```
### GLMEM: semantic similarity ~ response
```{r}
semmst_behav_data_rec_filtered_GLMEM <- semmst_behav_data_rec_filtered %>%
mutate(across(all_of(c("Vocabulary", "Digit_Symbol", "NonWord_sum", "arousal", "concreteness", "imageability", "meaningfulness")),
~ as.numeric(scale(.x)), # drop scale() attributes
.names = "{.col}_scale")) %>%
drop_na(Vocabulary) %>%
mutate(response = relevel(response, ref = "new"))
semmst_behav_data_rec_filtered_GLMEM_lure = semmst_behav_data_rec_filtered_GLMEM %>%
filter(trial_type %in% c("CLOSE", "DISTANT"))
model_simple_cont = glmer(response ~ cosine * trial_type * former_type +
(1|ID) +
(1|itemno),
family = binomial("probit"),
data = semmst_behav_data_rec_filtered_GLMEM_lure,
control = glmerControl(optimizer ='bobyqa', optCtrl=list(maxfun = 20000)), nAGQ=0)
model_complex_cont = glmer(response ~ cosine * trial_type * former_type +
age + sex + education + Vocabulary_scale + Digit_Symbol_scale +
NonWord_sum_scale + arousal_scale + meaningfulness_scale + concreteness_scale +
(1|ID) +
(1|itemno),
family = binomial("probit"),
data = semmst_behav_data_rec_filtered_GLMEM_lure,
control = glmerControl(optimizer ='bobyqa', optCtrl=list(maxfun = 20000)), nAGQ=0)
anova(model_simple_cont, model_complex_cont)
summary(model_complex_cont)
confint(model_complex_cont, parm="beta_", method="Wald")
# REPORT Estimate Std. Error z value Pr(>|z|) 2.5 % 97.5 % (Intercept)
## cosine 5.131681 1.806091 2.841 0.00449 ** -7.444086424 -0.35150302
## NonWord_sum_scale -0.213690 0.069192 -3.088 0.00201 ** -0.349304360 -0.07807487
## meaningfulness_scale 0.145144 0.073066 1.986 0.04698 * 0.001936827 0.28835127
## trial_typeDISTANT 2.011234 1.931225 1.041 0.29768
## former_typeTARGET -1.376078 1.970692 -0.698 0.48501
```
## Analysis 2 - ENC RS
### ANOVA - REPEAT vs. LURE
```{r}
# FILTER DATA
semmst_merged_data_H1 = semmst_merged_data_filtered %>%
filter(task == "Encoding") %>%
filter(cope_name %in% c("LureFirst", "ExactFirst", "LureRepeat", "ExactRepeat")) %>%
extract(
col = cope_name,
into = c("stim_type", "order"),
regex = "([A-Za-z]+?)(First|Repeat)$"
) %>%
mutate(
stim_type = as.factor(stim_type),
order = as.factor(order)
) %>%
mutate(stim_type = factor(stim_type, levels = c("Exact", "Lure")))
# ASSUMPTION CHECKS
## OUTLIER
semmst_merged_data_H1 %>%
group_by(stim_type, order, roi_cluster) %>%
identify_outliers(estimate) %>%
dplyr::select(ID, is.outlier, is.extreme) %>%
filter(is.extreme==TRUE)
semmst_merged_data_H1_nooutlier = semmst_merged_data_H1 %>%
group_by(stim_type, order, roi_cluster) %>%
filter(! ID %in% c("sub-434971")) %>% #, " sub-012421", "sub-632012", "sub-800472")) %>% # c("sub-434971", "sub-982347")) %>%
drop_na(NonWord_sum_scale) %>%
ungroup()
##NORMALITY
semmst_merged_data_H1_nooutlier %>%
group_by(stim_type, order) %>%
shapiro_test(estimate)
# ANOVA
semmst_merged_data_H1_nooutlier_AOV = semmst_merged_data_H1_nooutlier %>%
anova_test(dv = estimate, wid = ID, within = c(stim_type, order, roi_cluster), covariate = c(NonWord_sum_scale, Digit_Symbol_scale))
get_anova_table(semmst_merged_data_H1_nooutlier_AOV)
# REPORT Effect DFn DFd F p p<.05 ges
## 3 stim_type 1 25 16.271 4.54e-04 * 4.50e-02
## 4 order 1 25 50.948 1.76e-07 * 1.41e-01
## 4 stim_type:order 1 25 12.810 1.00e-03 * 3.20e-02
## 6 order:roi_cluster 1 25 5.677 2.50e-02 * 6.00e-03
# COVARIATES
## 16 Digit_Symbol_scale:stim_type:order 1 25 6.123 2.00e-02 * 1.60e-02
## 19 NonWord_sum_scale:order:roi_cluster 1 25 9.266 5.00e-03 * 1.00e-02
# POST-HOC
# pairwise comparisons - stim_type * order
semmst_merged_data_H1_nooutlier_PWC = semmst_merged_data_H1_nooutlier %>%
drop_na(Digit_Symbol) %>%
group_by(stim_type) %>%
pairwise_t_test(
estimate ~ order,
paired = TRUE,
detailed = TRUE,
) %>%
ungroup() %>%
# group_by(stim_type) %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
ungroup() %>%
mutate(label = paste0("p=", signif(p.adj, 3)))
semmst_merged_data_H1_nooutlier_PWC
# REPORT estimate .y. group1 group2 n1 n2 statistic p df conf.low conf.high method alternative p.adj p.adj.signif label
## Exact 0.190 estimate First Repeat 56 56 8.86 3.52e-12 55 0.147 0.232 T-test two.sided 7.04e-12 **** p=7.04e-12
## Lure 0.0720 estimate First Repeat 56 56 3.44 1 e- 3 55 0.0301 0.114 T-test two.sided 2 e- 3 ** p=0.002
# pairwise comparisons - order * roi_cluster
semmst_merged_data_H1_nooutlier_PWC = semmst_merged_data_H1_nooutlier %>%
drop_na(Digit_Symbol) %>%
group_by(order) %>%
pairwise_t_test(
estimate ~ stim_type,
paired = TRUE,
detailed = TRUE,
) %>%
ungroup() %>%
# group_by(stim_type) %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
ungroup() %>%
mutate(label = paste0("p=", signif(p.adj, 3)))
semmst_merged_data_H1_nooutlier_PWC
# REPORT estimate .y. group1 group2 n1 n2 statistic p df conf.low conf.high method alternative p.adj p.adj.signif labe
## First -0.0114 estimate Exact Lure 56 56 -0.734 0.466 55 -0.0427 0.0198 T-test two.sided 0.932 ns p=0.932
## Repeat -0.129 estimate Exact Lure 56 56 -5.45 0.00000124 55 -0.176 -0.0815 T-test two.sided 0.00000248 **** p=2.48e-06
```
### ANOVA - REPEAT vs. LURE and CORRECT vs. INCORRECT
For subsequent correctness we use those models, where only those runs were included, where non-zero amount of trials were in each bin.
```{r}
semmst_merged_data_filtered_subseq = semmst_merged_data_filtered %>%
dplyr::filter(task %in% c("EncodingSubseq", "EncodingSubseq_limited")) %>%
dplyr::group_by(ID) %>%
dplyr::filter(
task == "EncodingSubseq" &
!any(task == "EncodingSubseq_limited")
) %>%
dplyr::ungroup()
semmst_merged_data_HE = semmst_merged_data_filtered_subseq %>%
filter(cope_name %in% c("CorrectExactRepeat", "CorrectExactFirst", "CorrectLureRepeat", "CorrectLureFirst", "IncorrectExactFirst", "IncorrectExactRepeat", "IncorrectLureFirst", "IncorrectLureRepeat"))
```
```{r}
# STANDARD
semmst_merged_data_HE_test = semmst_merged_data_HE %>%
extract(col = cope_name,
into = c("correctness", "stim_type", "order"),
regex = "^(Correct|Incorrect)(Exact|Lure)(First|Repeat)$") %>%
mutate(
stim_type = as.factor(stim_type),
order = as.factor(order)
) %>%
mutate(stim_type = factor(stim_type, levels = c("Exact", "Lure"))) %>%
mutate(correctness = factor(correctness, levels = c("Correct", "Incorrect"))) %>%
droplevels()
# ASSUMPTION CHECKS
## OUTLIER
semmst_merged_data_HE_test %>%
group_by(order, correctness, roi_cluster) %>%
identify_outliers(estimate) %>%
dplyr::select(ID, roi_cluster, stim_type, order, correctness, is.outlier, is.extreme) %>%
filter(is.extreme == TRUE)
semmst_merged_data_HE_test_nooutlier = semmst_merged_data_HE_test %>%
#filter(!ID %in% c("sub-434971")) %>% #, "sub-124846")) %>% #c("sub-434971", "sub-632012", "sub-680997")) %>% #, "sub-193800")) %>%
ungroup()
##NORMALITY
semmst_merged_data_HE_test_nooutlier %>%
group_by(stim_type, order, correctness, roi, hemisphere) %>%
shapiro_test(estimate) %>% print(n=100)
# ANOVA
semmst_merged_data_HE_test_AOV = semmst_merged_data_HE_test_nooutlier %>%
anova_test(dv = estimate, wid = ID, within = c(stim_type, correctness, order, roi_cluster))#, covariate = c(NonWord_sum_scale))
get_anova_table(semmst_merged_data_HE_test_AOV)
# REPORT
## 1 stim_type 1 28 13.459000 1.00e-03 * 1.80e-02
## 3 order 1 28 43.614000 3.66e-07 * 4.60e-02
## 6 stim_type:order 1 28 7.165000 1.20e-02 * 1.50e-02
## 7 correctness:order 1 28 0.048000 8.28e-01 1.18e-04
## 11 stim_type:correctness:order 1 28 0.277000 6.03e-01 4.92e-04
# POST-HOC
# pairwise comparisons
semmst_merged_data_HE_test_nooutlier_PWC = semmst_merged_data_HE_test_nooutlier %>%
group_by(stim_type) %>%
pairwise_t_test(
estimate ~ order,
paired = TRUE,
detailed = TRUE,
) %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
ungroup() %>%
mutate(label = paste0("p=", signif(p.adj, 3)))
semmst_merged_data_HE_test_nooutlier_PWC %>% dplyr::select(-c(alternative, method, .y., )) %>% print(n=50)
# REPORT stim_type estimate group1 group2 n1 n2 statistic p df conf.low conf.high p.adj p.adj.signif label
## 1 Exact 0.164 First Repeat 116 116 6.32 0.00000000513 115 0.112 0.215 0.0000000103 **** p=1.03e-08
## 2 Lure 0.0454 First Repeat 116 116 1.72 0.088 115 -0.00681 0.0975 0.176 ns p=0.176
```
## Revision Analysis 3 - Exact, close, distant
```{r}
# FILTER DATA
semmst_merged_data_H1_clodis = semmst_merged_data_filtered %>%
filter(task == "EncodingFirst") %>%
filter(cope_name %in% c("ExactRepeat", "CloseRepeat", "DistantRepeat")) %>%
mutate(
roi_cluster = paste(hemisphere, roi, cluster, sep = "_"),
roi_hemisphere = paste(hemisphere, roi, sep = "_"),
condition = recode(
cope_name,
"ExactRepeat" = "Exact",
"CloseRepeat" = "Close",
"DistantRepeat" = "Distant"
)
) %>%
mutate(
condition = factor(condition, levels = c("Exact", "Close", "Distant")),
roi_cluster = factor(roi_cluster)
)
# ASSUMPTION CHECKS
## OUTLIER
semmst_merged_data_H1_clodis %>%
group_by(condition, roi_cluster) %>%
identify_outliers(estimate) %>%
dplyr::select(ID, condition, roi_cluster, is.outlier, is.extreme) %>%
filter(is.extreme == TRUE)
# REMOVE OUTLIERS IF NEEDED
semmst_merged_data_H1_clodis_nooutlier = semmst_merged_data_H1_clodis %>%
drop_na(estimate, NonWord_sum_scale, Digit_Symbol_scale) %>%
ungroup()
## NORMALITY
semmst_merged_data_H1_clodis_nooutlier %>%
group_by(condition, roi_cluster) %>%
shapiro_test(estimate)
# ANOVA
semmst_merged_data_H1_clodis_AOV = semmst_merged_data_H1_clodis_nooutlier %>%
anova_test(
dv = estimate,
wid = ID,
within = c(condition, roi_cluster),
)
get_anova_table(semmst_merged_data_H1_clodis_AOV)
# POST-HOC
# pairwise comparisons - condition within each ROI
semmst_merged_data_H1_clodis_PWC = semmst_merged_data_H1_clodis_nooutlier %>%
group_by(roi_cluster) %>%
pairwise_t_test(
estimate ~ condition,
paired = TRUE,
detailed = TRUE,
p.adjust.method = "none"
) %>%
ungroup() %>%
add_significance("p") %>%
mutate(
label = case_when(
p < .001 ~ "p < .001",
p < .01 ~ "p < .01",
p < .05 ~ paste0("p = ", sub("^0", "", sprintf("%.3f", p))),
TRUE ~ paste0("p = ", sub("^0", "", sprintf("%.2f", p)))
)
)
semmst_merged_data_H1_clodis_PWC
```
## Analysis 4 - ENC RS ~ LDI
### Prepare bias scores
```{r}
group_vars_H3 <- c("ID", "mask_type", "space_type", "contrast", "roi_cluster")
# baseline
semmst_merged_data_H3_neural_ps = semmst_merged_data_filtered %>%
filter(task %in% c("Encoding")) %>%
filter(cope_name %in% c("ExactFirst", "ExactRepeat", "LureFirst","LureRepeat", "CloseFirst", "CloseRepeat", "DistantFirst", "DistantRepeat")) %>%
dplyr::select(ID, space_type, cope_name, hemisphere, roi, mask_type, cluster, voxels, contrast, task, roi_cluster, roi_hemisphere, estimate, NonWord_sum_scale, Digit_Symbol_scale, rec_dprime:lure_incorrect) %>%
group_by(across(all_of(group_vars_H3))) %>%
pivot_wider(names_from = cope_name, values_from = estimate, values_fill = NA_real_) %>%
mutate(
# --- Non-RS bias scores (Exact denominator) ---
LureBiasScore = LureRepeat - ExactRepeat,
CloseBiasScore = CloseRepeat - ExactRepeat,
DistantBiasScore = DistantRepeat - ExactRepeat
) %>%
ungroup()
```
### Correlation - LURES
```{r}
# 1) Filter / define roi_group
df_corr_input_all <- semmst_merged_data_H3_neural_ps %>%
mutate(roi_group = roi_cluster) %>%
dplyr::select(
ID, roi_group, roi, hemisphere, cluster, voxels,
LureBiasScore, lure_dprime,
NonWord_sum_scale, Digit_Symbol_scale
) %>%
ungroup()
# 2) Mahalanobis outlier detection (per ROI)
alpha <- 0.99 # cutoff for chi-square (df=2)
md_tbl_all <- df_corr_input_all %>%
dplyr::group_by(roi_group) %>%
dplyr::group_modify(~{
d <- .x %>% dplyr::select(ID, LureBiasScore, lure_dprime) %>% tidyr::drop_na()
if (nrow(d) < 3) {
return(tibble::tibble(ID = character(), MD = numeric(), p = numeric(),
cutoff = numeric(), is_outlier = logical()))
}
X <- as.matrix(d[, c("LureBiasScore", "lure_dprime")])
center <- colMeans(X)
covmat <- stats::cov(X)
md <- stats::mahalanobis(X, center = center, cov = covmat)
tibble::tibble(
ID = d$ID,
MD = md,
p = 1 - stats::pchisq(md, df = 2),
cutoff = stats::qchisq(alpha, df = 2),
is_outlier = md > stats::qchisq(alpha, df = 2)
)
}) %>%
ungroup()
md_outliers_all <- md_tbl_all %>% dplyr::filter(is_outlier)
md_multi_all <- md_outliers_all %>% dplyr::count(ID, sort = TRUE) %>% dplyr::filter(n > 1)
md_tbl_all
md_outliers_all
md_multi_all
# 3) Remove outlier ID
df_corr_input_all_nooutlier <- df_corr_input_all %>%
dplyr::filter(ID != "sub-434971") %>%
droplevels() %>%
ungroup()
# 4) Fisher-z CI for Pearson r (95% hard-coded)
pearson_ci <- function(r, n) {
z <- atanh(r)
se <- 1 / sqrt(n - 3)
zcrit <- stats::qnorm(1 - (1 - 0.95) / 2)
c(low = tanh(z - zcrit * se), high = tanh(z + zcrit * se))
}
# 5) SIMPLE correlations only (Pearson) per ROI + Bonferroni
H3_corr_lure_simple <- df_corr_input_all_nooutlier %>%
dplyr::group_by(roi_group) %>%
dplyr::group_modify(~{
d <- .x %>% dplyr::select(LureBiasScore, lure_dprime) %>% tidyr::drop_na()
ct <- stats::cor.test(d$LureBiasScore, d$lure_dprime, method = "pearson")
r <- unname(ct$estimate)
p <- ct$p.value
n <- nrow(d)
ci <- pearson_ci(r, n)
tibble::tibble(
n = n,
r = r,
p = p,
ci = sprintf("[%.3f, %.3f]", ci["low"], ci["high"])
)
}) %>%
ungroup() %>%
mutate(
p_adj = p.adjust(p, method = "bonferroni")
) %>%
arrange(p_adj)
H3_corr_lure_simple
```
### Compare correlations
```{r}
# --- A) Build subject-level wide table (bias per ROI, plus lure_dprime per ID)
wide_bias <- df_corr_input_all_nooutlier %>%
group_by(ID, roi_group) %>%
summarise(
bias = mean(LureBiasScore, na.rm = TRUE),
lure_dprime = first(lure_dprime),
.groups = "drop"
) %>%
tidyr::pivot_wider(
names_from = roi_group,
values_from = bias,
names_repair = "minimal"
)
roi_cols <- setdiff(names(wide_bias), c("ID", "lure_dprime"))
# --- B) Pairwise dependent-correlation tests:
# compares r(lure_dprime, ROI1_bias) vs r(lure_dprime, ROI2_bias)
H3_pairwise_roi_corrdiff <- combn(roi_cols, 2, simplify = FALSE) %>%
purrr::map_dfr(function(pair){
roi1 <- pair[1]
roi2 <- pair[2]
d <- wide_bias %>%
select(lure_dprime, all_of(roi1), all_of(roi2)) %>%
drop_na()
n <- nrow(d)
r1 <- cor(d$lure_dprime, d[[roi1]], method = "pearson")
r2 <- cor(d$lure_dprime, d[[roi2]], method = "pearson")
r12 <- cor(d[[roi1]], d[[roi2]], method = "pearson")
# Williams/Steiger-style test for two dependent correlations sharing one variable
tst <- psych::r.test(n = n, r12 = r1, r13 = r2, r23 = r12)
tibble(
roi1 = roi1,
roi2 = roi2,
n = n,
r_roi1 = r1,
r_roi2 = r2,
r_roi1_roi2 = r12,
t = unname(tst$t),
p = unname(tst$p)
)
}) %>%
mutate(p_adj_bonf = p.adjust(p, method = "bonferroni")) %>%
arrange(p_adj_bonf, p)
H3_pairwise_roi_corrdiff
```
## Analysis 5 - REC MD
### ANOVA #1 - LureMD
```{r}
### FILTERING FR AT LEAST 5 items per condition per run
run_status <- semmst_behav_data_rec_filtered %>%
select(ID, run, trial_type, correctness) %>%
filter(trial_type %in% c("CLOSE", "DISTANT")) %>%
mutate(run = as.integer(str_extract(as.character(run), "\\d+"))) %>%
count(ID, run, correctness, name = "n") %>%
group_by(ID, run) %>%
complete(correctness, fill = list(n = 0)) %>%
summarise(
min_n = min(n),
run_ok = (min_n >= 3),
.groups = "drop"
) %>%
select(ID, run, run_ok) %>%
tidyr::pivot_wider(
names_from = run,
values_from = run_ok,
names_prefix = "run",
values_fill = FALSE
) %>%
mutate(
group = case_when(
run1 & run2 ~ "both_ok",
!run1 & run2 ~ "run1_low_only",
run1 & !run2 ~ "run2_low_only",
TRUE ~ "both_low"
)
)
# We use both runs only if both have n > 5 items in all bins, otherwise only 1 run or exclude
ids_both_ok <- run_status %>% filter(group == "both_ok") %>% pull(ID)
ids_run1_lowonly <- run_status %>% filter(group == "run1_low_only") %>% pull(ID)
ids_run2_lowonly <- run_status %>% filter(group == "run2_low_only") %>% pull(ID)
semmst_merged_data_filtered_lures = semmst_merged_data_filtered
df_LureCR_all <- semmst_merged_data_filtered_lures %>%
filter(task == "RecognitionRespAll", cope_name == "LureCR")
df_LureFA_both_ok <- semmst_merged_data_filtered_lures %>%
filter(task == "RecognitionRespAll", cope_name == "LureFA") %>%
semi_join(tibble(ID = ids_both_ok), by = "ID")
df_LureFA_use_run2 <- semmst_merged_data_filtered_lures %>%
filter(task == "RecognitionRespAllRun2", cope_name == "LureFA") %>%
semi_join(tibble(ID = ids_run1_lowonly), by = "ID")
df_LureFA_use_run1 <- semmst_merged_data_filtered_lures %>%
filter(task == "RecognitionRespAllRun1", cope_name == "LureFA") %>%
semi_join(tibble(ID = ids_run2_lowonly), by = "ID")
# Merge for testing
df_LureFA_all <- bind_rows(
df_LureFA_both_ok,
df_LureFA_use_run2,
df_LureFA_use_run1
)
semmst_merged_data_H4_lureFACR <- bind_rows(df_LureCR_all, df_LureFA_all)
keep_ids <- semmst_merged_data_H4_lureFACR %>%
group_by(ID, roi_cluster) %>%
summarise(
has_LureCR = any(cope_name == "LureCR"),
has_LureFA = any(cope_name == "LureFA"),
.groups = "drop"
) %>%
filter(has_LureCR & has_LureFA) %>% count(ID) %>%
filter(n > 1) %>%
pull(ID)
semmst_merged_data_H4_lureFACR <- semmst_merged_data_H4_lureFACR %>%
filter(ID %in% keep_ids) %>%
droplevels()
# ASSUMPTION CHECKS
## OUTLIER
semmst_merged_data_H4_lureFACR %>%
group_by(cope_name, roi_cluster, task) %>%
identify_outliers(estimate) %>%
dplyr::select(ID, roi_cluster, cope_name, is.outlier, is.extreme) %>%
filter(is.extreme==TRUE)
semmst_merged_data_H4_lureFACR_nooutlier = semmst_merged_data_H4_lureFACR %>%
#filter(ID != "sub-632012") %>%
ungroup()
semmst_merged_H4_AOV = semmst_merged_data_H4_lureFACR_nooutlier %>%
droplevels() %>%
anova_test(dv = estimate, wid = ID, within = c(cope_name, roi_cluster))#, covariate = c(NonWord_sum_scale))
get_anova_table(semmst_merged_H4_AOV)
# REPORT Effect DFn DFd F p p<.05 ges
## 1 cope_name 1 23 1.621 0.216 0.013
## 2 roi_cluster 1 23 0.584 0.453 0.005
## 3 cope_name:roi_cluster 1 23 1.079 0.310 0.006
# POST-HOC
# pairwise comparisons - stim_type * order
semmst_merged_H4_REC_lureFACR_PWC = semmst_merged_data_H4_lureFACR_nooutlier %>%
group_by(roi_cluster) %>%
pairwise_t_test(
estimate ~ cope_name,
paired = TRUE,
detailed = TRUE,
) %>%
ungroup() %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
ungroup() %>%
mutate(label = paste0("p=", signif(p.adj, 3)))
semmst_merged_H4_REC_lureFACR_PWC
```
### ANOVA #2 - REC CONDITIONS
```{r}
# Control analysis for CR only on the total sample
semmst_merged_data_filtered_H4_REC_conditions = semmst_merged_data_filtered %>%
filter(task == "RecognitionRespAll") %>%
filter(cope_name %in% c("LureCR", "FoilCR", "TargetHIT"))
# ASSUMPTION CHECKS
## OUTLIER
semmst_merged_data_filtered_H4_REC_conditions %>%
group_by(cope_name, roi_cluster) %>%
identify_outliers(estimate) %>%
dplyr::select(ID, roi_cluster, cope_name, is.outlier, is.extreme) %>%
filter(is.extreme==TRUE)
semmst_merged_data_filtered_H4_REC_conditions_nooutlier = semmst_merged_data_filtered_H4_REC_conditions %>%
ungroup()
semmst_merged_H4_REC_conditions_AOV = semmst_merged_data_filtered_H4_REC_conditions_nooutlier %>%
droplevels() %>%
anova_test(dv = estimate, wid = ID, within = c(cope_name, roi_cluster))#, covariate = c(NonWord_sum_scale))
get_anova_table(semmst_merged_H4_REC_conditions_AOV)
# Effect DFn DFd F p p<.05 ges
## 1 cope_name 2 58 6.967 0.002 * 0.040
## 2 roi_cluster 1 29 2.018 0.166 0.015
## 3 cope_name:roi_cluster 2 58 0.508 0.604 0.002
# POST-HOC
# pairwise comparisons - stim_type * order
semmst_merged_H4_REC_conditions_PWC = semmst_merged_data_filtered_H4_REC_conditions_nooutlier %>%
group_by(task) %>%
pairwise_t_test(
estimate ~ cope_name,
paired = TRUE,
detailed = TRUE,
) %>%
ungroup() %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
ungroup() %>%
mutate(label = paste0("p=", signif(p.adj, 3)))
semmst_merged_H4_REC_conditions_PWC
## roi_cluster estimate .y. group1 group2 n1 n2 statistic p df conf.low conf.high method alternative p.adj
## 1 L_HEAD_DGCA23_HEAD_SUB_14 0.138 estimate FoilCR LureCR 30 30 3.43 0.002 29 0.0558 0.221 T-test two.sided 0.012 *
## 4 R_HEAD_DGCA23_HEAD_SUB_15 0.103 estimate FoilCR LureCR 30 30 3.05 0.005 29 0.0341 0.173 T-test two.sided 0.03 *
# AGAINTS ZERO BASLEINE
semmst_merged_H4_REC_FoilLure_vs0 <- semmst_merged_data_filtered_H4_REC_conditions_nooutlier %>%
filter(cope_name %in% c("FoilCR", "LureCR")) %>%
group_by(cope_name) %>%
t_test(
estimate ~ 1,
mu = 0,
detailed = TRUE
) %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance("p.adj") %>%
mutate(label = paste0("p=", signif(p.adj, 3))) %>%
ungroup()
semmst_merged_H4_REC_FoilLure_vs0
### cope_name estimate .y. group1 group2 n statistic p df conf.low conf.high method alternative p.adj p.adj.signif label
## 1 FoilCR 0.0298 estimate 1 null model 60 0.976 0.333 59 -0.0313 0.0909 T-test two.sided 0.666 ns p=0.666
## 2 LureCR -0.0911 estimate 1 null model 60 -2.77 0.00758 59 -0.157 -0.0252 T-test two.sided 0.0152 * p=0.0152
```
# Figures
## Figure 1 - Behavioral figures (response, LDI)
### Figure 1A - response ratio
```{r}
knitr::opts_chunk$set(
fig.width = 11, # same as ggsave width
fig.height = 16, # same as ggsave height
dpi = 300,
fig.retina = 2, # crisp in HTML
out.width = "45%", # display smaller, but proportions/text scale together
dev = "ragg_png"
)
dodge <- 0.7
trial_order <- c("TARGET", "CLOSE", "DISTANT", "FOIL")
# 1) Build per-ID percentages so the two responses sum to 100% within each ID × trial_type
figA_base <- semmst_behav_data_rec_filtered %>%
filter(!is.na(ID), !is.na(trial_type), !is.na(response)) %>%
mutate(
trial_type = factor(trial_type, levels = trial_order),
response = factor(response)
)
resp_lvls <- levels(figA_base$response)
figA_pct <- figA_base %>%
dplyr::count(ID, trial_type, response, name = "n") %>%
tidyr::complete(response = factor(resp_lvls, levels = resp_lvls), fill = list(n = 0)) %>%
dplyr::mutate(
percent_response = 100 * n / sum(n),
.by = c(ID, trial_type)
)
# 2) Within-subject SE across ID for each (trial_type × response)
figA_barsum <- Rmisc::summarySEwithin(
data = figA_pct,
measurevar = "percent_response",
withinvars = c("trial_type", "response"),
idvar = "ID",
na.rm = TRUE
)
# Paul Tol "Light" (pastel, print-friendly): light blue / light orange
fill_vals <- setNames(c("#EE8866", "#77AADD"), levels(figA_barsum$response))
behav_fig_A <- ggplot(figA_barsum, aes(
x = trial_type, y = percent_response,
fill = response, group = response
)) +
geom_col(position = position_dodge(width = dodge), width = 0.70) +
geom_errorbar(
aes(ymin = percent_response - se, ymax = percent_response + se),
position = position_dodge(width = dodge), width = 0.2, size = 1.4
) +
scale_fill_manual(values = fill_vals) +
scale_y_continuous(limits = c(0, 100), expand = expansion(mult = c(0.01, 0.06))) +
scale_x_discrete(
drop = FALSE,
labels = c(
"TARGET" = "TARGET",
"CLOSE" = "CLOSE\nLURE",
"DISTANT" = "DISTANT\nLURE",
"FOIL" = "FOIL"
)
) +
labs(
x = "Trial type",
y = "Response percentages",
fill = "Response"
) +
theme_classic(base_size = 16) +
theme(
legend.position = c(0.15, 0.95),
)
behav_fig_A
```
```{r}
text_base_export=40
behav_fig_A_export <- behav_fig_A +
theme_classic(base_size = text_base_export) +
theme(
legend.position = c(0.18, 0.95),
legend.title = element_text(size = text_base_export * 0.95),
legend.text = element_text(size = text_base_export * 0.90),
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05)
)
ggsave(
"derivatives/figures/Figure1A.png",
plot = behav_fig_A_export,
width = 18, height = 13, units = "in",
dpi = 300, bg = "white", device = "png"
)
```
### Figure 1 D - D-primes
```{r}
# 1) Prep & adjustment (remove linear effects of covariates)
figD_plot_base <- semmst_behav_data_rec_LDI_ANCOVA %>%
filter(!is.na(ID), !is.na(dprime_type), !is.na(dprime_value)) %>%
mutate(
dprime_type = factor(dprime_type),
sex = factor(sex)
)
# order as: Close, Distant, Target
figD_contrast_order <- c("close_dprime", "distant_dprime", "rec_dprime")
figD_plot_base <- figD_plot_base %>%
mutate(dprime_type = fct_relevel(dprime_type, figD_contrast_order))
# covariate-only model for adjustment
figD_mod_cov <- lm(dprime_value ~ sex + age + Vocabulary_scale + Digit_Symbol_scale + NonWord_sum_scale,
data = figD_plot_base)
figD_plot_base <- figD_plot_base %>%
mutate(dprime_adj = dprime_value - predict(figD_mod_cov, newdata = figD_plot_base) + mean(dprime_value, na.rm = TRUE))
# 2) Labels & colors
figD_x_lab_map <- c(
"close_dprime" = "CLOSE\nLURE",
"distant_dprime" = "DISTANT\nLURE",
"rec_dprime" = "FOIL"
)
# Paul Tol (muted): blue / yellow / red — CVD-safe, good in print & grayscale
figD_fill_vals <- c("close_dprime"="#4477AA", "distant_dprime"="#DDCC77", "rec_dprime"="#117733")
# 3) Plot
behav_fig_D <- ggplot(figD_plot_base, aes(x = dprime_type, y = dprime_adj, fill = dprime_type)) +
geom_boxplot(width = 0.6, outlier.alpha = 0.4, linewidth = 1.8, color = "black") +
stat_summary(fun = mean, geom = "point", shape = 21, size = 2.2, fill = "white", color = "black") +
scale_x_discrete(labels = figD_x_lab_map, drop = FALSE) +
scale_fill_manual(values = figD_fill_vals, guide = "none") +
labs(
x = "Discriminability contrasts",
y = "Covariate-adjusted d'\n [p(\"old\"|target) - p(\"old\"|condition)]"
) +
theme_classic(base_size = 14)
behav_fig_D
```
```{r}
behav_fig_D_export <- behav_fig_D +
theme_classic(base_size = 40) +
theme(
axis.title = element_text(size = 42),
axis.text = element_text(size = 34)
) + labs(x = NULL)
ggsave(
"derivatives/figures/Figure1D.png",
plot = behav_fig_D_export,
width = 9, height = 16, units = "in",
dpi = 300,
bg = "white",
device = "png" # base png
)
```
### Figure 1 E - continous GLMEM
```{r}
# In this plot, we predict p|old by cosine similarity
# Base data: binary outcome + keep CLOSE & DISTANT only
semmst_behav_data_rec_filtered_GLMEM_plot <- semmst_behav_data_rec_filtered_GLMEM %>%
mutate(response = as.integer(response == "old")) %>%
filter(trial_type %in% c("CLOSE", "DISTANT"))
# If your column is 'trial_type', replace 'trial_type' with 'trial_type' above.
# Build a facetting variable: Overall, DISTANT-only, CLOSE-only
semmst_behav_data_rec_filtered_GLMEM_plot_facet <- bind_rows(
semmst_behav_data_rec_filtered_GLMEM_plot %>% mutate(panel = "ALL LURE"),
semmst_behav_data_rec_filtered_GLMEM_plot %>% filter(trial_type == "DISTANT") %>% mutate(panel = "DISTANT LURE"),
semmst_behav_data_rec_filtered_GLMEM_plot %>% filter(trial_type == "CLOSE") %>% mutate(panel = "CLOSE LURE")
) %>%
mutate(panel = factor(panel, levels = c("ALL LURE", "DISTANT LURE", "CLOSE LURE")))
# Okabe–Ito (CVD-safe): purple, yellow, blue
figE_cols <- c("ALL LURE" = "#CC79A7", "DISTANT LURE" = "#DDCC77", "CLOSE LURE" = "#4477AA")
behav_fig_E <- ggplot(semmst_behav_data_rec_filtered_GLMEM_plot_facet,
aes(x = cosine, y = response, color = panel, fill = panel)) +
geom_smooth(
method = "glm", method.args = list(family = "binomial"),
linewidth = 2, se = TRUE, alpha = 0.25
) +
facet_wrap(~ panel, nrow = 1, scales = "free_x") +
scale_x_continuous(
breaks = function(lims) round(lims[1] + c(0.1, 0.5, 0.9) * (lims[2] - lims[1]), 2),
labels = function(x) sprintf("%.2f", x)
) +
scale_y_continuous(labels = scales::percent, limits = c(0, 1)) +
scale_color_manual(values = figE_cols, guide = "none") +
scale_fill_manual(values = figE_cols, guide = "none") +
labs(x = "Cosine similarity", y = "Percentage of 'old' responses") +
theme_classic(base_size = 14) +
theme(
panel.grid.minor = element_blank(),
strip.text = element_text(face = "bold")
)
behav_fig_E
```
```{r}
behav_fig_E_export <- behav_fig_E +
theme_classic(base_size = 40) +
theme(
axis.title = element_text(size = 42),
axis.text = element_text(size = 32)
)
ggsave(
"derivatives/figures/Figure1E.png",
plot = behav_fig_E_export,
width = 16, height = 16, units = "in",
dpi = 300,
bg = "white",
device = "png" # base png
)
```
### Figure 1 C - Lure cosine similarities
```{r}
figC_plot_base <- semmst_behav_data_rec_filtered %>%
filter(trial_type %in% c("CLOSE", "DISTANT")) %>%
mutate(trial_type = factor(trial_type,
levels = c("CLOSE", "DISTANT"),
labels = c("CLOSE LURE", "DISTANT LURE"))) %>%
dplyr::select(trial_type, noun, cosine) %>%
unique()
# Paul Tol (muted): blue / yellow / red — CVD-safe, good in print & grayscale
figC_fill_vals <- c("CLOSE LURE"="#4477AA", "DISTANT LURE"="#DDCC77")
# 3) Plot
behav_fig_C_hists <- ggplot(figC_plot_base, aes(x = cosine, fill = trial_type)) +
geom_histogram(
bins = 20, # tweak bins/binwidth as you like
color = "black",
linewidth = 1.0
) +
facet_wrap(~ trial_type, ncol = 1) + # <-- vertically stacked (2 panels if 2 trial types)
scale_fill_manual(values = figC_fill_vals, guide = "none") +
labs(
x = "Cosine similarity [1.0 = identical]",
y = "Number of items in 20 equally spaced bins"
) +
theme_classic(base_size = 14) +
theme(
strip.background = element_blank(),
strip.text = element_text(face = "bold"),
panel.spacing = unit(0.7, "lines")
)
behav_fig_C_hists
```
```{r}
behav_fig_C_export <- behav_fig_C_hists +
theme_classic(base_size = 40) +
theme(
axis.title = element_text(size = 38),
axis.text = element_text(size = 34)
)
ggsave(
"derivatives/figures/Figure1C.png",
plot = behav_fig_C_export,
width = 12, height = 19, units = "in",
dpi = 300,
bg = "white",
device = "png" # base png
)
```
## Figure 2 - ENC RS specificity
```{r}
# BARPLOT
fig2_dodge <- 0.8
fig2_barsum <- Rmisc::summarySEwithin(
data = semmst_merged_data_H1_nooutlier,
measurevar = "estimate",
withinvars = c("stim_type","order", "roi_cluster"),
idvar = "ID",
na.rm = TRUE
) %>%
mutate(stim_type = factor(as.character(stim_type),
levels = c("Exact","Lure"),
labels = c("Exact","Modified")))
fig2_roi_lab <- semmst_merged_data_H1_nooutlier %>%
dplyr::distinct(roi_cluster, hemisphere, roi, cluster, voxels) %>%
dplyr::mutate(
roi_label = dplyr::case_when(
roi == "HEAD_DGCA23_HEAD_SUB" & hemisphere == "R" ~ "Right hippocampal head cluster",
roi == "HEAD_DGCA23_HEAD_SUB" & hemisphere == "L" ~ "Left hippocampal head cluster"
)
) %>%
dplyr::select(roi_cluster, roi_label)
# ------------------------------------------------------------
# NEW: paired tests of stim_type (Exact vs Modified) WITHIN each order,
# collapsed across roi_cluster (same p-values in both facets)
# ------------------------------------------------------------
fig2_stat.test <- semmst_merged_data_H1_nooutlier %>%
mutate(stim_type = factor(as.character(stim_type),
levels = c("Exact","Lure"),
labels = c("Exact","Modified"))) %>%
group_by(ID, order, stim_type) %>%
summarise(estimate = mean(estimate, na.rm = TRUE), .groups = "drop") %>%
group_by(order) %>%
rstatix::t_test(estimate ~ stim_type, paired = TRUE) %>%
ungroup() %>%
rstatix::adjust_pvalue(method = "bonferroni") %>%
rstatix::add_significance("p.adj") %>%
mutate(
label = dplyr::case_when(
p.adj < 0.001 ~ "p < .001",
p.adj < .01 ~ "p < .01",
TRUE ~ paste0("p = ", sub("^0", "", sprintf("%.2f", p.adj)))
)
)
# 1) Attach ROI labels to barsum
fig2_barsum <- fig2_barsum %>% dplyr::left_join(fig2_roi_lab, by = "roi_cluster")
# 2) replicate the same stats into both facets and set y.position per facet+order
fig2_stat.test <- fig2_stat.test %>%
tidyr::crossing(fig2_barsum %>% dplyr::distinct(roi_label)) %>%
dplyr::left_join(
fig2_barsum %>%
dplyr::group_by(order, roi_label) %>%
dplyr::summarise(y.position = max(estimate + se, na.rm = TRUE) * 1.08, .groups = "drop"),
by = c("order", "roi_label")
)
# 3) compute xmin/xmax to connect Exact ↔ Modified within each ORDER (dodge offset)
fig2_stim_lvls <- levels(factor(fig2_barsum$stim_type)) # should be c("Exact","Modified")
exact_idx <- which(fig2_stim_lvls == "Exact")
mod_idx <- which(fig2_stim_lvls == "Modified")
fig2_stat.test <- fig2_stat.test %>%
mutate(
order = factor(order, levels = c("First", "Repeat")),
offset = if_else(order == "First", -fig2_dodge/2, +fig2_dodge/2),
xmin = exact_idx + offset,
xmax = mod_idx + offset
)
# --- bar plot (fill by stim_type color, opacity by order), same dodging ---
fig2_plot <- ggplot(fig2_barsum, aes(
x = stim_type, y = estimate,
fill = stim_type, # color by stim type
group = `order` # ensures side-by-side bars per stim_type
)) +
geom_col(aes(alpha = order), position = position_dodge(width = fig2_dodge), width = 0.65) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
position = position_dodge(width = fig2_dodge), width = 0.15, size=1.2
) +
facet_wrap(~ roi_label) +
theme_classic(base_size = 13) +
labs(x = "Condition during encoding", y = "Mean % signal change", fill = "Stimulus", alpha = "Order") +
# accessible, pleasant green & violet
scale_fill_manual(values = c(
"Exact" = "#66BB6A", # soft green
"Modified" = "#9575CD" # soft violet
)) +
# subtle opacity difference for First vs Repeat
scale_alpha_manual(values = c(
"First" = 0.95,
"Repeat" = 0.55
)) +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.14))) +
custom_theme + guides(fill = "none") +
theme(
legend.position = c(0.88, 0.93) # a bit up
)
# --- add bracket + p (Exact vs Modified within each order; same in both facets) ---
fig2_plot +
ggpubr::stat_pvalue_manual(
fig2_stat.test,
label = "label",
xmin = "xmin",
xmax = "xmax",
y.position = "y.position",
tip.length = 0.01,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.shorten=1.2,
size = 5.4,
bracket.size = 1.0
)
```
```{r}
text_base_export=40
p_text_size_mm <- text_base_export * 0.20 # ~points → mm-ish (tweak 0.30–0.45)
p_bracket_mm <- p_text_size_mm * 0.15 # bracket thickness relative to text
fig2_plot_export <- fig2_plot +
theme_classic(base_size = text_base_export) +
theme(
legend.position = c(0.88, 0.88),
legend.title = element_text(size = text_base_export * 0.88),
legend.text = element_text(size = text_base_export * 0.90),
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05),
strip.text = element_blank(),
panel.spacing.x = unit(4, "in")
)
ggsave(
"derivatives/figures/Figure2B.png",
plot = fig2_plot_export,
width = 26, height = 13, units = "in",
dpi = 300, bg = "white", device = "png"
)
```
## Revised Figure 3 - exact, close, distant
```{r}
# BARPLOT SUMMARY
fig3_barsum <- Rmisc::summarySEwithin(
data = semmst_merged_data_H1_clodis_nooutlier,
measurevar = "estimate",
withinvars = c("condition", "roi_cluster"),
idvar = "ID",
na.rm = TRUE
) %>%
mutate(
condition = factor(as.character(condition), levels = c("Exact", "Close", "Distant"))
)
# ROI LABELS
fig3_roi_lab <- semmst_merged_data_H1_clodis_nooutlier %>%
dplyr::distinct(roi_cluster, hemisphere, roi, cluster, voxels) %>%
dplyr::mutate(
roi_label = dplyr::case_when(
roi == "HEAD_DGCA23_HEAD_SUB" & hemisphere == "R" ~ "Right hippocampal head cluster",
roi == "HEAD_DGCA23_HEAD_SUB" & hemisphere == "L" ~ "Left hippocampal head cluster",
TRUE ~ as.character(roi_cluster)
)
) %>%
dplyr::select(roi_cluster, roi_label)
# ATTACH ROI LABELS
fig3_barsum <- fig3_barsum %>%
dplyr::left_join(fig3_roi_lab, by = "roi_cluster")
# SIGNIFICANT PAIRWISE BRACKETS
fig3_stat.test <- semmst_merged_data_H1_clodis_PWC %>%
dplyr::filter(p < .05) %>%
dplyr::left_join(fig3_roi_lab, by = "roi_cluster")
# BRACKET HEIGHTS
fig3_ypos <- fig3_barsum %>%
dplyr::group_by(roi_label) %>%
dplyr::summarise(
y_max = max(estimate + se, na.rm = TRUE),
y_min = min(estimate - se, na.rm = TRUE),
y_span = y_max - y_min,
.groups = "drop"
) %>%
dplyr::mutate(
y_span = if_else(y_span == 0 | is.na(y_span), 0.1, y_span),
y_base = y_max + 0.10 * y_span,
y_step = 0.12 * y_span
)
# STACK BRACKETS WITHIN EACH ROI
fig3_stat.test <- fig3_stat.test %>%
dplyr::left_join(fig3_ypos, by = "roi_label") %>%
dplyr::group_by(roi_label) %>%
dplyr::arrange(group1, group2, .by_group = TRUE) %>%
dplyr::mutate(y.position = y_base + (dplyr::row_number() - 1) * y_step) %>%
dplyr::ungroup()
# BASE FIGURE
fig3_plot <- ggplot(fig3_barsum, aes(x = condition, y = estimate, fill = condition)) +
geom_col(width = 0.68) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
width = 0.12,
size = 1.2
) +
ggpubr::stat_pvalue_manual(
fig3_stat.test,
label = "label",
xmin = "group1",
xmax = "group2",
y.position = "y.position",
tip.length = 0.015,
hide.ns = FALSE,
inherit.aes = FALSE,
bracket.shorten = 0.03,
size = 6.6,
bracket.size = 1.35
) +
facet_wrap(~ roi_label, ncol = 2) +
labs(
x = "Conditions of repeats during encoding",
y = "Mean % signal change"
) +
scale_fill_manual(values = c(
"Exact" = "#66BB6A",
"Close" = "#4477AA",
"Distant" = "#DDCC77"
)) +
scale_y_continuous(expand = expansion(mult = c(0.03, 0.22))) +
theme_classic(base_size = 18) +
custom_theme +
guides(fill = "none")
fig3_plot
```
```{r}
# EXPORT
text_base_export = 40
fig3_plot_export <- ggplot(fig3_barsum, aes(x = condition, y = estimate, fill = condition)) +
geom_col(width = 0.68) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
width = 0.12,
linewidth = 1.8
) +
ggpubr::stat_pvalue_manual(
fig3_stat.test,
label = "label",
xmin = "group1",
xmax = "group2",
y.position = "y.position",
tip.length = 0.015,
hide.ns = FALSE,
inherit.aes = FALSE,
bracket.shorten = 0.03,
size = 10.5,
bracket.size = 2.1
) +
facet_wrap(~ roi_label, ncol = 2) +
labs(
x = "Conditions of repeats during encoding",
y = "Mean % signal change"
) +
scale_fill_manual(values = c(
"Exact" = "#66BB6A",
"Close" = "#4477AA",
"Distant" = "#DDCC77"
)) +
scale_y_continuous(expand = expansion(mult = c(0.03, 0.22))) +
theme_classic(base_size = text_base_export) +
custom_theme +
guides(fill = "none") +
theme(
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.95),
axis.title.x = element_text(size = text_base_export * 1.05, face = "plain"),
axis.title.y = element_text(size = text_base_export * 1.25, face = "plain"),
strip.text = element_text(size = text_base_export * 0.95, face = "plain"),
strip.background = element_rect(fill = "white", color = "black", linewidth = 2.2)
)
ggsave(
"derivatives/figures/Figure3.png",
plot = fig3_plot_export,
width = 22, height = 12.5, units = "in",
dpi = 300, bg = "white", device = "png"
)
```
## Figure 4 - RS ~ LDI
```{r}
# 1) Long table (ALL trials only)
figure4_df <- df_corr_input_all_nooutlier %>%
dplyr::select(ID, roi_group, roi, hemisphere, bias = LureBiasScore, lure_dprime) %>%
tidyr::drop_na(bias, lure_dprime) %>%
dplyr::mutate(
facet_lab = dplyr::case_when(
roi_group == "L_HEAD_DGCA23_HEAD_SUB_14" ~ "Left hippocampal head cluster",
roi_group == "R_HEAD_DGCA23_HEAD_SUB_15" ~ "Right hippocampal head cluster",
TRUE ~ as.character(roi_group)
)
)
# 2) Per-ROI correlation + Bonferroni + label text
figure4_labs <- figure4_df %>%
dplyr::group_by(roi_group, facet_lab) %>%
dplyr::summarise(
n = dplyr::n(),
r = unname(stats::cor(bias, lure_dprime, method = "pearson")),
p = stats::cor.test(bias, lure_dprime, method = "pearson")$p.value,
.groups = "drop"
) %>%
dplyr::mutate(
p_adj = p.adjust(p, method = "bonferroni"),
stars = dplyr::case_when(
p_adj < .001 ~ "***",
p_adj < .01 ~ "**",
p_adj < .05 ~ "*",
TRUE ~ ""
),
sig = if_else(p_adj < .05, "sig", "ns"),
ann = dplyr::case_when(
is.na(p_adj) ~ sprintf("r = %.2f, p = NA", r),
p_adj < .001 ~ sprintf("r = %.2f, p < .001%s", r, stars),
TRUE ~ sprintf("r = %.2f, p = %.3f%s", r, p_adj, stars)
)
)
figure4_full <- figure4_df %>%
left_join(figure4_labs %>% select(roi_group, sig), by = "roi_group")
# 3) Plot
fig4 = ggplot(figure4_full, aes(bias, lure_dprime)) +
geom_point(alpha = 0.7, size = 4.5) +
scale_color_manual(values = c(ns = "grey30", sig = "red"), guide = "none") +
facet_wrap(~ roi_group, scales = "free",
labeller = labeller(roi_group = setNames(figure4_labs$facet_lab, figure4_labs$roi_group))) +
labs(
x = "Neural pattern separation\nModified repeat - exact repeat\n(% signal change)",
y = "Behavioral mnemonic discrimination\nd' (lures vs targets)"
) +
theme_classic(base_size = 18) +
custom_theme
fig4 +
geom_smooth(aes(color = sig), method = "lm", se = FALSE, linewidth = 1.1) +
geom_text(
data = figure4_labs,
aes(x = -Inf, y = Inf, label = ann, color = sig),
inherit.aes = FALSE,
hjust = -0.05, vjust = 1.2, size = 5.5
)
```
```{r}
text_base_export=40
fig4_export <- fig4 +
geom_smooth(aes(color = sig), method = "lm", se = FALSE, linewidth = 2.6) +
geom_text(
data = figure4_labs,
aes(x = -Inf, y = Inf, label = ann, color = sig),
inherit.aes = FALSE,
hjust = -0.05, vjust = 1.2, size = 10.5
) +
theme_classic(base_size = text_base_export) +
theme(
legend.title = element_text(size = text_base_export * 0.95),
legend.text = element_text(size = text_base_export * 0.90),
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05)
)
ggsave(
"derivatives/figures/Figure4.png",
plot = fig4_export,
width = 20, height = 12, units = "in",
dpi = 300, bg = "white", device = "png"
)
```
## Figure 5 - ENC RS mask -> REC conditions
```{r}
fig5_dodge <- 0.8
level_map <- c("FoilCR", "LureCR", "TargetHIT")
level_lab <- c("Foil", "Lure", "Target")
# 1) Within-subject summary
fig5_barsum <- Rmisc::summarySEwithin(
data = semmst_merged_data_filtered_H4_REC_conditions_nooutlier,
measurevar = "estimate",
withinvars = c("cope_name", "roi_cluster"),
idvar = "ID",
na.rm = TRUE
)
fig5_roi_lab <- semmst_merged_data_filtered_H4_REC_conditions_nooutlier %>%
distinct(roi_cluster, hemisphere, roi, cluster, voxels) %>%
mutate(
roi_label = case_when(
roi == "HEAD_DGCA23_HEAD_SUB" & hemisphere == "R" ~ "Right hippocampal head cluster",
roi == "HEAD_DGCA23_HEAD_SUB" & hemisphere == "L" ~ "Left hippocampal head cluster",
TRUE ~ NA_character_
)
) %>%
select(roi_cluster, roi_label)
fig5_barsum <- fig5_barsum %>%
mutate(cope_name = factor(cope_name, levels = level_map, labels = level_lab)) %>%
left_join(fig5_roi_lab, by = "roi_cluster")
# 2) Base plot (no stats)
fig5_plot <- ggplot(fig5_barsum, aes(x = cope_name, y = estimate, fill = cope_name)) +
geom_col(position = position_dodge(width = fig5_dodge), width = 0.65) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
position = position_dodge(width = fig5_dodge),
width = 0.15,
size = 1.2
) +
facet_wrap(~ roi_label) +
theme_classic(base_size = 13) +
labs(x = "Condition during recognition", y = "Mean % signal change") +
scale_fill_manual(values = c(
"Target" = "#66BB6A",
"Lure" = "#9575CD",
"Foil" = "#EE8866"
)) +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.12))) +
custom_theme +
guides(fill = "none")
# 3) Global paired pairwise stats (collapsed across ROI)
fig5_stats_global <- semmst_merged_data_filtered_H4_REC_conditions_nooutlier %>%
mutate(cope_name = factor(cope_name, levels = level_map, labels = level_lab)) %>%
group_by(ID, cope_name) %>%
summarise(estimate = mean(estimate, na.rm = TRUE), .groups = "drop") %>%
pairwise_t_test(
estimate ~ cope_name,
paired = TRUE,
p.adjust.method = "bonferroni"
) %>%
mutate(
label = case_when(
p.adj < 0.001 ~ "p < .001",
p.adj < 0.01 ~ "p < .01",
p.adj < 0.05 ~ "p < .05",
TRUE ~ paste0("p = ", sub("^0", "", sprintf("%.3f", p.adj)))
)
)
# Global ymax across BOTH facets (shared bracket height)
ymax_global <- fig5_barsum %>%
summarise(ymax = max(estimate + se, na.rm = TRUE)) %>%
pull(ymax)
# Replicate the same stats into each facet; same y.position levels everywhere
fig5_stats_for_facets <- fig5_barsum %>%
distinct(roi_label) %>%
tidyr::crossing(fig5_stats_global) %>%
group_by(roi_label) %>%
arrange(group1, group2, .by_group = TRUE) %>%
mutate(
y.position = ymax_global * 1.08 + (row_number() - 1) * (0.10 * ymax_global)
) %>%
ungroup()
# 4) In-window plot WITH stats
fig5_plot_with_stats_global <- fig5_plot +
stat_pvalue_manual(
fig5_stats_for_facets,
label = "label",
xmin = "group1",
xmax = "group2",
y.position = "y.position",
tip.length = 0.01,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.size = 1.0,
size = 5.0
) +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.18)))
fig5_plot_with_stats_global
```
```{r}
text_base_export <- 40
p_text_export <- text_base_export * 0.28 # bigger p text on export only
fig5_plot_export <- fig5_plot +
stat_pvalue_manual(
fig5_stats_for_facets,
label = "label",
xmin = "group1",
xmax = "group2",
y.position = "y.position",
tip.length = 0.01,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.size = 1.3,
size = p_text_export
) +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.18))) +
theme_classic(base_size = text_base_export) +
theme(
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05),
strip.text = element_text(size = text_base_export * 0.95)
)
ggsave(
"derivatives/figures/Figure5.png",
plot = fig5_plot_export,
width = 22, height = 14, units = "in",
dpi = 300, bg = "white", device = "png"
)
```
# Supplementary Results
### Supplementary Results 1 - covariate effects
#### Behavioural GLMEM covariates
```{r}
model_complex_cont = glmer(response ~ cosine * trial_type * former_type +
age + sex + education + Vocabulary_scale + Digit_Symbol_scale +
NonWord_sum_scale + arousal_scale + meaningfulness_scale + concreteness_scale +
(1|ID) +
(1|itemno),
family = binomial("probit"),
data = semmst_behav_data_rec_filtered_GLMEM_lure,
control = glmerControl(optimizer ='bobyqa', optCtrl=list(maxfun = 20000)), nAGQ=0)
```
##### Supplementary Figure 2AB.
```{r}
# 1) Marginal (averaged-over-covariates) predictions
pred_nonword <- ggeffects::ggaverage(
model_complex_cont,
terms = "NonWord_sum_scale [n=100]" # smooth grid
) %>%
as.data.frame() %>%
mutate(effect = "Phonological awareness of participants")
pred_meaning <- ggeffects::ggaverage(
model_complex_cont,
terms = "meaningfulness_scale [n=100]"
) %>%
as.data.frame() %>%
mutate(effect = "Meaningfulness of items")
preds_2ABSI <- bind_rows(pred_nonword, pred_meaning)
# 2) Plot (2 facets next to each other)
fig_ABSI_behav <- ggplot(preds_2ABSI, aes(x = x, y = predicted)) +
geom_ribbon(aes(ymin = conf.low, ymax = conf.high), alpha = 0.18, fill="#4C72B0") +
geom_line(size = 1.2, colour="#4C72B0") +
facet_wrap(~ effect, nrow = 1, scales = "free_x") +
theme_classic(base_size = 13) +
labs(
x = NULL,
y = "p(response 'old'|lure)"
) +
scale_y_continuous(limits = c(0, 0.5), expand = expansion(mult = c(0.02, 0.06))) +
custom_theme
fig_ABSI_behav
```
```{r}
# 3) Export (matches your Figure2B export approach)
text_base_export <- 40
fig_ABSI_export <- fig_ABSI_behav +
theme_classic(base_size = text_base_export) +
theme(
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05)
)
ggsave(
"derivatives/figures/SI_figures/Figure2ABSI.png",
plot = fig_ABSI_export,
width = 21, height = 11, units = "in",
dpi = 300, bg = "white", device = "png"
)
```
#### Encoding ANCOVA MEDIAN SPLIT (Digit-Symbol)
```{r}
df <- semmst_merged_data_H1_nooutlier %>%
drop_na(Digit_Symbol) %>%
mutate(Digit_Symbol_scale = scale(Digit_Symbol))
df_E1 <- df %>%
mutate(
digit_group = if_else(Digit_Symbol_scale >= median(Digit_Symbol_scale, na.rm = TRUE), "High", "Low"),
digit_group = factor(digit_group, levels = c("Low", "High"))
)
semmst_merged_data_E1_digit_median_split = df_E1 %>%
group_by(stim_type, digit_group) %>%
pairwise_t_test(
estimate ~ order,
paired = TRUE,
detailed = TRUE,
) %>%
ungroup() %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
ungroup() %>%
mutate(label = paste0("p=", signif(p.adj, 3)))
semmst_merged_data_E1_digit_median_split
# stim_type digit_group estimate .y. group1 group2 n1 n2 statistic p df conf.low conf.high method alternative p.adj
## 1 Exact Low 0.221 estimate First Repeat 28 28 8.04 1.22e-8 27 0.164 0.277 T-test two.sided 4.88e-8 ****
## 2 Exact High 0.158 estimate First Repeat 28 28 4.92 3.82e-5 27 0.0923 0.225 T-test two.sided 1.53e-4 ***
## 3 Lure Low 0.0490 estimate First Repeat 28 28 1.54 1.36e-1 27 -0.0165 0.115 T-test two.sided 5.44e-1 ns
## 4 Lure High 0.0949 estimate First Repeat 28 28 3.53 2 e-3 27 0.0397 0.150 T-test two.sided 8 e-3 **
```
##### Supplementary Figure 2C.
```{r}
fig_mediansplit_CSI_dodge <- 0.8
# OPTIONAL but recommended if you have multiple rows per ID/condition (e.g., ROI clusters):
# collapse to one value per ID × stim_type × order × digit_group
df_E1_collapsed <- df_E1 %>%
group_by(ID, stim_type, order, digit_group) %>%
mutate(stim_type = factor(as.character(stim_type),
levels = c("Exact", "Lure"),
labels = c("Exact", "Modified"))) %>%
mutate(digit_group = factor(as.character(digit_group),
levels = c("Low", "High"),
labels = c("Low processing speed", "High processing speed")))
# --- summary for bars (within-subject correction for stim_type+order, between = digit_group) ---
fig_mediansplit_CSI_barsum <- Rmisc::summarySEwithin(
data = df_E1_collapsed,
measurevar = "estimate",
withinvars = c("stim_type", "order"),
betweenvars = "digit_group",
idvar = "ID",
na.rm = TRUE
)
# --- paired tests: First vs Repeat within each stim_type × digit_group ---
fig_mediansplit_CSI_stat.test <- df_E1_collapsed %>%
group_by(stim_type, digit_group) %>%
pairwise_t_test(
estimate ~ order,
paired = TRUE,
detailed = TRUE
) %>%
ungroup() %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance("p.adj") %>%
mutate(
label = case_when(
p.adj < 0.001 ~ "p < .001",
p.adj < 0.01 ~ "p < .01",
p.adj < 0.05 ~ "p < .05",
TRUE ~ paste0("p = ", sub("^0", "", sprintf("%.2f", p.adj)))
)
)
# y positions per facet + stim_type
fig_mediansplit_CSI_stat.test <- fig_mediansplit_CSI_stat.test %>%
left_join(
fig_mediansplit_CSI_barsum %>%
group_by(stim_type, digit_group) %>%
summarise(y.position = max(estimate + se, na.rm = TRUE) * 1.08, .groups = "drop"),
by = c("stim_type", "digit_group")
) %>%
mutate(
stim_idx = as.numeric(stim_type),
xmin = stim_idx - fig_mediansplit_CSI_dodge/2,
xmax = stim_idx + fig_mediansplit_CSI_dodge/2
)
# --- plot ---
fig_mediansplit_CSI_plot <- ggplot(
fig_mediansplit_CSI_barsum,
aes(x = stim_type, y = estimate, fill = stim_type, group = order)
) +
geom_col(
aes(alpha = order),
position = position_dodge(width = fig_mediansplit_CSI_dodge),
width = 0.65
) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
position = position_dodge(width = fig_mediansplit_CSI_dodge),
width = 0.15, size = 1.2
) +
facet_wrap(~ digit_group) +
theme_classic(base_size = 13) +
labs(
x = "Condition during encoding",
y = "Mean % signal change",
fill = "Stimulus",
alpha = "Order"
) +
scale_fill_manual(values = c(
"Exact" = "#66BB6A",
"Modified" = "#9575CD"
)) +
scale_alpha_manual(values = c(
"First" = 0.95,
"Repeat" = 0.55
)) +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.14))) +
custom_theme +
guides(fill = "none") +
theme(
legend.position = c(0.88, 0.13)
)
fig_mediansplit_CSI_plot +
ggpubr::stat_pvalue_manual(
fig_mediansplit_CSI_stat.test,
label = "label",
xmin = "xmin",
xmax = "xmax",
y.position = "y.position",
tip.length = 0.01,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.shorten = 1.2,
size = 5.4,
bracket.size = 1.0
)
```
```{r}
text_base_export <- 40
fig_mediansplit_CSI_plot_export <- fig_mediansplit_CSI_plot +
ggpubr::stat_pvalue_manual(
fig_mediansplit_CSI_stat.test,
label = "label",
xmin = "xmin",
xmax = "xmax",
y.position = "y.position",
tip.length = 0.01,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.shorten = 1.2,
size = 8.4,
bracket.size = 1.4
) +
theme_classic(base_size = text_base_export) +
theme(
legend.position = c(0.88, 0.12),
legend.title = element_text(size = text_base_export * 0.88),
legend.text = element_text(size = text_base_export * 0.90),
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05)
)
ggsave(
"derivatives/figures/SI_figures/Figure2CSI.png",
plot = fig_mediansplit_CSI_plot_export,
width = 21, height = 13, units = "in",
dpi = 300, bg = "white", device = "png"
)
```
### Revised Supplementary Results 2 - Exact, close, distant
```{r}
# HELPERS
parse_roi_clean_anat <- function(roi_string) {
roi_string <- as.character(roi_string)
dplyr::case_when(
str_detect(roi_string, regex("CA12|CA1/2|CA1_2|CA1-2|CA1", ignore_case = TRUE)) ~ "CA12",
str_detect(roi_string, regex("DGCA3|DGCA23|DG-CA3|DG-CA23|DG", ignore_case = TRUE)) ~ "DGCA3",
str_detect(roi_string, regex("SUB", ignore_case = TRUE)) ~ "SUB",
TRUE ~ NA_character_
)
}
# BODY DATA FROM ORIGINAL ANATOMICAL TABLE
H1_clodis_anat_body <- semmst_merged_data_filtered_anat %>%
filter(task == "EncodingFirst") %>%
filter(cope_name %in% c("ExactRepeat", "CloseRepeat", "DistantRepeat")) %>%
mutate(
roi_cluster = paste(hemisphere, roi, sep = "_"),
roi_type = if_else(str_detect(roi_cluster, "HEAD"), "head", "body")
) %>%
filter(roi_type == "body") %>%
mutate(
condition = recode(
cope_name,
"ExactRepeat" = "Exact",
"CloseRepeat" = "Close",
"DistantRepeat" = "Distant"
),
hemi_lab = case_when(
hemisphere == "L" ~ "Left hemisphere",
hemisphere == "R" ~ "Right hemisphere",
TRUE ~ as.character(hemisphere)
),
roi_clean = parse_roi_clean_anat(roi_cluster)
) %>%
dplyr::select(
ID, task, cope_name, condition, estimate,
hemisphere, hemi_lab, roi, roi_clean, roi_cluster, roi_type,
everything()
)
# HEAD DATA FROM SUPPLEMENTARY ANATOMICAL TABLE
H1_clodis_anat_head <- semmst_merged_data_supplement_filtered %>%
filter(contrast == "anatHC_berron") %>%
filter(task == "EncodingFirst") %>%
filter(cope_name %in% c("ExactRepeat", "CloseRepeat", "DistantRepeat")) %>%
mutate(
roi = str_remove(roi, "^_"),
roi = str_remove(roi, "^HEAD"),
roi = str_remove(roi, "^_"),
roi_type = "head",
roi_cluster = paste(hemisphere, roi, sep = "_")
) %>%
mutate(
condition = recode(
cope_name,
"ExactRepeat" = "Exact",
"CloseRepeat" = "Close",
"DistantRepeat" = "Distant"
),
hemi_lab = case_when(
hemisphere == "L" ~ "Left hemisphere",
hemisphere == "R" ~ "Right hemisphere",
TRUE ~ as.character(hemisphere)
),
roi_clean = parse_roi_clean_anat(roi_cluster)
) %>%
dplyr::select(
ID, task, cope_name, condition, estimate,
hemisphere, hemi_lab, roi, roi_clean, roi_cluster, roi_type,
everything()
)
# MERGE HEAD AND BODY
H1_clodis_anat <- bind_rows(
H1_clodis_anat_head,
H1_clodis_anat_body
) %>%
mutate(
condition = factor(condition, levels = c("Exact", "Close", "Distant")),
roi_type = factor(roi_type, levels = c("head", "body")),
hemi_lab = factor(hemi_lab, levels = c("Left hemisphere", "Right hemisphere")),
roi_clean = factor(roi_clean, levels = c("CA12", "DGCA3", "SUB"))
)
# ASSUMPTION CHECKS
## OUTLIER
H1_clodis_anat %>%
group_by(roi_type, hemi_lab, roi_clean, condition) %>%
identify_outliers(estimate) %>%
dplyr::select(ID, roi_type, hemi_lab, roi_clean, condition, is.outlier, is.extreme) %>%
filter(is.extreme == TRUE)
# REMOVE OUTLIERS IF NEEDED
H1_clodis_anat_nooutlier <- H1_clodis_anat %>%
filter(! ID %in% c("sub-632012", "sub-499854")) %>%
drop_na(estimate) %>%
ungroup()
## NORMALITY
H1_clodis_anat_nooutlier %>%
group_by(roi_type, hemi_lab, roi_clean, condition) %>%
shapiro_test(estimate)
# ANOVA
# condition effect is tested separately within each anatomical facet
H1_clodis_anat_AOV = H1_clodis_anat_nooutlier %>%
anova_test(
dv = estimate,
wid = ID,
within = c(condition, roi_type, hemi_lab, roi_clean),
)
get_anova_table(H1_clodis_anat_AOV)
# POST-HOC
# pairwise comparisons - FWE correction across conditions only within each facet
H1_clodis_anat_PWC <- H1_clodis_anat_nooutlier %>%
group_by(roi_type, hemi_lab, roi_clean) %>%
pairwise_t_test(
estimate ~ condition,
paired = TRUE,
detailed = TRUE,
p.adjust.method = "holm"
) %>%
ungroup() %>%
add_significance("p") %>%
mutate(
label = case_when(
p < .001 ~ "p < .001",
p < .01 ~ "p < .01",
p < .05 ~ paste0("p = ", sub("^0", "", sprintf("%.3f", p))),
TRUE ~ paste0("p = ", sub("^0", "", sprintf("%.2f", p)))
)
)
H1_clodis_anat_PWC %>% print(n=40)
```
##### Revised Supplementary Figure 3
```{r}
# BARPLOT SUMMARY
fig_anat_barsum <- Rmisc::summarySEwithin(
data = H1_clodis_anat_nooutlier,
measurevar = "estimate",
withinvars = c("condition", "roi_type", "hemi_lab", "roi_clean"),
idvar = "ID",
na.rm = TRUE
) %>%
mutate(
condition = factor(as.character(condition), levels = c("Exact", "Close", "Distant")),
roi_type = factor(roi_type, levels = c("head", "body")),
hemi_lab = factor(hemi_lab, levels = c("Left hemisphere", "Right hemisphere")),
roi_clean = factor(roi_clean, levels = c("CA12", "DGCA3", "SUB"))
)
# SIGNIFICANT FWE-CORRECTED PAIRWISE BRACKETS
fig_anat_stat.test <- H1_clodis_anat_PWC %>%
dplyr::filter(p < .05) %>%
mutate(
roi_type = factor(roi_type, levels = c("head", "body")),
hemi_lab = factor(hemi_lab, levels = c("Left hemisphere", "Right hemisphere")),
roi_clean = factor(roi_clean, levels = c("CA12", "DGCA3", "SUB")),
label = case_when(
p < .001 ~ "p < .001",
p < .01 ~ "p < .01",
p < .05 ~ paste0("p = ", sub("^0", "", sprintf("%.3f", p))),
TRUE ~ paste0("p = ", sub("^0", "", sprintf("%.2f", p)))
)
)
# BRACKET HEIGHTS
fig_anat_ypos <- fig_anat_barsum %>%
dplyr::group_by(roi_type, hemi_lab, roi_clean) %>%
dplyr::summarise(
y_max = max(estimate + se, na.rm = TRUE),
y_min = min(estimate - se, na.rm = TRUE),
y_span = y_max - y_min,
.groups = "drop"
) %>%
dplyr::mutate(
y_span = if_else(y_span == 0 | is.na(y_span), 0.1, y_span),
y_base = y_max + 0.10 * y_span,
y_step = 0.12 * y_span
)
# STACK BRACKETS WITHIN EACH FACET
fig_anat_stat.test <- fig_anat_stat.test %>%
dplyr::left_join(fig_anat_ypos, by = c("roi_type", "hemi_lab", "roi_clean")) %>%
dplyr::group_by(roi_type, hemi_lab, roi_clean) %>%
dplyr::arrange(group1, group2, .by_group = TRUE) %>%
dplyr::mutate(y.position = y_base + (dplyr::row_number() - 1) * y_step) %>%
dplyr::ungroup()
# PANEL FUNCTION
make_anat_panel <- function(which_type, title_txt,
base_size = 14,
bar_width = 0.68,
err_width = 0.12,
err_size = 1.1,
p_size = 4.2,
p_bracket = 0.9) {
p_dat <- fig_anat_barsum %>%
dplyr::filter(roi_type == which_type)
p_sig <- fig_anat_stat.test %>%
dplyr::filter(roi_type == which_type)
p <- ggplot(
p_dat,
aes(x = condition, y = estimate, fill = condition)
) +
geom_col(width = bar_width) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
width = err_width,
linewidth = err_size
) +
facet_grid(rows = vars(roi_clean), cols = vars(hemi_lab), scales = "free_y") +
scale_fill_manual(values = c(
"Exact" = "#66BB6A",
"Close" = "#4477AA",
"Distant" = "#DDCC77"
)) +
scale_y_continuous(expand = expansion(mult = c(0.03, 0.22))) +
coord_cartesian(clip = "off") +
labs(
x = "Conditions of repeats during encoding",
y = "Mean % signal change",
title = title_txt
) +
theme_classic(base_size = base_size) +
custom_theme +
guides(fill = "none") +
theme(
plot.title = element_text(hjust = 0.5, face = "bold"),
strip.background = element_blank(),
strip.text = element_text(face = "bold"),
panel.spacing.x = unit(1.25, "lines"),
panel.spacing.y = unit(0.6, "lines")
)
# ADD BRACKETS ONLY IF SIGNIFICANT TESTS EXIST
if (nrow(p_sig) > 0) {
p <- p +
ggpubr::stat_pvalue_manual(
p_sig,
label = "label",
xmin = "group1",
xmax = "group2",
y.position = "y.position",
tip.length = 0.01,
hide.ns = FALSE,
inherit.aes = FALSE,
bracket.shorten = 0.05,
size = p_size,
bracket.size = p_bracket
)
}
p
}
# VIEW
p_head_view <- make_anat_panel(
"head", "Hippocampal head",
base_size = 14,
bar_width = 0.68,
err_width = 0.12,
err_size = 1.1,
p_size = 4.2,
p_bracket = 0.9
)
p_body_view <- make_anat_panel(
"body", "Hippocampal body",
base_size = 14,
bar_width = 0.68,
err_width = 0.12,
err_size = 1.1,
p_size = 4.2,
p_bracket = 0.9
)
supp_fig_anat_bars <- p_head_view | p_body_view
supp_fig_anat_bars
```
```{r}
# EXPORT
text_base_export <- 24
FigureSI3_export_theme <- theme_classic(base_size = text_base_export) +
theme(
axis.text.x = element_text(size = text_base_export * 0.80, margin = margin(t = 6)),
axis.text.y = element_text(size = text_base_export * 0.80),
axis.title.x = element_text(size = text_base_export * 0.95, margin = margin(t = 10)),
axis.title.y = element_text(size = text_base_export * 0.95, margin = margin(r = 10)),
# more visible square facet boxes
strip.background = element_rect(
fill = "white",
colour = "black",
linewidth = 1.0
),
strip.text = element_text(
size = text_base_export * 0.76,
face = "bold",
margin = margin(4, 8, 4, 8)
),
plot.title = element_text(size = text_base_export * 1.00, face = "bold", hjust = 0.5),
legend.position = "none",
panel.spacing.x = unit(0.9, "lines"),
panel.spacing.y = unit(0.9, "lines"),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.border = element_blank(),
axis.line = element_line(linewidth = 0.5)
)
p_head_export <- make_anat_panel(
"head", "Hippocampal head",
base_size = text_base_export,
bar_width = 0.65,
err_width = 0.15,
err_size = 1.1,
# slightly smaller p labels/brackets
p_size = 5.8,
p_bracket = 0.65
) +
FigureSI3_export_theme +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.16)))
p_body_export <- make_anat_panel(
"body", "Hippocampal body",
base_size = text_base_export,
bar_width = 0.65,
err_width = 0.15,
err_size = 1.1,
# slightly smaller p labels/brackets
p_size = 5.8,
p_bracket = 0.65
) +
FigureSI3_export_theme +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.16)))
supp_fig_anat_bars_export <- p_head_export | p_body_export
ggsave(
"derivatives/figures/SI_figures/Figure3SI.png",
plot = supp_fig_anat_bars_export,
width = 23, height = 12, units = "in",
dpi = 300,
bg = "white",
device = ragg::agg_png
)
```
### Revised Supplementary Results 3 - anatomical masks
```{r}
# BODY ROIs from the original anatomical dataset
revision_results3_body_from_anat <- semmst_merged_data_filtered_anat %>%
filter(task == "Encoding") %>%
filter(cope_name %in% c("LureFirst", "ExactFirst", "LureRepeat", "ExactRepeat")) %>%
filter(!str_detect(roi, "HEAD")) %>%
extract(
col = cope_name,
into = c("stim_type", "order"),
regex = "([A-Za-z]+?)(First|Repeat)$"
) %>%
mutate(
stim_type = as.factor(stim_type),
order = as.factor(order)
) %>%
mutate(
stim_type = factor(stim_type, levels = c("Exact", "Lure")),
order = factor(order),
roi_type = "body"
)
# HEAD/other supplement anatomical data
revision_results3_supplement_anat <- semmst_merged_data_supplement_filtered %>%
filter(contrast == "anatHC_berron") %>%
filter(task == "Encoding") %>%
filter(cope_name %in% c("LureFirst", "ExactFirst", "LureRepeat", "ExactRepeat")) %>%
extract(
col = cope_name,
into = c("stim_type", "order"),
regex = "([A-Za-z]+?)(First|Repeat)$"
) %>%
mutate(
stim_type = as.factor(stim_type),
order = as.factor(order),
roi = str_remove(roi, "^_"),
roi = str_remove(roi, "^HEAD"),
roi = str_remove(roi, "^_"),
roi_type = if_else(str_detect(roi_cluster, "HEAD") | str_detect(roi, "^SUB|^CA1|^DGCA23"), "head", "body")
) %>%
mutate(
stim_type = factor(stim_type, levels = c("Exact", "Lure")),
order = factor(order)
)
# MERGE body anat + supplement anat
semmst_merged_data_supplement_filtered_revision_results3 = bind_rows(
revision_results3_body_from_anat,
revision_results3_supplement_anat
) %>%
mutate(
roi_type = factor(roi_type, levels = c("head", "body"))
)
# ASSUMPTION CHECKS
## OUTLIER
semmst_merged_data_supplement_filtered_revision_results3 %>%
group_by(stim_type, order, roi_type, roi, hemisphere) %>%
identify_outliers(estimate) %>%
dplyr::select(ID, roi_type, roi, hemisphere, is.outlier, is.extreme) %>%
filter(is.extreme == TRUE)
semmst_merged_data_supplement_filtered_revision_results3_nooutlier = semmst_merged_data_supplement_filtered_revision_results3 %>%
group_by(stim_type, order, roi_type, roi, hemisphere) %>%
filter(!ID %in% c("sub-499854", "sub-982347")) %>%
ungroup()
## NORMALITY
semmst_merged_data_supplement_filtered_revision_results3_nooutlier %>%
group_by(stim_type, order, roi_type) %>%
shapiro_test(estimate)
# ANOVA
semmst_merged_data_supplement_filtered_revision_results3_nooutlier_AOV = semmst_merged_data_supplement_filtered_revision_results3_nooutlier %>%
anova_test(
dv = estimate,
wid = ID,
within = c(stim_type, order, roi_type, roi, hemisphere)
)
get_anova_table(semmst_merged_data_supplement_filtered_revision_results3_nooutlier_AOV)
# REPORT Effect DFn DFd F p p<.05 ges
## 2 order 1.00 27.00 10.116 4.00e-03 * 1.70e-02
## 7 stim_type:roi_type 1.00 27.00 5.878 2.20e-02 * 2.00e-03
## 8 order:roi_type 1.00 27.00 28.916 1.11e-05 * 1.90e-02
## 9 stim_type:roi 2.00 54.00 4.116 2.20e-02 * 2.00e-03
## 10 order:roi 1.61 43.44 4.124 3.00e-02 * 2.00e-03
## 11 roi_type:roi 2.00 54.00 13.740 1.50e-05 * 2.70e-02
## 13 order:hemisphere 1.00 27.00 6.667 1.60e-02 * 2.00e-03
## 18 stim_type:roi_type:roi 2.00 54.00 3.655 3.20e-02 * 9.16e-04
## 25 roi_type:roi:hemisphere 2.00 54.00 3.282 4.50e-02 * 3.00e-03
## 29 stim_type:roi_type:roi:hemisphere 2.00 54.00 5.885 5.00e-03 * 1.00e-03
```
##### Revised Supplementary Figure 4.
```{r}
FigureR3_dodge <- 0.8
# within-subject summary per stim_type × order × roi_type × roi × hemisphere
FigureR3_barsum <- Rmisc::summarySEwithin(
data = semmst_merged_data_supplement_filtered_revision_results3_nooutlier,
measurevar = "estimate",
withinvars = c("stim_type", "order", "roi_type", "roi", "hemisphere"),
idvar = "ID",
na.rm = TRUE
) %>%
mutate(
stim_type = factor(stim_type, levels = c("Exact", "Lure")),
order = factor(order, levels = c("First", "Repeat")),
roi_type = factor(roi_type, levels = c("head", "body")),
roi = factor(roi),
hemisphere = factor(
hemisphere,
levels = c("L", "R"),
labels = c("Left hemisphere", "Right hemisphere")
)
) %>%
mutate(
stim_type = factor(
as.character(stim_type),
levels = c("Exact", "Lure"),
labels = c("Exact", "Modified")
)
)
FigureR3_cols <- c(
"Exact" = "#66BB6A",
"Modified" = "#9575CD"
)
FigureR3_make_panel <- function(df) {
ggplot(
df,
aes(
x = stim_type,
y = estimate,
fill = stim_type,
alpha = order,
group = order
)
) +
geom_col(
position = position_dodge(width = FigureR3_dodge),
width = 0.65
) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
position = position_dodge(width = FigureR3_dodge),
width = 0.15,
linewidth = 1.1
) +
facet_grid(rows = vars(roi), cols = vars(hemisphere)) +
theme_classic(base_size = 13) +
labs(
x = "Condition during encoding",
y = "Mean % signal change",
alpha = "Order"
) +
scale_fill_manual(values = FigureR3_cols) +
scale_alpha_manual(values = c("First" = 0.95, "Repeat" = 0.55)) +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.12))) +
guides(
fill = "none",
alpha = guide_legend(title = "Order")
) +
theme(
strip.background = element_blank(),
strip.text = element_text(face = "bold"),
panel.spacing = unit(0.6, "lines")
)
}
# Left panel (head): no legend
FigureR3_head_plot <- FigureR3_barsum %>%
filter(roi_type == "head") %>%
FigureR3_make_panel() +
ggtitle("Hippocampal head") +
theme(
plot.title = element_text(hjust = 0.5, face = "bold"),
legend.position = "none"
)
# Right panel (body): legend inset top-right
FigureR3_body_plot <- FigureR3_barsum %>%
filter(roi_type == "body") %>%
FigureR3_make_panel() +
ggtitle("Hippocampal body") +
theme(
plot.title = element_text(hjust = 0.5, face = "bold"),
legend.position = c(0.98, 0.98),
legend.justification = c(1, 1),
legend.background = element_rect(fill = alpha("white", 0.75), colour = NA),
legend.key = element_blank()
)
# Combine as two columns
FigureR3_plot <- (FigureR3_head_plot | FigureR3_body_plot)
FigureR3_plot
```
```{r}
text_base_export <- 24 # 20–28 usually looks best for this many facets
# IMPORTANT: do NOT set legend.position in the export theme
FigureR3_export_theme <- theme_classic(base_size = text_base_export) +
theme(
axis.text.x = element_text(size = text_base_export * 0.80, margin = margin(t = 6)),
axis.text.y = element_text(size = text_base_export * 0.80),
axis.title.x = element_text(size = text_base_export * 0.95, margin = margin(t = 10)),
axis.title.y = element_text(size = text_base_export * 0.95, margin = margin(r = 10)),
strip.text = element_text(size = text_base_export * 0.85, face = "bold"),
plot.title = element_text(size = text_base_export * 1.05, face = "bold", hjust = 0.5),
legend.title = element_text(size = text_base_export * 0.85),
legend.text = element_text(size = text_base_export * 0.80),
panel.spacing = unit(0.8, "lines")
)
# Apply export theme per panel + enforce legend behavior
FigureR3_head_export <- FigureR3_head_plot &
FigureR3_export_theme &
theme(legend.position = "none")
FigureR3_body_export <- FigureR3_body_plot &
FigureR3_export_theme &
theme(
legend.position = c(0.98, 0.98), # inside top-right of RIGHT panel
legend.justification = c(1, 1),
legend.background = element_rect(fill = alpha("white", 0.75), colour = NA),
legend.key = element_blank()
)
FigureR3_plot_export <- FigureR3_head_export | FigureR3_body_export
ggsave(
"derivatives/figures/SI_figures/Figure4SI.png",
plot = FigureR3_plot_export,
width = 23, height = 12, units = "in",
dpi = 300, bg = "white", device = ragg::agg_png
)
```
### Supplementary Results 4 - close vs distant REC
```{r}
semmst_merged_data_filtered_H4b_REC_clodis = semmst_merged_data_filtered %>%
filter(task == "RecognitionRespCloDis") %>%
filter(cope_name %in% c("DistantCR", "CloseCR"))
# ASSUMPTION CHECKS
## OUTLIER
semmst_merged_data_filtered_H4b_REC_clodis %>%
group_by(cope_name, roi_cluster) %>%
identify_outliers(estimate) %>%
dplyr::select(ID, roi_cluster, cope_name, is.outlier, is.extreme) %>%
filter(is.extreme==TRUE)
semmst_merged_data_filtered_H4b_REC_clodis_nooutlier = semmst_merged_data_filtered_H4b_REC_clodis %>%
ungroup()
semmst_merged_H4b_REC_clodis_PWC = semmst_merged_data_filtered_H4b_REC_clodis_nooutlier %>%
group_by(roi_cluster) %>%
pairwise_t_test(
estimate ~ cope_name,
paired = TRUE,
detailed = TRUE,
) %>%
ungroup() %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
ungroup() %>%
mutate(label = paste0("p=", signif(p.adj, 3)))
semmst_merged_H4b_REC_clodis_PWC
# REP roi_cluster estimate .y. group1 group2 n1 n2 statistic p df conf.low conf.high method alternative p.adj
## 1 L_HEAD_DGCA23_HEAD_SUB_14 -0.110 estimate CloseCR Distan… 30 30 -1.55 0.133 29 -0.255 0.0355 T-test two.sided 0.266 ns
## 2 R_HEAD_DGCA23_HEAD_SUB_15 -0.0498 estimate CloseCR Distan… 30 30 -1.07 0.294 29 -0.145 0.0455 T-test two.sided 0.588 ns
```
#### Supplementary Figure 5.
```{r}
# BARPLOT
figSI_clodisREC_dodge <- 0.8
figSI_clodisREC_barsum <- Rmisc::summarySEwithin(
data = semmst_merged_data_filtered_H4b_REC_clodis,
measurevar = "estimate",
withinvars = c("cope_name", "roi_cluster"),
idvar = "ID",
na.rm = TRUE
)
figSI_clodisREC_roi_lab <- semmst_merged_data_filtered_H4_REC_conditions_nooutlier %>%
distinct(roi_cluster, hemisphere, roi, cluster, voxels) %>%
mutate(
roi_label = case_when(
roi == "HEAD_DGCA23_HEAD_SUB" & hemisphere == "R" ~ "Right hippocampal head cluster",
roi == "HEAD_DGCA23_HEAD_SUB" & hemisphere == "L" ~ "Left hippocampal head cluser",
TRUE ~ NA_character_
)
) %>%
select(roi_cluster, roi_label)
figSI_clodisREC_level_map <- c("CloseCR", "DistantCR")
figSI_clodisREC_level_lab <- c("Close lure", "Distant lure")
figSI_clodisREC_barsum <- figSI_clodisREC_barsum %>%
mutate(cope_name = factor(cope_name, levels = figSI_clodisREC_level_map, labels = figSI_clodisREC_level_lab)) %>%
left_join(figSI_clodisREC_roi_lab, by = "roi_cluster")
figSI_clodisREC_plot <- ggplot(figSI_clodisREC_barsum, aes(x = cope_name, y = estimate, fill = cope_name)) +
geom_col(position = position_dodge(width = figSI_clodisREC_dodge), width = 0.65) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
position = position_dodge(width = figSI_clodisREC_dodge),
width = 0.15,
size = 1.2
) +
facet_wrap(~ roi_label) +
theme_classic(base_size = 13) +
labs(x = "Condition during recognition", y = "Mean % signal change") +
scale_fill_manual(values = c(
"Close lure" = "#4477AA",
"Distant lure" = "#DDCC77"
)) +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.12))) +
custom_theme +
guides(fill = "none")
figSI_clodisREC_plot
```
```{r}
text_base_export=40
figSI_clodisREC_plot_export <- figSI_clodisREC_plot +
theme_classic(base_size = text_base_export) +
theme(
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05),
strip.text = element_text(size = text_base_export * 0.95)
)
ggsave(
"derivatives/figures/SI_figures/Figure5SI.png",
plot = figSI_clodisREC_plot_export,
width = 22, height = 14, units = "in",
dpi = 300, bg = "white", device = "png"
)
```
### Supplementary Results 5 - Behaviour wo. covariates and outlier
```{r}
semmst_behav_data_rec_LDI_ANOVA = semmst_behav_data_rec_summary %>%
dplyr::select(ID, rec_dprime, close_dprime, distant_dprime) %>%
gather("dprime_type", "dprime_value", rec_dprime:distant_dprime) %>%
mutate(dprime_type = factor(dprime_type, levels = c("close_dprime", "rec_dprime", "distant_dprime"))) %>%
left_join(semmst_behav_data_rec_filtered %>% dplyr::select(ID, age, sex, education, Vocabulary, Digit_Symbol, NonWord_sum) %>% distinct(), by="ID")
semmst_behav_data_rec_LDI_ANCOVA = semmst_behav_data_rec_LDI_ANOVA %>%
left_join(semmst_behav_data_rec_filtered %>% dplyr::select(ID, age, sex, education, Vocabulary, Digit_Symbol, NonWord_sum) %>% distinct(), by="ID") %>%
mutate(Vocabulary_scale = scale(Vocabulary)) %>%
mutate(Digit_Symbol_scale = scale(Digit_Symbol)) %>%
mutate(NonWord_sum_scale = scale(NonWord_sum)) %>% drop_na()
# ASSUMPTION CHECKS
## OUTLIER
semmst_behav_data_rec_LDI_ANOVA %>%
group_by(dprime_type) %>%
identify_outliers(dprime_value)
##NORMALITY
semmst_behav_data_rec_LDI_ANOVA %>%
group_by(dprime_type) %>%
shapiro_test(dprime_value) %>% print(n=40)
ggqqplot(semmst_behav_data_rec_LDI_ANOVA, "dprime_value", ggtheme = theme_bw()) +
facet_grid(. ~ dprime_type, labeller = "label_both")
# ANOVA
ANOVA_behav_data_rec_LDI = semmst_behav_data_rec_LDI_ANOVA %>%
anova_test(dv = dprime_value, wid = ID, within = c(dprime_type), covariate = c(sex, age))
get_anova_table(ANOVA_behav_data_rec_LDI)
# REPORT
## dprime_type 2 54 159.192 2.28e-23 * 0.493
```
#### Supplementary Figure 6.
```{r}
# 1) Prep & adjustment (remove linear effects of covariates)
figSI_outlier_plot_base <- semmst_behav_data_rec_LDI_ANOVA %>%
filter(!is.na(ID), !is.na(dprime_type), !is.na(dprime_value)) %>%
mutate(
dprime_type = factor(dprime_type),
sex = factor(sex)
)
# order as: Close, Distant, Target
figSI_outlier_contrast_order <- c("close_dprime", "distant_dprime", "rec_dprime")
# order as: Close, Distant, Target
figSI_outlier_contrast_order <- c("close_dprime", "distant_dprime", "rec_dprime")
figSI_outlier_plot_base <- figSI_outlier_plot_base %>%
mutate(dprime_type = fct_relevel(dprime_type, figSI_outlier_contrast_order))
# covariate-only model for adjustment
figSI_outlier_mod_cov <- lm(dprime_value ~ sex + age,
data = figSI_outlier_plot_base)
figSI_outlier_plot_base <- figSI_outlier_plot_base %>%
mutate(dprime_adj = dprime_value - predict(figSI_outlier_mod_cov, newdata = figSI_outlier_plot_base) + mean(dprime_value, na.rm = TRUE))
# 2) Labels & colors
figSI_outlier_x_lab_map <- c(
"close_dprime" = "Close\nlure",
"distant_dprime" = "Distant\nlure",
"rec_dprime" = "Foil"
)
# Paul Tol (muted): blue / yellow / red — CVD-safe, good in print & grayscale
figSI_outlier_fill_vals <- c("close_dprime"="#4477AA", "distant_dprime"="#DDCC77", "rec_dprime"="#117733")
# 3) Plot
behav_figSI_outlier <- ggplot(figSI_outlier_plot_base, aes(x = dprime_type, y = dprime_adj, fill = dprime_type)) +
geom_boxplot(width = 0.6, outlier.alpha = 0.4, linewidth = 1.8, color = "black") +
stat_summary(fun = mean, geom = "point", shape = 21, size = 2.2, fill = "white", color = "black") +
scale_x_discrete(labels = figSI_outlier_x_lab_map, drop = FALSE) +
scale_fill_manual(values = figSI_outlier_fill_vals, guide = "none") +
labs(
x = "Discriminability contrasts",
y = "Covariate-adjusted d'\n [p(\"old\"|target) - p(\"old\"|condition)]"
) +
theme_classic(base_size = 14)
behav_figSI_outlier
```
```{r}
behav_figSI_outlier_export <- behav_figSI_outlier +
theme_classic(base_size = 40) +
theme(
axis.title = element_text(size = 42),
axis.text = element_text(size = 34)
) + labs(x = NULL)
ggsave(
"derivatives/figures/SI_figures/Figure6SI.png",
plot = behav_figSI_outlier_export,
width = 16, height = 12, units = "in",
dpi = 300,
bg = "white",
device = "png" # base png
)
```
### Supplementary Results 6 - Encoding wo. covariates and outlier
#### ANOVA
```{r}
# FILTER DATA
semmst_merged_data_H1 = semmst_merged_data_filtered %>%
filter(task == "Encoding") %>%
filter(cope_name %in% c("LureFirst", "ExactFirst", "LureRepeat", "ExactRepeat")) %>%
extract(
col = cope_name,
into = c("stim_type", "order"),
regex = "([A-Za-z]+?)(First|Repeat)$"
) %>%
mutate(
stim_type = as.factor(stim_type),
order = as.factor(order)
) %>%
mutate(stim_type = factor(stim_type, levels = c("Exact", "Lure")))
# ASSUMPTION CHECKS
## OUTLIER
semmst_merged_data_H1 %>%
group_by(stim_type, order, roi_cluster) %>%
identify_outliers(estimate) %>%
dplyr::select(ID, is.outlier, is.extreme) %>%
filter(is.extreme==TRUE)
semmst_merged_data_H1SI_nooutlier = semmst_merged_data_H1 %>%
group_by(stim_type, order, roi_cluster) %>%
#filter(! ID %in% c("sub-434971")) %>% #, " sub-012421", "sub-632012", "sub-800472")) %>% # c("sub-434971", "sub-982347")) %>%
ungroup()
##NORMALITY
semmst_merged_data_H1SI_nooutlier %>%
group_by(stim_type, order) %>%
shapiro_test(estimate)
# ANOVA
semmst_merged_data_H1SI_nooutlier_AOV = semmst_merged_data_H1SI_nooutlier %>%
anova_test(dv = estimate, wid = ID, within = c(stim_type, order, roi_cluster))#, covariate = c(NonWord_sum_scale, Digit_Symbol_scale))
get_anova_table(semmst_merged_data_H1SI_nooutlier_AOV)
# REPORT Effect DFn DFd F p p<.05 ges
## 1 stim_type 1 29 16.545 3.33e-04 * 3.30e-02
## 2 order 1 29 54.775 3.73e-08 * 1.07e-01
## 4 stim_type:order 1 29 4.123 5.20e-02 1.60e-02
```
#### Correlation
```{r}
group_vars_H3SI <- c("ID", "mask_type", "space_type", "contrast", "roi_cluster")
# baseline
semmst_merged_data_H3SI_neural_ps = semmst_merged_data_filtered %>%
filter(task %in% c("Encoding")) %>%
filter(cope_name %in% c("ExactFirst", "ExactRepeat", "LureFirst","LureRepeat", "CloseFirst", "CloseRepeat", "DistantFirst", "DistantRepeat")) %>%
dplyr::select(ID, space_type, cope_name, hemisphere, roi, mask_type, cluster, voxels, contrast, task, roi_cluster, roi_hemisphere, estimate, NonWord_sum_scale, Digit_Symbol_scale, rec_dprime:lure_incorrect) %>%
group_by(across(all_of(group_vars_H3SI))) %>%
pivot_wider(names_from = cope_name, values_from = estimate, values_fill = NA_real_) %>%
mutate(
# --- Non-RS bias scores (Exact denominator) ---
LureBiasScore = LureRepeat - ExactRepeat,
CloseBiasScore = CloseRepeat - ExactRepeat,
DistantBiasScore = DistantRepeat - ExactRepeat
) %>%
ungroup()
# 1) Filter to the Exact subset and define roi_group
df_corr_input_all_SI <- semmst_merged_data_H3SI_neural_ps %>%
mutate(roi_group = roi_cluster) %>%
dplyr::select(ID, roi_group, roi, hemisphere, cluster, voxels, LureBiasScore, lure_dprime, NonWord_sum_scale, Digit_Symbol_scale) %>%
ungroup()
df_corr_input_all_SI_nooutlier <- df_corr_input_all_SI %>%
droplevels() %>%
ungroup()
# Fisher-z CI for Pearson r (95% hard-coded)
pearson_ci <- function(r, n) {
z <- atanh(r)
se <- 1 / sqrt(n - 3)
zcrit <- qnorm(1 - (1 - 0.95) / 2)
c(low = tanh(z - zcrit * se), high = tanh(z + zcrit * se))
}
H3SI_corr_lure_simple <- df_corr_input_all_SI_nooutlier %>%
group_by(roi_group) %>%
group_modify(~{
d_all <- .x
# Pearson
d_s <- d_all %>% dplyr::select(LureBiasScore, lure_dprime) %>% tidyr::drop_na()
ct_s <- cor.test(d_s$LureBiasScore, d_s$lure_dprime, method = "pearson")
r_s <- unname(ct_s$estimate)
p_s <- ct_s$p.value
n_s <- nrow(d_s)
ci_s <- pearson_ci(r_s, n_s)
tibble::tibble(
n = n_s,
r = r_s,
p = p_s,
ci = sprintf("[%.3f, %.3f]", ci_s["low"], ci_s["high"])
)
}) %>%
ungroup() %>%
mutate(
p_adj = p.adjust(p, method = "bonferroni")
) %>%
arrange(p_adj, p)
H3SI_corr_lure_simple
```
#### Supplementary Figure 7B
```{r}
fig_corroutlier_BSI_dodge <- 0.8
fig_corroutlier_BSI_barsum <- Rmisc::summarySEwithin(
data = semmst_merged_data_H1SI_nooutlier,
measurevar = "estimate",
withinvars = c("stim_type","order","roi_cluster"),
idvar = "ID",
na.rm = TRUE
) %>%
mutate(stim_type = factor(as.character(stim_type),
levels = c("Exact","Lure"),
labels = c("Exact","Modified")))
fig_corroutlier_BSI_roi_lab <- semmst_merged_data_H1SI_nooutlier %>%
distinct(roi_cluster, hemisphere, roi, cluster, voxels) %>%
mutate(
roi_label = case_when(
roi == "HEAD_DGCA23_HEAD_SUB" & hemisphere == "R" ~ "Right DG-CA2/3-SUB cluster",
roi == "HEAD_DGCA23_HEAD_SUB" & hemisphere == "L" ~ "Left DG-CA2/3-SUB cluster",
TRUE ~ as.character(roi_cluster)
)
) %>%
select(roi_cluster, roi_label)
fig_corroutlier_BSI_barsum <- fig_corroutlier_BSI_barsum %>%
left_join(fig_corroutlier_BSI_roi_lab, by = "roi_cluster")
# bar plot
fig_corroutlier_BSI <- ggplot(fig_corroutlier_BSI_barsum, aes(
x = stim_type, y = estimate,
fill = stim_type,
group = order
)) +
geom_col(aes(alpha = order),
position = position_dodge(width = fig_corroutlier_BSI_dodge),
width = 0.65) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
position = position_dodge(width = fig_corroutlier_BSI_dodge),
width = 0.15,
size = 1.2
) +
facet_wrap(~ roi_label) +
theme_classic(base_size = 13) +
labs(x = "Condition during encoding", y = "Mean % signal change", fill = "Stimulus", alpha = "Order") +
scale_fill_manual(values = c(
"Exact" = "#66BB6A",
"Modified" = "#9575CD"
)) +
scale_alpha_manual(values = c(
"First" = 0.95,
"Repeat" = 0.55
)) +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.12))) +
custom_theme +
guides(fill = "none")
fig_corroutlier_BSI
```
```{r}
text_base_export <- 40
fig_corroutlier_BSI_export <- fig_corroutlier_BSI +
theme_classic(base_size = text_base_export) +
theme(
legend.position = c(0.88, 0.88),
legend.title = element_text(size = text_base_export * 0.88),
legend.text = element_text(size = text_base_export * 0.90),
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05),
strip.text = element_blank(),
panel.spacing.x = unit(4, "in")
)
ggsave(
"derivatives/figures/SI_figures/Figure7BSI.png",
plot = fig_corroutlier_BSI_export,
width = 26, height = 13, units = "in",
dpi = 300, bg = "white", device = "png"
)
```
#### Supplementary Figure 7A
```{r}
ID_HI <- "sub-434971"
semmst_merged_data_H1SI_nooutlier_data <- semmst_merged_data_H1SI_nooutlier %>%
filter(!is.na(estimate)) %>%
mutate(
highlight = (ID == ID_HI),
# Column labels you asked for
stim_col = case_when(
stim_type == "Exact" ~ "Exact repeat",
TRUE ~ "Modified repeat" # e.g., Lure -> Modified
),
# Row labels you asked for
roi_row = case_when(
str_detect(roi_cluster, "^L$|L") ~ "Left DG-CA2/3-Subiculum",
str_detect(roi_cluster, "^R$|R") ~ "Right DG-CA2/3-Subiculum",
TRUE ~ as.character(roi_cluster)
),
order_row = case_when(
order == "First" ~ "First presentation",
order == "Repeat" ~ "Repeat",
TRUE ~ as.character(order)
)
) %>%
mutate(
stim_col = factor(stim_col, levels = c("Exact repeat", "Modified repeat")),
roi_row = factor(roi_row, levels = c("Left DG-CA2/3-Subiculum", "Right DG-CA2/3-Subiculum")),
order_row = factor(order_row, levels = c("First presentation", "Repeat"))
)
# compute QQ points per facet (roi_row × order_row × stim_col) ---
semmst_merged_data_H1SI_nooutlier_qq <- semmst_merged_data_H1SI_nooutlier_data %>%
group_by(roi_row, order_row, stim_col) %>%
arrange(estimate, .by_group = TRUE) %>%
mutate(theoretical = qnorm(ppoints(n()))) %>%
ungroup()
# compute qqline (like R's qqline: through 25% and 75%) per facet ---
semmst_merged_data_H1SI_nooutlier_lines <- semmst_merged_data_H1SI_nooutlier_qq %>%
group_by(roi_row, order_row, stim_col) %>%
summarise(
y25 = quantile(estimate, 0.25, na.rm = TRUE),
y75 = quantile(estimate, 0.75, na.rm = TRUE),
x25 = qnorm(0.25),
x75 = qnorm(0.75),
slope = (y75 - y25) / (x75 - x25),
intercept = y25 - slope * x25,
.groups = "drop"
)
# plot (thin line, correct highlight, correct labels) ---
fig_corroutlier_SIA_figure = ggplot(semmst_merged_data_H1SI_nooutlier_qq, aes(x = theoretical, y = estimate)) +
geom_point(size = 2.4, alpha = 0.75, colour = "black") +
geom_point(
data = dplyr::filter(semmst_merged_data_H1SI_nooutlier_qq, highlight),
size = 2.6, alpha = 0.95, colour = "red"
) +
geom_abline(
data = semmst_merged_data_H1SI_nooutlier_lines,
aes(slope = slope, intercept = intercept),
linewidth = 1.2,
colour = "black"
) +
facet_grid(roi_row + order_row ~ stim_col, labeller = label_value) +
labs(
x = "Theoretical quantiles",
y = "Sample quantiles (% signal change)"
) +
theme_bw()
fig_corroutlier_SIA_figure
```
```{r}
text_base_export <- 34 # tweak (28–40); 34 usually good for dense facet grids
fig_corroutlier_SIA_export <- fig_corroutlier_SIA_figure +
# bump dot sizes (both black + highlighted red)
geom_point(size = 7.1, alpha = 0.75, colour = "black") +
geom_point(
data = dplyr::filter(semmst_merged_data_H1SI_nooutlier_qq, highlight),
size = 10.3, alpha = 0.95, colour = "red"
) +
# thicker qqline
geom_abline(
data = semmst_merged_data_H1SI_nooutlier_lines,
aes(slope = slope, intercept = intercept),
linewidth = 1.6,
colour = "black"
) +
theme_bw(base_size = text_base_export) +
theme(
axis.text.x = element_text(size = text_base_export * 0.85, margin = margin(t = 6)),
axis.text.y = element_text(size = text_base_export * 0.85),
axis.title.x = element_text(size = text_base_export * 0.95, margin = margin(t = 10)),
axis.title.y = element_text(size = text_base_export * 0.95, margin = margin(r = 10)),
strip.text = element_text(size = text_base_export * 0.80, face = "bold"),
panel.spacing = unit(0.65, "lines"),
plot.margin = margin(12, 12, 12, 12)
)
# install.packages("ragg")
ggsave(
"derivatives/figures/SI_figures/Figure7ASI.png",
plot = fig_corroutlier_SIA_export,
width = 16, height = 26, units = "in",
dpi = 300, bg = "white", device = ragg::agg_png
)
```
#### Supplementary Figure 7C
```{r}
ID_HI <- "sub-434971"
# 1) Long table (ALL trials only)
fig_corroutlierCSI_df <- df_corr_input_all_SI_nooutlier %>%
dplyr::select(ID, roi_group, roi, hemisphere, bias = LureBiasScore, lure_dprime) %>%
tidyr::drop_na(bias, lure_dprime) %>%
dplyr::mutate(
hemi_lab = dplyr::recode(hemisphere, L = "Left", R = "Right"),
roi_clean = roi %>%
gsub("HEAD_", "", .) %>%
gsub("_", "-", .) %>%
gsub("2", "", .),
facet_lab = paste(hemi_lab, roi_clean)
)
# 2) Facet labels (no stats)
fig_corroutlierCSI_labs <- fig_corroutlierCSI_df %>%
dplyr::distinct(roi_group, facet_lab)
# 3) Plot (base)
fig_corroutlierCSI <- ggplot(
fig_corroutlierCSI_df %>% mutate(highlight = (ID == ID_HI)),
aes(bias, lure_dprime)
) +
# all points
geom_point(
data = \(d) d %>% filter(!highlight),
alpha = 0.7, size = 4.5, colour = "grey20"
) +
# highlighted point on top
geom_point(
data = \(d) d %>% filter(highlight),
alpha = 1, size = 6.2, colour = "red"
) +
# regression line (single neutral colour)
geom_smooth(method = "lm", se = FALSE, linewidth = 1.1, colour = "grey20") +
facet_wrap(
~ roi_group, scales = "free",
labeller = labeller(roi_group = setNames(fig_corroutlierCSI_labs$facet_lab, fig_corroutlierCSI_labs$roi_group))
) +
labs(
x = "Modified repeat - exact repeat estimates",
y = "d' (lures vs targets)"
) +
theme_classic(base_size = 18) +
custom_theme +
theme(
panel.spacing = unit(4, "in") # big gap like your 6A; reduce if too much
)
fig_corroutlierCSI
```
```{r}
text_base_export <- 40
fig_corroutlierCSI_export <- fig_corroutlierCSI +
theme_classic(base_size = text_base_export) +
theme(
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05),
panel.spacing = unit(4, "in"),
strip.text = element_blank()
) +
# make the line thicker for export (optional)
geom_smooth(method = "lm", se = FALSE, linewidth = 2.6, colour = "grey20")
ggsave(
"derivatives/figures/SI_figures/Figure7CSI.png",
plot = fig_corroutlierCSI_export,
width = 26, height = 12, units = "in",
dpi = 300, bg = "white", device = "png"
)
```
### Supplementary Results 7 - Recognition per encoding status
```{r}
# FoilCR baseline per ID x roi_cluster
foil_baseline_df <- semmst_merged_data_filtered %>%
filter(task == "RecognitionRespAll",
cope_name == "FoilCR") %>%
select(ID, roi_cluster, foil_baseline = estimate)
# Target conditions + baseline correction
semmst_merged_data_filtered_supp4_REC_conditions_bc <- semmst_merged_data_supplement_filtered %>%
filter(task %in% c("RecognitionRespAllEncRepeat", "RecognitionRespAllEncSimilar"),
cope_name %in% c("LureCR", "TargetHIT")) %>%
left_join(foil_baseline_df, by = c("ID", "roi_cluster")) %>%
mutate(estimate_bc = estimate - foil_baseline)
# ASSUMPTION CHECKS
## OUTLIER
semmst_merged_data_filtered_supp4_REC_conditions_bc %>%
group_by(cope_name, task, roi_cluster) %>%
identify_outliers(estimate) %>%
dplyr::select(ID, roi_cluster, cope_name, is.outlier, is.extreme) %>%
filter(is.extreme==TRUE)
semmst_merged_data_filtered_supp4_REC_conditions_bc_nooutlier = semmst_merged_data_filtered_supp4_REC_conditions_bc %>%
#filter(!ID %in% c("sub-526930", "sub-468051")) %>%
ungroup()
semmst_merged_supp4_REC_conditions_bc_AOV = semmst_merged_data_filtered_supp4_REC_conditions_bc_nooutlier %>%
droplevels() %>%
anova_test(dv = estimate, wid = ID, within = c(cope_name, task, roi_cluster))#, covariate = c(NonWord_sum_scale))
get_anova_table(semmst_merged_supp4_REC_conditions_bc_AOV)
# Effect DFn DFd F p p<.05 ges
## 1 cope_name 1 29 2.597 0.118 8.00e-03
## 2 task 1 29 0.236 0.631 5.39e-04
## 4 cope_name:task 1 29 4.762 0.037 * 2.50e-02
semmst_merged_data_filtered_supp4_REC_conditions_bc_nooutlier %>%
ggplot(aes(x=cope_name, fill=task, y=estimate)) +
geom_boxplot(position = position_dodge2()) +
facet_wrap(~roi_cluster)
# POST-HOC
# pairwise comparisons - stim_type * order
semmst_merged_supp4_REC_conditions_bc_PWC = semmst_merged_data_filtered_supp4_REC_conditions_bc_nooutlier %>%
group_by(task) %>%
pairwise_t_test(
estimate ~ cope_name,
paired = TRUE,
detailed = TRUE,
) %>%
ungroup() %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
ungroup() %>%
mutate(label = paste0("p=", signif(p.adj, 3)))
semmst_merged_supp4_REC_conditions_bc_PWC
## roi_cluster estimate .y. group1 group2 n1 n2 statistic p df conf.low conf.high method alternative p.adj
## 1 L_HEAD_DGCA23_HEAD_SUB_14 0.138 estimate FoilCR LureCR 30 30 3.43 0.002 29 0.0558 0.221 T-test two.sided 0.012 *
## 4 R_HEAD_DGCA23_HEAD_SUB_15 0.103 estimate FoilCR LureCR 30 30 3.05 0.005 29 0.0341 0.173 T-test two.sided 0.03 *
# pairwise comparisons - stim_type * order
semmst_merged_supp4_REC_conditions_bc_PWC = semmst_merged_data_filtered_supp4_REC_conditions_bc_nooutlier %>%
group_by(cope_name) %>%
mutate(task = forcats::fct_rev(task)) %>%
pairwise_t_test(
estimate ~ task,
paired = TRUE,
detailed = TRUE,
) %>%
ungroup() %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
ungroup() %>%
mutate(label = paste0("p=", signif(p.adj, 3)))
semmst_merged_supp4_REC_conditions_bc_PWC
```
#### Revision Figure 4 - REC status
```{r}
#### Revision Figure 4 - REC conditions, baseline-corrected, collapsed across ROI ----
figR4_dodge <- 0.8
Figure4_revision_rec_bc_nofacet_level_map_task <- c(
"RecognitionRespAllEncRepeat",
"RecognitionRespAllEncSimilar"
)
Figure4_revision_rec_bc_nofacet_level_lab_task <- c(
"Exact repeat",
"Modified repeat"
)
Figure4_revision_rec_bc_nofacet_level_map_cope <- c(
"TargetHIT",
"LureCR"
)
Figure4_revision_rec_bc_nofacet_level_lab_cope <- c(
"Target hit",
"Lure correct rejection"
)
# 1) Base plotting data
Figure4_revision_rec_bc_nofacet_base <- semmst_merged_data_filtered_supp4_REC_conditions_bc_nooutlier %>%
filter(
task %in% Figure4_revision_rec_bc_nofacet_level_map_task,
cope_name %in% Figure4_revision_rec_bc_nofacet_level_map_cope,
roi_cluster %in% c("L_HEAD_DGCA23_HEAD_SUB_14", "R_HEAD_DGCA23_HEAD_SUB_15")
) %>%
mutate(
task = factor(
task,
levels = Figure4_revision_rec_bc_nofacet_level_map_task,
labels = Figure4_revision_rec_bc_nofacet_level_lab_task
),
cope_name = factor(
cope_name,
levels = Figure4_revision_rec_bc_nofacet_level_map_cope,
labels = Figure4_revision_rec_bc_nofacet_level_lab_cope
)
)
# 2) Collapse across ROI within participant
Figure4_revision_rec_bc_nofacet_idmean <- Figure4_revision_rec_bc_nofacet_base %>%
group_by(ID, task, cope_name) %>%
summarise(
estimate = mean(estimate, na.rm = TRUE),
.groups = "drop"
)
# 3) Within-subject summary for bar plot + SE
Figure4_revision_rec_bc_nofacet_barsum <- Rmisc::summarySEwithin(
data = Figure4_revision_rec_bc_nofacet_idmean,
measurevar = "estimate",
withinvars = c("task", "cope_name"),
idvar = "ID",
na.rm = TRUE
)
# 4A) Stats: Target vs Lure within each encoding condition
Figure4_revision_rec_bc_nofacet_stats_recwithinenc <- Figure4_revision_rec_bc_nofacet_idmean %>%
group_by(task) %>%
pairwise_t_test(
estimate ~ cope_name,
paired = TRUE,
p.adjust.method = "bonferroni"
) %>%
ungroup() %>%
mutate(
label = case_when(
p.adj < 0.001 ~ "p < .001",
p.adj < 0.01 ~ "p < .01",
p.adj < 0.05 ~ "p < .05",
TRUE ~ paste0("p = ", sub("^0", "", sprintf("%.3f", p.adj)))
),
x_num = case_when(
task == "Exact repeat" ~ 1,
task == "Modified repeat" ~ 2
)
)
Figure4_revision_rec_bc_nofacet_ypos_task <- Figure4_revision_rec_bc_nofacet_barsum %>%
group_by(task) %>%
summarise(
y_base = max(estimate + se, na.rm = TRUE),
.groups = "drop"
)
Figure4_revision_rec_bc_nofacet_stats_recwithinenc <- Figure4_revision_rec_bc_nofacet_stats_recwithinenc %>%
left_join(Figure4_revision_rec_bc_nofacet_ypos_task, by = "task") %>%
mutate(
xmin = x_num - 0.20,
xmax = x_num + 0.20,
y.position = y_base + 0.030
)
# 4B) Reverse stats: Repeat vs Similar within each recognition condition
Figure4_revision_rec_bc_nofacet_stats_encwithinrec <- Figure4_revision_rec_bc_nofacet_idmean %>%
group_by(cope_name) %>%
pairwise_t_test(
estimate ~ task,
paired = TRUE,
p.adjust.method = "bonferroni"
) %>%
ungroup() %>%
mutate(
label = case_when(
p.adj < 0.001 ~ "p < .001",
p.adj < 0.01 ~ "p < .01",
p.adj < 0.05 ~ "p < .05",
TRUE ~ paste0("p = ", sub("^0", "", sprintf("%.3f", p.adj)))
)
)
Figure4_revision_rec_bc_nofacet_ypos_cope <- Figure4_revision_rec_bc_nofacet_barsum %>%
group_by(cope_name) %>%
summarise(
y_base = max(estimate + se, na.rm = TRUE),
.groups = "drop"
)
Figure4_revision_rec_bc_nofacet_stats_encwithinrec <- Figure4_revision_rec_bc_nofacet_stats_encwithinrec %>%
left_join(Figure4_revision_rec_bc_nofacet_ypos_cope, by = "cope_name") %>%
mutate(
xmin = case_when(
cope_name == "Target hit" ~ 1 - 0.20,
cope_name == "Lure correct rejection" ~ 1 + 0.20
),
xmax = case_when(
cope_name == "Target hit" ~ 2 - 0.20,
cope_name == "Lure correct rejection" ~ 2 + 0.20
),
y.position = case_when(
cope_name == "Target hit" ~ y_base + 0.090,
cope_name == "Lure correct rejection" ~ y_base + 0.140
)
)
# 5) Plot
Figure4_revision_rec_bc_nofacet_plot <- ggplot(
Figure4_revision_rec_bc_nofacet_barsum,
aes(x = task, y = estimate, fill = cope_name)
) +
geom_col(
position = position_dodge(width = figR4_dodge),
width = 0.65
) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
position = position_dodge(width = figR4_dodge),
width = 0.15,
linewidth = 1.2
) +
stat_pvalue_manual(
Figure4_revision_rec_bc_nofacet_stats_recwithinenc,
label = "label",
xmin = "xmin",
xmax = "xmax",
y.position = "y.position",
tip.length = 0.01,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.size = 1.0,
size = 5.0
) +
stat_pvalue_manual(
Figure4_revision_rec_bc_nofacet_stats_encwithinrec,
label = "label",
xmin = "xmin",
xmax = "xmax",
y.position = "y.position",
tip.length = 0.01,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.size = 1.0,
size = 5.0
) +
scale_fill_manual(values = c(
"Target hit" = "#66BB6A",
"Lure correct rejection" = "#9575CD"
)) +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.28))) +
labs(
x = "Encoding condition",
y = "Baseline-corrected mean % signal change",
fill = "Recognition condition"
) +
theme_classic(base_size = 13) +
theme(
panel.grid.minor = element_blank()
)
Figure4_revision_rec_bc_nofacet_plot
```
EXPORT
```{r}
text_base_export <- 40
p_text_export <- text_base_export * 0.28
Figure4_revision_rec_bc_nofacet_plot_export <- ggplot(
Figure4_revision_rec_bc_nofacet_barsum,
aes(x = task, y = estimate, fill = cope_name)
) +
geom_col(
position = position_dodge(width = figR4_dodge),
width = 0.65
) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
position = position_dodge(width = figR4_dodge),
width = 0.15,
linewidth = 2.0
) +
stat_pvalue_manual(
Figure4_revision_rec_bc_nofacet_stats_recwithinenc,
label = "label",
xmin = "xmin",
xmax = "xmax",
y.position = "y.position",
tip.length = 0.01,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.size = 1.3,
size = p_text_export
) +
stat_pvalue_manual(
Figure4_revision_rec_bc_nofacet_stats_encwithinrec,
label = "label",
xmin = "xmin",
xmax = "xmax",
y.position = "y.position",
tip.length = 0.01,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.size = 1.3,
size = p_text_export
) +
scale_fill_manual(values = c(
"Target hit" = "#66BB6A",
"Lure correct rejection" = "#9575CD"
)) +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.32))) +
labs(
x = "Encoding condition",
y = "Mean % signal change\nduring recognition",
fill = "Recognition condition"
) +
theme_classic(base_size = text_base_export) +
theme(
axis.text.x = element_text(size = 34, margin = margin(t = 4)),
axis.text.y = element_text(size = 30),
axis.title.x = element_text(size = 38, margin = margin(t = 12)),
axis.title.y = element_text(size = 34, margin = margin(r = 18)),
legend.position = c(0.82, 0.96),
legend.justification = c(0.5, 1),
legend.title = element_text(size = 30),
legend.text = element_text(size = 26),
legend.background = element_rect(fill = scales::alpha("white", 0), color = NA),
plot.margin = margin(t = 24, r = 20, b = 20, l = 70)
)
ggsave(
"derivatives/figures/SI_figures/Figure8SI.png",
plot = Figure4_revision_rec_bc_nofacet_plot_export,
width = 16, height = 12, units = "in",
dpi = 300,
bg = "white",
device = "png"
)
```
### Revised Supplementary Figure 9 - Lure CR vs Lure FA
```{r}
## Figure H4 - Recognition specificity (Lure CR vs Lure FA)
# 1) Within-subject summary
FigureH4_barsum <- Rmisc::summarySEwithin(
data = semmst_merged_data_H4_lureFACR_nooutlier,
measurevar = "estimate",
withinvars = c("cope_name", "roi_cluster"),
idvar = "ID",
na.rm = TRUE
) %>%
mutate(
cope_name = factor(
as.character(cope_name),
levels = c("LureCR", "LureFA"),
labels = c("Correct rejection", "False alarm")
)
)
# 2) ROI labels
FigureH4_roi_lab <- semmst_merged_data_H4_lureFACR_nooutlier %>%
dplyr::distinct(roi_cluster) %>%
dplyr::mutate(
roi_label = dplyr::case_when(
roi_cluster == "L_HEAD_DGCA23_HEAD_SUB_14" ~ "Left hippocampal head cluster",
roi_cluster == "R_HEAD_DGCA23_HEAD_SUB_15" ~ "Right hippocampal head cluster",
TRUE ~ as.character(roi_cluster)
)
)
FigureH4_barsum <- FigureH4_barsum %>%
dplyr::left_join(FigureH4_roi_lab, by = "roi_cluster")
# 3) Plot
FigureH4_lure_col <- "#9575CD"
FigureH4_plot <- ggplot(
FigureH4_barsum,
aes(x = cope_name, y = estimate, alpha = cope_name)
) +
geom_col(
fill = FigureH4_lure_col,
width = 0.65,
show.legend = FALSE
) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
width = 0.15,
linewidth = 1.2
) +
facet_wrap(~ roi_label) +
theme_classic(base_size = 13) +
labs(
x = "Response during recognition",
y = "Mean % signal change"
) +
scale_alpha_manual(values = c(
"Correct rejection" = 0.95,
"False alarm" = 0.55
)) +
scale_y_continuous(expand = expansion(mult = c(0.02, 0.02))) +
custom_theme +
guides(alpha = "none")
FigureH4_plot
```
EXPORT
```{r}
text_base_export <- 40
FigureH4_plot_export <- FigureH4_plot +
theme_classic(base_size = text_base_export) +
theme(
axis.text.x = element_text(size = text_base_export * 0.82, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05),
strip.text = element_text(size = text_base_export * 0.82, face = "bold"),
panel.spacing.x = unit(0.8, "lines")
)
ggsave(
"derivatives/figures/SI_figures/Figure9SI.png",
plot = FigureH4_plot_export,
width = 20, height = 12, units = "in",
dpi = 300,
bg = "white",
device = "png"
)
```
# Revision Results
### Revision Results 1 - Encoding trial-wise RS (Response to reviewers)
```{r}
# 1) Neural data: keep all EncodingLSA* tasks, extract itemno from cope_name
semmst_merged_data_supplement_filtered_results3 = semmst_merged_data_supplement_filtered %>%
filter(stringr::str_detect(task, "^EncodingLSA")) %>%
mutate(
itemno = stringr::str_extract(as.character(cope_name), "\\d+"),
itemno = as.character(itemno)
) %>%
filter(!is.na(itemno)) %>%
dplyr::select(ID, task, roi_cluster, estimate, itemno) %>%
distinct(ID, roi_cluster, itemno, .keep_all = TRUE)
# 2) Encoding item-level behavioral data: only order == 2
semmst_behav_data_enc_itemwise_GLMEM <- semmst_behav_data_filtered %>%
filter(task == "enc") %>%
filter(trial_type != "FILLER") %>%
mutate(
across(
all_of(c("Vocabulary", "Digit_Symbol", "NonWord_sum",
"arousal", "concreteness", "imageability", "meaningfulness")),
~ as.numeric(scale(.x)),
.names = "{.col}_scale"
),
itemno = as.character(itemno),
order = as.character(order),
trial_type = factor(trial_type)
) %>%
filter(order == "2") %>%
drop_na(Vocabulary, itemno, cosine, trial_type) %>%
distinct(ID, itemno, .keep_all = TRUE)
# 3) Merge neural + encoding behavior by ID and itemno
semmst_merged_data_supplement_filtered_results3_glmem <- semmst_merged_data_supplement_filtered_results3 %>%
left_join(
semmst_behav_data_enc_itemwise_GLMEM %>%
dplyr::select(
ID, itemno,
noun, trial_type, cosine, order,
age, sex, education,
Vocabulary_scale, Digit_Symbol_scale, NonWord_sum_scale,
arousal_scale, meaningfulness_scale, concreteness_scale
),
by = c("ID", "itemno")
) %>%
drop_na(estimate, trial_type, cosine)
# quick checks
semmst_merged_data_supplement_filtered_results3_glmem %>%
count(roi_cluster, trial_type)
semmst_merged_data_supplement_filtered_results3_glmem %>%
dplyr::select(ID, itemno, task, roi_cluster, estimate, noun, trial_type, cosine, order) %>%
print()
```
Models
```{r}
model_simple_cosine_results3 <- lmer(
estimate ~ cosine * roi_cluster * trial_type +
(1 | ID) +
(1 | itemno),
data = semmst_merged_data_supplement_filtered_results3_glmem
)
model_complex_cosine_results3 <- lmer(
estimate ~ cosine * trial_type * roi_cluster +
Vocabulary_scale + Digit_Symbol_scale + NonWord_sum_scale +
arousal_scale + meaningfulness_scale + concreteness_scale +
(1 | ID) +
(1 | itemno),
data = semmst_merged_data_supplement_filtered_results3_glmem,
REML = FALSE,
control = lmerControl(
optimizer = "bobyqa",
optCtrl = list(maxfun = 20000)
)
)
anova(model_simple_cosine_results3, model_complex_cosine_results3)
summary(model_simple_cosine_results3)
```
#### Revision Figure 1
```{r}
# In this plot, we predict estimate by cosine similarity
FigureR1A_continuous_scatter_roi_base <- semmst_merged_data_supplement_filtered_results3_glmem %>%
filter(
roi_cluster %in% c("L_HEAD_DGCA23_HEAD_SUB_14", "R_HEAD_DGCA23_HEAD_SUB_15"),
trial_type %in% c("CLOSE", "DISTANT"),
!is.na(cosine),
!is.na(estimate)
) %>%
mutate(
trial_type = factor(trial_type, levels = c("CLOSE", "DISTANT")),
roi_label = case_when(
roi_cluster == "L_HEAD_DGCA23_HEAD_SUB_14" ~ "Left HC head cluster",
roi_cluster == "R_HEAD_DGCA23_HEAD_SUB_15" ~ "Right HC head cluster",
TRUE ~ as.character(roi_cluster)
),
roi_label = factor(
roi_label,
levels = c("Left HC head cluster", "Right HC head cluster")
)
)
# Okabe–Ito (CVD-safe): blue, yellow
FigureR1A_continuous_scatter_roi_cols <- c(
"CLOSE" = "#4477AA",
"DISTANT" = "#DDCC77"
)
FigureR1A_continuous_scatter_roi_plot <- ggplot(
FigureR1A_continuous_scatter_roi_base,
aes(x = cosine, y = estimate, color = trial_type, fill = trial_type)
) +
geom_point(
alpha = 0.30,
size = 1.8
) +
geom_smooth(
aes(group = trial_type),
method = "lm",
linewidth = 2,
se = TRUE,
alpha = 0.20
) +
facet_wrap(~ roi_label, ncol = 1) +
scale_x_continuous(
breaks = function(lims) round(lims[1] + c(0.1, 0.5, 0.9) * (lims[2] - lims[1]), 2),
labels = function(x) sprintf("%.2f", x)
) +
scale_color_manual(values = FigureR1A_continuous_scatter_roi_cols) +
scale_fill_manual(values = FigureR1A_continuous_scatter_roi_cols) +
labs(
x = "Cosine similarity",
y = "Mean % signal change",
color = "Encoding trial type",
fill = "Encoding trial type"
) +
theme_classic(base_size = 14) +
theme(
panel.grid.minor = element_blank(),
strip.text = element_text(face = "bold"),
legend.position = c(0.87, 0.88)
)
FigureR1A_continuous_scatter_roi_plot
```
##### EXPORT
```{r}
FigureR1A_continuous_scatter_roi_plot_export <- ggplot(
FigureR1A_continuous_scatter_roi_base,
aes(x = cosine, y = estimate, color = trial_type, fill = trial_type)
) +
geom_point(
alpha = 0.30,
size = 3.8
) +
geom_smooth(
aes(group = trial_type),
method = "lm",
linewidth = 3.2,
se = TRUE,
alpha = 0.20
) +
facet_wrap(~ roi_label, ncol = 1) +
scale_x_continuous(
breaks = function(lims) round(lims[1] + c(0.1, 0.5, 0.9) * (lims[2] - lims[1]), 2),
labels = function(x) sprintf("%.2f", x)
) +
scale_color_manual(values = FigureR1A_continuous_scatter_roi_cols) +
scale_fill_manual(values = FigureR1A_continuous_scatter_roi_cols) +
labs(
x = "Cosine similarity",
y = "Mean % signal change"
) +
theme_classic(base_size = 32) +
theme(
axis.title = element_text(size = 34),
axis.text = element_text(size = 24),
strip.text = element_text(size = 24, face = "bold"),
legend.position = "none"
)
ggsave(
"derivatives/figures/revision_figures/FigureR1A.png",
plot = FigureR1A_continuous_scatter_roi_plot_export,
width = 12, height = 18, units = "in",
dpi = 300,
bg = "white",
device = "png"
)
```
#### Revision Figure 1B
```{r}
FigureR1B_itemwise_base <- semmst_merged_data_supplement_filtered_results3_glmem %>%
filter(
roi_cluster %in% c("L_HEAD_DGCA23_HEAD_SUB_14", "R_HEAD_DGCA23_HEAD_SUB_15"),
trial_type %in% c("CLOSE", "DISTANT"),
!is.na(itemno),
!is.na(trial_type),
!is.na(cosine),
!is.na(estimate)
) %>%
mutate(
trial_type = factor(trial_type, levels = c("CLOSE", "DISTANT")),
roi_label = case_when(
roi_cluster == "L_HEAD_DGCA23_HEAD_SUB_14" ~ "Left HC head cluster",
roi_cluster == "R_HEAD_DGCA23_HEAD_SUB_15" ~ "Right HC head cluster",
TRUE ~ as.character(roi_cluster)
),
roi_label = factor(
roi_label,
levels = c("Left HC head cluster", "Right HC head cluster")
),
# Treat CLOSE and DISTANT versions of the same item separately
item_trial = paste(itemno, trial_type, sep = "_")
)
# Order x-axis categories by mean cosine similarity
FigureR1B_item_order <- FigureR1B_itemwise_base %>%
group_by(itemno, trial_type, item_trial) %>%
summarise(
mean_cosine = mean(cosine, na.rm = TRUE),
.groups = "drop"
) %>%
arrange(mean_cosine, itemno, trial_type) %>%
pull(item_trial)
FigureR1B_itemwise_base <- FigureR1B_itemwise_base %>%
mutate(
item_trial = factor(item_trial, levels = FigureR1B_item_order)
)
# Colors
FigureR1B_itemwise_cols <- c(
"CLOSE" = "#4477AA",
"DISTANT" = "#DDCC77"
)
## PLOT
FigureR1B_itemwise_plot <- ggplot(
FigureR1B_itemwise_base,
aes(x = item_trial, y = estimate, fill = trial_type, color = trial_type)
) +
geom_boxplot(
width = 0.70,
alpha = 0.45,
outlier.shape = NA
) +
geom_point(
aes(group = item_trial),
position = position_jitter(width = 0.16, height = 0),
alpha = 0.65,
size = 1.6
) +
facet_wrap(~ roi_label, ncol = 1) +
scale_fill_manual(values = FigureR1B_itemwise_cols) +
scale_color_manual(values = FigureR1B_itemwise_cols) +
labs(
x = "Items in ascending order of cosine similarity",
y = "Mean % signal change",
fill = "Encoding trial type",
color = "Encoding trial type"
) +
theme_classic(base_size = 14) +
theme(
panel.grid.minor = element_blank(),
strip.text = element_text(face = "bold"),
axis.text.x = element_blank(),
axis.ticks.x = element_blank(),
legend.position = "top"
)
FigureR1B_itemwise_plot
```
```{r}
FigureR1B_itemwise_plot_export <- ggplot(
FigureR1B_itemwise_base,
aes(x = item_trial, y = estimate, fill = trial_type, color = trial_type)
) +
geom_boxplot(
width = 0.72,
alpha = 0.45,
outlier.shape = NA,
linewidth = 1.1
) +
geom_point(
aes(group = item_trial),
position = position_jitter(width = 0.16, height = 0),
alpha = 0.70,
size = 2.8
) +
facet_wrap(~ roi_label, ncol = 1) +
scale_fill_manual(values = FigureR1B_itemwise_cols) +
scale_color_manual(values = FigureR1B_itemwise_cols) +
labs(
x = "Items in ascending order of cosine similarity",
y = "Mean % signal change"
) +
theme_classic(base_size = 32) +
theme(
axis.title = element_text(size = 34),
axis.text.x = element_blank(),
axis.ticks.x = element_blank(),
axis.text.y = element_text(size = 24),
strip.text = element_text(size = 24, face = "bold"),
legend.position = "none"
)
ggsave(
"derivatives/figures/revision_figures/FigureR1B.png",
plot = FigureR1B_itemwise_plot_export,
width = 16, height = 18, units = "in",
dpi = 300,
bg = "white",
device = "png"
)
```
### Revision Results 2 - yes vs no (Response to reviewers)
#### Behavioural yes-no results
```{r}
# Behaviour
semmst_behav_data_filtered_revision_results2A = semmst_behav_data_filtered %>%
filter(task == "enc") %>%
filter(trial_type != "FILLER") %>%
drop_na(response, order) %>%
mutate(
trial_type = case_when(
trial_type %in% c("CLOSE", "DISTANT") ~ "LURE",
TRUE ~ as.character(trial_type)
),
trial_type = factor(trial_type, levels = c("REPEAT", "LURE")),
response = tolower(as.character(response))
) %>%
filter(response %in% c("no-good", "good")) %>%
mutate(
response = factor(response, levels = c("no-good", "good")),
order = factor(order, levels = c(1, 2))
) %>%
droplevels()
# Percent response
semmst_behav_data_filtered_revision_results2A_good = semmst_behav_data_filtered_revision_results2A %>%
count(ID, trial_type, order, response, name = "n") %>%
tidyr::complete(ID, trial_type, order, response, fill = list(n = 0)) %>%
group_by(ID, trial_type, order) %>%
mutate(
percent_response = 100 * n / sum(n)
) %>%
ungroup() %>%
filter(response == "good") %>%
droplevels()
# OUTLIERS
semmst_behav_data_filtered_revision_results2A_good %>%
group_by(trial_type, order) %>%
identify_outliers(percent_response) %>%
dplyr::select(ID, trial_type, order, is.outlier, is.extreme) %>%
filter(is.extreme == TRUE)
# NORMALITY
semmst_behav_data_filtered_revision_results2A_good %>%
group_by(trial_type, order) %>%
shapiro_test(percent_response)
## ANOVA
semmst_behav_data_filtered_revision_results2A_AOV = semmst_behav_data_filtered_revision_results2A_good %>%
anova_test(
dv = percent_response,
wid = ID,
within = c(trial_type, order)
)
get_anova_table(semmst_behav_data_filtered_revision_results2A_AOV)
## POST-HOC
semmst_behav_data_filtered_revision_results2A_PWC_response = semmst_behav_data_filtered_revision_results2A_good %>%
group_by(trial_type) %>%
pairwise_t_test(
percent_response ~ order,
paired = TRUE,
detailed = TRUE
) %>%
ungroup() %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
mutate(label = paste0("p=", signif(p.adj, 3)))
semmst_behav_data_filtered_revision_results2A_PWC_response
```
#### Neural results
```{r}
# Prepare data
semmst_merged_data_revision_results2B = semmst_merged_data_supplement_filtered %>%
filter(task == "EncodingResp") %>%
filter(cope_name %in% c("YesFirst", "NoFirst", "YesRepeat", "NoRepeat")) %>%
filter(roi_cluster %in% c("L_HEAD_DGCA23_HEAD_SUB_14", "R_HEAD_DGCA23_HEAD_SUB_15")) %>%
filter(estimate != 0) %>%
extract(
col = cope_name,
into = c("response_type", "order"),
regex = "^(Yes|No)(First|Repeat)$"
) %>%
mutate(
response_type = factor(response_type, levels = c("No", "Yes")),
order = factor(order, levels = c("First", "Repeat"))
) %>%
droplevels()
# Keep only IDs with complete 2 x 2 x 2 cells
complete_ids_revision_results2B <- semmst_merged_data_revision_results2B %>%
count(ID, response_type, order, roi_cluster) %>%
tidyr::complete(
ID,
response_type = levels(semmst_merged_data_revision_results2B$response_type),
order = levels(semmst_merged_data_revision_results2B$order),
roi_cluster = unique(semmst_merged_data_revision_results2B$roi_cluster),
fill = list(n = 0)
) %>%
group_by(ID) %>%
summarise(all_cells_present = all(n > 0), .groups = "drop") %>%
filter(all_cells_present) %>%
pull(ID)
semmst_merged_data_revision_results2B = semmst_merged_data_revision_results2B %>%
filter(ID %in% complete_ids_revision_results2B) %>%
droplevels()
# OUTLIER
semmst_merged_data_revision_results2B %>%
group_by(response_type, order, roi_cluster) %>%
identify_outliers(estimate) %>%
dplyr::select(ID, roi_cluster, response_type, order, is.outlier, is.extreme) %>%
filter(is.extreme == TRUE)
# Exclude outliers
semmst_merged_data_revision_results2B_nooutlier = semmst_merged_data_revision_results2B %>%
filter(!ID %in% c("sub-434971")) %>%
ungroup()
# NORMALITY
semmst_merged_data_revision_results2B_nooutlier %>%
group_by(response_type, order, roi_cluster) %>%
shapiro_test(estimate) %>%
print(n = 100)
## ANOVA
semmst_merged_data_revision_results2B_AOV = semmst_merged_data_revision_results2B_nooutlier %>%
anova_test(
dv = estimate,
wid = ID,
within = c(response_type, order, roi_cluster)
)
get_anova_table(semmst_merged_data_revision_results2B_AOV)
## POST-HOC
semmst_merged_data_revision_results2B_PWC = semmst_merged_data_revision_results2B_nooutlier %>%
group_by(ID, order, response_type) %>%
summarise(
estimate = mean(estimate, na.rm = TRUE),
.groups = "drop"
) %>%
tidyr::pivot_wider(
names_from = response_type,
values_from = estimate
) %>%
drop_na(No, Yes) %>%
tidyr::pivot_longer(
cols = c(No, Yes),
names_to = "response_type",
values_to = "estimate"
) %>%
mutate(response_type = factor(response_type, levels = c("No", "Yes"))) %>%
group_by(response_type) %>%
pairwise_t_test(
estimate ~ order,
paired = TRUE,
detailed = TRUE
) %>%
ungroup() %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
mutate(label = paste0("p=", signif(p.adj, 3)))
semmst_merged_data_revision_results2B_PWC
```
#### Neural - behavioural correlation
```{r}
# Participant-specific "yes" rate for LURE, second presentation
revision_results2C_behav_lure_yesrate <- semmst_behav_data_filtered_revision_results2A_good %>%
filter(
trial_type == "LURE",
order == 2
) %>%
transmute(
ID,
lure_yes_rate = percent_response
) %>%
droplevels()
### Correlation - LURE second presentation
# 1) Filter / define roi_group
revision_results2C_corr_input_all <- semmst_merged_data_H3_neural_ps %>%
mutate(roi_group = roi_cluster) %>%
dplyr::select(
ID, roi_group, roi, hemisphere, cluster, voxels,
LureRepeat,
NonWord_sum_scale, Digit_Symbol_scale
) %>%
left_join(revision_results2C_behav_lure_yesrate, by = "ID") %>%
dplyr::select(
ID, roi_group, roi, hemisphere, cluster, voxels,
LureRepeat, lure_yes_rate,
NonWord_sum_scale, Digit_Symbol_scale
) %>%
ungroup()
# 2) Mahalanobis outlier detection (per ROI)
alpha <- 0.99 # cutoff for chi-square (df=2)
revision_results2C_md_tbl_all <- revision_results2C_corr_input_all %>%
dplyr::group_by(roi_group) %>%
dplyr::group_modify(~{
d <- .x %>%
dplyr::select(ID, LureRepeat, lure_yes_rate) %>%
tidyr::drop_na()
if (nrow(d) < 3) {
return(tibble::tibble(
ID = character(),
MD = numeric(),
p = numeric(),
cutoff = numeric(),
is_outlier = logical()
))
}
X <- as.matrix(d[, c("LureRepeat", "lure_yes_rate")])
center <- colMeans(X)
covmat <- stats::cov(X)
md <- stats::mahalanobis(X, center = center, cov = covmat)
tibble::tibble(
ID = d$ID,
MD = md,
p = 1 - stats::pchisq(md, df = 2),
cutoff = stats::qchisq(alpha, df = 2),
is_outlier = md > stats::qchisq(alpha, df = 2)
)
}) %>%
ungroup()
revision_results2C_md_outliers_all <- revision_results2C_md_tbl_all %>%
dplyr::filter(is_outlier)
revision_results2C_md_multi_all <- revision_results2C_md_outliers_all %>%
dplyr::count(ID, sort = TRUE) %>%
dplyr::filter(n > 1)
revision_results2C_md_tbl_all
revision_results2C_md_outliers_all
revision_results2C_md_multi_all
# 3) Remove outlier ID
revision_results2C_corr_input_all_nooutlier <- revision_results2C_corr_input_all %>%
dplyr::filter(ID != "sub-434971") %>%
droplevels() %>%
ungroup()
# 4) Fisher-z CI for Pearson r (95% hard-coded)
pearson_ci <- function(r, n) {
z <- atanh(r)
se <- 1 / sqrt(n - 3)
zcrit <- stats::qnorm(1 - (1 - 0.95) / 2)
c(low = tanh(z - zcrit * se), high = tanh(z + zcrit * se))
}
# 5) SIMPLE correlations only (Pearson) per ROI + Bonferroni
revision_results2C_corr_lure_yesrate_simple <- revision_results2C_corr_input_all_nooutlier %>%
dplyr::group_by(roi_group) %>%
dplyr::group_modify(~{
d <- .x %>%
dplyr::select(LureRepeat, lure_yes_rate) %>%
tidyr::drop_na()
ct <- stats::cor.test(d$LureRepeat, d$lure_yes_rate, method = "pearson")
r <- unname(ct$estimate)
p <- ct$p.value
n <- nrow(d)
ci <- pearson_ci(r, n)
tibble::tibble(
n = n,
r = r,
p = p,
ci = sprintf("[%.3f, %.3f]", ci["low"], ci["high"])
)
}) %>%
ungroup() %>%
mutate(
p_adj = p.adjust(p, method = "bonferroni")
) %>%
arrange(p_adj)
revision_results2C_corr_lure_yesrate_simple
```
#### Revision Figure 2A - Behavioural revision
```{r}
## BEHAVIOURAL FIGURE ------------------------------------------------------------
FigureR2A_dodge <- 0.8
# 1) Within-subject summary
FigureR2A_barsum <- Rmisc::summarySEwithin(
data = semmst_behav_data_filtered_revision_results2A_good,
measurevar = "percent_response",
withinvars = c("trial_type", "order"),
idvar = "ID",
na.rm = TRUE
) %>%
mutate(
trial_type = factor(
as.character(trial_type),
levels = c("REPEAT", "LURE"),
labels = c("Exact repeat", "Modified repeat")
),
order = factor(
as.character(order),
levels = c("1", "2"),
labels = c("First", "Repeat")
)
)
# 2) Only theoretically matched pairwise post-hocs
FigureR2A_stats_all <- semmst_behav_data_filtered_revision_results2A_good %>%
mutate(
cond = factor(
paste(as.character(trial_type), as.character(order), sep = "_"),
levels = c("REPEAT_1", "REPEAT_2", "LURE_1", "LURE_2")
)
) %>%
pairwise_t_test(
percent_response ~ cond,
paired = TRUE,
detailed = TRUE
) %>%
adjust_pvalue(method = "bonferroni") %>%
add_significance() %>%
mutate(
comparison = paste(as.character(group1), as.character(group2), sep = "__"),
label = case_when(
p.adj < 0.001 ~ "p < .001",
p.adj < 0.05 ~ paste0("p = ", sub("^0", "", sprintf("%.3f", p.adj))),
TRUE ~ NA_character_
)
) %>%
filter(comparison %in% c(
"REPEAT_1__REPEAT_2",
"LURE_1__LURE_2",
"REPEAT_1__LURE_1",
"REPEAT_2__LURE_2"
))
# 3) Map comparisons onto dodged bar centers
FigureR2A_xpos <- c(
"REPEAT_1" = 1 - FigureR2A_dodge / 4,
"REPEAT_2" = 1 + FigureR2A_dodge / 4,
"LURE_1" = 2 - FigureR2A_dodge / 4,
"LURE_2" = 2 + FigureR2A_dodge / 4
)
FigureR2A_y_base <- max(
FigureR2A_barsum$percent_response + FigureR2A_barsum$se,
na.rm = TRUE
)
FigureR2A_y_step <- 6
FigureR2A_stats_all <- FigureR2A_stats_all %>%
mutate(
xmin_plot = unname(FigureR2A_xpos[as.character(group1)]),
xmax_plot = unname(FigureR2A_xpos[as.character(group2)]),
y.position = case_when(
comparison == "REPEAT_1__REPEAT_2" ~ FigureR2A_y_base + FigureR2A_y_step * 1,
comparison == "LURE_1__LURE_2" ~ FigureR2A_y_base + FigureR2A_y_step * 2,
comparison == "REPEAT_1__LURE_1" ~ FigureR2A_y_base + FigureR2A_y_step * 3,
comparison == "REPEAT_2__LURE_2" ~ FigureR2A_y_base + FigureR2A_y_step * 4
)
) %>%
ungroup()
# only draw significant matched comparisons
FigureR2A_stats_draw <- FigureR2A_stats_all %>%
filter(!is.na(label), p.adj < 0.05) %>%
as.data.frame()
# 4) y-axis upper limit for brackets, but only label up to 100
FigureR2A_y_top <- max(
105,
FigureR2A_stats_draw$y.position,
na.rm = TRUE
)
# 5) Base plot
FigureR2A_plot <- ggplot(
FigureR2A_barsum,
aes(
x = trial_type,
y = percent_response,
fill = trial_type,
group = order
)
) +
geom_col(
aes(alpha = order),
position = position_dodge(width = FigureR2A_dodge),
width = 0.65
) +
geom_errorbar(
aes(ymin = percent_response - se, ymax = percent_response + se),
position = position_dodge(width = FigureR2A_dodge),
width = 0.15,
linewidth = 1.2
) +
ggpubr::stat_pvalue_manual(
FigureR2A_stats_draw,
label = "label",
xmin = "xmin_plot",
xmax = "xmax_plot",
y.position = "y.position",
tip.length = 0.01,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.shorten = 0,
bracket.size = 1.0,
size = 5.0,
step.increase = 0
) +
theme_classic(base_size = 13) +
labs(
x = "Condition during encoding",
y = "Percentage of 'accept' responses",
fill = "Trial type",
alpha = "Order"
) +
scale_fill_manual(values = c(
"Exact repeat" = "#66BB6A",
"Modified repeat" = "#9575CD"
)) +
scale_alpha_manual(values = c(
"First" = 0.95,
"Repeat" = 0.55
)) +
scale_y_continuous(
limits = c(0, FigureR2A_y_top),
breaks = c(0, 25, 50, 75, 100),
labels = c("0", "25", "50", "75", "100"),
expand = expansion(mult = c(0.02, 0))
) +
custom_theme +
guides(fill = "none") +
theme(
legend.position = c(0.05, 0.98),
legend.justification = c(0, 1),
legend.background = element_rect(fill = scales::alpha("white", 0), color = NA)
)
FigureR2A_plot
```
```{r}
# 6) Export
text_base_export <- 40
FigureR2A_plot_export <- ggplot(
FigureR2A_barsum,
aes(
x = trial_type,
y = percent_response,
fill = trial_type,
group = order
)
) +
geom_col(
aes(alpha = order),
position = position_dodge(width = FigureR2A_dodge),
width = 0.65
) +
geom_errorbar(
aes(ymin = percent_response - se, ymax = percent_response + se),
position = position_dodge(width = FigureR2A_dodge),
width = 0.15,
linewidth = 2.0
) +
ggpubr::stat_pvalue_manual(
FigureR2A_stats_draw,
label = "label",
xmin = "xmin_plot",
xmax = "xmax_plot",
y.position = "y.position",
tip.length = 0.015,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.shorten = 0,
bracket.size = 1.8,
size = text_base_export * 0.24,
step.increase = 0
) +
labs(
x = "Condition during encoding",
y = "Percentage of 'accept' responses",
fill = "Trial type",
alpha = "Order"
) +
scale_fill_manual(values = c(
"Exact repeat" = "#66BB6A",
"Modified repeat" = "#9575CD"
)) +
scale_alpha_manual(values = c(
"First" = 0.95,
"Repeat" = 0.55
)) +
scale_y_continuous(
limits = c(0, FigureR2A_y_top),
breaks = c(0, 25, 50, 75, 100),
labels = c("0", "25", "50", "75", "100"),
expand = expansion(mult = c(0.02, 0))
) +
custom_theme +
guides(fill = "none") +
theme_classic(base_size = text_base_export) +
theme(
legend.position = c(0.05, 0.98),
legend.justification = c(0, 1),
legend.background = element_rect(fill = scales::alpha("white", 0), color = NA),
legend.title = element_text(size = text_base_export * 0.82),
legend.text = element_text(size = text_base_export * 0.78),
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05)
)
ggsave(
"derivatives/figures/revision_figures/FigureR2A.png",
plot = FigureR2A_plot_export,
width = 16, height = 12, units = "in",
dpi = 300,
bg = "white",
device = "png"
)
```
#### Revision Figure 2B - Nerual results
```{r}
## NEURAL FIGURE ----------------------------------------------------------------
FigureR2B_dodge <- 0.8
# 1) Collapse across the two HC head clusters (to match the post-hoc tests)
FigureR2B_plotdata <- semmst_merged_data_revision_results2B_nooutlier %>%
group_by(ID, response_type, order) %>%
summarise(
estimate = mean(estimate, na.rm = TRUE),
.groups = "drop"
) %>%
mutate(
response_type = factor(
as.character(response_type),
levels = c("No", "Yes"),
labels = c("Not accept", "Accept")
),
order = factor(
as.character(order),
levels = c("First", "Repeat"),
labels = c("First", "Repeat")
)
)
# 2) Within-subject summary
FigureR2B_barsum <- Rmisc::summarySEwithin(
data = FigureR2B_plotdata,
measurevar = "estimate",
withinvars = c("response_type", "order"),
idvar = "ID",
na.rm = TRUE
)
# 3) Planned comparison: Not accept vs Accept during FIRST presentation only
FigureR2B_first_comp <- FigureR2B_plotdata %>%
filter(order == "First") %>%
tidyr::pivot_wider(
names_from = response_type,
values_from = estimate
) %>%
drop_na(`Not accept`, Accept)
FigureR2B_first_ttest <- t.test(
FigureR2B_first_comp$`Not accept`,
FigureR2B_first_comp$Accept,
paired = TRUE
)
FigureR2B_first_p <- FigureR2B_first_ttest$p.value
# 4) Bracket coordinates
FigureR2B_xpos <- c(
"Not accept_First" = 1 - FigureR2B_dodge / 4,
"Not accept_Repeat" = 1 + FigureR2B_dodge / 4,
"Accept_First" = 2 - FigureR2B_dodge / 4,
"Accept_Repeat" = 2 + FigureR2B_dodge / 4
)
FigureR2B_y_base <- max(
FigureR2B_barsum$estimate + FigureR2B_barsum$se,
na.rm = TRUE
)
FigureR2B_y_step <- 0.025
FigureR2B_stats_draw <- tibble::tibble(
group1 = "Not accept_First",
group2 = "Accept_First",
xmin_plot = unname(FigureR2B_xpos["Not accept_First"]),
xmax_plot = unname(FigureR2B_xpos["Accept_First"]),
y.position = FigureR2B_y_base + FigureR2B_y_step,
label = case_when(
FigureR2B_first_p < 0.001 ~ "p < .001",
FigureR2B_first_p < 0.1 ~ paste0("p = ", sub("^0", "", sprintf("%.3f", FigureR2B_first_p))),
TRUE ~ NA_character_
)
) %>%
filter(!is.na(label))
# 5) y-axis upper limit for bracket
FigureR2B_y_top <- max(
c(
FigureR2B_barsum$estimate + FigureR2B_barsum$se,
FigureR2B_stats_draw$y.position
),
na.rm = TRUE
) + 0.01
# 6) Base plot
FigureR2B_plot <- ggplot(
FigureR2B_barsum,
aes(
x = response_type,
y = estimate,
fill = response_type,
group = order
)
) +
geom_col(
aes(alpha = order),
position = position_dodge(width = FigureR2B_dodge),
width = 0.65
) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
position = position_dodge(width = FigureR2B_dodge),
width = 0.15,
linewidth = 1.2
) +
theme_classic(base_size = 13) +
labs(
x = "Response during encoding",
y = "Mean % signal change",
fill = "Response type",
alpha = "Order"
) +
scale_fill_manual(values = c(
"Accept" = "#4477AA",
"Not accept" = "#CC6677"
)) +
scale_alpha_manual(values = c(
"First" = 0.95,
"Repeat" = 0.55
)) +
scale_y_continuous(
limits = c(
min(0, min(FigureR2B_barsum$estimate - FigureR2B_barsum$se, na.rm = TRUE)),
FigureR2B_y_top
),
expand = expansion(mult = c(0.02, 0))
) +
custom_theme +
guides(fill = "none") +
theme(
legend.position = c(0.75, 0.98),
legend.justification = c(0, 1),
legend.background = element_rect(fill = scales::alpha("white", 0), color = NA)
)
# Add the single planned bracket only if p < .1
if (nrow(FigureR2B_stats_draw) > 0) {
FigureR2B_plot <- FigureR2B_plot +
ggpubr::stat_pvalue_manual(
FigureR2B_stats_draw,
label = "label",
xmin = "xmin_plot",
xmax = "xmax_plot",
y.position = "y.position",
tip.length = 0.01,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.shorten = 0,
bracket.size = 1.0,
size = 5.0
)
}
FigureR2B_plot
```
```{r}
# 7) Export
text_base_export <- 40
FigureR2B_plot_export <- ggplot(
FigureR2B_barsum,
aes(
x = response_type,
y = estimate,
fill = response_type,
group = order
)
) +
geom_col(
aes(alpha = order),
position = position_dodge(width = FigureR2B_dodge),
width = 0.65
) +
geom_errorbar(
aes(ymin = estimate - se, ymax = estimate + se),
position = position_dodge(width = FigureR2B_dodge),
width = 0.15,
linewidth = 2.0
) +
labs(
x = "Response during encoding",
y = "Mean % signal change",
fill = "Response type",
alpha = "Order"
) +
scale_fill_manual(values = c(
"Accept" = "#4477AA",
"Not accept" = "#CC6677"
)) +
scale_alpha_manual(values = c(
"First" = 0.95,
"Repeat" = 0.55
)) +
scale_y_continuous(
limits = c(
min(0, min(FigureR2B_barsum$estimate - FigureR2B_barsum$se, na.rm = TRUE)),
FigureR2B_y_top
),
expand = expansion(mult = c(0.02, 0))
) +
custom_theme +
guides(fill = "none") +
theme_classic(base_size = text_base_export) +
theme(
legend.position = c(0.75, 0.98),
legend.justification = c(0, 1),
legend.background = element_rect(fill = scales::alpha("white", 0), color = NA),
legend.title = element_text(size = text_base_export * 0.82),
legend.text = element_text(size = text_base_export * 0.78),
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05)
)
# Add the single planned bracket only if p < .1
if (nrow(FigureR2B_stats_draw) > 0) {
FigureR2B_plot_export <- FigureR2B_plot_export +
ggpubr::stat_pvalue_manual(
FigureR2B_stats_draw,
label = "label",
xmin = "xmin_plot",
xmax = "xmax_plot",
y.position = "y.position",
tip.length = 0.015,
hide.ns = TRUE,
inherit.aes = FALSE,
bracket.shorten = 0,
bracket.size = 1.8,
size = text_base_export * 0.24
)
}
ggsave(
"derivatives/figures/revision_figures/FigureR2B.png",
plot = FigureR2B_plot_export,
width = 16, height = 12, units = "in",
dpi = 300,
bg = "white",
device = "png"
)
```
#### Revision Figure 2C - Behavioural - neural correlation
```{r}
## Figure R2C - Neural response ~ behavioural "accept" rate
# 1) Long table
FigureR2C_df <- revision_results2C_corr_input_all_nooutlier %>%
dplyr::select(ID, roi_group, LureRepeat, lure_yes_rate) %>%
tidyr::drop_na(LureRepeat, lure_yes_rate) %>%
dplyr::mutate(
facet_lab = dplyr::case_when(
roi_group == "L_HEAD_DGCA23_HEAD_SUB_14" ~ "Left HC head cluster",
roi_group == "R_HEAD_DGCA23_HEAD_SUB_15" ~ "Right HC head cluster",
TRUE ~ as.character(roi_group)
)
)
# 2) Per-ROI correlation + Bonferroni + label text
FigureR2C_labs <- FigureR2C_df %>%
dplyr::group_by(roi_group, facet_lab) %>%
dplyr::summarise(
n = dplyr::n(),
r = unname(stats::cor(LureRepeat, lure_yes_rate, method = "pearson")),
p = stats::cor.test(LureRepeat, lure_yes_rate, method = "pearson")$p.value,
.groups = "drop"
) %>%
dplyr::mutate(
p_adj = p.adjust(p, method = "bonferroni"),
stars = dplyr::case_when(
p_adj < .001 ~ "***",
p_adj < .01 ~ "**",
p_adj < .05 ~ "*",
TRUE ~ ""
),
sig = if_else(p_adj < .05, "sig", "ns"),
ann = dplyr::case_when(
is.na(p_adj) ~ sprintf("r = %.2f, p = NA", r),
p_adj < .001 ~ sprintf("r = %.2f, p < .001%s", r, stars),
TRUE ~ sprintf("r = %.2f, p = %.3f%s", r, p_adj, stars)
)
)
FigureR2C_full <- FigureR2C_df %>%
left_join(FigureR2C_labs %>% dplyr::select(roi_group, sig), by = "roi_group")
# 3) Plot
FigureR2C_plot_base <- ggplot(FigureR2C_full, aes(LureRepeat, lure_yes_rate)) +
geom_point(alpha = 0.7, size = 4.5) +
scale_color_manual(values = c(ns = "grey30", sig = "red"), guide = "none") +
facet_wrap(
~ roi_group,
scales = "free",
labeller = labeller(roi_group = setNames(FigureR2C_labs$facet_lab, FigureR2C_labs$roi_group))
) +
labs(
x = "Neural % signal change to modified repeats",
y = "Percent rate of 'accept' responses\nfor modified repeats"
) +
theme_classic(base_size = 18) +
custom_theme
FigureR2C_plot <- FigureR2C_plot_base +
geom_smooth(aes(color = sig), method = "lm", se = FALSE, linewidth = 1.1) +
geom_text(
data = FigureR2C_labs,
aes(x = -Inf, y = Inf, label = ann, color = sig),
inherit.aes = FALSE,
hjust = -0.05, vjust = 1.2, size = 5.5
)
FigureR2C_plot
```
EXPORT
```{r}
text_base_export <- 40
FigureR2C_plot_export <- FigureR2C_plot_base +
geom_smooth(aes(color = sig), method = "lm", se = FALSE, linewidth = 2.6) +
geom_text(
data = FigureR2C_labs,
aes(x = -Inf, y = Inf, label = ann, color = sig),
inherit.aes = FALSE,
hjust = -0.05, vjust = 1.2, size = 10.5
) +
theme_classic(base_size = text_base_export) +
theme(
legend.title = element_text(size = text_base_export * 0.95),
legend.text = element_text(size = text_base_export * 0.90),
axis.text.x = element_text(size = text_base_export * 0.90, margin = margin(t = 2)),
axis.text.y = element_text(size = text_base_export * 0.90),
axis.title = element_text(size = text_base_export * 1.05)
)
ggsave(
"derivatives/figures/revision_figures/FigureR2C.png",
plot = FigureR2C_plot_export,
width = 20, height = 12, units = "in",
dpi = 300,
bg = "white",
device = "png"
)
```

smst_mr1_analysis.Rmd at commit b1f8280, no license · at the source

OverviewAuthors: Alex Ilyés1,2,3, Borbála Mónika Brosig2,4, György Mező5,6,7, Attila Keresztes2,3,8
ORCID iDs: Alex Ilyés, Borbála Mónika Brosig, György Mező
Doctoral School of Psychology, Eötvös Loránd University, Budapest 1075, Hungary
Brain Imaging Centre, Hungarian Research Network, Research Centre for Natural Sciences, Budapest 1117, Hungary
Institute of Psychology, Eötvös Loránd University, Budapest 1064, Hungary
Mental Health Sciences Division, Doctoral School, Semmelweis University, Budapest 1085, Hungary
Konkoly Observatory, Hungarian Research Network, Research Centre for Astronomy and Earth Sciences, Budapest 1121, Hungary
Wigner Data Center, Hungarian Research Network, Wigner Research Centre for Physics, Budapest 1121, Hungary
Hungarian Academy of Sciences Centre of Excellence, Hungarian Research Network, Research Centre for Astronomy and Earth Sciences, Budapest 1121, Hungary
Max Planck Partner Group of the Max Planck Institute for Human Development, Hippocampal Circuit and Code for Cognition Lab, Budapest 1117, Hungary
Institutions: Eötvös Loránd University (Hungary); HUN-REN Research Centre for Natural Sciences (Hungary); Semmelweis University (Hungary); HUN-REN Research Centre for Astronomy and Earth Sciences (Hungary); HUN-REN Wigner Research Centre for Physics (Hungary)
Journal: Proceedings of the National Academy of Sciences of the United States of America, volume 123, issue 31, article e2603114123
Dates: received 28 January 2026; accepted 18 June 2026; published online 28 July 2026; in print 4 August 2026
Type: Research article · Language: English
License: CC BY-NC-ND
Identifiers: DOI 10.1073/pnas.2603114123 · PMID 42520121 · PMCID PMC13438472 · OpenAlex W7171530764
Open access: hybrid, a free copy (OpenAlex)
Status: code verified
Categories: human (organism), cognitive (subfield)
Methods: Statistics, Preprocessing, fMRI & imaging
Keywords: pattern separation, episodic memory, semantic similarity, word2vec, repetition suppression
MeSH: Hippocampus*, Memory*, Adult, Brain Mapping, Female, Humans, Magnetic Resonance Imaging, Male, Memory, Episodic, Recognition, Psychology, Semantics, Young Adult (* major topic)
Journal subjects: Social Sciences, Psychological and Cognitive Sciences
Topic: Memory and Neural Mechanisms (Cognitive Neuroscience, Neuroscience), according to OpenAlex
Funding: Hungarian National Research, Development and Innovation Office (FK128648 and FK146496); Max Planck Society (Max Planck Partner Group); Hungarian Academy of Sciences (0708-21 515 AT); Eotvos Lorand University (Doctoral Projects 2022 Doctoral Projects 2023); University Research Scholarship Program of the Ministry for Culture and Innovation (EKÖP-25-4-I-ELTE-1020, EKÖP-24-2-I-ELTE-687)
Citations: not cited yet (Europe PMC); 144 references in the paper
AbstractThe abstract is not reproduced here: the paper's license (CC BY-NC-ND) does not allow it. Read it in the paper, at the publisher or on Europe PMC.

Repositories

Its files are read in the Code ↔ Paper reader above, with 9 matches between paragraphs and lines of code.

Zenodo 18395407

License: CC-BY-4.0

State: the link answers, verified on 27 September 2026

Evidence: files inventoried

Size: 1 file

Software Heritage: not checked

Found in: the references

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

Tools: NumPy (6 files), PsychoPy (5 files), pandas (4 files)

Availability: 1 check, the latest on 27 September 2026: the link answers (HTTP 200)

27 September 2026: the link answers (HTTP 200)

10 files

scripts/smst2_bids_participants_maker.py, Python, 216 lines
scripts/smst2_bids_tsv_maker.py, Python, 529 lines
scripts/smst2_bids_tsv_maker_ses02.py, Python, 359 lines
scripts/smst2_mr_summarize_ses01.py, Python, 65 lines
sem_mst_2_mr.py, Python, 4,024 lines
sem_mst_2_mr_enc_lastrun.py, Python, 2,712 lines
sem_mst_2_mr_lastrun.py, Python, 3,876 lines
sem_mst_2_mr_rec_lastrun.py, Python, 2,814 lines
sem_mst_2_mr_session2_mapping_lastrun.py, Python, 2,883 lines
README.md, Text, 35 lines

At the source:

Zenodo 20090797

License: CC-BY-4.0

State: the link answers, verified on 27 September 2026

Evidence: files inventoried

Size: 1 file

Software Heritage: not checked

Found in: the references

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

Tools: Curve Fitting Toolbox (2 files), afex (1 file), broom (1 file), easystats (1 file), emmeans (1 file), FSL (1 file), ggplot2 (1 file), ggpubr (1 file), lme4 (1 file), lmerTest (1 file), Matplotlib (1 file), NiBabel (1 file), NumPy (1 file), pandas (1 file), patchwork (1 file), psych (1 file), rstatix (1 file), SciPy (1 file), tidyverse (1 file)

Availability: 1 check, the latest on 27 September 2026: the link answers (HTTP 200)

27 September 2026: the link answers (HTTP 200)

6 files

scripts/smst_mr1_HCmask_snapshot.sh, Shell, 169 lines
scripts/smst_mr1_MNIcoordinates.py, Python, 479 lines
scripts/smst_mr1_analysis.Rmd, R, 5,188 lines
scripts/smst_mr1_powerfunction.m, MATLAB, 192 lines
scripts/smst_mr1_powerfunction_anat.m, MATLAB, 192 lines
README.md, Text, 19 lines

At the source:

ilyesalex/smst2_mr_experiment

License: none: the authors keep all their rights

State: the link answers, verified on 27 September 2026

Evidence: files inventoried

Commit: c73d841427fd9ac13d6434ad2bf9b971ab5c2e56, 27 January 2026

Languages: Python (9)

Size: 1,720 files, 9 scripts

Software Heritage: not archived

Found in: the Zenodo archive record

Holds: README

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

Tools: NumPy (6 files), PsychoPy (5 files), pandas (4 files)

Availability: 1 check, the latest on 27 September 2026: the link answers

27 September 2026: the link answers

10 files

scripts/smst2_bids_participants_maker.py, Python, 216 lines, 1 match
scripts/smst2_bids_tsv_maker.py, Python, 529 lines, 1 match
scripts/smst2_bids_tsv_maker_ses02.py, Python, 359 lines
scripts/smst2_mr_summarize_ses01.py, Python, 65 lines
sem_mst_2_mr.py, Python, 4,024 lines
sem_mst_2_mr_enc_lastrun.py, Python, 2,712 lines
sem_mst_2_mr_lastrun.py, Python, 3,876 lines
sem_mst_2_mr_rec_lastrun.py, Python, 2,814 lines
sem_mst_2_mr_session2_mapping_lastrun.py, Python, 2,883 lines
README.md, Text, 35 lines

At the source: github.com/IlyesAlex/smst2_mr_experiment/tree/fmri-experiment

ilyesalex/smst2_mr_analysis

License: none: the authors keep all their rights

State: the link answers, verified on 27 September 2026

Evidence: files inventoried

Commit: b1f8280750fbe099b86224e616a30e5db4c0f85a, 8 May 2026

Languages: MATLAB (2), Shell (1), Python (1), R (1)

Size: 64 files, 5 scripts

Software Heritage: not archived

Found in: the Zenodo archive record

Holds: README, 1 notebook

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

Availability: 1 check, the latest on 27 September 2026: the link answers

27 September 2026: the link answers

6 files

scripts/smst_mr1_HCmask_snapshot.sh, Shell, 169 lines, 1 match
scripts/smst_mr1_MNIcoordinates.py, Python, 479 lines
scripts/smst_mr1_analysis.Rmd, R, 5,188 lines, 6 matches
scripts/smst_mr1_powerfunction.m, MATLAB, 192 lines
scripts/smst_mr1_powerfunction_anat.m, MATLAB, 192 lines
README.md, Text, 19 lines

At the source: github.com/IlyesAlex/smst2_mr_analysis/tree/v2.0

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:

4 repositories of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
28 scripts, each with its path and the digest of its content;
9 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

Datasets cited

openneuro:ds007275, at OpenNeuro; found in the references

Code and data availability statement

The paper has a code and data availability statement. Its license (CC BY-NC-ND) does not allow reproducing it here; in short, from what the harvester recognized in it:

no repository, dataset or request procedure was recognized in it

Read it in the paper: doi.org/10.1073/pnas.2603114123.

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, 27 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 4 authors, 5 keywords, 12 MeSH terms, 5 funders, 123 references.

Cite

This paper

Ilyés, A., Brosig, B. M., Mező, G., & Keresztes, A. (2026). The human hippocampus can pattern separate memories by meaning. Proceedings of the National Academy of Sciences of the United States of America, 123(31), e2603114123. https://doi.org/10.1073/pnas.2603114123

BibTeX

@article{ilyes2026human, author = {Ilyés, Alex and Brosig, Borbála Mónika and Mező, György and Keresztes, Attila}, title = {{The human hippocampus can pattern separate memories by meaning}}, journal = {Proceedings of the National Academy of Sciences of the United States of America}, year = {2026}, month = jul, volume = {123}, number = {31}, pages = {e2603114123}, publisher = {National Academy of Sciences}, issn = {0027-8424}, doi = {10.1073/pnas.2603114123}, url = {https://doi.org/10.1073/pnas.2603114123}, pmid = {42520121}, pmcid = {PMC13438472} }

RIS

TY - JOUR AU - Ilyés, Alex AU - Brosig, Borbála Mónika AU - Mező, György AU - Keresztes, Attila TI - The human hippocampus can pattern separate memories by meaning T2 - Proceedings of the National Academy of Sciences of the United States of America J2 - Proc Natl Acad Sci U S A PY - 2026 DA - 2026/07/28 VL - 123 IS - 31 SP - e2603114123 SN - 0027-8424 PB - National Academy of Sciences DO - 10.1073/pnas.2603114123 UR - https://doi.org/10.1073/pnas.2603114123 LA - en ER -

CSL-JSON

{ "id": "10.1073/pnas.2603114123", "type": "article-journal", "title": "The human hippocampus can pattern separate memories by meaning", "container-title": "Proceedings of the National Academy of Sciences of the United States of America", "author": [ { "family": "Ilyés", "given": "Alex" }, { "family": "Brosig", "given": "Borbála Mónika" }, { "family": "Mező", "given": "György" }, { "family": "Keresztes", "given": "Attila" } ], "container-title-short": "Proc Natl Acad Sci U S A", "volume": "123", "issue": "31", "page": "e2603114123", "DOI": "10.1073/pnas.2603114123", "PMID": "42520121", "PMCID": "PMC13438472", "ISSN": "0027-8424", "publisher": "National Academy of Sciences", "URL": "https://doi.org/10.1073/pnas.2603114123", "language": "en", "issued": { "date-parts": [ [ 2026, 7, 28 ] ] } }

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:10.1038/s41467-026-73865-9 [code]: Histamine shapes the neurocomputational dynamics of human learning.
Journal: Nature communications
In common: PsychoPy, rstatix, easystats, 12 other tools, cognitive, 2 references
[2] doi:10.1093/braincomms/fcag255 [code]: Impaired consolidation of spatial memory during sleep in patients with leucine-rich glioma-inactivated 1-associated limbic encephalitis.
Journal: Brain communications
In common: afex, rstatix, emmeans, 6 other tools, cognitive, 6 references
[3] doi:10.1093/nc/niag029 [code]: A data-driven approach to identifying and evaluating connectivity-based neural correlates of conscious visual perception.
Journal: Neuroscience of consciousness
In common: afex, easystats, broom, 12 other tools, cognitive
[4] doi:10.1002/hbm.70512 [code]: Precision Imaging for Intraindividual Investigation of the Reward Response.
Journal: Human brain mapping
In common: PsychoPy, psych, easystats, 10 other tools, 2 references
[5] doi:10.1162/imag.a.1321 [code]: Phase similarity between similar objects indicates representational merging across retrieval training but not sleep.
Journal: Imaging neuroscience (Cambridge, Mass.)
In common: rstatix, easystats, broom, 10 other tools, cognitive, 1 reference
[6] doi:10.1126/sciadv.aec9291 [code]: Computational mechanisms of perception in autism revealed using games inspired by rodent operant tasks.
Journal: Science advances
In common: afex, psych, rstatix, 8 other tools, cognitive
[7] doi:10.1162/imag.a.1245 [code]: Towards precision EEG connectomics: Evaluating the benefits of dense sampling.
Journal: Imaging neuroscience (Cambridge, Mass.)
In common: psych, rstatix, lmerTest, 11 other tools, 1 reference
[8] doi:10.1016/j.nicl.2026.104012 [code]: Structural-functional multilayer brain network properties and outcome of combined repetitive transcranial magnetic stimulation and psychotherapy for obsessive-compulsive disorder.
Journal: NeuroImage. Clinical
In common: afex, psych, rstatix, 10 other tools, 1 reference
[9] doi:10.1038/s41398-026-04010-9 [code]: Bullying victimization and brain development: a longitudinal structural magnetic resonance imaging study from adolescence to early adulthood.
Journal: Translational psychiatry
In common: psych, rstatix, easystats, 8 other tools, 1 reference
[10] doi:10.1038/s41467-026-72934-3 [code]: Multi-focal ultrasound neuromodulation to the dorsal anterior cingulate cortex disrupts behavioural and neural pain processing.
Journal: Nature communications
In common: rstatix, broom, emmeans, 6 other tools, 4 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.

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: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

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

The human hippocampus can pattern separate memories by meaning.

Code ↔ Paper

Paper

The authors' code

Overview

Abstract

Repositories

Zenodo 18395407

Zenodo 20090797

ilyesalex/smst2_mr_experiment

ilyesalex/smst2_mr_analysis

Tracing map

Data

Datasets cited

Code and data availability statement

Versions

Version 1, 27 September 2026: the first record

Cite

This paper

Similar papers

Contribute

Claim this paper

Correct its record

Validate its tracing map

Add the badge to its README

Request its removal

Discussion, reproductions, activity