library(tidyverse)
library(here)
library(lmerTest)
library(MuMIn)
library(lme4)
library(dotenv)
library(broom)
library(broom.mixed)
library(effects)
library(emmeans)
env_file = here(".env")
default_project = "main"
if (file.exists(env_file)) {
load_dot_env(file = env_file)
PROJECT_VERSION <- Sys.getenv("PROJECT_VERSION")
if (PROJECT_VERSION == "") {
PROJECT_VERSION <- default_project
}
} else {
PROJECT_VERSION <- default_project
}
source("lmer_helpers.R")Load data
trial_metadata <- read.csv(here("data","metadata","level-trialtype_data.csv"))
trial_summary_data <- read.csv(here("data",PROJECT_VERSION, "processed_data","level-trials_data.csv"))
usable_trials <- trial_summary_data |>
# excluding possible scam participant
filter(exclude_participant_insufficient_data == 0 & trial_exclusion == 0 & exclude_participant == 0 & SubjectInfo.subjID != "PH2RNZ")
# Merging with similarity information and mean-centering main effects
trials_with_effect_vars <- usable_trials |>
left_join(trial_metadata) |>
mutate(age_in_months = SubjectInfo.testAge/30)Joining with `by = join_by(Trials.trialID, Trials.targetImage,
Trials.distractorImage, Trials.imagePair)`Sanity check - making sure all participants have at least 16 trials and that we have 83 participants
low_trial_count <- trials_with_effect_vars |> distinct(SubjectInfo.subjID,Trials.trialID) |> summarize(n=n(),.by=SubjectInfo.subjID) |> filter(n < 25)
nrow(trials_with_effect_vars |> distinct(SubjectInfo.subjID))[1] 91tidy_model <- function(main_effect){
table_data <- tidy(main_effect, effects = "fixed") %>%
mutate(
#p.value = 2 * (1 - pt(abs(statistic),df)), # Calculate p-values for lmer - just using default calculated ones
p.value.condensed = case_when(
p.value < .001 ~ "<.001",
p.value < .01 ~ "<.01",
p.value < .05 ~ "<.05",
TRUE ~ sprintf("%.3f", p.value)),
term = case_when(
term == "(Intercept)" ~ "Intercept",
term == "scale(age_in_months)" ~ "Age (scaled)",
#term == "scale(image_similarity)" ~ "Target-distractor image embedding similarity (scaled)",
TRUE ~ term
)
) %>%
rename(
Predictor = term,
"b" = estimate,
"SE" = std.error,
"t" = statistic, # Note: changed from z to t for lmer
"p" = p.value.condensed,
"p.full" = p.value
) %>%
mutate(across(c("b", "SE", "t"), ~round(., 2)))
return(table_data)
}Run mixed-effects model
Model 1: This is the model we said we’d run in our pre-reg
prereg_main_effect <- lmer(scale(corrected_target_looking) ~ scale(text_similarity)*scale(age_in_months)
+ (scale(text_similarity) | SubjectInfo.subjID)
+ (1|Trials.targetImage)
+ (1|Trials.imagePair),
data = trials_with_effect_vars)
summary(prereg_main_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(text_similarity) * scale(age_in_months) +
(scale(text_similarity) | SubjectInfo.subjID) + (1 | Trials.targetImage) +
(1 | Trials.imagePair)
Data: trials_with_effect_vars
REML criterion at convergence: 7006.3
Scaled residuals:
Min 1Q Median 3Q Max
-2.8148 -0.6245 -0.0230 0.6602 2.7597
Random effects:
Groups Name Variance Std.Dev. Corr
SubjectInfo.subjID (Intercept) 0.0174669 0.13216
scale(text_similarity) 0.0002581 0.01606 -0.25
Trials.targetImage (Intercept) 0.0147133 0.12130
Trials.imagePair (Intercept) 0.0017978 0.04240
Residual 0.9606505 0.98013
Number of obs: 2476, groups:
SubjectInfo.subjID, 91; Trials.targetImage, 24; Trials.imagePair, 16
Fixed effects:
Estimate Std. Error df
(Intercept) -0.00840 0.03659 23.37423
scale(text_similarity) -0.05291 0.02752 11.28920
scale(age_in_months) 0.05829 0.02411 90.79815
scale(text_similarity):scale(age_in_months) -0.03078 0.01978 91.78174
t value Pr(>|t|)
(Intercept) -0.230 0.8204
scale(text_similarity) -1.923 0.0801 .
scale(age_in_months) 2.418 0.0176 *
scale(text_similarity):scale(age_in_months) -1.556 0.1231
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__)
scl(txt_sm) -0.007
scl(g_n_mn) 0.002 -0.007
scl(_):(__) -0.007 -0.004 -0.003Currently, it looks like text_similarity is singular. I tried removing the correlation between similarity and subject but that did not work so removign the random intercept
text_similarity_effect <- lmer(scale(corrected_target_looking) ~ scale(text_similarity)*scale(age_in_months)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage)
+ (1|Trials.imagePair),
data = trials_with_effect_vars)
summary(text_similarity_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(text_similarity) * scale(age_in_months) +
(1 | SubjectInfo.subjID) + (1 | Trials.targetImage) + (1 |
Trials.imagePair)
Data: trials_with_effect_vars
REML criterion at convergence: 7006.3
Scaled residuals:
Min 1Q Median 3Q Max
-2.8124 -0.6239 -0.0226 0.6615 2.7627
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.01748 0.13222
Trials.targetImage (Intercept) 0.01470 0.12124
Trials.imagePair (Intercept) 0.00180 0.04243
Residual 0.96090 0.98026
Number of obs: 2476, groups:
SubjectInfo.subjID, 91; Trials.targetImage, 24; Trials.imagePair, 16
Fixed effects:
Estimate Std. Error df
(Intercept) -8.481e-03 3.659e-02 2.339e+01
scale(text_similarity) -5.291e-02 2.747e-02 1.164e+01
scale(age_in_months) 5.830e-02 2.412e-02 9.090e+01
scale(text_similarity):scale(age_in_months) -3.081e-02 1.971e-02 2.378e+03
t value Pr(>|t|)
(Intercept) -0.232 0.8187
scale(text_similarity) -1.926 0.0788 .
scale(age_in_months) 2.418 0.0176 *
scale(text_similarity):scale(age_in_months) -1.563 0.1181
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__)
scl(txt_sm) -0.001
scl(g_n_mn) 0.002 -0.007
scl(_):(__) -0.007 -0.005 0.010Singular fit debugs for image similarity
Swapping text similarity with image similarity:
image_similarity_effect <- lmer(scale(corrected_target_looking) ~ scale(image_similarity)*scale(age_in_months)
+ (scale(image_similarity) | SubjectInfo.subjID)
+ (1|Trials.targetImage)
+ (1|Trials.imagePair),
data = trials_with_effect_vars)boundary (singular) fit: see help('isSingular')summary(image_similarity_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(image_similarity) * scale(age_in_months) +
(scale(image_similarity) | SubjectInfo.subjID) + (1 | Trials.targetImage) +
(1 | Trials.imagePair)
Data: trials_with_effect_vars
REML criterion at convergence: 7004.4
Scaled residuals:
Min 1Q Median 3Q Max
-2.77998 -0.62742 -0.02154 0.66497 2.79679
Random effects:
Groups Name Variance Std.Dev. Corr
SubjectInfo.subjID (Intercept) 0.0180635 0.13440
scale(image_similarity) 0.0007765 0.02787 1.00
Trials.targetImage (Intercept) 0.0151305 0.12301
Trials.imagePair (Intercept) 0.0006291 0.02508
Residual 0.9595821 0.97958
Number of obs: 2476, groups:
SubjectInfo.subjID, 91; Trials.targetImage, 24; Trials.imagePair, 16
Fixed effects:
Estimate Std. Error df
(Intercept) -0.009043 0.035908 24.484218
scale(image_similarity) -0.063791 0.025625 12.185108
scale(age_in_months) 0.058504 0.024235 91.933938
scale(image_similarity):scale(age_in_months) -0.025213 0.019894 746.746833
t value Pr(>|t|)
(Intercept) -0.252 0.8033
scale(image_similarity) -2.489 0.0282 *
scale(age_in_months) 2.414 0.0178 *
scale(image_similarity):scale(age_in_months) -1.267 0.2054
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__)
scl(mg_sml) 0.043
scl(g_n_mn) 0.002 -0.002
scl(_):(__) 0.001 -0.005 0.092
optimizer (nloptwrap) convergence code: 0 (OK)
boundary (singular) fit: see help('isSingular')running into a singular fit, I tried removing each of the random effects but only removing the random slope of image_similarity and the random intercept for image pair fixed this. This is okay for now since we see a similar effect with the singular model and this model. (although the effect below is stronger)
image_similarity_effect <- lmer(scale(corrected_target_looking) ~ scale(image_similarity)*scale(age_in_months)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = trials_with_effect_vars)
summary(image_similarity_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(image_similarity) * scale(age_in_months) +
(1 | SubjectInfo.subjID) + (1 | Trials.targetImage)
Data: trials_with_effect_vars
REML criterion at convergence: 7005.3
Scaled residuals:
Min 1Q Median 3Q Max
-2.7952 -0.6299 -0.0274 0.6671 2.7858
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.01753 0.1324
Trials.targetImage (Intercept) 0.01554 0.1247
Residual 0.96098 0.9803
Number of obs: 2476, groups: SubjectInfo.subjID, 91; Trials.targetImage, 24
Fixed effects:
Estimate Std. Error df
(Intercept) -8.886e-03 3.553e-02 2.967e+01
scale(image_similarity) -6.297e-02 2.472e-02 1.565e+02
scale(age_in_months) 5.851e-02 2.412e-02 9.086e+01
scale(image_similarity):scale(age_in_months) -2.540e-02 1.969e-02 2.386e+03
t value Pr(>|t|)
(Intercept) -0.250 0.8042
scale(image_similarity) -2.547 0.0118 *
scale(age_in_months) 2.425 0.0173 *
scale(image_similarity):scale(age_in_months) -1.290 0.1972
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__)
scl(mg_sml) -0.002
scl(g_n_mn) 0.002 -0.003
scl(_):(__) 0.000 -0.006 0.006Adding in covariates
Now for the fun stuff: adding in our covariates using our original model – only doing this for text similarity for now because of the singular effects
main_image_effect <- lmer(scale(corrected_target_looking) ~ scale(image_similarity) * scale(age_in_months) + scale(AoA_Est_target)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = trials_with_effect_vars)
main_text_effect <- lmer(scale(corrected_target_looking) ~ scale(text_similarity) + scale(age_in_months) + scale(AoA_Est_target)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = trials_with_effect_vars)
summary(main_image_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(image_similarity) * scale(age_in_months) +
scale(AoA_Est_target) + (1 | SubjectInfo.subjID) + (1 | Trials.targetImage)
Data: trials_with_effect_vars
REML criterion at convergence: 7003.9
Scaled residuals:
Min 1Q Median 3Q Max
-2.82166 -0.62489 -0.01889 0.67371 2.81361
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.01720 0.13113
Trials.targetImage (Intercept) 0.00909 0.09534
Residual 0.96153 0.98058
Number of obs: 2476, groups: SubjectInfo.subjID, 91; Trials.targetImage, 24
Fixed effects:
Estimate Std. Error df
(Intercept) -8.364e-03 3.136e-02 2.742e+01
scale(image_similarity) -4.562e-02 2.409e-02 1.170e+02
scale(age_in_months) 5.785e-02 2.405e-02 9.097e+01
scale(AoA_Est_target) -7.943e-02 2.836e-02 2.308e+01
scale(image_similarity):scale(age_in_months) -2.563e-02 1.970e-02 2.386e+03
t value Pr(>|t|)
(Intercept) -0.267 0.7917
scale(image_similarity) -1.893 0.0608 .
scale(age_in_months) 2.405 0.0182 *
scale(AoA_Est_target) -2.801 0.0101 *
scale(image_similarity):scale(age_in_months) -1.301 0.1933
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__) s(AA_E
scl(mg_sml) -0.004
scl(g_n_mn) 0.003 -0.005
scl(AA_Es_) 0.010 -0.256 0.010
scl(_):(__) 0.000 -0.008 0.006 0.004summary(main_text_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(text_similarity) + scale(age_in_months) +
scale(AoA_Est_target) + (1 | SubjectInfo.subjID) + (1 | Trials.targetImage)
Data: trials_with_effect_vars
REML criterion at convergence: 6999.7
Scaled residuals:
Min 1Q Median 3Q Max
-2.81200 -0.61927 -0.02444 0.66947 2.78715
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.017067 0.13064
Trials.targetImage (Intercept) 0.008141 0.09023
Residual 0.962357 0.98100
Number of obs: 2476, groups: SubjectInfo.subjID, 91; Trials.targetImage, 24
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) -0.008134 0.030688 29.346063 -0.265 0.79281
scale(text_similarity) -0.044777 0.023607 90.933150 -1.897 0.06103 .
scale(age_in_months) 0.058185 0.024029 91.040497 2.421 0.01744 *
scale(AoA_Est_target) -0.087013 0.026941 23.867742 -3.230 0.00359 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__)
scl(txt_sm) -0.002
scl(g_n_mn) 0.003 -0.009
scl(AA_Es_) 0.009 -0.123 0.010r.squaredGLMM(main_image_effect) R2m R2c
[1,] 0.01431936 0.0405489r.squaredGLMM(main_text_effect) R2m R2c
[1,] 0.01365485 0.03883147Text similarity is still on the verge of significance
Checking if text similarity and image similarity are differently correlated with AoA
cor.test(trial_metadata$AoA_Est_target, trial_metadata$text_similarity)
Pearson's product-moment correlation
data: trial_metadata$AoA_Est_target and trial_metadata$text_similarity
t = 0.42649, df = 30, p-value = 0.6728
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
-0.2786047 0.4150887
sample estimates:
cor
0.07763097 cor.test(trial_metadata$AoA_Est_target, trial_metadata$image_similarity)
Pearson's product-moment correlation
data: trial_metadata$AoA_Est_target and trial_metadata$image_similarity
t = 1.4523, df = 30, p-value = 0.1568
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
-0.1014633 0.5553600
sample estimates:
cor
0.2562984 Well image similarity has a higher r but both are still insignificant.
image_model <- summary(main_image_effect)
image_modelLinear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(image_similarity) * scale(age_in_months) +
scale(AoA_Est_target) + (1 | SubjectInfo.subjID) + (1 | Trials.targetImage)
Data: trials_with_effect_vars
REML criterion at convergence: 7003.9
Scaled residuals:
Min 1Q Median 3Q Max
-2.82166 -0.62489 -0.01889 0.67371 2.81361
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.01720 0.13113
Trials.targetImage (Intercept) 0.00909 0.09534
Residual 0.96153 0.98058
Number of obs: 2476, groups: SubjectInfo.subjID, 91; Trials.targetImage, 24
Fixed effects:
Estimate Std. Error df
(Intercept) -8.364e-03 3.136e-02 2.742e+01
scale(image_similarity) -4.562e-02 2.409e-02 1.170e+02
scale(age_in_months) 5.785e-02 2.405e-02 9.097e+01
scale(AoA_Est_target) -7.943e-02 2.836e-02 2.308e+01
scale(image_similarity):scale(age_in_months) -2.563e-02 1.970e-02 2.386e+03
t value Pr(>|t|)
(Intercept) -0.267 0.7917
scale(image_similarity) -1.893 0.0608 .
scale(age_in_months) 2.405 0.0182 *
scale(AoA_Est_target) -2.801 0.0101 *
scale(image_similarity):scale(age_in_months) -1.301 0.1933
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__) s(AA_E
scl(mg_sml) -0.004
scl(g_n_mn) 0.003 -0.005
scl(AA_Es_) 0.010 -0.256 0.010
scl(_):(__) 0.000 -0.008 0.006 0.004vs_image_effect <- lmer(scale(corrected_target_looking) ~ scale(image_similarity)*scale(age_in_months)
+ scale(MeanSaliencyDiff)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = trials_with_effect_vars)
vs_text_effect <- lmer(scale(corrected_target_looking) ~ scale(text_similarity)*scale(age_in_months)
+ scale(MeanSaliencyDiff)
+ (scale(text_similarity) | SubjectInfo.subjID)
+ (1|Trials.targetImage)
+ (1|Trials.imagePair),
data = trials_with_effect_vars)
summary(vs_image_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(image_similarity) * scale(age_in_months) +
scale(MeanSaliencyDiff) + (1 | SubjectInfo.subjID) + (1 |
Trials.targetImage)
Data: trials_with_effect_vars
REML criterion at convergence: 7010.5
Scaled residuals:
Min 1Q Median 3Q Max
-2.79601 -0.62995 -0.02772 0.66689 2.78608
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.01753 0.1324
Trials.targetImage (Intercept) 0.01633 0.1278
Residual 0.96111 0.9804
Number of obs: 2476, groups: SubjectInfo.subjID, 91; Trials.targetImage, 24
Fixed effects:
Estimate Std. Error df
(Intercept) -8.977e-03 3.600e-02 2.878e+01
scale(image_similarity) -6.292e-02 2.492e-02 1.436e+02
scale(age_in_months) 5.850e-02 2.413e-02 9.087e+01
scale(MeanSaliencyDiff) 9.248e-04 2.938e-02 4.543e+01
scale(image_similarity):scale(age_in_months) -2.541e-02 1.970e-02 2.386e+03
t value Pr(>|t|)
(Intercept) -0.249 0.8049
scale(image_similarity) -2.524 0.0127 *
scale(age_in_months) 2.425 0.0173 *
scale(MeanSaliencyDiff) 0.031 0.9750
scale(image_similarity):scale(age_in_months) -1.290 0.1972
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__) s(MSD)
scl(mg_sml) -0.001
scl(g_n_mn) 0.002 -0.004
scl(MnSlnD) 0.024 0.068 -0.009
scl(_):(__) 0.000 -0.006 0.006 -0.010summary(vs_text_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(text_similarity) * scale(age_in_months) +
scale(MeanSaliencyDiff) + (scale(text_similarity) | SubjectInfo.subjID) +
(1 | Trials.targetImage) + (1 | Trials.imagePair)
Data: trials_with_effect_vars
REML criterion at convergence: 7011.5
Scaled residuals:
Min 1Q Median 3Q Max
-2.81386 -0.62430 -0.02362 0.66043 2.76315
Random effects:
Groups Name Variance Std.Dev. Corr
SubjectInfo.subjID (Intercept) 0.0174633 0.13215
scale(text_similarity) 0.0002475 0.01573 -0.26
Trials.targetImage (Intercept) 0.0155468 0.12469
Trials.imagePair (Intercept) 0.0017476 0.04180
Residual 0.9607953 0.98020
Number of obs: 2476, groups:
SubjectInfo.subjID, 91; Trials.targetImage, 24; Trials.imagePair, 16
Fixed effects:
Estimate Std. Error df
(Intercept) -0.008415 0.037035 22.750359
scale(text_similarity) -0.052770 0.027633 11.298025
scale(age_in_months) 0.058258 0.024111 90.804895
scale(MeanSaliencyDiff) 0.004043 0.029206 41.144733
scale(text_similarity):scale(age_in_months) -0.030773 0.019779 91.303360
t value Pr(>|t|)
(Intercept) -0.227 0.8223
scale(text_similarity) -1.910 0.0819 .
scale(age_in_months) 2.416 0.0177 *
scale(MeanSaliencyDiff) 0.138 0.8906
scale(text_similarity):scale(age_in_months) -1.556 0.1232
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__) s(MSD)
scl(txt_sm) -0.007
scl(g_n_mn) 0.002 -0.007
scl(MnSlnD) 0.023 0.022 -0.009
scl(_):(__) -0.007 -0.004 -0.003 -0.006Just a sanity check that adding our saliency metric as a covariate does not affect our similarity effects.
Alternate window analyses
Only using first instance of an item
first_instance_target <- trials_with_effect_vars |>
group_by(SubjectInfo.subjID, Trials.targetImage) |>
arrange(Trials.ordinal, .by_group = TRUE) |>
slice(1) |>
ungroup()
first_instance_primary_target <- first_instance_target |>
filter(Trials.trialType %in% c("easy", "hard"))
first_instance_image_pair <- trials_with_effect_vars |>
group_by(SubjectInfo.subjID, Trials.imagePair) |>
arrange(Trials.ordinal, .by_group = TRUE) |>
slice(1) |>
ungroup()
first_instance_effect <- lmer(scale(corrected_target_looking) ~ scale(image_similarity)*scale(age_in_months)*scale(AoA_Est_target)
+ scale(MeanSaliencyDiff)
+ (1 | Trials.ordinal)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = first_instance_target)
first_instance_pt_effect <- lmer(scale(corrected_target_looking) ~ scale(text_similarity)*scale(age_in_months)*scale(AoA_Est_target)
+ scale(MeanSaliencyDiff)
+ (1 | Trials.ordinal)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = first_instance_primary_target)
first_instance_image_pair_effect <- lmer(scale(corrected_target_looking) ~ scale(text_similarity)*scale(age_in_months)*scale(AoA_Est_target)
+ scale(MeanSaliencyDiff)
+ (1 | Trials.ordinal)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = first_instance_image_pair)
summary(first_instance_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(image_similarity) * scale(age_in_months) *
scale(AoA_Est_target) + scale(MeanSaliencyDiff) + (1 | Trials.ordinal) +
(1 | SubjectInfo.subjID) + (1 | Trials.targetImage)
Data: first_instance_target
REML criterion at convergence: 5497.6
Scaled residuals:
Min 1Q Median 3Q Max
-2.86068 -0.62838 -0.00198 0.66777 2.74935
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 2.257e-02 0.150232
Trials.ordinal (Intercept) 3.153e-05 0.005615
Trials.targetImage (Intercept) 1.029e-02 0.101429
Residual 9.558e-01 0.977657
Number of obs: 1936, groups:
SubjectInfo.subjID, 91; Trials.ordinal, 47; Trials.targetImage, 24
Fixed effects:
Estimate
(Intercept) -3.433e-03
scale(image_similarity) -4.836e-02
scale(age_in_months) 5.764e-02
scale(AoA_Est_target) -8.140e-02
scale(MeanSaliencyDiff) -2.871e-03
scale(image_similarity):scale(age_in_months) -2.427e-03
scale(image_similarity):scale(AoA_Est_target) 1.301e-03
scale(age_in_months):scale(AoA_Est_target) -4.547e-02
scale(image_similarity):scale(age_in_months):scale(AoA_Est_target) 1.359e-02
Std. Error
(Intercept) 3.586e-02
scale(image_similarity) 2.909e-02
scale(age_in_months) 2.842e-02
scale(AoA_Est_target) 3.202e-02
scale(MeanSaliencyDiff) 2.938e-02
scale(image_similarity):scale(age_in_months) 2.402e-02
scale(image_similarity):scale(AoA_Est_target) 3.045e-02
scale(age_in_months):scale(AoA_Est_target) 2.370e-02
scale(image_similarity):scale(age_in_months):scale(AoA_Est_target) 2.363e-02
df
(Intercept) 2.418e+01
scale(image_similarity) 6.720e+01
scale(age_in_months) 1.082e+02
scale(AoA_Est_target) 2.130e+01
scale(MeanSaliencyDiff) 3.085e+01
scale(image_similarity):scale(age_in_months) 1.872e+03
scale(image_similarity):scale(AoA_Est_target) 3.464e+01
scale(age_in_months):scale(AoA_Est_target) 1.840e+03
scale(image_similarity):scale(age_in_months):scale(AoA_Est_target) 1.846e+03
t value
(Intercept) -0.096
scale(image_similarity) -1.662
scale(age_in_months) 2.028
scale(AoA_Est_target) -2.542
scale(MeanSaliencyDiff) -0.098
scale(image_similarity):scale(age_in_months) -0.101
scale(image_similarity):scale(AoA_Est_target) 0.043
scale(age_in_months):scale(AoA_Est_target) -1.919
scale(image_similarity):scale(age_in_months):scale(AoA_Est_target) 0.575
Pr(>|t|)
(Intercept) 0.9245
scale(image_similarity) 0.1011
scale(age_in_months) 0.0450 *
scale(AoA_Est_target) 0.0188 *
scale(MeanSaliencyDiff) 0.9228
scale(image_similarity):scale(age_in_months) 0.9195
scale(image_similarity):scale(AoA_Est_target) 0.9662
scale(age_in_months):scale(AoA_Est_target) 0.0552 .
scale(image_similarity):scale(age_in_months):scale(AoA_Est_target) 0.5653
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__) s(AA_E s(MSD) sc(_):(__) s(_):(A s(__):
scl(mg_sml) -0.047
scl(g_n_mn) 0.001 -0.014
scl(AA_Es_) 0.025 -0.319 0.015
scl(MnSlnD) -0.023 -0.068 -0.015 0.062
scl(_):(__) -0.010 -0.022 -0.035 0.007 -0.006
s(_):(AA_E_ -0.292 0.158 0.003 -0.084 0.087 0.004
s(__):(AA_E 0.018 0.010 0.022 0.007 -0.015 -0.348 -0.013
s(_):(__):( 0.005 0.005 -0.283 -0.005 -0.007 0.171 -0.007 -0.071summary(first_instance_pt_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(text_similarity) * scale(age_in_months) *
scale(AoA_Est_target) + scale(MeanSaliencyDiff) + (1 | Trials.ordinal) +
(1 | SubjectInfo.subjID) + (1 | Trials.targetImage)
Data: first_instance_primary_target
REML criterion at convergence: 2041.8
Scaled residuals:
Min 1Q Median 3Q Max
-2.85231 -0.62731 -0.02125 0.65454 2.79976
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.019242 0.13872
Trials.ordinal (Intercept) 0.003945 0.06281
Trials.targetImage (Intercept) 0.007197 0.08484
Residual 0.959696 0.97964
Number of obs: 712, groups:
SubjectInfo.subjID, 91; Trials.ordinal, 35; Trials.targetImage, 8
Fixed effects:
Estimate
(Intercept) 0.02123
scale(text_similarity) -0.08484
scale(age_in_months) 0.04577
scale(AoA_Est_target) -0.13727
scale(MeanSaliencyDiff) 0.01551
scale(text_similarity):scale(age_in_months) 0.03462
scale(text_similarity):scale(AoA_Est_target) 0.04684
scale(age_in_months):scale(AoA_Est_target) -0.01635
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) -0.03583
Std. Error
(Intercept) 0.05490
scale(text_similarity) 0.05485
scale(age_in_months) 0.04323
scale(AoA_Est_target) 0.05126
scale(MeanSaliencyDiff) 0.05130
scale(text_similarity):scale(age_in_months) 0.04519
scale(text_similarity):scale(AoA_Est_target) 0.04804
scale(age_in_months):scale(AoA_Est_target) 0.04058
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 0.04419
df
(Intercept) 6.69518
scale(text_similarity) 700.93727
scale(age_in_months) 124.40097
scale(AoA_Est_target) 5.64017
scale(MeanSaliencyDiff) 19.87701
scale(text_similarity):scale(age_in_months) 699.96298
scale(text_similarity):scale(AoA_Est_target) 143.06829
scale(age_in_months):scale(AoA_Est_target) 630.44692
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 678.14119
t value
(Intercept) 0.387
scale(text_similarity) -1.547
scale(age_in_months) 1.059
scale(AoA_Est_target) -2.678
scale(MeanSaliencyDiff) 0.302
scale(text_similarity):scale(age_in_months) 0.766
scale(text_similarity):scale(AoA_Est_target) 0.975
scale(age_in_months):scale(AoA_Est_target) -0.403
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) -0.811
Pr(>|t|)
(Intercept) 0.7109
scale(text_similarity) 0.1224
scale(age_in_months) 0.2918
scale(AoA_Est_target) 0.0389 *
scale(MeanSaliencyDiff) 0.7655
scale(text_similarity):scale(age_in_months) 0.4438
scale(text_similarity):scale(AoA_Est_target) 0.3312
scale(age_in_months):scale(AoA_Est_target) 0.6872
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 0.4178
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__) s(AA_E s(MSD) sc(_):(__) s(_):(A s(__):
scl(txt_sm) -0.165
scl(g_n_mn) 0.029 -0.022
scl(AA_Es_) -0.050 0.358 -0.008
scl(MnSlnD) -0.082 0.513 -0.048 0.101
scl(_):(__) 0.004 -0.006 -0.162 0.029 0.003
s(_):(AA_E_ 0.356 -0.465 0.048 -0.137 -0.224 0.040
s(__):(AA_E -0.002 0.027 -0.051 0.027 -0.010 0.416 -0.002
s(_):(__):( 0.030 0.035 0.399 -0.001 -0.002 -0.415 0.014 -0.137summary(first_instance_image_pair_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(text_similarity) * scale(age_in_months) *
scale(AoA_Est_target) + scale(MeanSaliencyDiff) + (1 | Trials.ordinal) +
(1 | SubjectInfo.subjID) + (1 | Trials.targetImage)
Data: first_instance_image_pair
REML criterion at convergence: 4002.3
Scaled residuals:
Min 1Q Median 3Q Max
-2.83685 -0.64175 -0.02266 0.64978 2.79951
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.019286 0.1389
Trials.ordinal (Intercept) 0.003092 0.0556
Trials.targetImage (Intercept) 0.007483 0.0865
Residual 0.954090 0.9768
Number of obs: 1407, groups:
SubjectInfo.subjID, 91; Trials.ordinal, 33; Trials.targetImage, 24
Fixed effects:
Estimate
(Intercept) -4.019e-03
scale(text_similarity) -5.152e-02
scale(age_in_months) 4.710e-02
scale(AoA_Est_target) -1.145e-01
scale(MeanSaliencyDiff) -1.035e-02
scale(text_similarity):scale(age_in_months) -2.666e-02
scale(text_similarity):scale(AoA_Est_target) 1.694e-02
scale(age_in_months):scale(AoA_Est_target) -5.052e-02
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 1.458e-02
Std. Error
(Intercept) 3.746e-02
scale(text_similarity) 2.957e-02
scale(age_in_months) 2.999e-02
scale(AoA_Est_target) 3.235e-02
scale(MeanSaliencyDiff) 3.176e-02
scale(text_similarity):scale(age_in_months) 2.660e-02
scale(text_similarity):scale(AoA_Est_target) 2.888e-02
scale(age_in_months):scale(AoA_Est_target) 2.694e-02
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 2.512e-02
df
(Intercept) 1.982e+01
scale(text_similarity) 8.329e+01
scale(age_in_months) 9.181e+01
scale(AoA_Est_target) 2.422e+01
scale(MeanSaliencyDiff) 3.590e+01
scale(text_similarity):scale(age_in_months) 1.316e+03
scale(text_similarity):scale(AoA_Est_target) 6.172e+01
scale(age_in_months):scale(AoA_Est_target) 1.378e+03
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 1.386e+03
t value
(Intercept) -0.107
scale(text_similarity) -1.742
scale(age_in_months) 1.571
scale(AoA_Est_target) -3.539
scale(MeanSaliencyDiff) -0.326
scale(text_similarity):scale(age_in_months) -1.002
scale(text_similarity):scale(AoA_Est_target) 0.587
scale(age_in_months):scale(AoA_Est_target) -1.875
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 0.580
Pr(>|t|)
(Intercept) 0.91564
scale(text_similarity) 0.08517 .
scale(age_in_months) 0.11966
scale(AoA_Est_target) 0.00166 **
scale(MeanSaliencyDiff) 0.74640
scale(text_similarity):scale(age_in_months) 0.31632
scale(text_similarity):scale(AoA_Est_target) 0.55957
scale(age_in_months):scale(AoA_Est_target) 0.06100 .
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 0.56180
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__) s(AA_E s(MSD) sc(_):(__) s(_):(A s(__):
scl(txt_sm) -0.017
scl(g_n_mn) -0.001 0.002
scl(AA_Es_) -0.006 -0.083 0.018
scl(MnSlnD) -0.004 0.025 -0.001 0.075
scl(_):(__) 0.006 -0.021 -0.003 0.041 0.035
s(_):(AA_E_ -0.085 0.121 0.040 0.176 0.219 -0.030
s(__):(AA_E 0.019 0.037 -0.024 0.023 -0.040 -0.046 -0.024
s(_):(__):( 0.039 -0.039 -0.059 -0.018 -0.014 0.132 0.040 0.170Random order effects
main_image_effect_ordinal <- lmer(scale(corrected_target_looking) ~ scale(image_similarity)*scale(age_in_months)
+ scale(AoA_Est_target)
+ scale(MeanSaliencyDiff)
+ (1 | Trials.ordinal)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = trials_with_effect_vars)
summary(main_image_effect_ordinal)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(image_similarity) * scale(age_in_months) +
scale(AoA_Est_target) + scale(MeanSaliencyDiff) + (1 | Trials.ordinal) +
(1 | SubjectInfo.subjID) + (1 | Trials.targetImage)
Data: trials_with_effect_vars
REML criterion at convergence: 7009
Scaled residuals:
Min 1Q Median 3Q Max
-2.82655 -0.63385 -0.01613 0.67450 2.80043
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.017305 0.13155
Trials.ordinal (Intercept) 0.001863 0.04316
Trials.targetImage (Intercept) 0.009719 0.09859
Residual 0.959754 0.97967
Number of obs: 2476, groups:
SubjectInfo.subjID, 91; Trials.ordinal, 52; Trials.targetImage, 24
Fixed effects:
Estimate Std. Error df
(Intercept) -7.986e-03 3.269e-02 2.586e+01
scale(image_similarity) -4.590e-02 2.427e-02 1.143e+02
scale(age_in_months) 5.796e-02 2.407e-02 9.098e+01
scale(AoA_Est_target) -7.971e-02 2.881e-02 2.249e+01
scale(MeanSaliencyDiff) 2.927e-04 2.643e-02 3.585e+01
scale(image_similarity):scale(age_in_months) -2.523e-02 1.970e-02 2.384e+03
t value Pr(>|t|)
(Intercept) -0.244 0.8089
scale(image_similarity) -1.892 0.0611 .
scale(age_in_months) 2.408 0.0180 *
scale(AoA_Est_target) -2.766 0.0111 *
scale(MeanSaliencyDiff) 0.011 0.9912
scale(image_similarity):scale(age_in_months) -1.281 0.2005
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__) s(AA_E s(MSD)
scl(mg_sml) -0.004
scl(g_n_mn) 0.002 -0.005
scl(AA_Es_) 0.010 -0.255 0.010
scl(MnSlnD) 0.020 0.029 -0.009 0.019
scl(_):(__) 0.000 -0.008 0.006 0.003 -0.010Order does not explain much variance.
Target image order effects
ranked_trials <- trials_with_effect_vars |>
group_by(SubjectInfo.subjID, Trials.targetImage) |>
arrange(Trials.ordinal, .by_group = TRUE) |>
mutate(
slice_num = row_number(),
order = case_when(
slice_num == 1 ~ -0.5,
slice_num == 2 ~ 0.5,
TRUE ~ NA
)
)
main_image_effect_ranked <- lmer(scale(corrected_target_looking) ~ scale(image_similarity)*scale(age_in_months)
+ scale(AoA_Est_target)
+ (scale(order))
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = ranked_trials)
# |> filter(Trials.trialType %in% c("easy", "hard")
summary(main_image_effect_ranked)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(image_similarity) * scale(age_in_months) +
scale(AoA_Est_target) + (scale(order)) + (1 | SubjectInfo.subjID) +
(1 | Trials.targetImage)
Data: ranked_trials
REML criterion at convergence: 7006.8
Scaled residuals:
Min 1Q Median 3Q Max
-2.79650 -0.62416 -0.02258 0.66026 2.75160
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.016739 0.12938
Trials.targetImage (Intercept) 0.008273 0.09096
Residual 0.961445 0.98053
Number of obs: 2476, groups: SubjectInfo.subjID, 91; Trials.targetImage, 24
Fixed effects:
Estimate Std. Error df
(Intercept) -4.653e-03 3.078e-02 2.678e+01
scale(image_similarity) -4.488e-02 2.387e-02 1.114e+02
scale(age_in_months) 5.777e-02 2.395e-02 9.098e+01
scale(AoA_Est_target) -8.096e-02 2.777e-02 2.294e+01
scale(order) 3.662e-02 2.126e-02 5.941e+02
scale(image_similarity):scale(age_in_months) -2.504e-02 1.970e-02 2.385e+03
t value Pr(>|t|)
(Intercept) -0.151 0.8810
scale(image_similarity) -1.880 0.0627 .
scale(age_in_months) 2.413 0.0178 *
scale(AoA_Est_target) -2.916 0.0078 **
scale(order) 1.722 0.0855 .
scale(image_similarity):scale(age_in_months) -1.271 0.2038
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__) s(AA_E scl(r)
scl(mg_sml) -0.004
scl(g_n_mn) 0.003 -0.005
scl(AA_Es_) 0.008 -0.258 0.010
scale(ordr) 0.064 0.009 -0.002 -0.024
scl(_):(__) 0.001 -0.008 0.005 0.003 0.018Z-scoring embeddings
main_image_effect_zscored <- lmer(scale(corrected_target_looking) ~ scale(image_sim)*scale(age_in_months)*scale(AoA_Est_target)
+ scale(MeanSaliencyDiff)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = trials_with_effect_vars)
summary(main_image_effect_zscored)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula:
scale(corrected_target_looking) ~ scale(image_sim) * scale(age_in_months) *
scale(AoA_Est_target) + scale(MeanSaliencyDiff) + (1 | SubjectInfo.subjID) +
(1 | Trials.targetImage)
Data: trials_with_effect_vars
REML criterion at convergence: 7020.8
Scaled residuals:
Min 1Q Median 3Q Max
-2.76788 -0.62397 -0.02101 0.66876 2.77752
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.01740 0.1319
Trials.targetImage (Intercept) 0.01002 0.1001
Residual 0.96018 0.9799
Number of obs: 2476, groups: SubjectInfo.subjID, 91; Trials.targetImage, 24
Fixed effects:
Estimate
(Intercept) -1.716e-02
scale(image_sim) -5.102e-02
scale(age_in_months) 5.473e-02
scale(AoA_Est_target) -7.764e-02
scale(MeanSaliencyDiff) -7.578e-04
scale(image_sim):scale(age_in_months) -1.457e-02
scale(image_sim):scale(AoA_Est_target) 2.402e-02
scale(age_in_months):scale(AoA_Est_target) -4.268e-02
scale(image_sim):scale(age_in_months):scale(AoA_Est_target) 1.290e-02
Std. Error
(Intercept) 3.306e-02
scale(image_sim) 2.421e-02
scale(age_in_months) 2.464e-02
scale(AoA_Est_target) 2.904e-02
scale(MeanSaliencyDiff) 2.658e-02
scale(image_sim):scale(age_in_months) 2.051e-02
scale(image_sim):scale(AoA_Est_target) 2.501e-02
scale(age_in_months):scale(AoA_Est_target) 2.035e-02
scale(image_sim):scale(age_in_months):scale(AoA_Est_target) 2.013e-02
df t value
(Intercept) 2.646e+01 -0.519
scale(image_sim) 1.417e+02 -2.108
scale(age_in_months) 9.931e+01 2.221
scale(AoA_Est_target) 2.259e+01 -2.673
scale(MeanSaliencyDiff) 3.546e+01 -0.029
scale(image_sim):scale(age_in_months) 2.379e+03 -0.711
scale(image_sim):scale(AoA_Est_target) 6.964e+01 0.960
scale(age_in_months):scale(AoA_Est_target) 2.381e+03 -2.098
scale(image_sim):scale(age_in_months):scale(AoA_Est_target) 2.380e+03 0.641
Pr(>|t|)
(Intercept) 0.6080
scale(image_sim) 0.0368 *
scale(age_in_months) 0.0286 *
scale(AoA_Est_target) 0.0137 *
scale(MeanSaliencyDiff) 0.9774
scale(image_sim):scale(age_in_months) 0.4773
scale(image_sim):scale(AoA_Est_target) 0.3402
scale(age_in_months):scale(AoA_Est_target) 0.0361 *
scale(image_sim):scale(age_in_months):scale(AoA_Est_target) 0.5218
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__) s(AA_E s(MSD) sc(_):(__) s(_):(A s(__):
scal(mg_sm) 0.031
scl(g_n_mn) 0.001 -0.008
scl(AA_Es_) 0.000 -0.258 0.010
scl(MnSlnD) 0.025 0.046 -0.008 0.014
scl(_):(__) -0.006 -0.004 0.036 0.001 -0.006
s(_):(AA_E_ -0.248 -0.136 0.002 0.041 -0.022 -0.002
s(__):(AA_E 0.012 0.000 -0.014 0.008 -0.003 -0.260 -0.005
s(_):(__):( 0.003 -0.002 -0.210 -0.001 -0.008 -0.136 0.002 0.066Baseline window as a covariate
baseline_covariate_looking_text <- lmer(scale(mean_target_looking_critical_window) ~ scale(age_in_months)*scale(text_similarity)
+ scale(AoA_Est_target)
+ scale(MeanSaliencyDiff)
+ scale(mean_target_looking_baseline_window)
+ (scale(text_similarity) || SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = trials_with_effect_vars)
baseline_covariate_looking_image <- lmer(scale(mean_target_looking_critical_window) ~ scale(age_in_months)*scale(image_similarity)
+ scale(AoA_Est_target)
+ scale(MeanSaliencyDiff)
+ scale(mean_target_looking_baseline_window)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = trials_with_effect_vars)
summary(baseline_covariate_looking_text)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula: scale(mean_target_looking_critical_window) ~ scale(age_in_months) *
scale(text_similarity) + scale(AoA_Est_target) + scale(MeanSaliencyDiff) +
scale(mean_target_looking_baseline_window) + (scale(text_similarity) ||
SubjectInfo.subjID) + (1 | Trials.targetImage)
Data: trials_with_effect_vars
REML criterion at convergence: 6694.9
Scaled residuals:
Min 1Q Median 3Q Max
-2.73407 -0.72246 0.07168 0.76240 2.40301
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.031728 0.17812
SubjectInfo.subjID.1 scale(text_similarity) 0.002694 0.05191
Trials.targetImage (Intercept) 0.021091 0.14523
Residual 0.828964 0.91047
Number of obs: 2476, groups: SubjectInfo.subjID, 91; Trials.targetImage, 24
Fixed effects:
Estimate Std. Error df
(Intercept) 0.00178 0.03998 32.47778
scale(age_in_months) 0.06043 0.02616 88.75213
scale(text_similarity) -0.04923 0.02577 108.22062
scale(AoA_Est_target) -0.13464 0.03426 22.77036
scale(MeanSaliencyDiff) 0.02856 0.02983 51.98623
scale(mean_target_looking_baseline_window) 0.28182 0.01871 2433.15399
scale(age_in_months):scale(text_similarity) -0.03113 0.01912 86.90694
t value Pr(>|t|)
(Intercept) 0.045 0.964756
scale(age_in_months) 2.310 0.023190 *
scale(text_similarity) -1.910 0.058774 .
scale(AoA_Est_target) -3.930 0.000679 ***
scale(MeanSaliencyDiff) 0.957 0.342807
scale(mean_target_looking_baseline_window) 15.065 < 2e-16 ***
scale(age_in_months):scale(text_similarity) -1.628 0.107047
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) sc(__) scl(_) s(AA_E s(MSD) s(____
scl(g_n_mn) 0.004
scl(txt_sm) -0.001 -0.007
scl(AA_Es_) 0.013 0.007 -0.118
scl(MnSlnD) 0.026 -0.008 0.056 0.025
scl(mn____) -0.011 0.007 -0.005 0.027 -0.023
scl(__):(_) -0.006 0.008 -0.003 0.002 -0.005 -0.007summary(baseline_covariate_looking_image)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula: scale(mean_target_looking_critical_window) ~ scale(age_in_months) *
scale(image_similarity) + scale(AoA_Est_target) + scale(MeanSaliencyDiff) +
scale(mean_target_looking_baseline_window) + (1 | SubjectInfo.subjID) +
(1 | Trials.targetImage)
Data: trials_with_effect_vars
REML criterion at convergence: 6698.5
Scaled residuals:
Min 1Q Median 3Q Max
-2.7150 -0.7282 0.0709 0.7634 2.4479
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.03168 0.1780
Trials.targetImage (Intercept) 0.02198 0.1483
Residual 0.83259 0.9125
Number of obs: 2476, groups: SubjectInfo.subjID, 91; Trials.targetImage, 24
Fixed effects:
Estimate Std. Error df
(Intercept) 1.294e-03 4.046e-02 3.178e+01
scale(age_in_months) 6.047e-02 2.617e-02 8.873e+01
scale(image_similarity) -2.816e-02 2.519e-02 1.955e+02
scale(AoA_Est_target) -1.335e-01 3.545e-02 2.359e+01
scale(MeanSaliencyDiff) 2.846e-02 3.026e-02 4.938e+01
scale(mean_target_looking_baseline_window) 2.822e-01 1.874e-02 2.437e+03
scale(age_in_months):scale(image_similarity) -2.678e-02 1.834e-02 2.377e+03
t value Pr(>|t|)
(Intercept) 0.032 0.974690
scale(age_in_months) 2.311 0.023178 *
scale(image_similarity) -1.118 0.264932
scale(AoA_Est_target) -3.766 0.000971 ***
scale(MeanSaliencyDiff) 0.941 0.351522
scale(mean_target_looking_baseline_window) 15.061 < 2e-16 ***
scale(age_in_months):scale(image_similarity) -1.460 0.144504
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) sc(__) scl(_) s(AA_E s(MSD) s(____
scl(g_n_mn) 0.004
scl(mg_sml) -0.002 -0.006
scl(AA_Es_) 0.013 0.007 -0.227
scl(MnSlnD) 0.026 -0.008 0.105 0.007
scl(mn____) -0.010 0.007 -0.034 0.033 -0.025
scl(__):(_) 0.000 0.005 -0.005 0.002 -0.009 -0.005Similar predictions to our original model.
Adding window type as a covariate
trials_window_type_separated <- trials_with_effect_vars |>
pivot_longer(cols=c(mean_target_looking_critical_window, mean_target_looking_baseline_window), names_to="window_type", values_to="target_looking") |>
mutate(window_type = str_replace(window_type, "mean_target_looking_", "")) |>
mutate(trial_window_c = case_when(
window_type=="critical_window" ~ 0.5,
window_type=="baseline_window" ~ -0.5))
window_type_looking_text <- lmer(scale(target_looking) ~ scale(age_in_months)*trial_window_c*scale(text_similarity)
+ scale(AoA_Est_target)
+ scale(MeanSaliencyDiff)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = trials_window_type_separated)
window_type_looking_image <- lmer(scale(target_looking) ~ scale(age_in_months)*trial_window_c*scale(image_similarity)
+ scale(AoA_Est_target)
+ scale(MeanSaliencyDiff)
+ (scale(image_similarity) | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = trials_window_type_separated)
summary(window_type_looking_image)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula: scale(target_looking) ~ scale(age_in_months) * trial_window_c *
scale(image_similarity) + scale(AoA_Est_target) + scale(MeanSaliencyDiff) +
(scale(image_similarity) | SubjectInfo.subjID) + (1 | Trials.targetImage)
Data: trials_window_type_separated
REML criterion at convergence: 13896.5
Scaled residuals:
Min 1Q Median 3Q Max
-2.22883 -0.72198 0.02836 0.76614 2.04846
Random effects:
Groups Name Variance Std.Dev. Corr
SubjectInfo.subjID (Intercept) 0.0118259 0.10875
scale(image_similarity) 0.0008384 0.02895 -0.21
Trials.targetImage (Intercept) 0.0254849 0.15964
Residual 0.9417824 0.97045
Number of obs: 4952, groups: SubjectInfo.subjID, 91; Trials.targetImage, 24
Fixed effects:
Estimate
(Intercept) 1.795e-02
scale(age_in_months) 2.402e-02
trial_window_c 1.763e-01
scale(image_similarity) 1.671e-02
scale(AoA_Est_target) -1.088e-01
scale(MeanSaliencyDiff) 3.169e-02
scale(age_in_months):trial_window_c 6.997e-02
scale(age_in_months):scale(image_similarity) -9.385e-03
trial_window_c:scale(image_similarity) -7.275e-02
scale(age_in_months):trial_window_c:scale(image_similarity) -2.974e-02
Std. Error
(Intercept) 3.749e-02
scale(age_in_months) 1.791e-02
trial_window_c 2.758e-02
scale(image_similarity) 2.075e-02
scale(AoA_Est_target) 3.481e-02
scale(MeanSaliencyDiff) 2.694e-02
scale(age_in_months):trial_window_c 2.758e-02
scale(age_in_months):scale(image_similarity) 1.413e-02
trial_window_c:scale(image_similarity) 2.759e-02
scale(age_in_months):trial_window_c:scale(image_similarity) 2.752e-02
df t value
(Intercept) 2.559e+01 0.479
scale(age_in_months) 8.639e+01 1.341
trial_window_c 4.740e+03 6.393
scale(image_similarity) 1.817e+02 0.806
scale(AoA_Est_target) 2.320e+01 -3.126
scale(MeanSaliencyDiff) 7.023e+01 1.176
scale(age_in_months):trial_window_c 4.740e+03 2.537
scale(age_in_months):scale(image_similarity) 8.480e+01 -0.664
trial_window_c:scale(image_similarity) 4.740e+03 -2.637
scale(age_in_months):trial_window_c:scale(image_similarity) 4.740e+03 -1.080
Pr(>|t|)
(Intercept) 0.63606
scale(age_in_months) 0.18342
trial_window_c 1.78e-10 ***
scale(image_similarity) 0.42157
scale(AoA_Est_target) 0.00472 **
scale(MeanSaliencyDiff) 0.24356
scale(age_in_months):trial_window_c 0.01123 *
scale(age_in_months):scale(image_similarity) 0.50833
trial_window_c:scale(image_similarity) 0.00838 **
scale(age_in_months):trial_window_c:scale(image_similarity) 0.27998
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) sc(__) trl_w_ scl(_) s(AA_E s(MSD) sc(__):__ s(__):( t__:(_
scl(g_n_mn) 0.002
tril_wndw_c 0.000 0.000
scl(mg_sml) -0.009 -0.007 0.000
scl(AA_Es_) 0.015 0.006 0.000 -0.188
scl(MnSlnD) 0.028 -0.010 0.000 0.189 -0.005
scl(g__):__ 0.000 0.000 0.000 0.000 0.000 0.000
scl(__):(_) 0.000 -0.023 0.000 -0.003 0.002 -0.008 0.000
trl_wn_:(_) 0.000 0.000 0.000 0.000 0.000 0.000 -0.001 0.000
s(__):__:(_ 0.000 0.000 0.000 0.000 0.000 0.000 0.006 0.000 -0.012tidy_model(window_type_looking_text)# A tibble: 10 × 8
effect Predictor b SE t df p.full p
<chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <chr>
1 fixed Intercept 0.02 0.04 0.48 23.8 6.38e- 1 0.638
2 fixed Age (scaled) 0.02 0.02 1.35 86.4 1.80e- 1 0.180
3 fixed trial_window_c 0.18 0.03 6.4 4828. 1.66e-10 <.001
4 fixed scale(text_similarity) -0.02 0.02 -1.17 272. 2.44e- 1 0.244
5 fixed scale(AoA_Est_target) -0.1 0.03 -2.85 20.8 9.71e- 3 <.01
6 fixed scale(MeanSaliencyDiff) 0.02 0.03 0.85 74.5 3.99e- 1 0.399
7 fixed scale(age_in_months):trial_wi… 0.07 0.03 2.55 4828. 1.09e- 2 <.05
8 fixed scale(age_in_months):scale(te… -0.01 0.01 -0.81 4860. 4.20e- 1 0.420
9 fixed trial_window_c:scale(text_sim… -0.06 0.03 -2.09 4828. 3.70e- 2 <.05
10 fixed scale(age_in_months):trial_wi… -0.04 0.03 -1.39 4828. 1.65e- 1 0.165The interaction makes sense, plotting the predictions here to try to understand what’s going on
trials_window_type_separated$predicted <- predict(window_type_looking_image)
# Plot interaction effect
ggplot(trials_window_type_separated, aes(x = image_similarity, y = predicted, color = factor(window_type))) +
geom_point(alpha = 0.5) + # Add points for raw data
geom_smooth(method = "lm", se = TRUE) + # Add regression lines
labs(title = "Interaction Between Trial Window & Image Similarity",
x = "Scaled Image Similarity",
y = "Predicted Target Looking",
color = "Trial Window") +
theme_minimal()`geom_smooth()` using formula = 'y ~ x'
This is also making me wonder whether we see any signal in how infants look at images in the baseline window. ## Baseline window looking
baseline_looking_image <- lmer(scale(mean_target_looking_baseline_window) ~ scale(AoA_Est_target)
+ scale(MeanSaliencyDiff)
+ (1 | SubjectInfo.subjID),
data = trials_with_effect_vars)boundary (singular) fit: see help('isSingular')summary(baseline_looking_image)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula: scale(mean_target_looking_baseline_window) ~ scale(AoA_Est_target) +
scale(MeanSaliencyDiff) + (1 | SubjectInfo.subjID)
Data: trials_with_effect_vars
REML criterion at convergence: 7031.3
Scaled residuals:
Min 1Q Median 3Q Max
-1.75217 -0.73601 -0.01836 0.76794 1.76245
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.0000 0.0000
Residual 0.9959 0.9979
Number of obs: 2476, groups: SubjectInfo.subjID, 91
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) -1.231e-15 2.006e-02 2.473e+03 0.000 1.00000
scale(AoA_Est_target) -4.386e-02 2.006e-02 2.473e+03 -2.187 0.02887 *
scale(MeanSaliencyDiff) 5.471e-02 2.006e-02 2.473e+03 2.727 0.00643 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) s(AA_E
scl(AA_Es_) 0.000
scl(MnSlnD) 0.000 0.005
optimizer (nloptwrap) convergence code: 0 (OK)
boundary (singular) fit: see help('isSingular')Can’t get this to not be not singular but still fun to see that saliency is predictive.
Using a shorter critical window
short_image_effect <- lmer(scale(corrected_target_looking_short) ~ scale(image_similarity)*scale(age_in_months)
+ scale(AoA_Est_target)
+ scale(MeanSaliencyDiff)
+ (1 | SubjectInfo.subjID)
+ (1|Trials.targetImage),
data = trials_with_effect_vars)
short_text_effect <- lmer(scale(corrected_target_looking_short) ~ scale(text_similarity)*scale(age_in_months)*scale(AoA_Est_target)
+ (1 | SubjectInfo.subjID)
+ (1 | Trials.targetImage),
data = trials_with_effect_vars)
summary(short_text_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula: scale(corrected_target_looking_short) ~ scale(text_similarity) *
scale(age_in_months) * scale(AoA_Est_target) + (1 | SubjectInfo.subjID) +
(1 | Trials.targetImage)
Data: trials_with_effect_vars
REML criterion at convergence: 7014.4
Scaled residuals:
Min 1Q Median 3Q Max
-2.64342 -0.64400 -0.00667 0.67117 2.67872
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.012906 0.11360
Trials.targetImage (Intercept) 0.008015 0.08952
Residual 0.963283 0.98147
Number of obs: 2476, groups: SubjectInfo.subjID, 91; Trials.targetImage, 24
Fixed effects:
Estimate
(Intercept) -0.01308
scale(text_similarity) -0.03885
scale(age_in_months) 0.07433
scale(AoA_Est_target) -0.07705
scale(text_similarity):scale(age_in_months) -0.02155
scale(text_similarity):scale(AoA_Est_target) 0.03069
scale(age_in_months):scale(AoA_Est_target) -0.04273
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 0.01107
Std. Error
(Intercept) 0.03002
scale(text_similarity) 0.02367
scale(age_in_months) 0.02313
scale(AoA_Est_target) 0.02731
scale(text_similarity):scale(age_in_months) 0.02019
scale(text_similarity):scale(AoA_Est_target) 0.02261
scale(age_in_months):scale(AoA_Est_target) 0.02018
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 0.01851
df
(Intercept) 26.89591
scale(text_similarity) 104.62642
scale(age_in_months) 91.05618
scale(AoA_Est_target) 24.18882
scale(text_similarity):scale(age_in_months) 2380.05572
scale(text_similarity):scale(AoA_Est_target) 61.81199
scale(age_in_months):scale(AoA_Est_target) 2387.00281
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 2379.59345
t value
(Intercept) -0.436
scale(text_similarity) -1.641
scale(age_in_months) 3.213
scale(AoA_Est_target) -2.822
scale(text_similarity):scale(age_in_months) -1.067
scale(text_similarity):scale(AoA_Est_target) 1.357
scale(age_in_months):scale(AoA_Est_target) -2.118
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 0.598
Pr(>|t|)
(Intercept) 0.66661
scale(text_similarity) 0.10376
scale(age_in_months) 0.00181 **
scale(AoA_Est_target) 0.00940 **
scale(text_similarity):scale(age_in_months) 0.28600
scale(text_similarity):scale(AoA_Est_target) 0.17959
scale(age_in_months):scale(AoA_Est_target) 0.03432 *
scale(text_similarity):scale(age_in_months):scale(AoA_Est_target) 0.54989
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__) s(AA_E sc(_):(__) s(_):(A s(__):
scl(txt_sm) -0.012
scl(g_n_mn) 0.002 -0.010
scl(AA_Es_) -0.011 -0.104 0.012
scl(_):(__) -0.009 -0.004 -0.004 0.002
s(_):(AA_E_ -0.110 0.091 0.003 0.181 0.010
s(__):(AA_E 0.013 0.002 -0.015 0.004 -0.051 -0.013
s(_):(__):( 0.003 0.009 -0.074 -0.011 0.192 0.000 0.204summary(short_image_effect)Linear mixed model fit by REML. t-tests use Satterthwaite's method [
lmerModLmerTest]
Formula: scale(corrected_target_looking_short) ~ scale(image_similarity) *
scale(age_in_months) + scale(AoA_Est_target) + scale(MeanSaliencyDiff) +
(1 | SubjectInfo.subjID) + (1 | Trials.targetImage)
Data: trials_with_effect_vars
REML criterion at convergence: 7009.8
Scaled residuals:
Min 1Q Median 3Q Max
-2.70462 -0.65711 -0.01135 0.66973 2.73256
Random effects:
Groups Name Variance Std.Dev.
SubjectInfo.subjID (Intercept) 0.012707 0.11273
Trials.targetImage (Intercept) 0.008632 0.09291
Residual 0.965452 0.98257
Number of obs: 2476, groups: SubjectInfo.subjID, 91; Trials.targetImage, 24
Fixed effects:
Estimate Std. Error df
(Intercept) -8.088e-03 3.027e-02 2.554e+01
scale(image_similarity) -4.193e-02 2.401e-02 1.137e+02
scale(age_in_months) 7.523e-02 2.304e-02 9.017e+01
scale(AoA_Est_target) -7.588e-02 2.806e-02 2.347e+01
scale(MeanSaliencyDiff) 1.022e-02 2.588e-02 3.653e+01
scale(image_similarity):scale(age_in_months) -2.373e-02 1.973e-02 2.387e+03
t value Pr(>|t|)
(Intercept) -0.267 0.79145
scale(image_similarity) -1.746 0.08343 .
scale(age_in_months) 3.265 0.00155 **
scale(AoA_Est_target) -2.704 0.01252 *
scale(MeanSaliencyDiff) 0.395 0.69507
scale(image_similarity):scale(age_in_months) -1.203 0.22928
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) scl(_) sc(__) s(AA_E s(MSD)
scl(mg_sml) -0.004
scl(g_n_mn) 0.002 -0.005
scl(AA_Es_) 0.010 -0.257 0.010
scl(MnSlnD) 0.020 0.024 -0.009 0.019
scl(_):(__) 0.000 -0.009 0.006 0.004 -0.010