Sensorial Metacognition, Gabor pattern detection

Type 2 signal detection theory analysis using meta-d’ (Fondecyt 2017)

cat("\014")     # clean terminal



rm(list = ls()) # clean workspace
try(dev.off(), silent = TRUE) # close all plots
library(afex)
library(emmeans)
library(ggplot2)
library(GGally)

theme_set(
  theme_minimal()
)

master_dir <- '~/Insync/Drive/00EEG/Proyectos/Huepe/fdcyt_2017/gabormeta_huepe/behavioural'
data_dir   <- paste(master_dir, 'results',  sep = '/')
gabor_meta_data_name <- paste(data_dir, 'signal_params_r.csv', sep='/')
gabor_meta_data      <- read.csv(gabor_meta_data_name, header = TRUE)
gabor_meta_data$Sex     <- factor(gabor_meta_data$Sex)
gabor_meta_data$Group   <- factor(gabor_meta_data$Group)
gabor_meta_data$Stress  <- factor(gabor_meta_data$Stress)
gabor_meta_data$session <- factor(gabor_meta_data$session)
levels(gabor_meta_data$Sex)    <- list(female  = "F", male  = "M")
levels(gabor_meta_data$Group)  <- list(non_vulnerable  = "CN", vulnerable  = "EX")
levels(gabor_meta_data$Stress) <- list(not_stressed  = "NS", stressed  = "SS")
levels(gabor_meta_data$session) <- list(first  = "1", second  = "2")
gabor_meta_data_clean <- gabor_meta_data[gabor_meta_data$da > 0, ]
gabor_meta_data_clean <- gabor_meta_data_clean[gabor_meta_data_clean$meta_da > 0, ]
stress_response_name  <- paste(master_dir, 'stress_response.csv', sep='/')
stress_response_data  <- read.csv(stress_response_name, header = TRUE)
gabor_meta_data_clean <- merge(gabor_meta_data_clean, stress_response_data, by = "full.id")
gabor_meta_data_clean$stress_response <- factor(gabor_meta_data_clean$stress_response)
gabor_meta_data_clean$eliminate       <- factor(gabor_meta_data_clean$eliminate)
gabor_meta_data_clean <- gabor_meta_data_clean[gabor_meta_data_clean$eliminate == "no", ]
n_occur <- data.frame(table(gabor_meta_data_clean$full.id))
gabor_meta_data_clean <- gabor_meta_data_clean[gabor_meta_data_clean$full.id %in% n_occur$Var1[n_occur$Freq > 1],]
write.csv(gabor_meta_data_clean, file = paste(data_dir, "signal_params_r_clean.csv",  sep = '/'), row.names = FALSE)

1 General Description

Observations with negative d’ and/or negative meta-d’ have been eliminated.

Also subjects with weird cortisol levels (according to young Rojas Thomas) were eliminated.

options(width = 100)
mytable <- xtabs(~ Group + stress_response, data = gabor_meta_data_clean) / length(unique(gabor_meta_data_clean$session))
ftable(addmargins(mytable))

               stress_response no yes Sum
Group                                    
non_vulnerable                 11   5  16
vulnerable                      6  12  18
Sum                            17  17  34

summary(gabor_meta_data_clean)

   full.id             Subject        session         da                s        meta_da       
 Length:68          Min.   :10.00   first :34   Min.   :0.04083   Min.   :1   Min.   :0.08672  
 Class :character   1st Qu.:23.00   second:34   1st Qu.:1.60497   1st Qu.:1   1st Qu.:1.42272  
 Mode  :character   Median :38.50               Median :2.27427   Median :1   Median :2.15641  
                    Mean   :38.12               Mean   :2.37402   Mean   :1   Mean   :2.23626  
                    3rd Qu.:50.00               3rd Qu.:3.11005   3rd Qu.:1   3rd Qu.:2.94001  
                    Max.   :76.00               Max.   :4.97992   Max.   :1   Max.   :5.78546  
     M_diff             M_ratio           meta_ca           threshold            nR_S1a     
 Min.   :-2.772576   Min.   :0.03987   Min.   :-0.76953   Min.   :0.006576   Min.   :  0.0  
 1st Qu.:-0.459988   1st Qu.:0.80516   1st Qu.:-0.18449   1st Qu.:0.007388   1st Qu.: 62.0  
 Median :-0.004952   Median :0.99864   Median :-0.09727   Median :0.008121   Median :123.0  
 Mean   :-0.137766   Mean   :1.08311   Mean   :-0.06619   Mean   :0.008546   Mean   :129.9  
 3rd Qu.: 0.388688   3rd Qu.:1.20317   3rd Qu.:-0.03053   3rd Qu.:0.008920   3rd Qu.:198.0  
 Max.   : 1.598223   Max.   :5.58541   Max.   : 2.45080   Max.   :0.014718   Max.   :267.0  
     nR_S1b           nR_S1c           nR_S1d           nR_S2a            nR_S2b      
 Min.   :  0.00   Min.   :  0.00   Min.   : 0.000   Min.   :  0.000   Min.   :  0.00  
 1st Qu.: 30.00   1st Qu.: 10.75   1st Qu.: 1.000   1st Qu.:  0.000   1st Qu.:  7.00  
 Median : 83.00   Median : 23.00   Median : 3.500   Median :  1.500   Median : 21.00  
 Mean   : 78.03   Mean   : 34.68   Mean   : 7.971   Mean   :  8.544   Mean   : 26.69  
 3rd Qu.:126.25   3rd Qu.: 47.50   3rd Qu.: 9.000   3rd Qu.:  6.000   3rd Qu.: 39.00  
 Max.   :200.00   Max.   :133.00   Max.   :80.000   Max.   :229.000   Max.   :154.00  
     nR_S2c           nR_S2d          name               Sex                Group   
 Min.   :  0.00   Min.   :  1.0   Length:68          female:28   non_vulnerable:32  
 1st Qu.: 23.00   1st Qu.: 77.5   Class :character   male  :40   vulnerable    :36  
 Median : 78.00   Median :140.0   Mode  :character                                  
 Mean   : 79.57   Mean   :134.6                                                     
 3rd Qu.:128.75   3rd Qu.:194.5                                                     
 Max.   :238.00   Max.   :264.0                                                     
          Stress   stress_response eliminate
 not_stressed:30   no :34          no :68   
 stressed    :38   yes:34          yes: 0   
                                            
                                            
                                            
                                            

1.1 SDT parameters by stress response

options(width = 100)
sdt_params <- c('da', 'meta_da', 'M_diff')
sdt_params_pairs <- ggpairs(gabor_meta_data_clean,
                       columns = sdt_params,
                       aes(colour = stress_response, alpha = .25),
                       progress = FALSE,
                       lower = list(continuous = wrap("points")))
suppressWarnings(print(sdt_params_pairs))

summary(gabor_meta_data_clean[sdt_params])

       da             meta_da            M_diff         
 Min.   :0.04083   Min.   :0.08672   Min.   :-2.772576  
 1st Qu.:1.60497   1st Qu.:1.42272   1st Qu.:-0.459988  
 Median :2.27427   Median :2.15641   Median :-0.004952  
 Mean   :2.37402   Mean   :2.23626   Mean   :-0.137766  
 3rd Qu.:3.11005   3rd Qu.:2.94001   3rd Qu.: 0.388688  
 Max.   :4.97992   Max.   :5.78546   Max.   : 1.598223  

2 Analysis by stress response

2.1 Type 1 sensitivity, d’

options(width = 100)
d_prime_boxplot <- ggplot(gabor_meta_data_clean, aes(x = session, y = da, color = stress_response)) +
  geom_boxplot(outlier.shape = NA) +
  geom_point(position = position_jitterdodge()) +
  ggtitle("d'")
# d_prime_boxplot

rep_anova_d_prime = aov_ez("Subject", "da", data = gabor_meta_data_clean, within = c("session"), between = c("stress_response", "Group"))

Contrasts set to contr.sum for the following variables: stress_response, Group

nice(rep_anova_d_prime)

Anova Table (Type 3 tests)

Response: da
                         Effect    df  MSE    F   ges p.value
1               stress_response 1, 30 2.24 0.99  .026    .328
2                         Group 1, 30 2.24 1.54  .040    .224
3         stress_response:Group 1, 30 2.24 0.08  .002    .784
4                       session 1, 30 0.51 0.77  .005    .387
5       stress_response:session 1, 30 0.51 0.67  .004    .420
6                 Group:session 1, 30 0.51 0.02 <.001    .893
7 stress_response:Group:session 1, 30 0.51 0.38  .002    .541
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘+’ 0.1 ‘ ’ 1

d_prime_afex_plot <-
  afex_plot(
    rep_anova_d_prime,
    x = "session",
    trace = "stress_response",
    error = "between",
    error_arg = list(width = .15),
    dodge = -.5,
    mapping = c("color"),
    point_arg = list(size = 4)
  )

Warning: Panel(s) show a mixed within-between-design.
Error bars do not allow comparisons across all means.
Suppress error bars with: error = "none"

d_prime_afex_plot

# emmeans(rep_anova_d_prime, c("stress_response", "session"), contr = "pairwise")

2.2 Metacognitive sensitivity, meta-d’

options(width = 100)
meta_d_prime_boxplot <- ggplot(gabor_meta_data_clean, aes(x = session, y = meta_da, color = stress_response)) +
  geom_boxplot(outlier.shape = NA) +
  geom_point(position = position_jitterdodge()) +
  ggtitle("meta d'")
# meta_d_prime_boxplot

rep_anova_meta_d_prime = aov_ez("Subject", "meta_da", data = gabor_meta_data_clean, within = c("session"), between = c("stress_response", "Group"))

Contrasts set to contr.sum for the following variables: stress_response, Group

nice(rep_anova_meta_d_prime)

Anova Table (Type 3 tests)

Response: meta_da
                         Effect    df  MSE      F  ges p.value
1               stress_response 1, 30 2.21   1.16 .033    .290
2                         Group 1, 30 2.21 3.13 + .083    .087
3         stress_response:Group 1, 30 2.21   0.10 .003    .760
4                       session 1, 30 0.33 7.50 * .031    .010
5       stress_response:session 1, 30 0.33   0.56 .002    .460
6                 Group:session 1, 30 0.33   1.24 .005    .274
7 stress_response:Group:session 1, 30 0.33   0.38 .002    .543
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘+’ 0.1 ‘ ’ 1

meta_d_prime_afex_plot <-
  afex_plot(
    rep_anova_meta_d_prime,
    x = "session",
    trace = "stress_response",
    panel = "Group",
    error = "between",
    error_arg = list(width = .15),
    dodge = -.5,
    mapping = c("color"),
    point_arg = list(size = 4)
  )

Warning: Panel(s) show a mixed within-between-design.
Error bars do not allow comparisons across all means.
Suppress error bars with: error = "none"

meta_d_prime_afex_plot

# emmeans(rep_anova_meta_d_prime, c("Group"), contr = "pairwise")
emmeans(rep_anova_meta_d_prime, c("session"), contr = "pairwise")

$emmeans
 session emmean    SE df lower.CL upper.CL
 first     2.44 0.223 30     1.98     2.89
 second    2.03 0.190 30     1.64     2.42

Results are averaged over the levels of: stress_response, Group 
Confidence level used: 0.95 

$contrasts
 contrast       estimate    SE df t.ratio p.value
 first - second    0.407 0.149 30   2.738  0.0103

Results are averaged over the levels of: stress_response, Group 

2.3 Metacognitive efficiency, M_diff

options(width = 100)
mdiff_boxplot <- ggplot(gabor_meta_data_clean, aes(x = session, y = M_diff, color = stress_response)) +
  geom_boxplot(outlier.shape = NA) +
  geom_point(position = position_jitterdodge()) +
  ggtitle("meta-d' minus d'")
# mdiff_boxplot

rep_anova_m_diff = aov_ez("Subject", "M_diff", data = gabor_meta_data_clean, within = c("session"), between = c("stress_response", "Group"))

Contrasts set to contr.sum for the following variables: stress_response, Group

nice(rep_anova_m_diff)

Anova Table (Type 3 tests)

Response: M_diff
                         Effect    df  MSE    F   ges p.value
1               stress_response 1, 30 1.21 0.01 <.001    .920
2                         Group 1, 30 1.21 0.49  .011    .491
3         stress_response:Group 1, 30 1.21 0.00 <.001    .968
4                       session 1, 30 0.52 1.68  .017    .205
5       stress_response:session 1, 30 0.52 1.96  .019    .172
6                 Group:session 1, 30 0.52 1.03  .010    .318
7 stress_response:Group:session 1, 30 0.52 0.02 <.001    .900
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘+’ 0.1 ‘ ’ 1

m_diff_afex_plot <-
  afex_plot(
    rep_anova_m_diff,
    x = "session",
    trace = "stress_response",
    error = "within",
    error_arg = list(width = .15),
    dodge = -.5,
    mapping = c("color"),
    point_arg = list(size = 4)
  )

Warning: Panel(s) show a mixed within-between-design.
Error bars do not allow comparisons across all means.
Suppress error bars with: error = "none"

m_diff_afex_plot

# emmeans(rep_anova_m_diff, c("stress_response", "session"), contr = "pairwise")