11indecsionmodel.Rmd

---
title: "Indecision model "
output: html_notebook
editor_options: 
  chunk_output_type: inline
---

### Reading libraries and parameters. Preparing the data

```{r message=FALSE}
library(tidyverse)
library(cowplot)

list.files("R", full.names = TRUE) %>% walk(source)
source("graphical_parameters.R")
source("parameters.R")


load(file = "logdata/sym_averages_s_vs_d_best.RData")
load(file = "logdata/sym_curves_s_vs_d_best.RData")
load(file = "logdata/best.RData")
load(file = "logdata/best_sym_absent_lapses_same_slope.RData")
load(file = "logdata/best_sym_no_absent_lapses_same_slope.RData")
load(file = "logdata/best_sym_no_absent_lapses_no_same_slope.RData")
load(file = "logdata/best_sym_no_same_guess.RData")

load(file = "logdata/sym_averages_s_vs_d_best2.RData")
load(file = "logdata/sym_curves_s_vs_d_best2.RData")
load(file = "logdata/best2.RData")



```

### Experiment 1


#### Absent lapses same slope 

```{r}
fun_sym_absent_lapses_same_slope_gp <-  dat_sym  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    
    function(x, p) 0.01 + (1 - 2 * 0.01) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[5] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3]))),
    function(x, p) 0.01 + (1 - 2 * 0.01) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[5]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3])))

    )))

fit_sym_absent_lapses_same_slope_gp <- quickpsy(dat_sym %>% 
                                                  semi_join(
                                                    best %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, vertical)) %>% 
                                                  semi_join(best_sym_absent_lapses_same_slope), 
                               orientation, response, 
                grouping = .(subject, vertical, references),
                fun = fun_sym_absent_lapses_same_slope_gp,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(-1.5, 0), c(0, 1)),
                bootstrap = "none")

model_selection_aic(fit_sym_absent_lapses_same_slope_gp$aic,
                    fit_sym_absent_lapses_same_slope$aic %>% 
                      semi_join(best %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, vertical)) %>% 
                      semi_join(best_sym_absent_lapses_same_slope))

### symetric crit 
fun_sym_absent_lapses_same_slope_gp2 <-  dat_sym  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    
    function(x, p) 0.01 + (1 - 2 * 0.01) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[4] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x,  p[2] + p[1], p[3]))),
    function(x, p) 0.01 + (1 - 2 * 0.01) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[4]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x,  p[2] + p[1], p[3])))

    )))

fit_sym_absent_lapses_same_slope_gp2 <- quickpsy(dat_sym %>% 
                                                  semi_join(
                                                    best %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, vertical)) %>% 
                                                  semi_join(best_sym_absent_lapses_same_slope), 
                               orientation, response, 
                grouping = .(subject, vertical, references),
                fun = fun_sym_absent_lapses_same_slope_gp2,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(0, 1)),
                bootstrap = "none")

model_selection_aic(fit_sym_absent_lapses_same_slope_gp2$aic,
                    fit_sym_absent_lapses_same_slope$aic %>% 
                      semi_join(best %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, vertical)) %>% 
                      semi_join(best_sym_absent_lapses_same_slope))
```



```{r fig.height=12, fig.width=10}
ggplot() + facet_wrap(subject ~ vertical, ncol = 3) +
  geom_point(data = sym_averages_s_vs_d_best %>% filter(best == "decision" | best == "full"), 
             aes(x = orientation, y = prob, color = references)) +
  geom_line(data = sym_curves_s_vs_d_best %>% filter(best == "decision"| best == "full"), 
            aes(x = x, y = y, color = references, lty = "SDT")) +
    geom_line(data = fit_sym_absent_lapses_same_slope_gp$curves, 
            aes(x = x, y = y, color = references, lty = "Indecision model")) +
      geom_line(data = fit_sym_absent_lapses_same_slope_gp2$curves, 
            aes(x = x, y = y, color = references, lty = "Indecision model 2")) +
  theme_grey() + theme(legend.position = "top") 
```


#### Same guess same slope 

```{r}
fun_sym_same_guess_same_slope_gp <-  dat_sym  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    
    function(x, p) p[6] + (1 - 2 * p[6]) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[5] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3]))),
    function(x, p) p[6] + (1 - 2 * p[6]) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[5]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3])))

    )))

fit_sym_same_guess_same_slope_gp <- quickpsy(dat_sym %>% 
                                                  semi_join(
                                                    best %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, vertical)), 
                               orientation, response, 
                grouping = .(subject, vertical, references),
                fun = fun_sym_same_guess_same_slope_gp,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(-1.5, 0), c(0, 1), pini_lapse),
                bootstrap = "none")



model_selection_aic(fit_sym_same_guess_same_slope_gp$aic %>% 
                      semi_join(best_sym_no_absent_lapses_same_slope),
                    fit_sym_same_guess_same_slope$aic %>% 
                      semi_join(best %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, vertical)) %>% 
                      semi_join(best_sym_no_absent_lapses_same_slope))

### symetric crit 
fun_sym_same_guess_same_slope_gp2 <-  dat_sym  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    
    function(x, p) p[5] + (1 - 2 * p[5]) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[4] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3]))),
    function(x, p) p[5] + (1 - 2 * p[5]) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[4]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3])))

    )))

fit_sym_same_guess_same_slope_gp2 <- quickpsy(dat_sym %>% 
                                                  semi_join(
                                                    best %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, vertical)), 
                               orientation, response, 
                grouping = .(subject, vertical, references),
                fun = fun_sym_same_guess_same_slope_gp2,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(0, 1), pini_lapse),
                bootstrap = "none")


model_selection_aic(fit_sym_same_guess_same_slope_gp2$aic %>% 
                      semi_join(best_sym_no_absent_lapses_same_slope),
                    fit_sym_same_guess_same_slope$aic %>% 
                      semi_join(best %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, vertical)) %>% 
                      semi_join(best_sym_no_absent_lapses_same_slope))

```

```{r fig.height=12, fig.width=10}
ggplot() + facet_wrap(subject ~ vertical, ncol = 3) +
  geom_point(data = sym_averages_s_vs_d_best %>% filter(best == "decision" | best == "full"), 
             aes(x = orientation, y = prob, color = references)) +
  geom_line(data = sym_curves_s_vs_d_best %>% filter(best == "decision"| best == "full"), 
            aes(x = x, y = y, color = references, lty = "SDT")) +
  geom_line(data = fit_sym_same_guess_same_slope_gp$curves %>% 
                semi_join(best_sym_no_absent_lapses_same_slope), 
            aes(x = x, y = y, color = references, lty = "Indecision model")) +
  geom_line(data = fit_sym_same_guess_same_slope_gp2$curves %>% 
              semi_join(best_sym_no_absent_lapses_same_slope),
            aes(x = x, y = y, color = references, lty = "Indecision model 2")) +
  theme_grey() + theme(legend.position = "top") 
```


#### Sym same guess

```{r}



fun_sym_same_guess_gp <-  dat_sym  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    
    function(x, p) p[6] + (1 - 2 * p[6]) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[5] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3]))),
    function(x, p) p[6] + (1 - 2 * p[6]) *
    (pnorm(x,  -p[1] + p[2], p[7]) + 
       (1 - p[5]) * (pnorm(- x, - p[2] + p[1], p[7]) - pnorm(-x, - p[4] + p[1], p[7])))

    )))

fit_sym_same_guess_gp <- quickpsy(dat_sym %>% 
                                                  semi_join(
                                                    best %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, vertical)), 
                               orientation, response, 
                grouping = .(subject, vertical, references),
                fun = fun_sym_same_guess_gp,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(-1.5, 0), 
                              c(0, 1), pini_lapse, pini_scale),
                bootstrap = "none")


model_selection_aic(fit_sym_same_guess_gp$aic %>% 
                      semi_join(best_sym_no_absent_lapses_no_same_slope),
                    fit_sym_same_guess$aic %>% 
                      semi_join(best %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, vertical)) %>% 
                      semi_join(best_sym_no_absent_lapses_no_same_slope))

### symetric crit 

fun_sym_same_guess_gp2 <-  dat_sym  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    
    function(x, p) p[5] + (1 - 2 * p[5]) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[4] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3]))),
    function(x, p) p[5] + (1 - 2 * p[5]) *
    (pnorm(x,  -p[1] + p[2], p[7]) + 
       (1 - p[4]) * (pnorm(- x, - p[2] + p[1], p[6]) - pnorm(-x, + p[2] + p[1], p[6])))

    )))

fit_sym_same_guess_gp2 <- quickpsy(dat_sym %>% 
                                                  semi_join(
                                                    best %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, vertical)), 
                               orientation, response, 
                grouping = .(subject, vertical, references),
                fun = fun_sym_same_guess_gp2,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(-1.5, 0), 
                              c(0, 1), pini_lapse, pini_scale),
                bootstrap = "none")


model_selection_aic(fit_sym_same_guess_gp2$aic %>% 
                      semi_join(best_sym_no_absent_lapses_no_same_slope),
                    fit_sym_same_guess$aic %>% 
                      semi_join(best %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, vertical)) %>% 
                      semi_join(best_sym_no_absent_lapses_no_same_slope))

```

```{r fig.height=12, fig.width=10}
ggplot() + facet_wrap(subject ~ vertical, ncol = 3) +
  geom_point(data = sym_averages_s_vs_d_best %>% filter(best == "decision" | best == "full"), 
             aes(x = orientation, y = prob, color = references)) +
  geom_line(data = sym_curves_s_vs_d_best %>% filter(best == "decision"| best == "full"), 
            aes(x = x, y = y, color = references, lty = "SDT")) +
  geom_line(data = fit_sym_same_guess_gp$curves %>% 
                semi_join(best_sym_no_absent_lapses_no_same_slope), 
            aes(x = x, y = y, color = references, lty = "Indecision model")) +
  geom_line(data = fit_sym_same_guess_gp2$curves %>%
              semi_join(best_sym_no_absent_lapses_no_same_slope),
            aes(x = x, y = y, color = references, lty = "Indecision model 2")) +
  theme_grey() + theme(legend.position = "top") 
```


#### Same guess same slope

```{r}

fun_sym_guess_same_slope_gp <-  dat_sym  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    
    function(x, p) p[6] + (1 - 2 * p[6]) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[5] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3]))),
    function(x, p) p[7] + (1 - 2 * p[7]) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[5]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3])))

    )))

fit_sym_guess_same_slope_gp <- quickpsy(dat_sym %>% 
                                                  semi_join(
                                                    best %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, vertical)), 
                               orientation, response, 
                grouping = .(subject, vertical, references),
                fun = fun_sym_guess_same_slope_gp,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(-1.5, 0), 
                              c(0, 1), pini_lapse, pini_lapse),
                bootstrap = "none")


model_selection_aic(fit_sym_guess_same_slope_gp$aic %>% 
                      semi_join(best_sym_no_same_guess),
                    fit_sym_guess_same_slope$aic %>% 
                      semi_join(best %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, vertical)) %>% 
                      semi_join(best_sym_no_same_guess))

### symetric crit 

fun_sym_guess_same_slope_gp2 <-  dat_sym  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    
    function(x, p) p[5] + (1 - 2 * p[5]) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[4] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3]))),
    function(x, p) p[6] + (1 - 2 * p[6]) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[4]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3])))

    )))

fit_sym_guess_same_slope_gp2 <- quickpsy(dat_sym %>% 
                                                  semi_join(
                                                    best %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, vertical)), 
                               orientation, response, 
                grouping = .(subject, vertical, references),
                fun = fun_sym_guess_same_slope_gp2,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, 
                              c(0, 1), pini_lapse, pini_lapse),
                bootstrap = "none")


model_selection_aic(fit_sym_guess_same_slope_gp2$aic %>% 
                      semi_join(best_sym_no_same_guess),
                    fit_sym_guess_same_slope$aic %>% 
                      semi_join(best %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, vertical)) %>% 
                      semi_join(best_sym_no_same_guess))

```

```{r fig.height=12, fig.width=10}
ggplot() + facet_wrap(subject ~ vertical, ncol = 3) +
  geom_point(data = sym_averages_s_vs_d_best %>% filter(best == "decision" | best == "full"), 
             aes(x = orientation, y = prob, color = references)) +
  geom_line(data = sym_curves_s_vs_d_best %>% filter(best == "decision"| best == "full"), 
            aes(x = x, y = y, color = references, lty = "SDT")) +
  geom_line(data = fit_sym_guess_same_slope_gp$curves %>% 
                semi_join(best_sym_no_same_guess), 
            aes(x = x, y = y, color = references, lty = "Indecision model")) +
  geom_line(data = fit_sym_guess_same_slope_gp2$curves %>%
              semi_join(best_sym_no_same_guess),
            aes(x = x, y = y, color = references, lty = "Indecision model 2")) +
  theme_grey() + theme(legend.position = "top") 
```


### Experiment 2


#### Absent lapses same slope 

```{r}
fun_sym_absent_lapses_same_slope2_gp <-  dat2  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    
    function(x, p) 0.01 + (1 - 2 * 0.01) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[5] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3]))),
    function(x, p) 0.01 + (1 - 2 * 0.01) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[5]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3])))

    )))

fit_sym_absent_lapses_same_slope2_gp <- quickpsy(dat2 %>% 
                                                  semi_join(
                                                    best2 %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, cond_per_block)), 
                               orientation, response, 
                grouping = .(subject, references, cond_per_block),
                fun = fun_sym_absent_lapses_same_slope_gp,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(-1.5, 0), c(0, 1)),
                bootstrap = "none")

model_selection_aic(fit_sym_absent_lapses_same_slope2_gp$aic %>% 
                      semi_join(best_sym_absent_lapses_same_slope2),
                    fit_sym_absent_lapses_same_slope2$aic %>% 
                      semi_join(best2 %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, cond_per_block)) %>% 
                      semi_join(best_sym_absent_lapses_same_slope2))

### symetric crit 
fun_sym_absent_lapses_same_slope2_gp2 <-  dat2  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    
    function(x, p) 0.01 + (1 - 2 * 0.01) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[4] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3]))),
    function(x, p) 0.01 + (1 - 2 * 0.01) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[4]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3])))

    )))

fit_sym_absent_lapses_same_slope2_gp2 <- quickpsy(dat2 %>% 
                                                  semi_join(
                                                    best2 %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, cond_per_block)), 
                               orientation, response, 
                grouping = .(subject, references, cond_per_block),
                fun = fun_sym_absent_lapses_same_slope_gp2,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(0, 1)),
                bootstrap = "none")

model_selection_aic(fit_sym_absent_lapses_same_slope2_gp2$aic %>% 
                      semi_join(best_sym_absent_lapses_same_slope2),
                    fit_sym_absent_lapses_same_slope2$aic %>% 
                      semi_join(best2 %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, cond_per_block)) %>% 
                      semi_join(best_sym_absent_lapses_same_slope2))

```

```{r fig.height=12, fig.width=10}
ggplot() + facet_wrap(subject ~ cond_per_block, ncol = 3) +
  geom_point(data = sym_averages_s_vs_d_best2 %>% filter(best == "decision" | best == "full"), 
             aes(x = orientation, y = prob, color = references)) +
  geom_line(data = sym_curves_s_vs_d_best2 %>% filter(best == "decision"| best == "full"), 
            aes(x = x, y = y, color = references, lty = "SDT")) +
  geom_line(data = fit_sym_absent_lapses_same_slope2_gp$curves %>% 
                semi_join(best_sym_absent_lapses_same_slope2), 
            aes(x = x, y = y, color = references, lty = "Indecision model")) +
  geom_line(data = fit_sym_absent_lapses_same_slope2_gp2$curves %>%
              semi_join(best_sym_absent_lapses_same_slope2),
            aes(x = x, y = y, color = references, lty = "Indecision model 2")) +
  theme_grey() + theme(legend.position = "top") 
```

#### Sym absent lapses

```{r}


fun_sym_absent_lapses2_gp <-  dat2  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    
    function(x, p) 0.01 + (1 - 2 * 0.01) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[5] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3]))),
    function(x, p) 0.01 + (1 - 2 * 0.01) *
    (pnorm(x,  -p[1] + p[2], p[6]) + 
       (1 - p[5]) * (pnorm(- x, - p[2] + p[1], p[6]) - pnorm(-x, - p[4] + p[1], p[6])))

    )))

fit_sym_absent_lapses2_gp <- quickpsy(dat2 %>% 
                                                  semi_join(
                                                    best2 %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, cond_per_block)), 
                               orientation, response, 
                grouping = .(subject, references, cond_per_block),
                fun = fun_sym_absent_lapses_same_slope_gp,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(-1.5, 0), c(0, 1),
                              pini_scale),
                bootstrap = "none")

model_selection_aic(fit_sym_absent_lapses2_gp$aic %>% 
                      semi_join(best_sym_absent_lapses_no_same_slope2),
                    fit_sym_absent_lapses2$aic %>% 
                      semi_join(best2 %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, cond_per_block)) %>% 
                      semi_join(best_sym_absent_lapses_no_same_slope2))

### symetric crit 

fun_sym_absent_lapses2_gp2 <-  dat2  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    
    function(x, p) 0.01 + (1 - 2 * 0.01) *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[4] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3]))),
    function(x, p) 0.01 + (1 - 2 * 0.01) *
    (pnorm(x,  -p[1] + p[2], p[5]) + 
       (1 - p[4]) * (pnorm(- x, - p[2] + p[1], p[6]) - pnorm(-x, + p[2] + p[1], p[5])))

    )))

fit_sym_absent_lapses2_gp2 <- quickpsy(dat2 %>% 
                                                  semi_join(
                                                    best2 %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, cond_per_block)), 
                               orientation, response, 
                grouping = .(subject, references, cond_per_block),
                fun = fun_sym_absent_lapses_same_slope_gp2,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale,  c(0, 1),
                              pini_scale),
                bootstrap = "none")

model_selection_aic(fit_sym_absent_lapses2_gp2$aic %>% 
                      semi_join(best_sym_absent_lapses_no_same_slope2),
                    fit_sym_absent_lapses2$aic %>% 
                      semi_join(best2 %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, cond_per_block)) %>% 
                      semi_join(best_sym_absent_lapses_no_same_slope2))

```

```{r fig.height=12, fig.width=10}
ggplot() + facet_wrap(subject ~ cond_per_block, ncol = 3) +
  geom_point(data = sym_averages_s_vs_d_best2 %>% filter(best == "decision" | best == "full"), 
             aes(x = orientation, y = prob, color = references)) +
  geom_line(data = sym_curves_s_vs_d_best2 %>% filter(best == "decision"| best == "full"), 
            aes(x = x, y = y, color = references, lty = "SDT")) +
  geom_line(data = fit_sym_absent_lapses2_gp$curves %>% 
                semi_join(best_sym_absent_lapses_no_same_slope2), 
            aes(x = x, y = y, color = references, lty = "Indecision model")) +
  geom_line(data = fit_sym_absent_lapses2_gp2$curves %>%
              semi_join(best_sym_absent_lapses_no_same_slope2),
            aes(x = x, y = y, color = references, lty = "Indecision model 2")) +
  theme_grey() + theme(legend.position = "top") 
```



#### Sym same guess same slope 

```{r}

fun_sym_same_guess_same_slope2_gp <-  dat2  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    function(x, p) p[6] + (1 - 2 * p[6])  *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[5] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3]))),
    function(x, p) p[6] + (1 - 2 * p[6])  *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[5]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3])))

    )))

fit_sym_same_guess_same_slope2_gp <- quickpsy(dat2 %>% 
                                                  semi_join(
                                                    best2 %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, cond_per_block)), 
                               orientation, response, 
                grouping = .(subject, references, cond_per_block),
                fun = fun_sym_same_guess_same_slope2_gp,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(-1.5, 0), c(0, 1), pini_lapse),
                bootstrap = "none")

model_selection_aic(fit_sym_same_guess_same_slope2_gp$aic %>% 
                      semi_join(best_sym_no_absent_lapses_same_slope2),
                    fit_sym_same_guess_same_slope2$aic %>% 
                      semi_join(best2 %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, cond_per_block)) %>% 
                      semi_join(best_sym_no_absent_lapses_same_slope2))

### symetric crit 

fun_sym_same_guess_same_slope2_gp2 <-  dat2  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    function(x, p) p[5] + (1 - 2 * p[5])  *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[4] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3]))),
    function(x, p) p[5] + (1 - 2 * p[5])  *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[4]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3])))

    )))

fit_sym_same_guess_same_slope2_gp2 <- quickpsy(dat2 %>% 
                                                  semi_join(
                                                    best2 %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, cond_per_block)), 
                               orientation, response, 
                grouping = .(subject, references, cond_per_block),
                fun = fun_sym_same_guess_same_slope2_gp2,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(0, 1), pini_lapse),
                bootstrap = "none")

model_selection_aic(fit_sym_same_guess_same_slope2_gp2$aic %>% 
                      semi_join(best_sym_no_absent_lapses_same_slope2),
                    fit_sym_same_guess_same_slope2$aic %>% 
                      semi_join(best2 %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, cond_per_block)) %>% 
                      semi_join(best_sym_no_absent_lapses_same_slope2))


```
```{r fig.height=12, fig.width=10}
ggplot() + facet_wrap(subject ~ cond_per_block, ncol = 3) +
  geom_point(data = sym_averages_s_vs_d_best2 %>% filter(best == "decision" | best == "full"), 
             aes(x = orientation, y = prob, color = references)) +
  geom_line(data = sym_curves_s_vs_d_best2 %>% filter(best == "decision"| best == "full"), 
            aes(x = x, y = y, color = references, lty = "SDT")) +
  geom_line(data = fit_sym_same_guess_same_slope2_gp$curves %>% 
                semi_join(best_sym_no_absent_lapses_same_slope2), 
            aes(x = x, y = y, color = references, lty = "Indecision model")) +
  geom_line(data = fit_sym_same_guess_same_slope2_gp2$curves %>%
              semi_join(best_sym_absent_lapses_no_same_slope2),
            aes(x = x, y = y, color = references, lty = "Indecision model 2")) +
  theme_grey() + theme(legend.position = "top") 
```

#### sym guess same slope  

```{r}
fun_sym_guess_same_slope2_gp <-  dat2  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    function(x, p) p[6] + (1 - 2 * p[6])  *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[5] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3]))),
    function(x, p) p[7] + (1 - 2 * p[7])  *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[5]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3])))

    )))

fit_sym_guess_same_slope2_gp <- quickpsy(dat2 %>% 
                                                  semi_join(
                                                    best2 %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, cond_per_block)), 
                               orientation, response, 
                grouping = .(subject, references, cond_per_block),
                fun = fun_sym_guess_same_slope2_gp,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(-1.5, 0), 
                              c(0, 1), pini_lapse, pini_lapse),
                bootstrap = "none")

model_selection_aic(fit_sym_guess_same_slope2_gp$aic %>% 
                      semi_join(best_sym_no_same_guess2),
                    fit_sym_guess_same_slope2$aic %>% 
                      semi_join(best2 %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, cond_per_block)) %>% 
                      semi_join(best_sym_no_same_guess2))

### symetric crit 
fun_sym_guess_same_slope2_gp2 <-  dat2  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    function(x, p) p[5] + (1 - 2 * p[5])  *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[4] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3]))),
    function(x, p) p[6] + (1 - 2 * p[6])  *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[4]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3])))

    )))

fit_sym_guess_same_slope2_gp2 <- quickpsy(dat2 %>% 
                                                  semi_join(
                                                    best2 %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, cond_per_block)), 
                               orientation, response, 
                grouping = .(subject, references, cond_per_block),
                fun = fun_sym_guess_same_slope2_gp2,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale,  
                              c(0, 1), pini_lapse, pini_lapse),
                bootstrap = "none")

model_selection_aic(fit_sym_guess_same_slope2_gp2$aic %>% 
                      semi_join(best_sym_no_same_guess2),
                    fit_sym_guess_same_slope2$aic %>% 
                      semi_join(best2 %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, cond_per_block)) %>% 
                      semi_join(best_sym_no_same_guess2))


```

```{r fig.height=12, fig.width=10}
ggplot() + facet_wrap(subject ~ cond_per_block, ncol = 3) +
  geom_point(data = sym_averages_s_vs_d_best2 %>% filter(best == "decision" | best == "full"), 
             aes(x = orientation, y = prob, color = references)) +
  geom_line(data = sym_curves_s_vs_d_best2 %>% filter(best == "decision"| best == "full"), 
            aes(x = x, y = y, color = references, lty = "SDT")) +
  geom_line(data = fit_sym_guess_same_slope2_gp$curves %>% 
                semi_join(best_sym_no_same_guess2), 
            aes(x = x, y = y, color = references, lty = "Indecision model")) +
  geom_line(data = fit_sym_guess_same_slope2_gp2$curves %>%
              semi_join(best_sym_no_same_guess2),
            aes(x = x, y = y, color = references, lty = "Indecision model 2")) +
  theme_grey() + theme(legend.position = "top") 
```


#### sym no two guess same slope 

```{r}

fun_sym_no_two_guess_same_slope2_gp <-  dat2  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    function(x, p) p[6] + (1 - 2 * p[7])  *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[5] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3]))),
    function(x, p) p[8] + (1 - 2 * p[9])  *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[5]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, - p[4] + p[1], p[3])))

    )))

fit_sym_no_two_guess_same_slope2_gp <- quickpsy(dat2 %>% 
                                                  semi_join(
                                                    best2 %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, cond_per_block)), 
                               orientation, response, 
                grouping = .(subject, references, cond_per_block),
                fun = fun_sym_no_two_guess_same_slope2_gp,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale, c(-1.5, 0), 
                              c(0, 1), pini_lapse, pini_lapse, 
                              pini_lapse, pini_lapse),
                bootstrap = "none")

model_selection_aic(fit_sym_no_two_guess_same_slope2_gp$aic %>% 
                      semi_join(best_sym_no_two_guess_same_slope2),
                    fit_sym_no_two_guess_same_slope2$aic %>% 
                      semi_join(best2 %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, cond_per_block)) %>% 
                      semi_join(best_sym_no_two_guess_same_slope2))

### symetric crit 

fun_sym_no_two_guess_same_slope2_gp2 <-  dat2  %>% 
  distinct(references) %>% 
  bind_cols(tibble(fun = c(
    function(x, p) p[5] + (1 - 2 * p[6])  *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       p[4] * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3]))),
    function(x, p) p[7] + (1 - 2 * p[8])  *
    (pnorm(x,  -p[1] + p[2], p[3]) + 
       (1 - p[4]) * (pnorm(- x, - p[2] + p[1], p[3]) - pnorm(-x, + p[2] + p[1], p[3])))

    )))

fit_sym_no_two_guess_same_slope2_gp2 <- quickpsy(dat2 %>% 
                                                  semi_join(
                                                    best2 %>% 
                                                      filter(best == "decision" | best == "full") %>% 
                                                      select(subject, cond_per_block)), 
                               orientation, response, 
                grouping = .(subject, references, cond_per_block),
                fun = fun_sym_no_two_guess_same_slope2_gp2,
                xmin = -3, xmax = 3,
                parini = list(pini_origin, c(0, 1.5), pini_scale,  
                              c(0, 1), pini_lapse, pini_lapse, 
                              pini_lapse, pini_lapse),
                bootstrap = "none")

model_selection_aic(fit_sym_no_two_guess_same_slope2_gp2$aic %>% 
                      semi_join(best_sym_no_two_guess_same_slope2),
                    fit_sym_no_two_guess_same_slope2$aic %>% 
                      semi_join(best2 %>% 
                                  filter(best == "decision" | best == "full") %>% 
                                  select(subject, cond_per_block)) %>% 
                      semi_join(best_sym_no_two_guess_same_slope2))



```

```{r fig.height=12, fig.width=10}
ggplot() + facet_wrap(subject ~ cond_per_block, ncol = 3) +
  geom_point(data = sym_averages_s_vs_d_best2 %>% filter(best == "decision" | best == "full"), 
             aes(x = orientation, y = prob, color = references)) +
  geom_line(data = sym_curves_s_vs_d_best2 %>% filter(best == "decision"| best == "full"), 
            aes(x = x, y = y, color = references, lty = "SDT")) +
  geom_line(data = fit_sym_no_two_guess_same_slope2_gp$curves %>% 
                semi_join(best_sym_no_two_guess_same_slope2), 
            aes(x = x, y = y, color = references, lty = "Indecision model")) +
  geom_line(data = fit_sym_no_two_guess_same_slope2_gp2$curves %>%
              semi_join(best_sym_no_two_guess_same_slope2),
            aes(x = x, y = y, color = references, lty = "Indecision model 2")) +
  theme_grey() + theme(legend.position = "top") 
```