Modeling psychophysical data: appropriate visualization to account for experimental design

Question

Background information:

I am modeling psychophysical data at the trial level for the first time. The data stem from a visual line bisection task with constant stimulus intensity, the response variable is bivariate (right vs. left) and is effect coded (0.5 vs. -0.5), and besides the experimental predictors, the model contains the stimulus category (true right vs. left).

According to Knoblauch & Maloney (2012) "Modeling Psychophysical Data in R", estimates of the model parameter in the linear model should so correspond to dprime values as they were more traditionally estimated separately for each participant.

(To ensure the basis of my model is correct I run the model (without the experimental predictors) for a single block for some single participants and compared the parameter estimates to the dprime values as I calculated them on the subject level. They were identical.)

Problem:

I am trying to visualize the model estimates (dprimes) and the factorial design at the same time in one plot.

I was able to plot values predicted by the model (using sjplot/ggeffects/afex). However, the predicted values are always on the scale of the response variable (which is, as mentioned above, effect coded with 0.5/-0.5) and so these values do not corresepond to the dprime values obtained the traditional way. The relations between the conditonal values make some sense, however not the absolute values.

I also drew whisker-plots to visualize the model estimates - but with these plots it seems impossible to visualize the factorial design at the same time.

Here's the model, just in case helps to understand why I'd like to visualize the factorial design along with the results and why I'd prefer a solution that does not involve summing up model parameters by hand (though I was able to extract them using broom.mixed::tidy(mymodel)). There are 4 experimental predictors in the model, plus the stimulus category. All predictors are factors and have only 2 levels. Two of the experimental predictors are within subject factors, the other two are between subjects factors:

my.model <- lme4::glmer(response ~ Condition + Starttime + PrePostTraining + FirstSecondInterval + stimulus +
                                   Condition:Starttime:PrePostTraining:FirstSecondInterval +
                                    (PrePostTraining + FirstSecondInterval|subjectCode),
                                family = binomial("probit"),
                                data = mydata)

Heres a draft plot to give an example of the plot I have in mind. However, this example plot was built with dprime values obtained the traditional way on subject level, and not with model parameters obtained at trial level as I intend to do now:

Concrete questions:

1. Would you at all consider plotting model estimates and the factorial design in one plot a good way of visualizing data in the case of dprime as the DV?
2. Is there a good way to do it?
3. If the only way to do it is by summing up intercepts and estimates by hand/code, how to deal with the predictor of the stimulus category?
4. To plot the model estimates and the factorial design at the same time is probably not the only possibility to visualize such experimental data with dprime as the DV and the analysis at the trial level. Which other possibilities could be recommended? Inhowfar is it (not) meaningful to visualize predicted values in the case of dprime as the DV?

Any hints and answers would be appreciated very much.