Some notes on answers to the practical questions
Not too much to say here - the format of the data files should be fairly transparent. The only possibly slightly confusing thing is that the stimulus and button_selected columns do double duty - so for example there is a text label under stimulus on matcher trials, but it’s an image name on director trials.
The key part of the code looks like this:
var observation_trials = jsPsych.randomization.repeat(
[
observation_trial_obj4_long,
observation_trial_obj4_short,
observation_trial_obj5_long,
observation_trial_obj5_short,
],
[6, 6, 2, 2]
);
We have previously created 4 observation trials, featuring our two objects each with their short and long label; here we repeat and randomise them, such that we get 6 occurrences of object 4 and its long label, 6 occurrences of object 4 and its short label, etc. So if you wanted to change those numbers, e.g. to double the amount of exposure, you could simply change this to:
var observation_trials = jsPsych.randomization.repeat(
[
observation_trial_obj4_long,
observation_trial_obj4_short,
observation_trial_obj5_long,
observation_trial_obj5_short,
],
[12, 12, 4, 4]
);
director_trial function in dyadic_interaction.js, which creates the three sub-trials that make up a director trial, and change the value assigned to n_clicks_required).At the moment we specify the number of clicks required in the on_finish of the 2nd sub-trial of the director trials, where the participants select a label (line 541 of dyadic_interaction.js).
n_clicks_required = label_selected.length; //this determines how many times we click in the loop
So at the moment, longer labels require more clicks. You could change that by just setting n_clicks_required to 1, so that short and long labels can both be produced with a single click:
n_clicks_required = 1;
In that case it might make sense to remove the looping trial that handles the repeated-clicking step, and just have participants select a label to send, but that is more work for you.
I think the key thing here is the relative effort of short and long labels - at the moment it is 3 clicks vs 7 clicks, so we want the difference to be even bigger. E.g. we could keep the 3-character labels as 3 clicks, but make the 7-character labels require 10 clicks:
if (label_selected.length==3) {
n_clicks_required = 3;
}
else {
n_clicks_required = 10;
}
That’s just an if-statement setting the pre-created n_clicks_required variable to the value we want, depending on the length of the selected label. There are other more creative things you could do here!
dyadic_interaction_server.py, which looks like this: … This creates a variable, target_list, which consists of 6 occurrences of object4 and 2 of object5. Even if you have never seen python before, hopefully you can guess how to edit this list to change the relative proportions of the two objects or the total number of trials.The relevant line is:
target_list = ['object4','object4','object4','object5']*2
That’s a list of targets - object4 3 times, object5 once - that is then doubled (giving 6 occurrences of object4 and 2 of object5). So if you want to change the overall number you could change the *2 bit to e.g. *1 or *4. If you want to change the proportion you have to change the list - e.g. to change it to 2 occurrences of object5 to every 3 of object4, you could do:
target_list = ['object4','object4','object4','object5','object5']*2
I am going to go for the first option, two pairs of objects, each pair shares a short label. First, I am going to run the server for this extended version on a different port, so I can have the basic and extended versions both running (the server code is slightly different). I will use “ws2” for the extended one, so I have to change the my_port_number variable so my clients connect to the correct server.
var my_port_number = "/ws2/";
Next I have to change my lists of labels and objects (excuse my lack of inspiration for the new labels):
/*
The director's label choices depend on the object.
*/
var object_4_labels = ["zop", "zopekil"];
var object_5_labels = ["zop", "zopudon"];
var object_1_labels = ["dax", "daxedop"];
var object_2_labels = ["dax", "daxalot"];
/*
The matcher always chooses between these four objects.
*/
var possible_objects = ["object1", "object2", "object4", "object5"];
Next we have to provide observation trials for the new objects - we can just extend the code that generates the existing observation trials:
/*
Note that our observation trials have been extended to cover the new objects.
*/
var observation_trial_obj4_long = make_observation_trial("object4", "zopekil");
var observation_trial_obj4_short = make_observation_trial("object4", "zop");
var observation_trial_obj5_long = make_observation_trial("object5", "zopudon");
var observation_trial_obj5_short = make_observation_trial("object5", "zop");
var observation_trial_obj1_long = make_observation_trial("object1", "daxedop");
var observation_trial_obj1_short = make_observation_trial("object1", "dax");
var observation_trial_obj2_long = make_observation_trial("object2", "daxalot");
var observation_trial_obj2_short = make_observation_trial("object2", "dax");
/*
Repeat and then shuffle to produce our observation trials - object 1 and object 4 are
3 times as frequent as objects 2 and 5, but both objects occur equally frequently with their
short and long labels.
*/
var observation_trials = jsPsych.randomization.repeat(
[
observation_trial_obj4_long,
observation_trial_obj4_short,
observation_trial_obj5_long,
observation_trial_obj5_short,
observation_trial_obj1_long,
observation_trial_obj1_short,
observation_trial_obj2_long,
observation_trial_obj2_short
],
[6, 6, 2, 2, //counts for objects 4 and 5
6, 6, 2, 2, //counts for objects 1 and 2
]
);
The only other change on the javscript side (apart from some tiny rewordings in the instructions) is to extend the if-else statements at the start of director_trial so that it knows where to find the labels to present the director when the target is object1 or object2.
/*
A small addition here - need to show object-specific labels to the director
*/
function director_trial(target_object, partner_id) {
end_waiting();
//look up label choices depending on object
if (target_object == "object4") {
label_choices = object_4_labels;
} else if (target_object == "object5") {
label_choices = object_5_labels;
}
//adding if-statement to retrieve object1 and object2 labels
else if (target_object == "object1") {
label_choices = object_1_labels;
}
else if (target_object == "object2") {
label_choices = object_2_labels;
}
...
And that’s it on the client side - the code was already pretty general. The final thing to do is to make two small changes to the server, so that it runs on port 9002 (which matches up with ws2 that the clients connect on) and so it has interaction trials for all the objects:
# Extended target list here
target_list = ['object4','object4','object4','object5','object1','object1','object1','object2']*2
#much much later...
PORT=9002 #this will run on port 9002
That’s it. If you want to see this in action you can download my code through the following three links:
Drop these into your dyadic_interaction folder alongside the original version and they will be able to access the stimuli etc folders that are already there; I am already running the modified server on ws2/9002 but if you want to run your own copy, put dyadic_interaction_server_extended.py in dyadic_interaction/server.
All aspects of this work are licensed under a Creative Commons Attribution 4.0 International License.