Hi!
I would like to ask if it’s possible to do in Psychopy Builder an auditory identification task with rating (hear a sound, choose the correspondent word and rate the similarity) like the one in following image:
(image of a window of a test ran in TP 3.1 Software retrived from Rauber, Rato, Kluge & Santos: TP 3.1 Software: A Tool for Designing Audio, Visual, and Audiovisual Perceptual Training Tasks and Perception Tests , 2013)
The goal is to have accurate answers (not mouse coordinates), since the experience has to run in different computers.
If it’s possible, could you tell me which of the PsychoPy3 Demos should I look at?
I don’t know a specific tutorial but the task seems doable in the builder. You can use polygon component and text component to create the buttons on screen, a sound component to play the sound and a mouse component to collect the responses. Then you can create a loop (Flow — PsychoPy v2020.2) to specify what sound is played in each trial and what and when are the possible answers on screen. After that you can even add another routine within the loop in witch a slider component is use to rate the similarity.
I suggest to just start trying messing with the builders!
Dear tandy,
Here is my question in a more detailed way:
In the first routine of the trial, I used a rating type of response: they listen to the sound and chose from the categories (numbers 1 to 9) a word:
Although technically this test runs perfectly, the design is not the best one: I would prefer that they choose directly a word and not through a number (because this indirect choice burdens the memory).
I tried to change the categories from numbers to the words, but there is this problem:
Plus, it would also be better it the words in these categories would appear in random order, not always the same order.
I started to ‘play’ with the option of rectangles/text or images with clickable mouse, but in the excel with the data I have the mouse coordinates and not an accurate response, plus I’m afraid that in different computers with different screen definitions things won’t match up…
Can you advise me which is the best direction: images+mouse? rectangles/text+mouse? rating categories? other?
(important information! in the practice trial, I will have to add feedbacks)
You can use a polygon component to make the response rectangle (e.g polygon_word1, polygon_word2 etc.) and a text component to show the word inside (e.g. text_word1, text_word2) so you’ll have 9 polygon and 9 text component. You don’t need to worry about different screen size if you use the dafault psychopy unit of measurement (“unit”) to place the rectangle, since it will always be relative to the size of the screen the experiment is played on: 1 unit is the height of the screen and the coordination (0, 0) are the middle of the screen. You can specify a list of clickable stimuli in a mouse component (so a list of all the polygon) and have the routine end only after a valid mouse response (o a rectanlge is being clicked). This should also ensure that in the data are saved as you want them. You can randomise the position of the text, putting something like $text1 as the text of the textbox component and adding a text1 column to the excel or cvs file you use for the loop. There should be no problem adding feedback, I’m sure there are tutorial around you could look into.
Another solution would be to prepare several different images where you manually change the order of the word and just change the image presented in the same way using a loop and a excel/csv file. You could still use the mouse to collect the responses by placing some polygon behind the images (you should test this before creating all the images, as I’m not 100% sure it will work).
It works but has some limitations, some of them I think I can improve if I use images, but before I continue ‘playing’ with Builder I have at pressing question: is there a way, after in excel, to convert the coordinates from the mouse response click? how can I know fast and easily to which word corresponded the participant mouse click?
(I hope my question is clear).
In this case I have a mouse component that can click on 2 polygons (basically yes or no). With this settings in the datafile I have a column called “mouse_tutorial.clicked_name” where the name of the polygon (different_polygon_tutorial or same_polygon_tutorial) is being saved alongside xy coordinates, witch mouse button was pressed and RTs.
Hi, tandy, I think I understand your solution, but could you send me a picture of your settings data excel file?
because my file for now is just like this:
and this test is an open-choice answer test, there’s no write or wrong answer, so I can’t really pre-relate a sound with a word, 2 participants can choose different words for the same sound.
Gabriela
Even though you don’t have a correct answer in your experiment you will know what response (polygon) the participant gave, what word was in that polygon and what sound was played in each trial with the right loop file, especially if you use images to randomise the words position.
It works!!!
Now I’ll substitute the text with images, and randomise these too, and I’ll get back to you with my results.
One question about screen definitions:
In my experience I set the screen definitions to ‘deg’ and the components of the routine to ‘norm’. I also tried to with ‘norm’ for everything and it looks good, at least in my computer. Should I maintain like this?
(And I didn’t ‘calibrate’ my monitor, I was to afraid of doing something wrong…)
G.
