| Reference | Downloads | Github

Am I using the analysis function properly to get the JND?

Win10, 1.84
Standard Standalone

Hi all,
I am trying to use the staircase to find the JND for a phonetic feature (degree of nasality, without nasality being contrastive in the language, so this is a degree perception, not a category perception).
I have already used the staircase in PsychoPy and it’s working great, but I’ve always used it to find the 50/50 threshold between two categories. For this new experiment, I’ll need to use a 2AFC in which the subject will hear a standard and then assess which of the 2 following sounds had more nasality.
My first question is this - if I’m going to use the built-in PsychoPy analysis functions, can I simply use a 1-up/1-down procedure like I do with my previous yes/no staircase procedures? Or do I need to use 1-up/3-down (or something else). Since PsychoPy uses all of the data (rather than just the reversals) to build the pyschometric function, then it should be fine to use 1-up/1-down, but I’m certainly not sure on that.
My second question is this: In the analysis of the resulting data, am I using the formula correctly (modified from my use of it in a 50/50 threshold):

fit2 = data.FitWeibull(levelsList, responseList, expectedMin = 0.5)
MyJND = fit2.inverse(0.80)

My understanding is that this will give the 80% detect a difference threshold for the continuum of nasality I’m using. Is that right? Is there a reason to use 80% detection as the marker of JND rather than 75% or 60%?
Thanks for your help,

I’m too rusty on this to answer most of your questions, but for this one: plot your data.

You need to know if the fit corresponds to your empirical proportions, and you’ll be able to directly read off the graph to verify the interpretation of the threshold.

Thanks - I assume the demo plotting script that comes with PsychoPy is suitable for answering this question (I’m looking at it now and it seems to plot the raw data so there should be no problem).