Creating an adaptive task based on average accuracy

EJacobs

Hi!

I want to create an adaptive arithmetic task, where participants get arithmetic problems of a certain difficulty level based on their average accuracy on the task. More specifically, I have divided the arithmetic problems in 6 difficulty categories and I want each participant to end the task with an average accuracy around 80%. This means that they have to get an easier problem when their average accuracy goes below the treshold of 80% and a more difficult problem when their average accuracy goes beyond this treshold during the task. Could someone give me some guidelines on the best way to do this? Thank you!

eduard

Hi @EJacobs,

you need some inline python coding for that, because you need to dynamically choose the problem that is presented to participants.

I think I would do following.

1) Create separate lists of the problems for the different difficulties (easy, medium, hard list)

2) Choose a problem from the currently active list, present it, and collect the response

3) Compute the accuracy (if var.acc is not automatically tracking it, you can compute it yourself, by incrementing the trial count on every trial, and the correct_response count for every correct response, and then divide)

4) After every x trials, check the accuracy, if it is too high, choose a harder problem list. If it is too low, choose an easier one.

Does this make sense? I guess, it is best to get started and try to set this up yourself. If you get stuck in the process and can't find the problem in the documentation (unlikely ;) ), you can ask us here again, and we can provide some more specific pointers.

Eduard

EJacobs

Thank you for your guidance @eduard!

Just for clarification: is the staircase plugin also a possibility to create the adaptive task that I described? If so, could you provide some guidance on how to set the parameters? I can't seem to find any documentation or tutorials on this. Thank you!

eduard

Perhaps, I am not familiar with its functionality at the moment, but I doubt. It might help adjusting the difficult of certain stimulus settings, but I don't think it is able to choose different stimuli altogether. But that's just a guess. You can try it, and see for yourself.

Eduard