Auditory and visual digit span

[becky-gilbert]

I'm planning to contribute auditory and visual digit span tasks. These will be based on the most common method that I'm aware of: 3 trials per list length, starting with short (3 item) lists. List lengths increase by 1 item, until the participant responds incorrectly to all 3 trials of a given list length.

I'm not sure which response method to use: textbox or number pad. Maybe two different versions - one for each response type.

Feedback and suggestions welcome! I'll try to submit the pull request in the next few weeks.

[steffejr]

I have a version of auditory span task written in psychopy. It is intended to be done with an RA who records the verbal responses. What do you mean by textbox versus number pad?

When I did it, I used a staircase up/down design to get the span. I am a big fan of adaptive difficulty and there is a paper showing that it provides a better estimate of span. Once you finish the standard version I can make a staircase version also.

[becky-gilbert]

Cool, thanks @steffejr! I'll still plan to submit my simple stop rule version, because I think it's useful in some cases. But I agree that adaptive difficulty is often better, and it's definitely worth having a separate staircase version.

I have a version of auditory span task written in psychopy. It is intended to be done with an RA who records the verbal responses. What do you mean by textbox versus number pad?

With an online version of digit span, I suppose it's possible to record spoken responses (for instance using one of the audio-response plugins for jsPsych, although this one is currently in development and not part of the official library yet). But in my experience, the participant's response is usually recorded some other way in online versions. The two ways I've done this are (1) using a textbox (e.g. survey-text plugin), where the participant types the in the whole digit sequence, or (2) with a visual number pad on the screen, where the participant clicks on each digit, in order. The experiment factory's digit span demo uses a number pad for responses, if you want to see what I mean.

I'd be interested to know if you have any thoughts on the best method for recording responses. I don't have a strong preference. I guess one difference is that, with the textbox, the participant can see and review/edit their whole response before submitting it.

And with both the textbox and number pad response options, it's possible to require that each response contains exactly the expected/correct number of digits. But I prefer to keep the task similar to the in-person spoken response version (at least the one that I'm familiar with), where the participant is free to report more or fewer digits than were actually in the presented sequence.

[steffejr]

Helo @becky-gilbert
I just tried the digit span demo. The number pad works well. I think it is as close as we can get to speaking responses. When you tell an RA an answer you do not get to "check" what you said. And there is no limit to the length of the number string you are "allowed" to respond.

[becky-gilbert]

Thanks for the feedback @steffejr! That all makes sense to me. I'll submit a version that uses a number pad to get the response, and with no requirement/limit for the number of digits given in the response.

[steffejr]

Hello Becky, I don't know if you saw this but it may be useful...
https://svanhedger.github.io/jspsych/

[becky-gilbert]

I had not - looks very useful! Thanks!!