Reducing delay onset

[Steve Lukito]

Hi all, 

We have a problem with delayed event onset of stimulus in our task. I attach our original sample task, a multi-sensory oddball EEG task where a child needs to respond to a picture of a cat. We ran the task on a Windows 10 PC, with duplicate monitor setup. Primary monitor (refresh rate 114Hz) is set to the participant screen, secondary monitor (75Hz) is on the experimenter room. We used E-prime 2.0 and NetStation 5.3. We also use an SRbox for the response 

The trial consists of:

- Stimulus slide

- ISI slide with jittered duration

The initial Stimulus.OnsetDelay was around 180ms. We managed to reduce this to between 80-90ms. with ISI.OnsetDelay of 15ms. But we are stuck at those numbers. We have done the following changes to the original sample task:

- Set the pre-release same as duration of the stimulus. 

- Changed the stimulus duration to be more appropriate for the screen refresh rate (75Hz)

- Installed an ASIO4ALL driver, set E-Prime Devices Sound API to ASIO.

- Used smaller images appropriate for the screen and not used "Stretch".

- Mapped every audio stimuli to the appropriate codec using the configcodex.

- We disabled newsletter/weather updates of the Windows PC.

We experimented with the following but saw no improvement:

- removed Usepackage of the NStation from the e-prime routine.

- experimented by changing the refresh rate of the screens.

- we switched the order of the ISI and the stimulus slide, the delay on the stimulus slide becomes very small became 15ms but the ISI onset delay remained 80ms.  

We are unsure what else to do. Can anyone advise?

Kind regards,

Steve

Postdoctoral research associate

King's College London, United Kingdom.  

[Michiel Spape]

Hi Steve,

I have worked with oddball designs and netstation, so I hope I can help. The issue is, I believe, with regards to the NetStation at the beginning and end of the ‘trial’. What you have to remember, however, is that from the point of view of an oddball design, a trial is not a stimulus, but of course a number of stimuli. Now, NetStation packages are designed with the idea of:

Trial Start …. Pre …. Fix …. Critical Stimulus … Reaction …. Feedback

You know, the basic E-Prime style. So, in this situation, the ideal timing is to set everything exactly correct at the very beginning, make sure that everything runs on time, and then only after everything useful is already done, send the information to NetStation. This is good for that sort of design, but not the oddball design, because now at the beginning and end of every stimulus it starts communicating a lot to NetStation, while you probably want to have the interval between stimulus onsets more or less equal, right?

 

So, there are two strategies:

1. To start the trial at the very beginning and simply dump all the stimuli within one procedure [i.e. have stimslide1, isi, stimslide2, isi, stimslide3, …]. Looks ugly and is a lot of manual work, but will very likely resolve your timing issue.
2. To take it as a given that there is an amount of delay and make all the durations shorter to compensate for the error. That’s a bit of a hit and miss, but you can try to work it out. Remember, onsetdelay isn’t as important as the actual time of appearance, so it’s more convenient to declare a long variable in the user script (LastTime) and put an inline after your stimulus:

 

debug.print “SOA: “ & Stimulus.OnsetTime – LastTime

LastTime = Stimulus.OnsetTime

 

And then you fidget around with the Stimulus duration until it actually starts to look like what you’re aiming for. Remember that for your design all types of stimuli should have equal types of duration. Easy fixes for that is to also just use empty audio files to present so the same stuff is happening even if you don’t want to hear anything.

 

Typically, I use a combination of 1 and 2 and show about 5 images at a time, but with a slow PC I still get markedly more error after the 5, just because of the netstation. I think that perhaps a faster PC might help, but I don’t have one :)

 

Hope that helps!

Michiel

 

//Michiel Spape, PhD

//Associate Professor

//Institute for Cultural Innovation & Centre for Cognitive and Brain Sciences

//University of Macau, Macau SAR

--
You received this message because you are subscribed to the Google Groups "E-Prime" group.
To unsubscribe from this group and stop receiving emails from it, send an email to e-prime+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/e-prime/0295e6d8-1c2f-401f-95f9-60fe118e8d4en%40googlegroups.com.