By plush - 12/1/2014
Dear friends, I'm having a problem with a script in which participants have to press a picture to respond. Regularly, the moment I move the cursor, Inquisit will end the trial, without having had the opportunity to click a button. . I have embedded a screen capture to show you what happens. I already tested whether have double quotes on the valid responses would make a change (e.g., "steal" or steal) but this does not seem to make a difference.
Any suggestions would be appreciated!
|
By Dave - 12/2/2014
Unfortunately I cannot see any screenshot that would illustrate the problem. In addition, you'll also want to provide the actual code (at least for that <trial> element).
In general, moving the mouse would only considered to be a validresponse (and thus terminate the trial) if
/ inputdevice = mouse
/ validresponse = (anyresponse)
or if the validresponse attribute explicitly includes mousemove among the valid responses.
The way to define a stimulus element (<text>, <picture>, etc.) as clickable object / response in a <trial> is
<trial mytrial>
/ stimulusframes = [1=objectA, objectB]
/ inputdevice = mouse
/ validresponse = (objectA, objectB)
...
</trial>
without any double quotes.
|
By plush - 12/2/2014
Thanks for your response Dave, the trial-code is given below; and I attached the script to this reply.
Looking at the code below, Inquisit shouldn't skip the trial after making a mere mouse movement. Yet, this is what happens once in every four times I start Inquisit as participant 1. To illustrate this, Another youtube link to show you what happens to be as clear as possible.
attempt1: (allegedly seeking and connecting to another person, instruction screen, act-trial without error, response trial, ctrl-q)
attempt2: (allegedly seeking and connecting to another person, instruction screen, act-trial without error, response trial, ctrl-q)
attempt3: (allegedly seeking and connecting to another person, instruction screen, act-trial WITH error, responsetrial skipped)
<trial act>
/stimulustimes = [1=now;2000=steal,split]
/validresponse = (steal, split)
/responsetrial = (steal, waiting1steal)
/responsetrial = (split, waiting1split)
/inputdevice = mouse
</trial>
|
By Dave - 12/2/2014
Thanks for the video illustration and the script. A couple of thoughts and remarks:
(1) Just to make sure we're on the same page: What's the Inquisit version installed on your Mac. If it's not the latest release (4.0.7.0), please update it and let me know if the issue persists.
(2) Also, what's the exact OSX version you're running?
(3) You are correct in that the syntax *should indeed not* terminate the trial upon mere mouse movement.
(4) Based on the observation that you did not close the Inquisit application between runs, I suspect a memory issue may be responsible. I.e., the trial response from the *previous* run might not have been cleared properly and still linger around in memory. The trial in the current run then mistakenly thinks a valid response has already occurred. (a) If this theory holds, you should *not* see this issue when shutting down Inquisit between runs. (b) It would also be interesting to see a data file from a botched run, particularly what's reported in the 'response' column for the respective trial.
Thanks!
|
By plush - 12/3/2014
Hi Dave, in response to your questions:
1) Yes I have the latest version installed
2) OSX YOSEMITE 10.10.1
3) good to know!
4a) Quitting the program in between sessions did not prevent the error from occurring once in every few times.
4b) I've attached the data file. I ran this on a different mac with version 4.04, there were no problems then. Then I updated to 4.07 and the error occurred (see the last run 19:37:57). However the data file is not really informative. The last row is an open-ended question in which I entered the response 'error'. So the row before that one is the row in which the error occurs.
Best,
Tom
|
By Dave - 12/3/2014
Thanks for the data file. That file is confounding, to be honest. It frequently lists 0 (which indicates 'no response') in the response column for your 'act' trials. That appears to be the case for *all* runs you conducted under 4.0.7.0, including those that ostensibly "worked". It is, however, also the case for *some* runs you conducted under 4.0.4.0, which reportedly all worked as expected. I'm somewhat baffled by this.
To boot, the data file shows "0" and in one instance "steal" rated as correct response in the correct column, while "split" is rated as incorrect. However, the trial definition (from the script you attached) provides no basis for that. No /correctresponse is defined.
<trial act>
/stimulustimes = [1=now;2000=steal,split]
/validresponse = (steal, split)
/responsetrial = (steal, waiting1steal)
/responsetrial = (split, waiting1split)
/inputdevice = mouse
</trial>
Is the script you conducted those tests with 100% identical to the one you attached here previously?
|
|