Millisecond Forums

Thanks Dave.

> I'm not seeing any marker values of 0, which I would expect due to the ontrialend code.
> ...It seems like maybe no recording is happening between trials/surveypages?  

That is correct.

> So when the trial ends the values.markernum and port both get updated, but there won't be
> another recording until after the stimulusframes for the next trial triggers the port.

Yes. /ontrialend does _not_ display stimuli or send markers, it merely executes arbitrary logic (e.g. setting an item number, etc.). Nothing is (supposed be) sent or displayed via /ontrialend.

> Without seeing any 0 markers, all I know is that sometime between that third timestamp and the fourth timestamp the script changes between surveypages.

Yes (as I think noted previously) stimuli are cleared / erased between surveypages, which should usually take on the order of a single display frame. That is the origin of the "gap," i.e. the time between erasing the stimuli after page N and the start of the stimulus presentation sequence of page N+1 (where the next marker is sent in its 1st frame).

> Can I assume that the screen blanks soon after that third timestamp, and furthermore that the new stimulus comes on-screen with the fourth timestamp?

Yes.

> [...] how do I handle this eyetracking data at the beginning and end of my trials when I don't know what's actually on-screen?

Not sure I understand this. Your <surveypage> sends a single marker, starting in the 1st frame of the page's stimulus display sequence:

<surveypage practiceSurveypage>
...
/ stimulusframes = [1=greybackground, practicePictures, xdat]
</surveypage>

The marker signal then remains raised for the duration of the trial / surveypage.

Hope this clarifies.

I have datastreams=eyetracker on my expt.  I have a survey with surveypages.
Each surveypage has a port in its stimulusframes, with ontrialbegin and ontrialend attributes that change the marker's value and set that marker to the port:
<surveypage practiceSurveypage>
/ ontrialbegin = [values.markernum = picture.practicePictures.currentitemnumber + 200;]
/ ontrialbegin = [port.xdat.setitem(values.markernum, 1);]
/ ontrialend = [values.markernum = "00000000";]
/ ontrialend = [port.xdat.setitem(values.markernum, 1);]
/ stimulusframes = [1=greybackground, practicePictures, xdat]
</surveypage>

It is saving eyetracker data like this snippet:

timestamp	marker
1452228498	200
1452231831	200
1452235165	200
1452271831	201
1452275165	201
1452278498	201

So I'm getting 300 eyetracker samples per second (the timestamp is in microseconds and increments by 3333, 3333, 3334, repeating) as long as the marker is staying the same...so throughout a surveypage.
But when one page changes to another, there's a gap...in this case it's 36666 microseconds.

Can you help me understand what's going on?

I'm not seeing any marker values of 0, which I would expect due to the ontrialend code.
...It seems like maybe no recording is happening between trials/surveypages?  So when the trial ends the values.markernum and port both get updated, but there won't be another recording until after the stimulusframes for the next trial triggers the port.

Without seeing any 0 markers, all I know is that sometime between that third timestamp and the fourth timestamp the script changes between surveypages.
Can I assume that the screen blanks soon after that third timestamp, and furthermore that the new stimulus comes on-screen with the fourth timestamp?

Or is it just a mystery what's happening at these time scales?  ...and if it's a mystery how do I handle this eyetracking data at the beginning and end of my trials when I don't know what's actually on-screen?

Thanks for your help.

In the page about the tobii extension to the eyetracker element, there are some attributes without descriptions, e.g., calibrationmode, illuminationmode, lowblinkmode.
http://www.millisecond.com/support/docs/v5/html/language/elements/eyetrackertobii.htm
It looks like they are unimplemented.  Is that right?
Are they planned?
It would be nice to have some control over the calibration that pops up at the start of the script.  For example, if the eyetracker is already calibrated for this participant, then just skip it (...if such a thing is possible).

Also, looks like a typo on this tobii programming page http://www.millisecond.com/support/docs/v5/html/howto/tobii.htm which a link that says "Special attributes and properties enabled by the SR Research plugin"  but the link goes to the tobii-specific extensions page.
And the same page says "Once connected, the plugin recieves and records the stream of gaze point data from the eye tracker."  But it seems I have to set datastreams (on the expt or block or trial, for example), to actually save eyetracking data to a file...which is fine, but doesn't seem to be mentioned anywhere on the eyetracker page (or tobii extensions or tobii programming).

Thanks.

Yes, it connects if I specify the ipaddress.  That's something of a hassle because we have more than one eyetracker and an equal number of laptops and we travel with them.  We'll have to make sure to keep the laptops & eyetrackers paired up to avoid having to edit the iqx file.
But I'm relieved that it works.

Hm, it actually would be nice to see if I can send markers to the tracker instead of connecting to it through the eyetracker element, since I really just want the markers.  I'll try looking into that if I get some time.

Thanks.

> Aha, I need to set the ipaddress attribute.  Previously I have not had to do so.  Why is it necessary now?

I take it that explicitly specifying the tracker's /ipaddress resolved the connection problem, correct? If so, that's great. I have no spontaneous idea why automatic detection and connection failed, it might be an artifact of the particular network configuration / environment.

P.S.: Apologies for the cross-posts, I had not seen your earlier reply.

One other thing / possible workaround if the primary aim is sending markers to the eyetracker. As far as I can see from the Spectrum's specification, the unit has a DB9 connector which can be used for TTL input to the unit. Thus it may be possible to have Inquisit send markers to the device via the COM / serial interface.

(Regarding the Spectrum and the connection / detection issue, let me consult with our eyetracking specialist, who'll be back in office after Labor Day.)

Hmm, does connecting succeed if you explicitly specify the tracker's IP address? You can do so via the /ipaddress attribute in the <eyetracker> element, e.g.

<eyetracker>
/ plugin = "tobii"
/ ipaddress = "192.168.0.10"
</eyetracker>

Aha, I need to set the ipaddress attribute.  Previously I have not had to do so.  Why is it necessary now?