Sign up Calendar Latest Topics
 
 
 


Reply
  Author   Comment  
SeanC

Junior Member
Registered:
Posts: 13
Reply with quote  #1 

Hi, 

I am using the old version of MonkeyLogic and had a question about the accuracy of timestamps. 

I am trying to find the latency of saccades relative to the onset of a visual stimulus. In the timing file, the onset of the relevant stimulus is indicated by an eventmarker (as an argument within the toggleobject function). I am then using the BhvInfo.CodeTimes and BhvInfo.CodeNumbers to find when the visual stimulus was presented on each trial. I then timelock the InBhvInfo.AnalogData.EyeSignal to the value in BhvInfo.CodeTimes. I then process the eye signal to detect saccades. However, I need to know whether this method of timelocking is going to provide me with a reliable latency measure? 

To elaborate, we are using a screen with a refresh rate of 60Hz and are sampling eye data at 240Hz. We cannot recall how the timestamps in CodeTimes relate to what actually happened on each trial. For instance, will there be a jitter in the actual stimulus presentation relative to the timestamp - due to the refresh rate of the monitor (i.e. 1000/60 = 17ms of jitter)? Or does MonkeyLogic only trigger this event code when the stimulus is presented on the screen? Has this been extensively tested, or would I need to test my configuration using a diode? 

A short summary of the eventmarkers in the timing file, and the analysis, is shown here:

eventmarker(12);    %Fixation finished

toggleobject([Cue,stopCue],'EventMarker',62); %%% Turn on relevant stimulus

On an example trial, code 62 has a timing of 2078ms. I therefore take the eyesignal as InBhvInfo.AnalogData.EyeSignal(2078:end,😉

There is a 4-21ms (i.e. 17ms jitter) between when code 12 and code 62 is produced on each trial. I am aligning my eyetracking responses to code 62, assuming this is a reliable measure of when the stimulus appeared on the screen. 

 

Thanks for your advice, 

 

Sean

0
Jaewon

Administrator
Registered:
Posts: 971
Reply with quote  #2 
I am not responsible for the original MonkeyLogic, but the numbers you got look okay to me and your analysis is fine.
0
SeanC

Junior Member
Registered:
Posts: 13
Reply with quote  #3 

Hi Jaewon,

Thanks for getting back to me on this, it is reassuring to know the analysis is fine. However, if possible, please could you provide a brief explanation (or point me in the direction of where I can find one) of what features of the way MonkeyLogic is coded/ setup means this analysis works fine? A reviewer of our paper is concerned we should have used a photo-diode to help verify the timing of the task and make sure the saccade latency measures were reliable. So a brief explanation of why this should not be necessary with MonkeyLogic would be really helpful. 

Thanks

Sean

0
Jaewon

Administrator
Registered:
Posts: 971
Reply with quote  #4 
I have no objection to the reviewer's point, since each setup (i.e., the response characteristic of your monitor) can be different. You can run a test with a photodiode and show the reviewer the intervals between EventMarker 62 and the rising time of the photodiode signals in your system.

Performance data of the original ML can be found in the papers in the following link.
https://www.brown.edu/Research/monkeylogic/temporalprecision.html

There are some data from the original ML in my paper, too. I also explained what can affect the latency of visual stimuli there. It is mostly the screen resolution. If you didn't use HD or 4K, you should be fine.
https://www.ncbi.nlm.nih.gov/pubmed/31071345
0
SeanC

Junior Member
Registered:
Posts: 13
Reply with quote  #5 
Thanks!
0
Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.