Analysing Eyetracker data

Author

Ian

Opticka supports 4 different eyetrackers, and for Eyelink, iRec and Tobii, there are analysis classes. In general, opticka sends markers compatible with the Eyelink pipeline, this is TRIALID # at the start of a trial and TRIALRESULT # at the end. We can use these markers to parse out trials with the TRIALID number defining which variable was shown on tht trial and TRIALRESULT number defining correct/incorrect etc. We can also pass other messages to give more details for each trial, like the uuid of the state machine state etc.

Lets take the Tobii. The data is stored in the MAT file at the end of a session. We can load it using:

t = tobiiAnalysis('fileName', '/path/to/file');

Then we can parse the data, this loads the raw data, extracts the trials and variables, and calculates saccades etc. If https://github.com/CogPlatform/NystromHolmqvist2010 is installed, we can use this to provide better smoothing / data replacement for blinks / saccade detection etc.

>> t.parse
---> runExperiment loadobj: NEW:Pupil-2024-4-9-17-36-2<runExperiment#1B48E001A> (UUID: 1B48E001A)

   > metaStimulus object present
   > SIF: /home/cog5/Code/opticka/CoreProtocols/PupillaryReflex.m 
   > stateInfoFile: /home/cog5/Code/opticka/CoreProtocols/PupillaryReflex.m assigned
   > loaded taskSequence
   > Devices:   tobii
   > loaded screenManager


... runExperiment (Opticka file) found
===>>> Loaded Tobii<tobiiManager#1B5229A6A> containing 327 messages and 56081 samples
===>>> Variables contained in the task:
    {'1 = alpha:0'  }
    {'2 = alpha:0.1'}
    {'3 = alpha:1'  }
    {'4 = alpha:1'  }
    {'5 = alpha:1'  }
    {'6 = alpha:1'  }
    {'7 = alpha:1'  }
    {'8 = alpha:1'  }
:#: Loading Raw MAT Data took 671 ms
Parsing Tobii Experiment Events: [====================] Done. [0 seconds]    
:#: Parsing Tobii Events into 42 Trials took 0.08 secs | min-t = -2.09 max-t = 1.27
No ROI specified...
No TOI specified...
Loading trials to compute microsaccades: [====================] Done. [0 seconds]  
:#: Parsing MicroSaccades took 97 ms
--->>> Full saccadic analysis of Trial 1:
--->>> Full saccadic analysis of Trial 2:
...

We get 42 trials, with a time range from -2.09 to +1.27 seconds. Each trial’s data is stored in t.trials and if there were variables they are stored in t.vars. For example if you want to plot all trials from variable 3:

>> t.plotRange = [-2 1]; % set the plotting range in seconds
>> t.plot(t.vars(3).idx);

If you want to plot all correct trials:

>> t.plot(t.correct.idx);

You can visually step through each trial using t.explore. To plot the NystromHolmqvist2010 analysis results only:

t.plotNH(10); % plot the 10th trial

The NystromHolmqvist2010 data is stored for each trial, e.g. t.trials(10).data.

Other Links