Contact Us

Use the form on the right to contact us.

You can edit the text in this area, and change where the contact form on the right submits to, by entering edit mode using the modes on the bottom right. 

           

123 Street Avenue, City Town, 99999

(123) 555-6789

email@address.com

 

You can set your address, phone number, email and site description in the settings tab.
Link to read me page with more information.

Eye-Tracking Help

Running an Experiment on Eyelink                                                     Sanjay Manohar

 

There are two computers:

1) Stimulus PC: running Windows and Matlab, connected to the main CRT screen, and

2) Eyelink PC: in DOS, running the eyelink software

Subjects

Record age and handedness. Ask if colour-blind.
If wearing glasses / contacts, check prescription in dioptres.

Setting up

Turn on power to both computers
Turn on Eyelink at mains, ensure power plugged in and yellow light glowing.
Boot the Stimulus PC into Windows (may need to use keyboard to select)
Log in to Stimulus PC “cogneuro” “kilo1000”
Boot Eyelink PC into DOS
Check there is enough free space on Eyelink PC, type “dir” and check free space > 50,000,000 bytes.
Type “elcl” on Eyelink PC to start eyelink software.
Start Matlab on Stimulus PC and navigate to directory that has experiment
Start experiment by typing “result = “ then the name of experiment at the Matlab prompt.
The Eyelink PC should now show the camera setup screen.
Brief the participant on the task, with example sheet or powerpoint. Get participant comfortable, ask them to adjust their seat. Then raise or lower the monitor table so their chin is near the rest. Adjust chin rest height with side-screws, and adjust forehead-rest by raising or lowering the camera, by carefully releasing the side-pins.
Warn them that you are going to turn the lights down. Turn off all lights and close the door.
Adjust focus lever on camera. The iris should look crisp on the eyelink PC.
Can you see the eye fully? If not, adjust the mirror on the eyelink using the screws in front, and gently moving the mirror up and down.
Press “autothreshold” to get the blue circle to enshroud the pupil, and the yellow circle to highlight the corneal reflex.
If eye is not immediately “captured”, adjust “Pupil +/-” if necessary, to avoid blue on the eyelashes encroaching upon pupil. Similarly with corneal reflex adjustment, so that it avoids the medial canthus or any reflective studs on glasses etc.
There should now be a steady crosshair visible on the eyelink PC. Select “Calibrate”.
The Stimulus PC should now show a circle in the centre of the screen. You can press “accept fixation” on the eyelink PC if the participant seems to know what they are doing, and are not too confusible. If the participant is elderly or brain damaged etc, best to first turn off “auto-calibrate” on the Eyelink PC, and the manually press “Accept fixation” for each of the 9 calibration points, when the participant has correctly fixated each dot.
At the end of calibration press Exit Setup on Eyelink PC.

 

Running experiments

The stimulus PC should now show “Start of Experiment” etc.

Between blocks, turn the lights up, and give the subject a minimum of 1 minute break. That is a good time to complete questionnaires etc. Then turn the lights down and resume; it is not usually necessary to fully recalibrate, as drift-correction is performed online in most tasks.
If recalibration becomes necessary, some experiments are configured that if the “C” key on the Stimulus PC is held down at the end of the trial, the calibration will start again. Otherwise, exit the experiment by holding “escape”, and restart it using the command “result = “ name of experiment “(result”). Adding result in parentheses after the task name should resume from where you left off.
At the end of the experiment, right-click the “result” variable, and save it as a file with an appropriate name.

 

Troubleshooting

Signs of a problem:

Crosshair vanishes (e.g. for more that half a second) during trial,
Trial takes longer than usual to complete, or there is a pause before the foreperiod
Orange square appears in middle of screen: Ask participant to keep eyes on centre of screen. After 4 seconds, if eye is detected correctly, experiment will continue.
Eye position appears to jump around erratically or oscillate on the Eyelink PC.
A large blue region is visible above the pupil on the camera image
A large yellow spot, or yellow grids, are visible outside the eye on the camera image

 

If Subject loses fixation:

Is their head forwards enough? Instruct to keep head fully forwards against forehead rest
Remove mascara / other eye makup
Try the other eye – loosen the screw of the camera,  slide to other eye, and re-tighten.
Ensure no hair in front of face
Ask them to brush their eyelashes upwards
Is the subject sleepy? Are their eyes half-closing? Turn the temperature down with the air conditioning. Offer a drink. Remind them about task / incentives. Ask them to walk outside for 3 minutes in the breaks.
Instruct to keep eyes wide open. Older participants may have a “ptosis” - naturally droopy eyelids. Eyelash problem is much worse with droopy eyelids.
Do they need their glasses on?

1.    Try with no glasses, unless your task requires visual discrimination.

2.    Do they have contact lenses? Tracking is much better with contacts.

3.    If glasses required, clean the glasses.

4.    Try adjusting the glasses position, up or down the nose.

5.    If the reflection remains bad, try adjusting the height of the camera and the mirror simultaneously, so that the camera is “looking around” the reflection.

If the Eyelink is getting confused with eye vs non-eye objects, you can set a “bounding box” for the eye position, so that it only looks in a fixed rectangle. Click on the pupil to set the location of the box, which should appear superimposed on the picture.

 

Impossible subjects: 

These participants' data will probably need to be discarded.

Nystagmus
Albinos / very pale iris
Very long eyelashes
Sleepy subjects: if eyelid comes down over pupil you will lose the trace.