Before starting the collection of eye tracking and other biometric measures simultaneously, also known as co-registration, it is necessary to ensure that the setup will allow the synchronization of the data streams for later analysis. There are two main types of co-registration setups. The first type uses only one software to record all data steams simultaneously. With this setup, the software ensures the synchronization of data streams and therefore the user does not need to take care of it. The possibility of using this type of setup will be mainly determined by the availability of a co-registration software to be compatible with the biometric and eye tracking systems that are going to be used. In the second type of setup there is one dedicated recording software for eye tracking data and a different one for the other biometric data. Since the recordings are done independently and usually with different computers, we need a method to ensure that the recordings will be accurately synchronized in time. In this article, we are going to focus on this second type of setup and the main method for ensuring a good timing synchronization between biometrics data: shared events.
Definition
Shared events are triggers or events that were produced and registered, theoretically, at the same time by your eye tracking system and biometric system. These events will be used as known common points in time between your data streams and will allow their synchronization and analysis after the recording.
Timing accuracy of shared events
What we mean by timing accuracy of the events is the ability of the setup to send and receive a shared event as close in time as possible in the eye tracking and the other biometric data streams. The timing accuracy needed will depend on your intended application and what biometric data you are recording together with eye tracking data. For instance, in the case of EEG, the timing accuracy to timestamp common events should be as high as possible, in the order of one to few milliseconds. This is mainly due to the high temporal resolution of EEG (brain responses are measured with millisecond accuracy). Eye movements, usually categorized as artifacts in the EEG data, are also measured with millisecond temporal resolution. It is important to be able to align the eye tracking data and EEG data with high timing accuracy, so that the eye movements registered in the EEG and eye tracking data are perfectly aligned and the results derived from the synchronized data are correct. On the other hand, if we want to co-register Galvanic Skin Response (GSR) together with eye tracking, the timing accuracy of the shared events is not as crucial as in the case of EEG, mostly because of the high latency of the skin conductance response to an event (1 to 4 seconds).
The timing accuracy of the events highly depends on the capability of your setup to send and timestamp them in both data streams with high timing accuracy. Tobii Pro recommends that you always check the event timing accuracy of the method you choose to send and timestamp your shared events. For example, by making a co-registration recording where you send events to all data streams at known intervals and then check the time distribution of those events. If you are interested in co-registering EEG, you can also use tools like the EYE-EEG plugin (EEGLAB), which will output information about the synchronization quality of your co-registration recording.
Sender is the method you use to send the events to your data streams. The most common options are:
Receiver is the device that receives and timestamps the events in each data stream. There will be one receiver for eye tracking and one for the other biometric measure. The receivers can be:
Event types are the means and signal characteristics of your events. There are two main types of events used when high timing accuracy is needed:
The main information of a shared event is the time when it was timestamped. This is what will allow the synchronization of our data streams. However, usually events can also contain numeric information in the case of TTL and any string information in the case of TCP/IP. It is quite advisable to send your events at specific points of times during your presentation. In this way, your events will contain three types of information:
This approach is quite useful if you want to analyze your data separately (eye tracking data in the eye tracking software and other biometric data in a dedicated software with appropriate biometric analysis tools) as both data streams will contain all the experimental information related with trial onset times.