Eye and body tracking in the lab, in the wild, and in the clinic

• Author / Creator

  Stone, Scott

• In behavioural psychology experiments, we try to ask and answer questions about real life phenomena. Arguably, this is a hard question to answer due to the difficulty in actually collecting the data. One concession that might be made is to ask a similar question in a much more controlled environment. For example, if we want to know the average response time to look at a salient target, we may design a study that attempts to isolate this response. Perhaps a dot is shown on a computer screen, and we measure the average response to initiate a saccade. This is valid science, but it is hard to say that it represents real life. Real life is dynamic, and as a consequence, much more noisy. We don’t sit in rooms with isoluminant lighting at exactly 40cm from the screen and try to move ‘as fast and accurately as possible’ in our everyday life. In this thesis, I challenge these assumptions through the use of eye and body tracking techniques. Three studies were conducted, each increasingly giving up experimental control to collect more naturalistic data.

  In the first study, I demonstrate that tightly-synchronized simultaneous collection of laboratory-grade eye- and body-tracking can be used to generate accurate 3D gaze vectors that have sub-centimetre accuracy in optimal conditions. Generating 3D gaze vectors is a challenging problem because there are no widely used established calibration routines. This study aimed to answer this question. Participants completed four different calibration routines, where the pupil position data was used to predict gaze position in 3D space. A validation procedure was used to assess model performance when the experimenters know exactly when and where the participant is supposed to be looking. Finally, participants completed a previously validated ecologically valid task known as the Pasta Task to assess the model's performance during real-world behaviour. This study provides tangible advice for researchers interested in collecting naturalistic simultaneous eye and body tracking data.

  In the second study, I used a consumer-grade eye tracker and mouse tracking to record the behaviour of participants as they navigated through a menu interface modeled after the popular video game, Mass Effect 3. Typically user experience researchers use qualitative measures such as interviews, surveys, and technical reports to uncover problematic areas of user experience known as friction points. Here, I use eye and hand tracking to attempt to quantify friction as a dynamic process. I demonstrate a methodology that allows for the detection of friction points based on gaze and movement signals. I conducted both an in-person (local) cohort as well as a remote cohort collected using webcam based eye tracking. I show that many of the same patterns of friction that occur in the local cohort also occur in the remote cohort, suggesting that the increased noise (due to decreased experimental control) in the remote cohort did not overpower the signal. The study provides evidence that many of the techniques and tools used in the laboratory can be adapted for use in digital environments.

  In the third study, I assessed the feasibility of using eye tracking in a clinical setting to augment the neurological assessment of vertigo. Vertigo is a condition that manifests as extreme dizziness due to imbalance in the vestibular system. Diagnosis is challenging because vertigo is not a single disease entity but the cardinal symptom of different diseases of varying etiology ranging from benign to deadly. A typical neurological battery will assess the vision of the patient to test oculomotor function, which is critical for diagnosis. Here, we created a simple systematic screen-based set of stimuli that approximates the neurological assessment given to a patient with vertigo symptoms. We used a portable eye tracker to collect eye position during the approximately five minute stimulus presentation. Data was collected by someone without any background in eye tracking, testing the claims of modern eye tracker manufacturers regarding ease-of-use and portability. The assessment was designed to be systematic and easy to use, completing in around 5 minutes. In general, control participants produced high quality data whereas patients produced poor quality data. We created analysis pipelines for the data and speculate on the use of such a device to increase the efficacy of a physician.

• Subjects / Keywords

  • eye tracking
  • body tracking
  • real-world

• Graduation date

  Fall 2023

• Type of Item

  Thesis

• Degree

  Doctor of Philosophy

• DOI

  https://doi.org/10.7939/r3-qrzq-rm44

• License

  This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.