Gaze in Action: Head-mounted Eye Tracking of Children's Dynamic Visual Attention During Naturalistic Behavior

Abstract

Young children's visual environments are dynamic, changing moment-by-moment as children physically and visually explore spaces and objects and interact with people around them. Head-mounted eye tracking offers a unique opportunity to capture children's dynamic egocentric views and how they allocate visual attention within those views. This protocol provides guiding principles and practical recommendations for researchers using head-mounted eye trackers in both laboratory and more naturalistic settings. Head-mounted eye tracking complements other experimental methods by enhancing opportunities for data collection in more ecologically valid contexts through increased portability and freedom of head and body movements compared to screen-based eye tracking. This protocol can also be integrated with other technologies, such as motion tracking and heart-rate monitoring, to provide a high-density multimodal dataset for examining natural behavior, learning, and development than previously possible. This paper illustrates the types of data generated from head-mounted eye tracking in a study designed to investigate visual attention in one natural context for toddlers: free-flowing toy play with a parent. Successful use of this protocol will allow researchers to collect data that can be used to answer questions not only about visual attention, but also about a broad range of other perceptual, cognitive, and social skills and their development.

Figure 1.. Head-mounted eye tracking employed in three different contexts:

(A) tabletop toy play, (B) toy play on the floor, and (C) reading a picture book.

Figure 2.. Setting up the head-mounted eye-tracking system.

(A) A researcher positioning an eye tracker on an infant. (B) A well-positioned eye tracker on an infant. (C) Good eye image with large centered pupil and clear corneal reflection (CR). (D, E, F) Examples of bad eye images.

Figure 3.. Three different ways of obtaining calibration points.

Two views of each moment are shown; top: third-person view, bottom: child’s first-person view. Arrows in the third-person view illustrate the direction of a laser beam. Inset boxes in the upper right of the child’s view show good eye images at each moment used for calibration and pink crosshairs indicate point of gaze based on the completed calibration. (A) Calibration point generated by an experimenter using a finger and laser pointer to direct attention to an object on the floor. (B) Calibration point generated by an experimenter using a laser pointer to direct attention to dots on a surface. (C) Calibration point during toy play with a parent in which the child’s attention is directed to a held object.

Figure 4.. Example plots used to assess calibration quality.

Individual dots represent per-frame x-y point of gaze (POG) coordinates in the scene camera image, as determined by the calibration algorithm. (A) Good calibration quality for a child toy-play experiment, indicated by roughly circular density of POG that is centered and low (child POG is typically directed slightly downward when looking at toys the child is holding), and roughly evenly distributed POG in the remaining scene camera image. (B) Poor calibration quality, indicated by elongated and tilted density of POG that is off-centered, and poorly distributed POG in the remaining scene camera image. (C) Poor calibration quality and/or poor initial positioning of the scene camera, indicated by off-centered POG.

Figure 5.. Two children’s eye-gaze data and statistics.

(A) Sample ROI streams for Child 1 and Child 2 during 60 s of the interaction. Each colored block in the streams represents continuous frames in which the child looked at an ROI for either a specific toy or the parent’s face. White space represents frames in which the child did not look at any of the ROIs. (B) Proportion of time looking at the parent’s face and 10 toy ROIs, for both children. Proportion was computed by summing the durations of all looks to each ROI, and dividing the summed durations by the total session time of 6 minutes. (C) Mean duration of looks to the parent’s face and ten toy ROIs, for both children. Mean duration was computed by averaging the durations of individual looks to each ROI during the 6-minute interaction.