How to Build a Dichoptic Presentation System That Includes an Eye Tracker

Summary

We recently proposed a method that allows dichoptic visual stimulus presentation and binocular eye tracking simultaneously¹. The key is the combination of an infrared eye tracker and the corresponding infrared-transparent mirrors. This manuscript provides an in depth protocol for initial setup and everyday operation.

Abstract

The presentation of different stimuli to the two eyes, dichoptic presentation, is essential for studies involving 3D vision and interocular suppression. There is a growing literature on the unique experimental value of pupillary and oculomotor measures, especially for research on interocular suppression. Although obtaining eye-tracking measures would thus benefit studies that use dichoptic presentation, the hardware essential for dichoptic presentation (e.g. mirrors) often interferes with high-quality eye tracking, especially when using a video-based eye tracker. We recently described an experimental setup that combines a standard dichoptic presentation system with an infrared eye tracker by using infrared-transparent mirrors¹. The setup is compatible with standard monitors and eye trackers, easy to implement, and affordable (on the order of US$1,000). Relative to existing methods it has the benefits of not requiring special equipment and posing few limits on the nature and quality of the visual stimulus. Here we provide a visual guide to the construction and use of our setup.

Introduction

Under normal viewing conditions each of our eyes receives a slightly different visual input. This input is then processed to produce one coherent, three-dimensional representation of the world. Dichoptic presentation, the practice of independently controlling the input presented to each of the two eyes, thus enables researchers to study how humans reconstruct a three-dimensional representation from two two-dimensional retinal images^2,3,4. In addition, if the two eyes' images are too dissimilar, this interocular combination fails, and observers instead report perception of only one of the images at a time while the other remains suppressed, in phenomena such as binocular rivalry⁵ and continuous flash suppression⁶. Researchers of such interocular suppression, too, use dichoptic presentation, in this case to examine questions related to topics like the neural locus of awareness⁷, perceptual selection^8,9, and unconscious processing¹⁰.

Gaze and pupil dynamics are recorded for multiple purposes in research on human behavior and perception. Gaze direction can inform about, for instance, attention allocation^11,10,13 and decision making¹⁴, while pupil size can reveal aspects of visual processing^15,16, task engagement¹⁷, or fluid intelligence¹⁸.

Combining eye tracking with dichoptic presentation is useful in research into, for instance, three dimensional (3D) perception^19,20,21,22 or ocular responses to visual input during interocular suppression^23,24,25. For example, eye movements have been found to reveal unconscious processing without subjective perception during continuous flash suppression²³. Clinical visual researchers can use the ability to track both eyes during dichoptic presentation to investigate ocular diseases that affect the two eyes asymmetrically, for example, to monitor the monocular and binocular visual distortions occurring in amblyopia²⁶ and maculopathy²⁷.

We recently described a setup¹ that allows for the combination of high quality video-based eye tracking and dichoptic stimulation with little limitation on the size or color of the stimuli, and we evaluated its performance. Below we will summarize the construction and use of this setup.

Protocol

This protocol has been approved by the Institutional Review Boards of Michigan State University.

1. Building the system

1. Rationale
   1. Prepare the mirror setup, a variant of the classic Wheatstone stereoscope²⁸ illustrated in Figure 1, consisting of two mirrors positioned at a 45° angle relative to the participant's midline. The mirrors reflect stimuli from two screens that are positioned at opposite ends of a table, facing each other.
   2. Seat a participant in front of the mirrors and have them view a different screen, reflected via a different mirror, with each eye. For best results, use a head rest for stabilizing the participant's head.
   3. Position an infrared-sensitive video-based eye tracker, including a camera and an illuminator, in front of the participant but behind the mirrors. The eye tracker is represented by a box in Figure 1.
      NOTE: One challenge when trying to track the eyes in normal setups of this type, is that the eyes are blocked by the mirrors.
   4. Use two front-surface mirrors, often advertised as "cold mirrors" (incident angle: 45°), that feature near-complete reflectance of visible wavelengths and near-complete transmission of near-infrared wavelengths (see Table 1 for detailed information about the mirrors).
      NOTE: Such mirrors can be obtained via companies supplying optical equipment for scientific and industrial purposes, which usually list components like these as 'cold mirrors' or as a type of 'dichroic mirrors' (see more detail in Materials/Equipment Table).

      		Setup 1	Setup 2
      Mirrors	Dimensions	10.10 × 12.70 cm	10.10 × 12.70 cm
      	Reflectance	400 ~ 690 nm	425 ~ 650 nm
      	Transmission	750 ~ 1200 nm	800 ~ 1200 nm
      Eye Tracker	Brand	Research-end Eye Tracker	Customer-grade Eye Tracker
      	Transmission	890 ~ 940 nm	Around 850 nm

      
      Table 1. Details of two versions of the setup with which we have worked.
      The eye tracker's transmission wave length range is covered by the mirrors' transmission range at a 45° incidence angle, but outside their reflectance range.
2. Structure of the Setup
   1. Build the setup on top of a desk. Besides the mirrors and eye tracker, it consists only of three custom-built elements made of fiberboard (see Figure 2 for an assembly guide) and two flat-screen monitors on monitor-arms available from normal office supply stores.
   2. Fiberboard elements
      1. Build the framework of the setup from three components of fiberboards: one central component and two reference boards on each side (see Figure 1 for general positioning, Table 2 for detailed dimensions, and Figure 2 for an assembly guide of each component). Paint all these pieces in matte black to reduce light scatter.
         NOTE: The central component (see Figure 2B and 2D) holds the mirrors and eye tracker. Both are on the same plateau, thus keeping the eye tracker at participants' eye level.
      2. Place the top element of this component such that it leaves 8 cm in depth in the front of the desk. Such an arrangement allows enough room for the participant's face when stabilized on the head rest and avoids condensation on the mirrors during expiration, while minimizing the distance between the participant's eyes and the mirrors to maximize the possible use of the participant's visual field.
      3. Position the two reference boards straight below the monitors (see Figure 1 for positioning and Figure 2 panels A and C for an assembly guide) for easy manual calibration of the screens. Note that the apparent offset in Figure 1 between screen and board is due to limited depth cues in the image; the boards are straight below the monitors on both sides.
      4. Exactly align the long horizontals with the edges of the desk, while the long verticals leave 4 cm beyond the front of the desk for ease of stabilizing a calibration board (see below) to these boards. The two small verticals will ensure the long vertical staying vertical as the reference for the monitors.
      5. Optionally, use a separate piece of fiberboard as a calibration board (see Figure 3). In this case, after obtaining an optimal position of a monitor, position the calibration board against the reference board and indicate the positions of both the reference board and the monitor on the calibration board while it is in place (in the example of Figure 3, wooden slats provide these indications).
      6. Whenever this desired monitor position is lost (accidentally or because other experiments require a different position), retrieve this position by using the markings on the calibration board to put the calibration board back in the same place relative to the reference board that has a fixed position on the desk. Move the monitor again to line up with the appropriate markings (see step 2.1.1. for details).

         Component	Dimensions (cm)	Number	Remark
         Central Component	80 × 25 × 2	1	Horizontal top
         	23 × 25 × 2	1	Horizontal bottom
         	21 × 32 × 2	1	Central vertical
         	32 × 25 × 2	1	Front-facing vertical
         Reference Boards	61 × 11 × 2	2	Long horizontal
         	66 × 29 × 2	2	Long vertical
         	11 × 15 × 2	4	Small vertical

         
         Table 2. Details of the fiberboard components.
   3. Monitors and Mirrors
      1. Position the setup on top of a standard office desk.
      2. Mount two flat-screen monitors on standard monitor arms clamped to the side of the desk (clamping both the reference board and the desk). These arms allow translation in three dimensions as well as rotation in the plane of the screen. Conventional CRT-monitors are clearly also compatible with the setup, but would not afford the same flexibility in terms of positioning and repositioning.
      3. Mount the mirrors on mirror mounts that are sold for the purpose by the same suppliers that stock cold mirrors. Connect these mounts to the fiber board holding the mirror at participants' eye level. Position the mirrors to touch at a 90° angle in the center, right before the participant's nose.
   4. Remaining elements
      NOTE: Some experiments require participants not seeing the screens from the corner of their eyes, so that a direct line of sight to the screens (dashed lines in Figure 4A) should be avoided.
      1. In that case, create "blinders" made of black cardboard and foam-padded hole straps painted in black, and attach them to the posts of the head rest (see Figure 4B). Adjust the blinders in height and angle to accommodate individual participants. If the wall in front of the participant has high reflectance, hanging a piece of black fabric will help remedy this.

2. Using the system

1. Hardware calibration
   NOTE: The purpose of calibration is to achieve satisfactory alignment of the two monitors for ease of fusion of the two monitors' images for each participant. This can be achieved in two steps: hardware calibration (described here) and software calibration (described below).
   1. When using a calibration board, as described above, align it with one of the reference boards, holding it in place with a C-clamp if needed, and then move the corresponding monitor to line up with the desired reference lines on the calibration board. The monitors should be parallel to each other, and each should be straight above its reference board.
   2. When using blinders, move them to the participant's eye level and rotate them slightly toward the midline, i.e. more inward, compared to the orientation of the monitors. Make sure that each eye can see the whole visual stimulus in the mirror without seeing any of it directly. Turning the blinders toward rather than away from the midline will minimize participants' exposure to other visual input.
2. Software calibration
   1. Since participants may vary in their eye position relative to the mirrors despite the use of a head rest, calibrate further before doing experiments. This part is most easily done in the software, i.e. without moving the setup's parts any further. There are two possible methods.
      1. For the first, present a dot on each of the two screens in alternation, and instruct the participant to eliminate the perceived position change by moving the dot on one of the screens (or both in opposite directions).
      2. For the second method, instruct the participant to align the frames of experimental stimuli instead of two dots so that both eyes' visual fields critical to the particular experiment are aligned.
   2. After applying either method, center the stimuli in the experiment on the resulting on-screen positions. Other aspects of setting up displays and stimuli for dichoptic presentation in general can be found elsewhere⁵.

Results

After the calibration described in the protocol, we performed a calibration-validation procedure without problems with the mirrors in place. The effectiveness of the method is clearly illustrated by Figure 5, which shows the camera's image (using a research end eye tracking system) with the mirrors in place. The two sets of parallel lines along participants' nose and the lines above the eye brows are the edges of the mirrors but, nevertheless, the face is as ...

Discussion

We present a step-by-step guide for the construction and use of an experimental setup that allows simultaneous tracking of the both eyes and dichoptic presentation of visual stimuli. In many situations where dichoptic stimulation is used, the critical issue preventing effective eye tracking is that the mirrors for dichoptic presentation block the sight of video-based eye trackers. This is resolved here by using infrared-transparent mirrors and an infrared-sensitive eye tracker. This setup allows researchers of 3D vision,...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Mirrors in Setup 1	Edmund Optics	#64-452	dimensions 10.10 × 12.70 cm; Reflectance: 400 ~ 690 nm; Transmission: 750 ~ 1200nm
Mirrors in Setup 2	Edmund Optics	Item discontinued	dimensions 10.10 × 12.70 cm; Reflectance: 425 ~ 650 nm; Transmission: 800 ~ 1200nm
Other Mirror Option	Edmund Optics	#62-634	dimensions 12.50 × 12.50 cm; Reflectance: 425 ~ 650 nm; Transmission: 800 ~ 1200nm
Eye Tracker in Setup 1	SR Research Ltd., Mississauga, Ontario, Canada	Eyelink 1000	Transmission: 890 ~ 940 nm
Eye Tracker in Setup 2	The Eye Tribe Aps, Copenhagen, Denmark	Eye Tribe (item discontinued)	Transmission: around 850 nm