The definition
Virtual reality replaces what a person sees and hears with a computer-generated environment that responds to their movements and lets them act inside it (Milgram and Kishino, 1994). The principle comes from the head-mounted display Ivan Sutherland described in 1968: a perspective image that changes as the user moves.
Sutherland’s headset, built at Harvard with part of its funding from ARPA, put two miniature cathode ray tubes in front of the eyes and measured the position of the head with either an ultrasonic sensor or a mechanical arm hanging from the ceiling. The arm supplied the six motions needed to track translation and rotation, and the sensors followed the head within a volume about six feet across and three feet high. Custom hardware recomputed the perspective of every line, up to 3,000 lines at 30 frames per second, and the synthetic image covered a 40-degree field of view. Sutherland wrote that stereo matters, though less than the change in the image when the observer moves their head, and current headsets follow the same logic: they track the head, often the hands as well, and redraw the scene for each eye. His display also let the wearer see the room through half-silvered mirrors, so today it would be filed among the optical see-through displays, on the augmented reality side.
The border with its neighbors is drawn by what the person stops seeing. When Ronald Azuma defined augmented reality in 1997, he described the virtual kind by contrast: while immersed, the user cannot see the surroundings. If the headset shows the room, through clear optics or through the video of its cameras, and adds digital content on top, the system belongs to augmented or mixed reality.
Proposals often treat two words as synonyms, and a buyer gains by keeping them apart. Immersion is what the system delivers, and it can be written into a specification: frame rate, tracking latency, field of view, image quality and how much of the body is tracked, among the parameters Mel Slater lists. Presence is the person’s response. Slater calls it place illusion, the sense of being there while knowing one is not, and adds plausibility illusion, the sense that what happens in the scene is actually happening. Immersion sets the boundaries within which place illusion can occur, while plausibility depends on whether the events in the scene relate to the person and are credible against what they expect.
Slater also separates the actions a person takes to perceive, such as turning the head or bending down to look under something, from the actions that change the environment, such as grabbing an object with a tracked hand. The first set is what divides immersive systems from non-immersive ones. Immersive virtual reality on a standalone headset runs with no computer or cables, like the four Fyware Safe simulators on Meta Quest 3, which also need no internet connection. On a PC-tethered headset it can carry heavier scenes. Non-immersive virtual reality runs on a screen with a mouse and keyboard: the person acts on the scene, but turning the head does not change the image.
For a buyer, the type follows from what the body has to do. A procedure that depends on bending down or on judging the distance to a suspended load needs a headset that tracks the head in position and in rotation. When it depends on the order of the steps and on what gets touched, a screen can be enough and saves buying headsets.