High immersion does not necessarily produce high presence

What it is

Tracking, field of view, stereo, and refresh can all sit at the current headset ceiling and presence reports can still be low: an empty grey box, standing with no purpose, someone talking in the lab, the wearer wondering whether the strap is slipping. High immersion does not entail high presence. Even a complete supply of sensorimotor contingencies is only feedstock. Whether “I am there” takes hold still depends on whether the content can carry the hypothesis, whether attention stays in the environment, and whether the person is being pulled back by the apparatus itself.

The converse shows up as an anecdote that is easy to dismiss: a motivated short film in a cardboard viewer can outscore an empty gallery on a high-end headset. Entailment fails in both directions.

Why it happens

Presence is a perceptual hypothesis about the current environment, not a linear readout of hardware tier. For the hypothesis to hold, events in the scene have to add up — objects with weight and occlusion, agents with reasons, one’s own actions with consequences. That is the plausibility layer, and it is not written on the GPU. Hardware only decides how wide a sensorimotor base the hypothesis can be built on; a wide base does not mean the hypothesis is adopted.

Attention is a second, independent path. A questionnaire, a cable, the next meeting, pressure on the nose pad, all pull attention out of the environment. Once it is gone, the widest field of view is still a display strapped to a face. Cummings and Bailenson’s meta-analysis gave the magnitude: tracking level, stereoscopy, and field of view have small-to-medium effects on presence — nowhere near “one spec step, one report step.” Spending the whole budget on the immersion inventory and assuming presence will follow treats a correlation as a sufficient condition.

Studying it

The cleanest design holds immersion fixed and splits content or situation: same headset and tracking, manipulate whether the scene has event logic, whether the task has a goal, whether an experimenter is talking in the room. The complement holds content fixed and tiers immersion: the same scene on cardboard, 3DoF, and 6DoF, and looks for where presence gains saturate.

Independent variables: content coherence, presence or absence of a task goal, quality of social agents, frequency of device reminders (cables, calibration prompts). Dependent variables: IPQ subscales (spatial presence and involvement read separately), behavioural responses to a virtual threat, the share of free report spent “there” versus “in the lab.”

A meta-analysis or a multi-lab replication is worth more than a single “we shipped a better headset and the score rose three points.” A rise may be novelty, not entailment confirmed.

Where it stops holding

When immersion is near the floor, the ceiling is set by the system first: a pinhole viewer barely supplies input that could be taken as an environment, so raising immersion will move presence, and non-entailment looks like entailment near the floor. First-session wow inflates any high-end kit; scores drop on repeat exposure, so day-one data cannot prove that more immersion produced more presence. Clinical exposure and pain-distraction applications want particular affective and attentional states; a high presence score is not the success criterion and can even be harmful. Abstract data spaces (node graphs, dashboards) never claimed to be a place; signing them off with a presence questionnaire has no object.

Applying it

• Do not accept headset tier as a presence test. Review content on its own: are there events that can be taken as really happening, and does the task give a reason to leave attention in the environment.
• Lab and booth demos should not stack hard interruptions (live narration, calibration pop-ups, cable warnings) on a high-immersion condition. To demonstrate presence, clear those interruptions from the experience segment rather than adding another effect layer.
• Empty, task-free “look at the engine” builds are for checking smoothness and tracking. Do not score them with a presence questionnaire and write the number into the product story.
• How to check: run the same content on current hardware and on one tier down. If the presence scores are close and the higher tier costs an order of magnitude more, move budget to content and task rather than climbing another immersion step.

Related

• Same group: N1.01.1 Immersion is a system property; presence is a subjective experience · N1.01.3 What breaks presence is often inconsistency in the details
• Nearby: N1.03 Field of View · N4.08 Spatial Scale and Size Perception
• Search terms: immersion does not entail presence · plausibility illusion · presence meta-analysis