Without a known-size referent, vision alone cannot judge absolute scale

What it is

An unseen cube in an empty white room, asked “how tall,” scatters answers across an order of magnitude. Familiar size is the anchor vision uses to give objects an absolute scale: a door, a chair, a hand, a credit card. Strip those known objects from a virtual scene and retinal angle is only “larger than what”; metres cannot be read off the picture.

This is not “it looks like a toy so presence drops,” and not whether earlier scenes will bias later ones. It asks only whether this glance has a referent to calibrate with.

Why it happens

Visual angle is joint in true size and distance. Without familiar size, the same angle can be a small thing nearby or a large thing far away. Real rooms are full of body-scale objects, so calibration is almost automatic. A virtual scene can be an untextured infinite plane, abstract geometry, or architecture with doors and windows deliberately removed, and the calibration cues are emptied out. Headsets also crop some periphery and ground texture; even “how long is my own foot” is sometimes out of view.

People grab whatever anchor remains: their own hand, controller length, default eye height. If content has rewritten those anchors — a giant’s hand, a child’s eye height — absolute scale will be wrong, and confidently so. With no anchor, judgments jump among guesses rather than converging on a stable metre.

Studying it

Use verbal size estimates or a matching task: show an object with no familiar companions, have people report metres or adjust a virtual ruler; then add a door, a human figure, a hand, and see whether the spread of estimates narrows. Manipulate referent type and whether the person’s own body is visible. Record absolute error and variance, not only the mean — no-anchor conditions often explode in variance while the mean happens to sit close.

Control binocular disparity and motion parallax, or those depth cues will fill in and the test is no longer about familiar size. Sit the object on a ground plane of known eye height and the ground itself becomes a scale cue; the empty-room condition is contaminated.

Where it stops holding

With a self-avatar and visible hands and feet, the body is the strongest familiar size and an abstract scene will not fully lose its anchor. Objects not shared across cultures — a particular socket, a particular street lamp — cannot be universal referents. At very near grasping distances, the hand’s visual angle and proprioception take over and familiar-size weight drops. Swap “how tall” for “can I walk under it” and the task is already affordance; people will try with the body rather than report a metre.

Applying it

• In scenes where people must judge true size (furniture preview, architectural review, clinical dimensions), place at least one body-scale referent: a door, a chair, a standing person, and let people see their own hands.
• If an abstract or artistic scene deliberately withholds absolute scale, do not ask “how many metres” in the UI; ask relative relations.
• Do not rely on a ground grid as a ruler — if the spacing is unlabelled, people will read it as an arbitrary scale.
• How to check: strip doors, windows, furniture, and have someone new to the scene estimate an unknown object’s height. If spread crosses an order of magnitude, vision did not supply absolute scale. Put a standard door back and the spread should narrow.

Related

• Same group: N4.08.2 A world miniature lets people operate a large environment from a god view · N4.08.3 Extreme deviation from real scale weakens presence · N4.08.4 Scaling the self turns far-field operations into near-field ones · N4.08.5 Once scale is established in a scene, it keeps biasing size judgments of new objects · N4.08.6 Virtual scale changes perception of distance and of one's own size · N4.08.7 Inconsistent scale breaks action expectations
• Nearby: N1.01 Immersion and Presence · N2.04 Direct Grab
• Search terms: familiar size · absolute scale · size constancy