3D volume encoding carries the largest error

What it is

Conditional error in 3D volume encoding arises when readers estimate magnitude from the apparent volume of spheres, cubes, or other solids while also resolving three-dimensional scale, two-dimensional projection, viewpoint, and occlusion. In classic channel orderings and several chart experiments, it sits near the high-error end for particular magnitude tasks. “Largest” in the title is an empirical shorthand for those comparisons, not an absolute law across every task, graphic, and reader.

Why it happens

Volume grows with the cube of linear scale: doubling a sphere's radius with value creates eight times the geometrical volume. Even with correct volume mapping, a screen supplies only a projection. Perspective can shrink equally sized distant objects, occlusion removes contours, and lighting or rotation changes apparent shape. Readers may rely on height, projected area, or familiar-object cues, so error can compress differences or take other context-dependent directions; it is not evidence that everyone applies a cube root.

Studying it

Separate solids whose volume encodes value, depth decoration added to two-dimensional data, and data with genuine three-dimensional spatial structure. For magnitude encoding, vary volume ratio, linear scale, projection, viewpoint, occlusion, and interaction; measure magnitude estimates, difference thresholds, and task accuracy. Zacks and colleagues found that extraneous perspective cues reduced bar-reading accuracy, with effects contingent on delay and surrounding marks. Hughes's constant-stimuli study found larger difference thresholds for 3D bars in its tested setup. Report the rendering and task rather than only “3D is worse.”

Where it stops holding

A 3D bar whose value controls height still uses length as its core channel; its projection and context problems are not volume encoding. Molecular structure, medical volumes, terrain, and spatial assembly require three-dimensional relations, where rotation, slices, and depth cues can expose information absent in a flat chart. Spence also found conditions in which readers accommodated nonvarying extra dimensions. Ordering, identification, and spatial navigation may produce other outcomes, so compare total task error instead of judging by dimensionality alone.

Applying it

• For intrinsically two-dimensional data and precise comparison, start with common-scale position, length, or facets. Remove perspective, extrusion, and solids that carry no information.
• If volume genuinely carries a nonnegative magnitude, map by volume rather than radius or edge length, provide values and scale references, and check that order survives viewpoint changes.
• For genuine 3D data, provide rotation, slicing, view reset, and accessible 2D or tabular alternatives. Do not make a critical object identifiable only through an occluded volume.
• Compare 2D and 3D candidates on target-task error, time, and failure type. Record projection, camera, occlusion, device, and reader experience so the conclusion can be audited.

Related

• Same group: U1.04.1 Perceived area systematically underestimates actual area · U1.04.2 Mapping values to diameter inflates them quadratically
• Adjacent: U1.01.5 The ranking is an accuracy ceiling; alignment and mark size erode it · U2.06.5 Precise third-variable comparison belongs in facets or a separate chart
• Search terms: volume perception · perspective cues · 3D chart judgment