Skip to content
PaperFren

When half a face is hidden, do we imagine it as more average?

Open paper intelligence

When people imagined the hidden half of a face, they consistently pictured it as more average in shape than it really was, especially for faces that looked unusual.

Source

Incomplete faces are completed using a more average face

Kramer RSS, Jones AL · Cognitive research: principles and implications · 2022

doi.org/10.1186/s41235-022-00429-yRead the full paper ↗6 citationscc by

Study at a glance

Design
Human experiment — Participants saw half a face and adjusted the missing half along a shape continuum from caricature (-50%) to anti-caricature (+50%, towards the same-sex average) to show how they imagined the real face; upper- vs lower-half condition between subjects
N
N=110 · 110 participants in the main task (56 adjusted the lower face, 54 the upper face); a separate online sample of 160 rated the 60 original faces for typicality or attractiveness
Population
Mostly young White adult volunteers and university participants in the UK, judging 60 White young-adult face photographs
Outcome
Morph percentage selected as 'how the person actually looks' (0% = original), and whether this shift was predicted by the face's perceived and geometrically measured typicality

Structured fields used in claim comparison tables when every cited study has a complete layer.

What they did

The researchers showed 110 participants 60 photographs with either the upper or the lower half of the face blanked out. On each trial, participants used arrow keys to scroll through versions of the missing half that ranged from exaggerated (caricatured) to shifted towards the average face shape, and stopped on the version they thought matched the real person. A separate group of raters judged how typical and how attractive each original face was, and the researchers also computed each face's geometric distance from the average shape.

What they found

Instead of choosing the true original, participants picked versions shifted towards the average: on average 19.16% of the way along the continuum for the lower half and 16.53% for the upper half, both very large effects. Faces that raters judged as less typical were completed with an even bigger shift towards the average, while the geometric typicality measure gave weaker and less consistent effects. Perceived typicality and attractiveness were strongly correlated across faces.

The limits

What it doesn't show

The study did not measure attractiveness of the completed faces, so it cannot say directly whether masks make faces look more attractive; and because making a face more average also tends to make it more attractive, the authors cannot tell whether people imagined faces as more average specifically or just more attractive. Only face shape was morphed, not skin texture, and all stimuli (and most participants) were White, so the own-race average may behave differently for other-race faces. The upper- and lower-half conditions were not compared statistically because the study was not powered for that, and the task is an artificial lab procedure rather than real encounters with masked people.

Key terms

Face space / prototype
The idea that we store faces relative to a mental average built from all the faces we have seen; each face is coded by how it differs from that average.
Anti-caricature
A version of a face morphed towards the average so its distinctive features are toned down; a caricature does the opposite.
Positivity bias for incomplete faces
The earlier finding that partly hidden or incomplete face images tend to be rated as more attractive than the complete originals.
Typicality
How close a face is to an average face; here measured both by raters' judgements and by geometric distance from the average shape.
Regression towards the mean
The statistical tendency for extreme scores to move closer to the average on a second measurement, which can mimic a real effect when faces are split into high and low groups.

Flashcards

1 / 11

0 of 11 answers reviewed

Research intelligence for this paper

See its role on concept claims, tensions it is part of, placement history, and related discoveries.

Open paper intelligence

Quiz yourself

1 / 6

When participants reconstructed the hidden half of a face, what did they tend to choose?

Common questions

Why would filling in a face with the average make it look more attractive?

Average faces are generally judged as attractive, so if the brain fills the missing part using its stored average, the completed face ends up more average and therefore often more attractive.

Does this mean everyone looks better in a face mask?

Not necessarily. The study measured how people imagined the hidden half, not how attractive they rated masked faces, and earlier studies disagree about whether already-attractive faces gain or lose attractiveness with masks.

Why include a geometric measure of typicality as well as ratings?

Because the morph steps were smaller for faces already close to the average, the geometric measure let the authors control for this stimulus artefact when testing whether perceived typicality predicted the shift.

More on Face processing