How do we look at our own face compared with a stranger's?
People spent relatively more time looking at the lower half of faces they judged to be their own, and those with a sharper mental image of their own face looked more at the eye region of their own face.
Source
Looking at My Own Face: Visual Processing Strategies in Self-Other Face Recognition
Study at a glance
- Design
- Human experiment — Within-subject self-face recognition task with individually made self-to-novel face morphs (15 morph levels, two runs of 150 trials) and simultaneous eye-tracking of upper vs lower face regions; AQ measured afterwards.
- N
- N=33 · 33 healthy adults (2 male); some individual-difference correlations excluded influential data points.
- Population
- Right-handed university students and community adults around the University of Reading, mean age 20.67 years, mostly female.
- Outcome
- Gaze duration and proportion to upper (eyes) vs lower (mouth) face regions for faces judged 'self' vs 'other'; slope of the psychometric self-recognition curve; Autism Spectrum Quotient scores.
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
33 adults were photographed and their face was morphed with a same-sex, same-ethnicity stranger in 15 steps from fully self to fully other. On each trial a morph appeared briefly and participants pressed a key to say 'self' or 'other' while an eye tracker recorded where they looked. The steepness of each person's self-recognition curve served as a measure of how distinct their self-face representation was, and participants also completed the Autism Spectrum Quotient.
What they found
Faces judged as 'self' drew a greater proportion of gaze to the lower (mouth) region than faces judged as 'other' (effect size r = 0.42), because looking at the lower face lasted longer while looking at the eyes was similar for both identities. People with steeper self-recognition curves looked relatively longer at the upper face of 'self' versus 'other' faces (Kendall's tau = 0.23). Autistic traits were not related to where people looked on the face, but higher traits were linked to less total looking time at both self and other faces.
The limits
What it doesn't show
The sample was small and almost entirely female (2 men), so gender differences could not be examined and the correlations are modest. The design is correlational for individual differences, so it cannot show whether a distinct self-face representation causes the gaze pattern or vice versa. No familiar-other face was included, so the self-face effects may simply be familiarity effects; faces were static, central, closed-mouth images. Autistic traits were measured in a non-clinical sample in which no one reached the clinical cut-off, so results may not apply to diagnosed autism.
Key terms
- Face morph continuum
- A series of images blending two faces in graded proportions, used to find where perception switches from one identity to the other.
- Psychometric function slope
- How sharply responses switch from one category to another across a stimulus continuum; a steeper slope here means a more distinct self-face representation.
- Region of interest (ROI)
- A predefined area of an image, such as the upper or lower face, within which eye-tracking fixations are counted.
- Autism Spectrum Quotient (AQ)
- A self-report questionnaire measuring autistic traits, which vary continuously across the general population.
- Social motivation theory of autism
- The proposal that reduced attention to faces and eyes in autism reflects lower reward value of social stimuli.
Flashcards
0 of 10 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
Which face region received a larger share of gaze when faces were judged as 'self'?
Common questions
Why morph the participant's face with a stranger's instead of just showing photos?
Morphing creates ambiguous images, so the researchers could measure how sharply each person's 'self' judgements switch to 'other', giving a quantitative index of self-face representation.
Does this show autistic people look less at their own face?
Not specifically. Higher autistic traits went with less looking at faces in general, both self and other, and none of the participants reached the clinical cut-off, so diagnosed autism was not tested.
Why did people look more at the lower part of their own face?
Looking at the eyes was similar for both identities, so the authors suggest people explore more features of a highly familiar and possibly rewarding face; familiar-other faces were not tested to confirm this.
More on Face processing