Association of idiosyncratic eye-movement patterns with holistic processing of faces as measured by the composite face effect and the face inversion effect

ABSTRACT

Faces are said to be processed holistically, with the composite face paradigm and the face inversion paradigm being two widely used tasks to demonstrate this mode of processing. However, recent studies have found that individuals differ in their eye movements to faces, and little is known about whether these differences in eye movements are associated with holistic processing. Here, participants were asked to complete the inversion task and the top-cued (or bottom-cued) composite task. Using a Hidden Markov Model approach to analyze eye movements, participants were clustered into two groups on the basis of their eye movements in identifying upright faces, with an upper-focused group who preferred to look at the eye region, and a lower-focused group who preferred to look at the centre of faces. These two groups did not show any differences in the size of the inversion effect. But in the composite task, the upper-focused group showed a stronger composite effect for matching upper halves of faces than lower halves, whereas the composite effect shown by the lower-focused group had a similar magnitude no matter which half was being matched. Thus, holistic face processing as measured by the composite effect is associated with one’s preferred face-scanning pattern.

KEYWORDS:

• Individual difference
• eye movements
• hidden markov model
• holistic processing
• face inversion effect
• composite face effect

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 For completeness, we report the additional main effects and interactions here: The main effect of alignment was marginally significant, F (1,54) = 3.11, p = 0.083, ${\eta }_p^2$ = 0.054, and the main effect of eye-movement pattern was not significant, F (1,54) = 0.77, p = 0.39, ${\eta }_p^2$ = 0.014. The interaction between cue half and eye-movement pattern was significant, F (1,54) = 13.01, p < 0.001, ${\eta }_p^2$ = 0.1, and the three-way interaction among eye-movement pattern, cue half, and alignment was significant, F (1,54) = 5.29, p = 0.025, ${\eta }_p^2$ = 0.09. The interaction between congruency and cue half was marginally significant, F (1,54) = 3.40, p = 0.07, ${\eta }_p^2$ = 0.06, as was the three-way interaction among cue half, congruency and alignment, F (1,54) = 3.16, p = 0.081, ${\eta }_p^2$ = 0.06.

Additional information

Funding

This work was supported by General Research Fund of the Hong Kong Research Grants Council: [Grant Number LU17608519].