Study 3: We used OpenFace to test whether people were just matching based on trustworthiness. Even after accounting for perceived trustworthiness, ingroup faces were objectively more similar to their creators, confirming a role for self-image.

(4/5)

Study 2: Ingroup faces resembled their creators. An independent sample could match ingroup faces to photographs of the people who generated them better than chance, suggesting self-image guided the visual representations.

(3/5)

Study 1: People with higher self-esteem generated ingroup faces that were perceived as more trustworthy to others. Self-evaluation shaped how “us” looked even in novel group settings.

(2/5)

In Study 3, we merely assigned participants to groups without the competitive framing. Sound symbolism effects on group perception reemerged. However, sound–shape congruence rates were still lower than in the no-group condition in Study 1. (6/7)

The Bouba–Kiki effect is a well-documented phenomenon in sound symbolism. People reliably associate round-sounding words like “Bouba” with round shapes and sharp-sounding words like “Kiki” with angular shapes. We asked: What happens when those sounds are used as group names? (1/7)