Abstract
Introduction. Although the face aftereffect phenomenon is known for face identity, age and emotion, it is unknown how the diversity of the face training set influences aftereffect formation. Observers saw a homogeneous set of distorted faces (middle-aged Caucasian males) and a diverse set of distorted faces (male and female; Caucasian, Asian, Latino and African American) during adaptation. Evidence of aftereffects was tested. Methods. 33 participants underwent an opposing aftereffects paradigm viewing a set of diverse and a set of Caucasian male faces. Participants viewed manipulated face images in one of two adaptation conditions: 1) the diverse face set with features contracted, and the homogeneous face set with features expanded or 2) the opposite. During pretest, participants viewed 64 face pairs, one contracted by 10% and one expanded by 10% and selected which face they found more attractive. During adaptation, faces were distorted by 60%. Post-adaptation testing was identical to pre-adaptation testing. Results. Using a mean change score (contracted faces selected as more attractive before vs. after adaptation) as the dependent variable, the two adaptation conditions were significantly different (F(1,31)=13.072, p= <.05). Adaptation to the contracted diverse face set led to evidence of aftereffects in the expected directions (t(31)=2.685, p= <0.05), but the significant differences observed after adaptation to the contracted homogeneous face set was opposite of the expect direction for both conditions (t(31)=3.061, p= <0.05). Aftereffects were created only by adapting to the diverse face set, not the homogeneous face set. These results indicate that adapting to the diverse face set after a featural manipulation led to aftereffects for all faces (the diverse and the homogeneous face sets) in the same direction. This result suggests that diversity within a face set influenced aftereffect formation.