Abstract
Background: Allocentric kin recognition – recognition that individuals are close genetic relatives – plays an important role in social organization and kin selection (Hamilton, 1964). Facial judgments – including kin recognition – are typically modeled as a form of cue combination but with the identity of cues currently unknown. In two experiments, we examined human ability to classify pairs of children as siblings or not siblings and tested whether kinship information in different facial regions combined as statistically independent cues. We used complementary masks to occlude a face region or present it in isolation (Exp. 1 eye region; Exp. 2: mouth region) and tested whether we could predict performance with the unmasked face from performance in the two masked conditions. Methods: 30 pairs of photographs, each portraying a frontal view of a child’s face. Half were siblings, half unrelated. A total of 122 adult observers (Exp. 1) or 121 (Exp. 2) judged whether each pair were siblings in one of three conditions. We summarized performance in each condition by signal detection d’ estimates. Conditions: Experiment 1: Eye region masked (EM), eye region only (EO), full face (FF). Experiment 2: mouth region masked (MM), mouth region only (MO), full face (FF). Results: Experiment 1: Performance was above chance in all three conditions. FF: d’ = 1.067, EM d’ = 0.677, EO d’ = 0.991. 2: Performance was above chance in all three conditions. FF: d’ = 1.132, MM d’ = 1.249, MO d’ = 0.416. We tested additivity of kinship information in facial regions: whether the sum of d’ values squared for the masked conditions was equal to the square of the d’ for the masked face. We did not reject the hypothesis of additivity p > 0.05) for either experiment.
Meeting abstract presented at VSS 2015