The development of facial identity discrimination through learned attention

Learned attention models of perceptual discrimination predict that with age, sensitivity will increase for dimensions of stimuli useful for discrimination. We tested this prediction by examining the face dimensions 4- to 6-month-olds (n = 77), 9- to 12-month-olds (n = 66), and adults (n = 73) use for discriminating human, monkey, and sheep faces systematically varying in outer features (contour), inner features (eyes, mouth), or configuration (feature spacing). We controlled interindividual variability across species by varying faces within natural ranges and measured stimulus variability using computational image similarity. We found the most improvement with age in human face discrimination, and older participants discriminated more species and used more facial properties for discrimination, consistent with learned attention models. Older infants and adults discriminated human, monkey, and sheep faces; however, they used different facial properties for primates and sheep. Learned attention models may provide insight into the mechanisms underlying perceptual narrowing.