Infants' Individuation of Faces by Gender

Abstract

By 3 months of age, infants can perceptually distinguish faces based upon differences in gender. However, it is still unknown when infants begin using these perceptual differences to represent faces in a conceptual, kind-based manner. The current study examined this issue by using a violation-of-expectation manual search individuation paradigm to assess 12- and 24-month-old infants' kind-based representations of faces varying by gender. While infants of both ages successfully individuated human faces from non-face shapes in a control condition, only the 24-month-old infants' reaching behaviors provided evidence of their individuating male from female faces. The current findings help specify when infants begin to represent male and female faces as being conceptually distinct and may serve as a starting point for socio-cognitive biases observed later in development.

Keywords: face processing; gender; individuation; infants; manual search task.

Figure 1

Example of face identities. Condition 1 infants were presented with human versus non-human images on foam-core blocks. Faces were either male or female paired with non-human image. Condition 2, infants were presented with male versus female faces. In both conditions infants either saw Caucasian or African American faces. There were a total of 6 different face identities for each face race group.

Figure 2

Manual search box task. Infants completed 8 trials, 4 trials per face block. There were three types of face outcomes. The image on the left depicts the expected-empty trial. Such that infants saw one face go in and only found one face within the box. The image on the right depicts the unexpected-empty and expected-empty-final trials. Unexpected-empty trials included infants seeing two faces go inside of the box but only finding one face. Expected-empty-final trials followed unexpected-empty trials. Infants were presented with the second face, leaving the box completely empty.

Figure 3

Depiction of the age separated two-way mixed measures ANOVA. Twelve-month-old infants reached significantly longer during the unexpected-empty relative to the expected-empty and expected-empty final outcomes. Additionally, infants showed longer reaching during the expected-empty than during the expected-empty final outcome. Twenty-four-month-old infants reached significantly longer during the unexpected- than during the expected-empty final outcome. Yellow circles represent each individual reaching data point. Error bars represent the standard error of the mean. The solid line represents a significant difference, which was not maintained following Bonferroni correction; * represents p < 0.05 Bonferroni corrected.

Figure 4

Depiction of the significant interaction of face outcome by face stimuli race found for the 24-month-old reaching duration. Infants reached significantly longer during unexpected-empty outcome than during the expected-empty outcome for Caucasian face stimuli but not for African American stimuli. Infants reached significantly longer during unexpected-empty versus expected-empty final outcome for African American face stimuli. Yellow circles represent each individual reaching data point. Error bars represent the standard error of the mean; * represents p < 0.05 Bonferroni corrected.

Figure 5

A depiction of age differences in reaching duration patterns for Condition 2. Twelve-month-old infants reached longer during unexpected-empty versus expected-empty final. In contrast, 24-month-old infants, but not 12-month-old infants, reached significantly longer during the unexpected-empty compared to expected-empty and expected-empty final. Error bars represent standard error of the mean, * represents p < 0.05 Bonferroni corrected.

Figure 6

The graph depicts differences in reaching duration between face outcome types separated by face stimuli race. Infants who saw Caucasian faces reached significantly longer during the unexpected-empty than during the expected-empty and expected-empty final outcome. Infants who saw African American faces did not show significant differences in reaching duration. Error bars represent the standard error of the mean; * represents p < 0.05 Bonferroni corrected.