Poster #1 - Bimodal familiarization to other-race faces improves discrimination performance in 12-month-old infants

• In Event: 1-115 - Poster Session 03
  In Poster Session: PS 03 Section - Attention, Learning, Memory

Integrative Statement

The Other-Race effect is a well-established phenomenon, wherein people have much worse abilities in accurately perceiving faces which belong to ethnicities other than those which they grew up seeing. This effect begins to appear in infants during the second half of the first year of life. Before this point, infants perceive both same-race and other-race faces with a similar level of accuracy, but afterwards, infants decline in their ability to perceive other-race faces, while improving their ability to accurately perceive same-race faces. It has been hypothesized that this process of perceptual narrowing is driven by statistical learning, whereby the infant brain becomes more adapted to discriminating among faces which surround them in their everyday environment, at the expense of face types which are encountered more rarely. One key part of this hypothesis is that the greater the variety of faces belonging to a single type which infants see, the better their face processing system will be able to calibrate to this face type. Exposing infants to a greater variety of other-race faces should therefore improve discrimination among other-race faces
The goal of our experiment was to test this possibility by familiarizing 40 Caucasian (German) 12-month-old infants with no prior exposure to persons of Asian descent to a set of morphs created from two original other-race faces, and then testing these infants on their ability to discriminate among those faces.
Our stimuli consisted of two original photographs of Asian (Chinese) faces, as well as 19 morphs which were creating by combining the original faces at 5% intervals (e.g. the exact middle point between Face A and Face B would be the 50%/50% morph). During the familiarization stage infants saw 159 face morphs, with each morph being presented for 750ms. Twenty infants were familiarized unimodally (less variety, with the 50%/50% morph appearing most often during familiarization), and twenty were familiarized bimodally (more variety, with the 25%/75% and 75%/25% morphs appearing most often during familiarization), see Figure 1. Under the statistical learning hypothesis, we expected that infants in the bimodal familiarization condition would outperform infants in the unimodal condition in a discrimination test thanks to being exposed to a greater variety of morph variations.
Test trials consisted of infants seeing a prime face (100% face A or 100% face B) for 750ms, and a target face (100% face A or 100% face B) for up to 20s or until infants looked away for 2s. Equal numbers of match (same prime and target faces) and mismatch (different prime and target faces) trials were shown in randomized counter-balanced order.
Results showed that bimodally familiarized infants could discriminate face A from face B as indicated by different looking times toward the target faces in the match versus mismatch trials (p<.01), while unimodally familiarized infants could not discriminate between the faces, as indicated by similar looking times towards the target faces in the match and mismatch trials (Figure 2).
Our results therefore support the statistical learning conception of the Other-Race effect, and further implications will be discussed.

Group Authors

Research Unit "Crossing the borders"

Authors

• Anna Krasotkina, Justus Liebig University Giessen
  Presenting Author

• Antonia Goetz, Potsdam University
  Non-Presenting Author

• Barbara Hoehle, Potsdam University
  Non-Presenting Author

• Gudrun Schwarzer, Department of Developmental Psychology, Justus-Liebig-University Giessen
  Non-Presenting Author