Mapping the Other-Race-Effect in Face Recognition Using a Three-Experiment Test Battery

The fact that people are better at recognizing faces of their own race than others is called the other-race-effect (ORE). Most studies use only a single test to map and determine the characteristics of the ORE, however. Here, we investigated how two groups of fifteen age-matched Korean and German participants recognize Asian and Caucasian faces with three experiments as part of testing a new battery for characterizing face-processing performance. Participants first underwent the standard Cambridge face memory test in which they had to learn Caucasian target faces at varying noise levels which then were to be recognized in a forced-choice task. In this task, German participants performed significantly better than Koreans (83% versus 72%). The second experiment used a standard old-new recognition task with 20 Caucasian and 20 Asian faces (courtesy of the tarrlab@CMU). Here, Koreans were better with Asian faces (d'-difference = 1.23) whereas Germans only showed a tendency towards an ORE (d'-difference = 0.44). In the third experiment, participants had to rate the similarity of Caucasian face pairs which varied parametrically along featural and configural dimensions using the morphable faces from the MPI face-database. Here, we found that Korean participants were significantly less sensitive to featural changes than German participants. In conclusion, we were able to demonstrate an ORE for most of our experimental conditions. Interestingly, data from the third experiment suggest that the ORE may be due more to lessened sensitivity to featural than to configural processing for other-race faces. Future studies will extend this new test battery to prosopagnosics