Investigating the other-race categorization advantage within the context of learning

Abstract

It is well established that people process own-race and other-race faces differently. When required to recognise or differentiate between specific facial identities, better performance is achieved with own-race faces compared to other-race faces, a finding termed the other-race effect (ORE). However, when asked to categorize faces, performance is better for other-race faces than for own-race faces, a finding termed the other-race categorization advantage (ORCA). This work explored how these race effects might vary as a function of face familiarity. A race categorization task was chosen in which stimuli were covered with two-dimensional Gaussian noise and participants were required to classify faces as Asian or Caucasian. Two experiments were conducted. In Experiment 1, the signal (face) alpha level on each trial was gradually increased to modulate the visibility of the target face such that the faces emerged from noise. Participants ended the trial by making a response as soon as possible. Faces were repeated across blocks to include face familiarity. The ORCA was present in the task in both accuracy data and levels of alpha, with participants responding more accurately for other-race faces as well as requiring less visible face to categorize other-race faces. Despite the repetition of faces, no learning took place in this task. In Experiment 2, the alpha levels of the face within the dynamic noise were predetermined, providing 4 possible levels (0.5, 0.6, 0.7 and 0.8) of visibility. Participants were required to respond as quickly as possible, with reaction time (RT) being the main dependent measure. Faces were again repeated across blocks. The ORCA was present in both accuracy and RT data. A learning effect for own-race faces was obtained such that RT to categorize own-race faces decreased while the RT to categorize other-race faces increased. A follow- up control experiment presented a new set of faces in each block and concluded that the learning effect obtained in Experiment 2 was due to the individuation of the faces presented during the task, not more general task or race-template learning. Taken together these results suggest that the ORCA is a robust phenomenon that can be consistently measured in novel, online tasks. The results of Experiment 2 further suggest that with repetition of faces, faster individuation of own-race faces takes place, reducing the influence of the ORCA and leading to patterns of performance more consistent with the ORE.