Abstract
Great strides have been made in recent years in understanding how we perceive faces in terms of underlying "face spaces". Our work is focused on the relatively unexplored notion of default regions of these spaces, toward which our face representations may be attracted. Picture a person standing before you, asking for directions. Now consider: what did that person look like? And where did that information come from? Here we used a novel instantiation of the method of serial reproduction to explore the effective 'default settings' in face space. As a first case study, we explored the perception of race. A single face was briefly presented to each observer, with its race selected from a smooth continuum between White and Black (with all faces matched for mean luminance). The observer then simply reproduced that face, by using a slider to morph a test face along this continuum. Their response was then used as the face initially presented to the next observer, and so on down the line in each reproduction chain. This method has been shown (albeit in very different contexts) to reveal the contents of people's default assumptions about the world. In our experiments, White observers' reproduction chains consistently and steadily converged onto faces that were significantly Whiter than both the original face and the continuum's midpoint. Indeed, even chains beginning near the Black end of the continuum inevitably ended up well inside the White region. These results highlight a default region of face space for White observers, which biases downstream perception and memory. In further experiments, we report extensions both to other cultures (e.g. exploring White and Indian observers' face-space defaults in race continua from White to South Asian faces) and to other types of features (in particular, continua that vary faces' age, gender, and perceived animacy).
Meeting abstract presented at VSS 2015