Refined reverse correlation : a technique for investigating the power of faces

People effortlessly and rapidly form a first impression of an individual’s personality based on their facial appearance. Forming an impression based on facial cues can have real world implications, for example, for the outcome of elections, courtroom decisions or work-place interviews. Research using traditional methods has, however, failed to identify the facial features that are related to specific personality traits in a reliable and valid way. This challenge can be overcome using a reverse correlation method. Here I present a refinement of the traditional reverse correlation image classification technique. Over the course of four projects I highlight the different possibilities that the refined technique offers. In the first project I will present how the technique was used to extract the facial prototype of someone that is likely to be ostracized. In the second project, I show how we extracted prototypes that evoke different emotions, applied them to real facial photographs and set the different prototypes in relation with each other. The third project offers insights into how the technique was used to investigate self-perception without any external standard of comparison except the participants’ own face. Finally, I present a fourth project where the technique was used to investigate whether the belief about how two personality traits co-occur on a conceptual level is reflected in the facial characteristics that are used to form an impression from faces. The here presented refined technique adds to the traditional reverse correlation technique in that internal representations can be visualized without visible artifacts, that the extracted prototypes can be applied to real photographs, and set in relation with each other. The discussion focuses on the reliability and validity of the method and presents future research possibilities.