Abstract
Faces seem to enjoy special advantages for human observers to process with remarkable speed and accuracy. For example, a face target may pop-out in a visual search task (cf. Hershler & Hochstein, 2005; VanRullen, 2006). However, it is debated whether special low-level image properties of the faces have led to this effect. To address this question, we tested human observers ability to process Mooney faces in a visual search task among non-face Mooney image distractors. Mooney images are two-toned images where a median luminance is found and all parts of the image with that luminance and higher were made white and all parts of the image with the lower luminance were made black. Many low-level visual features, such as luminance and contrast, were fully controlled for by using Mooney images. We present evidence that upright Mooney faces pop out in a visual search task, whereas inverted Mooney faces do not. Observers were able to detect target Mooney faces among non-face distractor Mooney images when the target face was upright faster than when the target face was inverted. Observers also acknowledged the absence of the target on catch trials more quickly when the target face was upright than inverted. Since one cannot rely on any local features to differentiate upright and inverted Mooney faces, our results indicate that instead of any low-level image properties, holistical representation of faces crucially underlies rapid and efficient visual processing of face images.
Meeting abstract presented at VSS 2013