Face recognition is a complex process that many perceptual factors can affect the face recognition judgment. Distinctiveness has been identified one of the perceptual factors that affect recognition. Many researches have attempted to improve face recognition by manipulating the image stimuli, such as using caricatures that exaggerate distinctive features of a face. However, the caricature effect was not found consistent throughout those studies. Another technique of manipulating the stimuli is the prototypical process, which averages a set of images and eliminating any transient features that are not found consistent throughout the set of images. This averaging technique is hypothesized as one of the explanation to account for the inconsistent results of the caricature advantage. 40 British university students took part in the celebrity name-verification task. Four conditions were created in the task, as photo veridical, photo caricature, prototype veridical and prototype caricature. The experiment found the prototype effect was significant but failed to find a significant caricature advantage. The results showed that participants performed best in prototype caricature. A proposal is made in attempt to incorporate the prototype and the distractors on images into the multidimensional face space framework (MDS). It is suggested that all images are under the influence of distractors, such as lighting and make up, that are not consistently found in all images. These distractors are regarded as noises that interfere the process of recognition. By reducing the noises via the averaging process, the recognition of face, in terms of accuracy and reaction time, is improved on these prototypes. Therefore, this may account for why some researches failed to find the caricature advantage. The results of the experiment hopes to shed light on how to improve the image stimuli quality, and to investigate how the prototypes affect other perceptual factors such as familiarity and best likeness.