Mano Tsakiris of the Department of Psychology at Royal Holloway, University of London recruited a total of 18 female participants. Before the experiment, a digital photograph was taken of each, and Photoshop was used to replace the hair and ears with a black background template. Then, a computerised algorithm called Fantamorph was used to merge each photo with  the unfamiliar face of a woman of the same age. This resulted in 100 images, ranging from 100% self (the participant's face on a black background) to 0% self (the unfamiliar face on a black background), with a gradual merging of the two in between, in 1% steps.    

The participants were then placed in front of a laptop computer, and asked to watch a short film clip consisting of 100 frames, each of which represented an incremental 1% change from one face to the other. In one condition, the film showed the participant's face morphing into the unfamiliar face, and each participant was asked to press a key when they perceived the face as looking more like the other one than their own. In another condition, the film showed the unfamiliar face morphing into theirs, and they were required to indicate when they perceived it to look more like their own face than the other.

They were then asked to watch another film clip, of 2 minutes in duration, showing a morphed face being touched on the cheek every 2 seconds by a paintbrush. The face on the screen contained 50% of the participant's face merged with 50% the unfamiliar face. As they watched the film, identical brushstrokes were apllied to their own face, either at the same time as the face in the film clip was brushed, or with a 1 second discrepancy. Afterwards, each participant was shown one of the original morphing film clips, and again asked to indicate with a key press when the face on the screen looked more like their own face than the unfamiliar one.

During the initial test trials with the film clips of morphing faces, they stopped the film earlier when asked to judge when the face looked more like the other, and later when asked to judge when the face looked more like their own: the "self-to-other" film was stopped on average after 44 frames on average, and the "other-to-self" film on average after 62 frames, revealing a tendency to judge the observed face as looking more like their own than the other.

Tsakiris found that the tactile stimuli applied during the experimental condition had a significant effect on the participants' face self-recognition. Following the trials in which the brushstrokes were synchronous with those in the film, the participants stopped the "self-to-other" film clips much later, and the "other-to-self" clips much later, than during the test trials. The synchronous brush strokes had reduced the participants' bias towards their own faces. By contrast, there was no significant difference between the test trials and those which followed the condition in which asynchronous brushstrokes were applied.  

Thus, the synchronous visual and tactile stimuli significantly altered the participants' ability to recognize their own faces. This manipulation is exactly the same as the rubber hand illusion,  which involves simultaneously applying identical brushstrokes to the participant's hand and to the rubber hand. This induces in the participant the experience that the rubber hand is a prt of their own body. The new study demonstrates that the brain's representation of one's own face is highly susceptible to the same sorts of cues and, rather than being stable, is constantly updated by incoming visual and tactile stimuli, and possibly others too.

Related:

• The body swap illusion
• Rubber hand feels real for amputees
  
• Distorting the body image alters perception of pain 
• Phantom sensations exorcized by changes in body position
  

Tsakiris, M. (2008). Looking for Myself: Current Multisensory Input Alters Self-Face Recognition. PLoS One 3 (12) DOI: 10.1371/journal.pone.0004040.