Virtual Human Interaction Lab – Stanford University


{ Facial Identity Capture and Presidential Candidate Preference }

Paper was presented at the 55th Annual Conference of the International Communication Association.

Jeremy N. Bailenson, Shanto Iyengar, & Nick Yee
Department of Communication, Stanford University

Today’s media features the pervasive use of digital representations of people in video feeds, static photographs, and scans used for mass printing of direct mail. In the current work we demonstrate the effectiveness of algorithmic transformations which capitalize on human beings’ biologically driven disposition to prefer faces similar to them [1]. By altering pictures of Kerry and Bush to capture features of potential voters, we manipulated the outcome of the 2004 Presidential election.

Facial images in media are typically altered using touch ups and airbrushing, changes in resolution, contrast, and scale, as well as more creative transformations of various facial features. We utilized pixel blending software to examine facial identity capture, making one person’s face appear more similar to someone else’s.

Similarity between two people instills altruism [2] and trust [3]. Biological explanations for this effect argue that phenotype matching (implicit recognition of subtle physical cues) is a mechanism organisms use to identify genetically-related kin. Indeed, different areas of the brain process facial images morphed with the self than images morphed with familiar others [4]. Social explanations argue that people use physical similarity as a proxy for compatible interests and values [5].k

It is inevitable that political candidates, advertisers, educators, and others who seek social influence will resort to methods of dynamically transforming appearance [6]. Currently, political candidates tailor the information content of their targeted mailings and televised messages to particular demographic groupings [7]. Increasingly, they are also in a position to vary salient attributes of their physical appearance, e.g. their weight, dress style, facial expression, or skin tone, depending on the audience in question.

To test the effect of facial identity capture on vote choice, we passively acquired digital photographs of a national random sample of voting aged citizens. One week before the 2004 presidential election, participants completed a survey of their attitudes concerning George Bush and John Kerry while viewing photographs of both candidates side by side (See Figure 1). For a random one-third of the subjects, their own faces were morphed with Kerry while unfamiliar faces were morphed with Bush. For a different one-third, their own faces were morphed with Bush while unfamiliar faces were morphed with Kerry. The remaining one-third of the sample viewed un-morphed pictures of the candidates.

Bush/Kerry Morphs

Figure 1: Two subjects, (Panels A and B), the morph of Subject 1 and Bush (Panel C), the morph of Subject 2 and Kerry (Panel D), and the vote intention score by condition (Panel E). The difference in vote intention for Bush and Kerry by condition was statistically significant (p < .05).

Post-experiment interviews demonstrated that not a single person detected that his or her image had been morphed into the photograph of the candidate. Participants were more likely to vote for the candidate morphed with their own face than the candidate morphed with an unfamiliar face. The effects of facial identity capture on candidate support were concentrated among weak partisans and independents; for ‘card carrying’ members of the Democratic and Republican parties, the manipulation made little difference.

The use of facial identity capture was sufficient to change the outcome of the presidential election by a double-digit margin, according to a national random sample. In the case of presidential elections, it is well documented that the candidates’ party affiliation, their positions on major issues, their personal traits, as well as the state of the economy affect vote choice. Our results demonstrate that implicit facial similarity should be added to this list.

References

1. Hauber, M. E. & Sherman, P. W. 2001 Self-referent phenotype matching: theoretical considerations and empirical evidence. Trends in Neuroscience. 24, 609–616.

2. Gaertner, S.L., and J.F. Dovidio. 1977. The subtlety of white racism, arousal and helping behavior, Journal of Personality and Social Psychology 35: 691-707.

3. DeBruine, L. M. (2002). Facial resemblance enhances trust. Proceedings of the Royal Society of London B, 269, 1307-1312.

4. Turk, D.J., Heatherton, T.F., Kelley, W.M., Funnell, M.G., Gazzaniga, M.S., & Macrae, C.N. (2002). Mike or me? Self recognition in a split-brain patient. Nature Neuroscience, 5, 841-842.

5. Zajonc, R.B., Adelmann, P.K., Murphy, S.T., and Niendenthal, P.M. (1987). Convergence in the physical appearance of spouses. Motivation and Emotion, 11, 335–346.

6. Bailenson, J.N., Beall, A.C., Loomis, J., Blascovich, J., & Turk, M. (2004). Transformed Social Interaction: Decoupling Representation from Behavior and Form in Collaborative Virtual Environments. PRESENCE: Teleoperators and Virtual Environments, 13(4), 428-441.

7. Iyengar, S., D. L. Lowenstein, and S. Masket. 2001. The stealth campaign: Experimental studies of slate mail in California, The Journal of Law and Politics 17: 295-332.