1

Predicting political elections from rapid face judgments

Alexander Todorov

Princeton University

FABBS, Science café, Washington DC

October 24, 2007

Basic findings

• People are extremely efficient at making trait judgments (e.g., competent, trustworthy) from faces

• Rapid, unreflective judgments of competence based solely on facial appearance predict election outcomes

2

Choleric Melancholic Sanguine Phlegmatic

Between 1772 and 1940 more than 150 editions

“the nearer the eyebrows are to the eyes, the more earnest, deep, and firm the character”

Darwin was almost denied the chance to take the historic Beagle voyage on account of his nose. Apparently, the Captain [a fan of Lavater] did not believe that a person with such a nose would “possess sufficient energy and determination.”

3

“whether they are or are not sensible of it, all men [and women] are daily influenced by physiognomy”

Evaluating faces: Judging the book by its cover

• 100 ms exposure is sufficient for a variety of person judgments– Competence– Trustworthiness– Aggressiveness – Likeability

• Additional time exposure increases confidence in judgments

• Single glance impressions

Willis & Todorov (2006). Psychological Science.

4

Speed of inferences from faces

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0 50 100 150 200 250 300 350 400 450 500Time exposure (ms)

Engell, Haxby, & Todorov (2007). Journal of Cognitive Neuroscience.

5

Trait judgments from faces

• Rapid

• Spontaneous

• Non-deliberative

• Requiring minimal attention

• Linked to neural systems underlying processing of emotionally significant stimuli

Should this matter for voting decisions?

• Not on many counts– From a rational perspective, candidate information should

overwrite fleeting initial impressions– From an ideological perspective, party affiliation should

sway such impressions– From a voter’s perspective, decisions are justified in terms

of the candidate’s positions not their appearance

• Yet, our mental life is often guided by rapid, snap decisions that may not appear on the consciousness radar

6

Basic paradigm

• Participants are presented with the faces of the winner and the runner-up– Democratic and Republican candidates for prospective

predictions (2004 & 2006)

• Political races with highly familiar politicians are excluded (e.g., Hillary Clinton)

• If a participant recognizes any of the faces, their judgments for this race are not included in the analysis

• All predictions are based on judgments of facial appearance and no other knowledge

A. Todorov et al., Science 308, 1623 -1626 (2005)

Predicting Senate elections

7

Senate elections (2000, 2002, 2004, & 2006)

Senate elections, 71.8% correctly predicted

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Perceived competence of candidate

Replications and extensions

• Judgments based on 10 s silent video clips of gubernatorial debates– D. Benjamin & J. Shapiro, Economics, Harvard

• Judgments of morphed faces of heads of state– A. Little et al., Psychology, University of Liverpool

• Judgments of competence from faces of Mexican politicians– C. Lawson & G. Lenz, Political Science, MIT

8

Automaticity of competence judgments

Ballew & Todorov (2007). PNAS.

A B

Which person is more competent?

Predicting gubernatorial elections

60.0%Deliberation

67.3%Unlimited time

65.5%Response deadline (2 s)

67.3%250 ms exposure

61.8%100 ms exposure

Correctly predicted racesExperimental condition

Ballew & Todorov (2007). PNAS.

9

Why competence?

• People think that this is the most important attribute for a politician

• Not an entirely irrational process

• Just looking in the wrong place for the right information

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

Com

pete

nt

Depen

dabl

e

Hones

t

Org

anized

Anxio

us

Calm

Like

able

Criti

cal

Sym

path

etic

Conve

ntio

nal

Extrov

erte

d

Compl

ex

Reser

ved

Reg

ress

ion

weig

hts

Effects are specific for competence

10

Is it really competence?

• Effect of perceived competence is not due to– Differences in ethnicity and gender of candidates– Age differences– Attractiveness – Face familiarity

• Not an effect of other judgments either– Trustworthiness does not predict outcomes– Likeability does not predict outcomes– Etc.

What about Presidential elections?

• Work by Little, Burriss, Jones, & Roberts (University of Liverpool)

• Using morphing technique to remove recognition effects

• Elections in UK, USA, Australia, and New Zealand

11

Little et al. (2007). Evolution and human behavior.

Little et al. (2007). Evolution and human behavior.

12

Simulated voting and competence judgments

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Inferred competence from faces

Todorov, Mandisodza, Goren, & Hall (2005). Science.

Context dependence of decisions

• Importance of trait attributes can shift as a function of context– In war time, we value different attributes than

in peace time

• Events defining times of voting can change the perceived importance of trait attributes

13

In a war situation, who would you vote for?

In a peace situation, who would you vote for?

14

Masculine

Dominant

Strong leader

Attractive

Likable

Forgiving

Intelligent

Little et al. (2007). Evolution and Human Behavior.

Difference between Bush and

Kerry applied to novel face

15

0

10

20

30

40

50

60

70

80

Vote War Peace

% V

ote

Bush

Kerry

Little et al. (2007). Evolution and human behavior.

How does it work in the real world

• Some caveats – Appearance is not all you need

– Clearly one needs the backing of one of the major parties

– Pre-selection of candidates

– Most likely, impressions from appearance do notaffect partisans and informed voters

16

Relevant evidence

• At the level of voting decision, perhaps implicitly affecting undecided voters– Evidence from Lawson & Lenz (MIT, Political Science)

• At the level of party (elite) decisions, perhaps affecting the choice of candidates and fund raising ability– Evidence from Atkinson, Enos, & Hill (UCLA, Political

Science)

Conclusions • Inferences of competence from faces predict election

outcomes

• The inferences are fairly automatic

• The effect is highly specific for competence

• But perhaps “fitting the face to the context” is as important as having a competent appearance

17

AcknowledgmentsAnesu Mandisodza, Amir Goren, Crystal Hall,Janine Willis, Chas Ballew, Valerie Loehr, Manish Pakrashi, Andy Engell, Chris Olivola, Nick Oosterhof, Chris Said, Sean Baron, Sara Verosky

Thank you!