Modeling first impressions from highly variable facial images

Richard J W Vernon, Clare A M Sutherland, Andrew W. Young, Tom Hartley

Research output: Contribution to journalArticle

68 Citations (Scopus)

Abstract

First impressions of social traits, such as trustworthiness or dominance, are reliably perceived in faces, and despite their questionable validity they can have considerable real-world consequences. We sought to uncover the information driving such judgments, using an attribute-based approach. Attributes (physical facial features) were objectively measured from feature positions and colors in a database of highly variable "ambient" face photographs, and then used as input for a neural network to model factor dimensions (approachability, youthful-attractiveness, and dominance) thought to underlie social attributions. A linear model based on this approach was able to account for 58% of the variance in raters' impressions of previously unseen faces, and factor-attribute correlations could be used to rank attributes by their importance to each factor. Reversing this process, neural networks were then used to predict facial attributes and corresponding image properties from specific combinations of factor scores. In this way, the factors driving social trait impressions could be visualized as a series of computer-generated cartoon face-like images, depicting how attributes change along each dimension. This study shows that despite enormous variation in ambient images of faces, a substantial proportion of the variance in first impressions can be accounted for through linear changes in objectively defined features.

Original languageEnglish
Pages (from-to)E3353–E3361
Number of pages9
JournalProceedings of the National Academy of Sciences of the United States of America
Volume111
Issue number32
DOIs
Publication statusPublished - 12 Aug 2014
Externally publishedYes

Fingerprint

Cartoons
Neural Networks (Computer)
Linear Models
Color
Databases
Sociological Factors

Cite this

@article{210629e30d9a41f194b0ce60f36776b4,
title = "Modeling first impressions from highly variable facial images",
abstract = "First impressions of social traits, such as trustworthiness or dominance, are reliably perceived in faces, and despite their questionable validity they can have considerable real-world consequences. We sought to uncover the information driving such judgments, using an attribute-based approach. Attributes (physical facial features) were objectively measured from feature positions and colors in a database of highly variable {"}ambient{"} face photographs, and then used as input for a neural network to model factor dimensions (approachability, youthful-attractiveness, and dominance) thought to underlie social attributions. A linear model based on this approach was able to account for 58{\%} of the variance in raters' impressions of previously unseen faces, and factor-attribute correlations could be used to rank attributes by their importance to each factor. Reversing this process, neural networks were then used to predict facial attributes and corresponding image properties from specific combinations of factor scores. In this way, the factors driving social trait impressions could be visualized as a series of computer-generated cartoon face-like images, depicting how attributes change along each dimension. This study shows that despite enormous variation in ambient images of faces, a substantial proportion of the variance in first impressions can be accounted for through linear changes in objectively defined features.",
keywords = "Face perception, Impression formation, Person perception, Social cognition",
author = "Vernon, {Richard J W} and Sutherland, {Clare A M} and Young, {Andrew W.} and Tom Hartley",
year = "2014",
month = "8",
day = "12",
doi = "10.1073/pnas.1409860111",
language = "English",
volume = "111",
pages = "E3353–E3361",
journal = "National Academy of Sciences, Proceedings",
issn = "0027-8424",
publisher = "NATL ACAD SCIENCES",
number = "32",

}

Modeling first impressions from highly variable facial images. / Vernon, Richard J W; Sutherland, Clare A M; Young, Andrew W.; Hartley, Tom.

In: Proceedings of the National Academy of Sciences of the United States of America, Vol. 111, No. 32, 12.08.2014, p. E3353–E3361.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Modeling first impressions from highly variable facial images

AU - Vernon, Richard J W

AU - Sutherland, Clare A M

AU - Young, Andrew W.

AU - Hartley, Tom

PY - 2014/8/12

Y1 - 2014/8/12

N2 - First impressions of social traits, such as trustworthiness or dominance, are reliably perceived in faces, and despite their questionable validity they can have considerable real-world consequences. We sought to uncover the information driving such judgments, using an attribute-based approach. Attributes (physical facial features) were objectively measured from feature positions and colors in a database of highly variable "ambient" face photographs, and then used as input for a neural network to model factor dimensions (approachability, youthful-attractiveness, and dominance) thought to underlie social attributions. A linear model based on this approach was able to account for 58% of the variance in raters' impressions of previously unseen faces, and factor-attribute correlations could be used to rank attributes by their importance to each factor. Reversing this process, neural networks were then used to predict facial attributes and corresponding image properties from specific combinations of factor scores. In this way, the factors driving social trait impressions could be visualized as a series of computer-generated cartoon face-like images, depicting how attributes change along each dimension. This study shows that despite enormous variation in ambient images of faces, a substantial proportion of the variance in first impressions can be accounted for through linear changes in objectively defined features.

AB - First impressions of social traits, such as trustworthiness or dominance, are reliably perceived in faces, and despite their questionable validity they can have considerable real-world consequences. We sought to uncover the information driving such judgments, using an attribute-based approach. Attributes (physical facial features) were objectively measured from feature positions and colors in a database of highly variable "ambient" face photographs, and then used as input for a neural network to model factor dimensions (approachability, youthful-attractiveness, and dominance) thought to underlie social attributions. A linear model based on this approach was able to account for 58% of the variance in raters' impressions of previously unseen faces, and factor-attribute correlations could be used to rank attributes by their importance to each factor. Reversing this process, neural networks were then used to predict facial attributes and corresponding image properties from specific combinations of factor scores. In this way, the factors driving social trait impressions could be visualized as a series of computer-generated cartoon face-like images, depicting how attributes change along each dimension. This study shows that despite enormous variation in ambient images of faces, a substantial proportion of the variance in first impressions can be accounted for through linear changes in objectively defined features.

KW - Face perception

KW - Impression formation

KW - Person perception

KW - Social cognition

UR - http://www.scopus.com/inward/record.url?scp=84905964293&partnerID=8YFLogxK

U2 - 10.1073/pnas.1409860111

DO - 10.1073/pnas.1409860111

M3 - Article

VL - 111

SP - E3353–E3361

JO - National Academy of Sciences, Proceedings

JF - National Academy of Sciences, Proceedings

SN - 0027-8424

IS - 32

ER -