Why robots, computers, and other agents are anthropomorphized: A meta-analysis

Date

2021-09-10

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

People increasingly come in contact with artificial agents such as computers, robots, and autonomous vehicles and frequently anthropomorphize these agents by imbuing them with human characteristics. In some cases, anthropomorphism benefits human-agent-interaction, and other times it creates problems. Accordingly, it would serve the designers of human-agent interactions to know which factors influence anthropomorphism. Key perspectives on anthropomorphism hypothesize that it is affected by nine situational factors: social disconnection, anticipated agent interaction, perceived agent predictability, perceived negative outcomes, power of perceiver, similarity between perceiver and agent, stereotypes, perceived threat, and reminder of harm performed by ingroup.

Meta-analyses of 119 studies investigated the effect that each situational factor had on the anthropomorphism of artificial agents. Perceived agent predictability, similarity, and stereotypes significantly affected anthropomorphism and in their anticipated directions. The other five situational factors either produced mixed results or have been manipulated in too few studies for us to meta-analyze. Positive agent personality was identified as a potential additional situational factor that affects anthropomorphism. Significant moderators of the relationship between situational factors and anthropomorphism included participant culture, interaction level, and measurement type.

Findings have both theoretical and practical implications. While all existing perspectives on anthropomorphism were partially supported, none accounted for all of the situational factors found to influence artificial agent anthropomorphism. Any comprehensive theories on anthropomorphism should account for the influence of perceived agent predictability, similarity, stereotypes, and positive agent personality on artificial agent anthropomorphism. Practically speaking, findings provide guidance about which factors reliably influence anthropomorphism and could be leveraged when designing artificial agents.


Embargo status: Restricted until 01/2027. To request the author grant access, click on the PDF link to the left.

Description

Keywords

anthropomorphism, mind perception, dehumanization, human-agent interaction, human-robot interaction, human-computer interaction, meta-analysis

Citation