Does accent trump skin color in guiding children's social preferences? Evidence from Brazil's natural lab

Emma Cohen*, Edwin J.C. van Leeuwen, Andrezza Barbosa, Daniel B.M. Haun

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review


Previous research has shown significant effects of race and accent on children's developing social preferences. Accounts of the primacy of accent biases in the evolution and ontogeny of discriminant cooperation have been proposed, but lack systematic cross-cultural investigation. We report three controlled studies conducted with 5−10 year old children across four towns in the Brazilian Amazon, selected for their variation in racial and accent homogeneity/heterogeneity. Study 1 investigated participants’ (N = 289) decisions about friendship and sharing across color-contrasted pairs of target individuals: Black-White, Black-Pardo (Brown), Pardo-White. Study 2 (N = 283) investigated effects of both color and accent (Local vs Non-Local) on friendship and sharing decisions. Overall, there was a significant bias toward the lighter colored individual. A significant preference for local accent mitigates but does not override the color bias, except in the site characterized by both racial and accent heterogeneity. Results also vary by participant age and color. Study 3 (N = 235) reports results of an accent discrimination task that shows an overall increase in accuracy with age. The research suggests that cooperative preferences based on accent and race develop differently in response to locally relevant parameters of racial and linguistic variation.

Original languageEnglish
Article number101111
Pages (from-to)1-24
JournalCognitive Development
Publication statusPublished - Oct 2021
Externally publishedYes


  • Accent
  • Brazil
  • Cross-cultural
  • Race
  • Social behavior


Dive into the research topics of 'Does accent trump skin color in guiding children's social preferences? Evidence from Brazil's natural lab'. Together they form a unique fingerprint.

Cite this