The role of language in developing abstract, multidimensional emotion concepts

• In Event: The development of emotion organization from infancy to adolescence

Abstract

People regularly use emotion words like “angry” and “sad” to communicate about their own and others’ emotions. Although one might assume that these words have shared conceptual meanings, individual differences in adults’ understanding and use of emotion words shape how they perceive and experience emotions (e.g., Kashdan, Barrett, & McKnight, 2014). But how do emotion concepts develop, and what psychological mechanisms might explain these developmental trajectories? Here, we investigated how two aspects of emotion representation (i.e., abstraction and multidimensionality) develop in a cross-sectional study of N = 203 participants aged 4-25.

First, we investigated how the abstractness of children’s emotion representations varies across age. There is robust evidence that the ability to engage in abstract reasoning (i.e., identifying patterns that transcend concrete details; e.g., Dumontheil, 2014) develops across childhood and adolescence. To test whether emotion concepts undergo similar developments in abstractness, we developed an emotion vocabulary assessment in which participants provided verbal definitions of 24 emotion words, and we scored participants’ responses both for comprehension (i.e., the ability to provide a fitting commonsense definition) and abstractness (i.e., the extent to which the definition attended to concrete situational details vs. more abstract general principles).

Smoothing-spline regression analyses revealed that comprehension of emotion words increased rapidly across childhood and plateaued around age 11. However, the abstractness of participants’ responses showed protracted development, only reaching a plateau around age 18 (Figure 1). Analyses of strategy use revealed that children primarily defined emotions in concrete, situation-bound ways (e.g., defining anger as ‘when my sister takes my toys’), but this strategy decreased across age and was replaced with the more abstract strategy of providing definitions that generalize across specific situations (e.g., defining anger as arising ‘when one’s goals are blocked’).

Secondly, we explored the development of children’s emotion representations along the two primary axes of valence (positive—negative) and arousal (activating—deactivating; Russell, 1980). Adapting prior work on emotion concept representation, we asked participants to rate how similarly they represent the meanings of 10 emotion words. Multidimensional scaling analyses were used to construct models of the dimensions underlying each participant’s emotion representations, allowing us to capture how the spread of emotions along the valence and arousal axes changed with age.

Focus on the valence dimension of emotions decreased with age and focus on the arousal dimension increased with age, suggesting that children’s emotion representations initially emphasize valence but become more multidimensional with age (Figure 2). We tested three potential mechanisms of this developmental increase. Mediation analyses demonstrated that increased verbal knowledge (i.e., WASI vocabulary scores) explained increased multidimensionality in emotion representation across development, over and above WASI fluid reasoning scores and the general ability to simultaneously represent two perceptual dimensions.

These results reveal that childhood emotion representations focus on the concrete situational causes of emotions, and emotion concepts cluster in terms of valence. However, as vocabulary expands across development, adolescents and young adults learn to represent emotions as abstract states that generalize across specific situations and that they can be organized along dimensions beyond valence.

Authors

• Erik C. Nook, Princeton University
  Presenting Author

• Caitlin M. Stavish, University of Washington
  Non-Presenting Author

• Stephanie F. Sasse
  Non-Presenting Author

• Hilary Lambert, University Of Washington
  Non-Presenting Author

• Patrick Mair, Harvard University
  Non-Presenting Author

• Katie McLaughlin, Harvard University
  Non-Presenting Author

• Leah H. Somerville, Harvard University
  Non-Presenting Author