Perceptual Similarity and Labels Impact Facial Expression Categorization in Infancy

• In Event: 2-005 - How Cognitive Science Can Inform a New Perspective for Understanding Emotional Development

Integrative Statement

Research has found that preschoolers categorize emotions in terms of two broad dimensions: valence (positive v. negative) and arousal (high v. low; Russell, 1994). These broad categories narrow into discrete/basic emotion categories throughout development as children learn emotion labels (Widen & Russell, 2008). However, little is known about how emotional expressions are categorized in infancy. The current studies explored (a) whether infants perceive different facial expressions (anger, sad, disgust) as belonging to a superordinate category of negative valence, and (b) whether verbal labels facilitate the formation of this category.
STUDY 1 habituated 14- and 18-month-olds (n=48 per age) to facial expressions from the Radboud Face Database (Langner et al., 2010). After habituation, infants were shown four test trials (see Figure 1). If infants formed a superordinate category of negative valence, their looking time to the negative familiar event should not differ from their looking time to the negative novel face and the negative novel emotion. Their looking time to each of these three negative emotions should also be significantly lower than their looking time to the positive novel emotion. A Test Trial effect (p<.001) indicated that infants did not form a superordinate category.
STUDY 2 (n=48 per age) tested whether infants would form a superordinate category of negative valence if the novel LABEL “toma” was added to each habituation event. A Test Trial x Condition effect (p<.001) showed that infants formed a superordinate category in the DISGUST-SAD condition only. However, when data from STUDY 1 and STUDY 2 were combined, there were no significant study interactions (ps>.05).
STUDY 3 and 4 replicated and extended the labeling effect found in the first two studies. These studies used stimuli from the NimStim Database (Tottenham et al., 2009) and controlled for the amount of teeth displayed on the faces (Caron et al., 1985). In STUDY 1 and 2, the anger and happy expressions displayed teeth, while the disgust and sadness expressions did not. In STUDY 3 and 4, none of the expressions displayed teeth (Figure 1). Given that (a) there were no age effects in the first two studies, and (b) only the DISGUST-SAD condition showed evidence of categorization, only 14-month-olds and the DISGUST-SAD condition were tested further.
STUDY 3 (n=24) followed the same procedure as STUDY 1 (with no labels). There was no significant Test Trial effect (p>.05), indicating that infants did not form a superordinate category. STUDY 4 (n=17 currently) added labels to the task in the same manner as STUDY 2. A significant Test Trial effect (p=.006) indicated that infants formed a superordinate category. When data from STUDY 3 and STUDY 4 were combined, there was a significant Trials x Study interaction (p<.05). Looking times and statistics for all studies are presented in Table 1.
These results indicate that (a) preverbal infants do not spontaneously form superordinate categories of negative valence, but (b) verbal labels facilitate the formation of these categories. Results will also be discussed in terms of the perceptual similarity of the facial expressions.

Authors

• Ashley Ruba, University of Washington
  Presenting Author

• Andrew Meltzoff, University of Washington
  Non-Presenting Author

• Betty Repacholi, University of Washington
  Non-Presenting Author