What Do They See and Hear? 6-month-olds’ Natural Emotional Input from Faces and Language

• In Event: The Role of Input in Early Emotion Learning

Abstract

Understanding emotions is critical for children’s social interactions. It allows children to interpret the emotions of others and respond to them appropriately. Research with US and European children has revealed that some aspects of emotion understanding are early developing, including emotion perception. But currently there is virtually no data about the emotional input that infants see in their everyday lives. The absence of descriptive data on infants’ real-world exposure to emotion information constitutes a serious gap in the literature, and limits our ability to derive and test theoretically meaningful hypotheses about the early development of emotion perception. The present study addresses this gap by using an existing dataset to explore base rates of infants’ real-world experience with emotion information at 6 months.

Hour-long naturalistic video recordings from the Study of Environmental Effects of Developing Linguistic Skills (SEEDLings) corpus (Bergelson, 2016) were coded, using 6-month-old infants’ head cameras to characterize infants’ emotional input. The final sample included 36 6-month-old infants (Mage=6.13 months, SDage=0.11 months; 17 female). Videos were coded first for when a face was visible, then for which emotional facial configurations were within infants’ view using the Facial Action Coding System (FACS; identifying which specific facial muscles move). Audio was transcribed and the Linguistic Inquiry and Word Count emotion dictionary (LIWC; Pennebaker et al., 2015) was used to identify the frequency of emotion language within each transcript, which included both words for emotions (e.g., happy) and words with affective connotations (e.g., smile).

On average, there was at least a portion of a face within infants’ field of view 29.10% of the time, or a mean of 17.59 minutes per infant (SD=9.57, range=0.98-41.33) out of the 1-hour session. However, a full, close-up face was only visible to 6-month-old infants 3.44% of the time, or an average of 2.09 minutes (SD=1.86, range=0.11-7.22). During these instances where a full face was in view, infants saw an average of 36.06 unique facial configurations (SD=21.26, range=6-106). They also heard an average of 102.54 positive emotional words (SD=77.80, range=8-321), and 30.81 negative emotional words (SD=24.68, range=2-109) throughout the full video, although the same word could be repeated within each video. Unsurprisingly, infants who had at least a partial face in view for more time saw more unique emotional configurations (r=.743, p<.001). Interestingly, these infants also heard a greater number of both positive (r=.358, p=.032) and negative emotion words (r=.488, p=.003). Further, infants who heard more negative emotion words also saw a greater number of facial configurations (r=.330, p=.0496), but this relation was non-significant for positive emotion words (r=.294, p=.082).

Overall, these results help provide the groundwork for future work on emotional development by documenting the amount of natural emotional information infants are exposed to in their everyday lives, suggesting that emotional information both within and across infants may be highly variable. Further, rates of emotional input were correlated across audio and visual channels, which ultimately might support the kind of inferential learning that has been hypothesized to aid emotion understanding development.

Authors

• Marissa Ogren, Rutgers University - Newark
  Presenting Author

• Lauren Leotti, Rutgers University - Newark

• Katie Hoemann, KU Leuven

• Lisa M Oakes, University of California - Davis

• Lisa Feldman Barrett, Northeastern University

• Vanessa LoBue, Rutgers University