Probability as Input for Children’s Emotion Inferences

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

Abstract

How do children infer people’s emotions? We propose that with age, children increasingly view emotions as depending on how outcomes relate to other things. When using theory of mind to infer people’s emotions (Harris, 2008; Wellman & Banerjee, 1991), children consider how outcomes relate to people’s mental states. Beyond this, children also come to infer emotions by relating outcomes to information about prior probability. We discuss three experiments that support this account.

We first examined how probability information impacts children’s surprise inferences. In Experiment 1 (n=120), 4-7-year-olds saw scenarios where two characters each received a red gumball from different gumball machines; Figure 1. One machine contained mostly red gumballs (36 red, 4 black), and the other contained mostly black gumballs (opposite distribution). Children judged which character was more surprised to receive a red gumball. We found a significant effect of age, p=.005; 7-year-olds mostly chose the character with a low chance of getting a red gumball, 6-year-olds trended in this direction, 5-year-olds responded at chance, and 4-year-olds mostly chose the character with a high chance of getting a red gumball.

Next, we wanted to see if we could improve 5- and 6-year-olds’ performance. In Experiment 2 (n=120), before 5-6-year-olds made their surprise judgments, they were prompted to consider the characters’ chances of getting a red gumball, the characters’ beliefs about getting a red gumball, or a control prompt. We found that 6-year-olds’ performance differed by condition, p=.019, but 5-year-olds’ did not, p=.656. Six-year-olds’ were more likely to choose the correct character when prompted to consider probability than when prompted to consider beliefs or when asked the control prompt, ps<.003, but no differences were found between the belief and control prompts, p>.5. These findings suggest that by age 6, children infer people’s surprise by relating their outcomes to the prior probability of receiving those outcomes.

We next examined how probability information impacts children’s happiness inferences. In Experiment 3 (n=180), 4-6-year-olds saw a scenario where a girl received 2 “yummy” and 2 “yucky” gumballs from either a machine with mostly “yummy” gumballs or a machine with mostly “yucky” gumballs; Figure 2. Children rated how the girl felt about getting these gumballs using a 7-point scale ranging from extremely sad to extremely happy. We found only a significant age by condition interaction, p=.014. Five- and six-year-olds rated the girl as being significantly happier when the machine contained mostly yucky gumballs than when the machine contained mostly yummy gumballs, ps≤.043. Four-year-olds showed no differences between the two conditions, p=.243. The findings from this experiment suggests that children infer people’s happiness by comparing their outcomes to the prior probability of receiving a better outcome.

Authors

• Tiffany Doan, University of Toronto
  Presenting Author

• Ori Friedman, University of Waterloo

• Stephanie Denison, University of Waterloo