To Learn or not to Learn? Surprise-Induced Learning Following Violations of Social Expectations

• In Event: Learning From Violation of Expectations in the Domain of Agents

Abstract

From early in life, infants hold basic expectations about the physical and social world around them. These early expectations help guide learning— Research finds that infants learn better about objects that defy physical principles, compared to objects that behave in accord with these principles (Stahl & Feigenson, 2015). But what about violations in the social domain? While much work shows that infants have expectations about the actions and mental states of others, it remains unknown whether seeing events that violate psychological expectations has any consequences for infants’ thinking. Here we asked whether infants experience a boost in learning after observing violations of psychological expectations.

In Experiment 1 we asked whether 16-month-old infants (N = 28) would show surprise-induced learning following violation of expectations about the emotional responses of others. Infants saw a person presented with a delicious-looking but novel dessert and a novel insect. The person looked at both, then turned towards one of the items and produced an emotional response, either saying “Yum!” to a dessert or “Yuck!” to an insect (Expected Emotion), or saying “Yuck!” to a dessert or “Yum!” to an insect (Surprising Emotion). A different person then labeled the novel dessert/ insect (“That’s a gaffa!”). Finally, infants’ learning was tested by showing them previously-labeled object (e.g., dessert) and the unlabeled object (e.g., insect) while a hidden experimenter asked, “Where’s the gaffa?” Infants exhibited longer looking and better learning following the Surprising Emotion (“Yum!” rather than “Yuck!” to an insect, or “Yuck!” rather than “Yum!” to a dessert), compared to the Expected Emotion.

Next, in Experiment 2, we asked whether 18-month-old infants (N = 25) would also show enhanced learning following a violation of social expectations concerning an agent’s preference. We presented infants with videos based on seminal experiments by Woodward (1998): infants first were familiarized to a person who repeatedly reached for one novel object over another when given the option to choose between the two. Then on the critical test trial, infants saw the person either again reach for the previously preferred object (Expected Reach) or switched to the previously non-chosen object (Surprising Reach). After the person had grasped the test object, a different person uttered a novel word for it (e.g., “That’s a gaffa!”). As in Experiment 1, infants’ learning was tested by showing them the two objects shown before while a hidden experimenter said, “Where’s the gaffa?” Infants showed significantly better learning following the Surprising Reach, compared to the Expected Reach.

These findings replicate many previous studies that show that infants have expectations about the social world, including about people’s emotional responses and preferences. In addition, our findings suggest that these expectations can form a foundation for gathering more information about the world around them—that is, infants appear to use violations of social expectations as special opportunities to learn.

Authors

• Qiong Cao, Johns Hopkins University
  Presenting Author

• Alexis Sierra Smith-Flores, University of California - San Diego

• Jasmin Perez, Johns Hopkins University

• Lisa Feigenson, Johns Hopkins University