Online testing of classic responses to violation-of-expectation events in infancy: Measuring looking & learning

• In Event: Poster Session 11
  In Poster Session: PS 11 Topic - Cognitive Processes

Abstract

Much research on infant cognition hinges on the fundamental finding that in laboratory settings, infants look longer at impossible than possible events (e.g., Baillargeon et al., 1985, Perez & Feigenson, 2020). Additional work finds that infants don’t just look longer at surprising events—they also learn better about objects that violated their expectations (Stahl & Feigenson, 2015). But do these reactions to expectancy violations extend outside of the controlled laboratory setting? In particular, can they be observed within the context of infants’ own homes? Given the ongoing COVID-19 pandemic, investigating the feasibility of online violation-of-expectation (VOE) testing is imperative, especially using methods that can be compared to in-laboratory findings. Here, we conducted two live experiments administered by an experimenter over Zoom, one testing 15-month-olds (Exp. 1, N = 24) and the other 11-month-olds (Exp. 2, N = 20), to ask whether infants exhibit reliable VOE responses, and/ or surprise-induced learning.
In Experiment 1, 13- to 18-month-old infants (M=15.56 months, SD=1.95) saw a video in which an experimenter pushed a novel object along a supporting surface. For half of the infants the object was pushed to the edge, where it remained fully supported and did not fall (Expected). For the other infants, the object was pushed completely over the edge, where it appeared to hover in mid-air (Surprising). Infants’ looking at these outcomes was recorded for 15 seconds. Next, infants were taught a novel label for the object. The experimenter pointed at it, saying, “That’s a gaffa! Yeah, a gaffa!” Finally, in the test trial, infants saw the target object on one side of the screen and a novel distractor object on the other. After a silent baseline period, the experimenter said, “Where’s the gaffa? Look at the gaffa!” and infants’ looking was measured for 10 seconds. Longer looking at the target (above baseline preference) indicates successful label learning. We found that infants looked longer at the Surprising than the Expected object outcome (Mdifference = 2.40s, SEM = 1.05s; t (22) = 2.27, p = .03; Fig.1). In addition, infants learned better following the Surprising than the Expected outcome (Mdifference =.38, SEM = .09; t (22) = 3.83, p < .01; Fig.2).
Next, in Experiment 2, we sought to replicate these findings with younger infants; data collection is in progress. The procedure is identical to that of Experiment 1: we first measure infants’ looking at either an Expected or Surprising outcome of an object event. Next, we tested surprise-induced learning by teaching infants a hidden auditory property of the target object, rather than a novel label (to more closely match previous methods; Stahl & Feigenson, 2015). Preliminary results suggest that these younger, 11-month olds also detect object violations presented online, looking longer at Surprising than Expected outcomes (Mdifference = 2.90s, SEM = 1.52s; Fig. 1). Data analysis to assess surprise-induced learning is ongoing.
Overall, our findings demonstrate that classic responses to violation-of-expectation events can be measured remotely, and hence that online testing is a promising format for future infant research.

Authors

• Jasmin Perez, Johns Hopkins University
  Presenting Author

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

• Michelle Hong Zhang, Johns Hopkins University
  Non-Presenting Author

• Lisa Feigenson, Johns Hopkins University
  Non-Presenting Author