Infants use Others' (Incomplete) Representations Instead of Their own to Track Multiple Occluded Objects

• In Event: 3-224 - The Nature of Infants' and Preschoolers' Sensitivity to Other People's Object-representations

Integrative Statement

Infants can maintain representations of multiple objects that are no longer in view (Kibbe, 2015). Infants also are sensitive to others’ visual access to objects: they can keep track of an agent’s belief about a hidden object (Kovacs et al, 2010) using the same brain areas that support their own representations (Kampis et al., 2015). What happens when an agent’s knowledge about objects is different from infants’? Do infants represent others’ representations instead of or in addition to their own?
Here, we investigated these questions by taking advantage of a well-known limit: if the number of objects infants must track exceeds three, infants’ memory fails catastrophically, and they are unable to maintain representations of even a subset of the hidden objects (e.g. Feigenson & Carey, 2003, 2005). We modified the foraging task of Feigenson & Carey (2005), in which infants are tasked with tracking different numbers of crackers hidden across two locations, and are invited to crawl to one of the locations. Infants’ choice to crawl to the location with the greater number of crackers is taken to indicate “successful” object tracking. In our modified version, 12-month-old infants (n=66, M=12.63, SD=0.9) observed that a second experimenter (the “agent”) watched as some of the crackers were hidden, but became distracted and looked away while the others were hidden, resulting in an incomplete representation (Figure 1). We asked whether infants’ location choices would be driven by their own representations, or by the agent’s (incomplete) representation.
In Experiment 1, infants saw 1 and 4 crackers hidden (exceeding infants’ object tracking limits), while the agent saw 1 and 2 (within infants’ limits). If infants rely on their own representations, their memory should fail catastrophically, and they should choose the locations at chance. However, if their location choice is driven by the agent’s representation, they should succeed. Indeed, we found that 13/16 infants (81%) succeeded (chance=50%; binomial p=0.02). In Experiment 2, we confirmed that infants’ success was due to the agent’s visual access to the objects: when the agent’s eyes were closed during cracker distribution, infants crawled at chance (8/16 (50%) succeeded, p=1.0). In Experiment 3, we found that infants’ success in Experiment 1 was not due to tracking all four objects (e.g. via chunking); when infants saw 2 and 4 hidden while the agent saw 2 and 2, infants crawled at chance (9/16 (56%) succeeded; p=1.0), suggesting their choice was driven by the agent’s representation. In ongoing Experiment 4, we asked whether the agent’s representation would drive infants’ choices even when the number of hidden objects was within infants’ object-tracking limits. When infants observed 1 and 2 hidden, while the agent observed 1 and 1, infants crawled at chance (3/9 (33%) succeeded). When the agent had her eyes closed throughout, infants showed the typical success pattern (7/9 infants (77%) succeeded). Figure 2 summarizes the results of experiments 1-4.
Together, these results suggest infants may default to other agents’ representations to guide their behavior, even when those representations conflict with their own and with reality.

Authors

• Melissa M. Kibbe, Boston University
  Presenting Author

• Jessica R. Beal, Boston University
  Non-Presenting Author