Poster #5 - Objects’ categorizability impacts infants’ encoding of object features

• In Event: Poster Session 17

Abstract

Infants can track the featural identities of simple objects (e.g. a triangle; Kibbe & Leslie, 2011), but also track objects’ categorical identities instead of their featural identities when objects are from contrasting superordinate categories (e.g. human/object; Kibbe & Leslie, 2019; Zosh & Feigenson, 2012). Objects’ categorizability may support infants’ ability to track object identities when objects are featurally complex. In two experiments, we asked how the categorizability of objects impacts infants’ ability to track object identities in a manual search task.

In both Experiments, infants completed 8 Test trials (Figure 1) in which the experimenter placed an object on top of an opaque box, hid the object inside a covered opening in the box’s front, and then allowed infants to retrieve an object from the box. In No Switch trials (4 trials), infants retrieved the object that was hidden. In Switch trials (4 trials), infants retrieved a different object (surreptitiously swapped with the original by the experimenter). A 10-second measurement period followed in which we measured infants’ searching in the box. If infants tracked the object’s identity, they should search longer on Switch compared to No Switch trials.

In Experiment 1 (n=18 16-18-month olds, M=17;20; 9 girls), we hid objects from categories familiar to infants: dolls and cars (Figure 1). Infants completed two blocks of trials, with block order counterbalanced across infants. In the Between-Category block, on Switch trials infants retrieved a different object from a different category than the one that was hidden (e.g. car hidden, doll retrieved). In the Within-Category Block, on Switch trials infants retrieved a featurally distinct object from the same category (e.g. yellow car hidden, blue car retrieved). We observed no main effect of Category Switch Type (Between-Category or Within-Category, p=.37), but we did observe a significant interaction between Trial Type (Switch or No Switch) and Block Order (p=.023); infants who completed the Within-Category block first searched longer on Switch trials in both blocks, while infants who completed the Between-Category block first searched longer in Switch trials only in the Between-Category block (Figure 2).

In Experiment 2 (n=18 16-18-month olds, M=17;15; 10 girls), we constructed stimuli that matched the overall surface features with the objects from Experiment 1, but were not from identifiable categories (Figure 1). The experiment proceeded as in Experiment 2. Infants’ search times did not differ significantly between Switch and No Switch trials (all effects p>.41). Comparisons between Experiments 1 and 2 showed infants’ search patterns differed significantly between the two experiments (p=.001).

Together, these results suggest that infants may track object identities based on the categorical level that is most task-relevant. In Experiment 1, infants who observed between-category switches first may have processed objects based on superordinate kinds (doll and car) largely ignoring object features, while infants who observed within-category switches first may have used the objects’ shared category membership to support tracking object features. Combined with Experiment 2, these results suggest that the categorizability of objects may support infants’ ability to track object features, particularly when objects are visually complex.

Authors

• Melissa M. Kibbe, Boston University
  Presenting Author

• Aimee E Stahl, The College of New Jersey