Individual Submission Summary
Share...

Direct link:

Object Exploration Influences Three-dimensional Object Completion Preferences in Children

Sat, March 25, 8:30 to 9:15am, Salt Palace Convention Center, Floor: 2, Meeting Room 254 A

Abstract

From a young age, humans are remarkably capable of making three-dimensional representations of objects in their environments, despite the fact that objects can only be viewed from a single limited perspective at a given moment, and are often occluded in cluttered scenes. The inverse problem of inferring full 3D object form may be mitigated by relying on certain three-dimensional object completion preferences.

In our work, we observe that humans explicitly prefer to complete 3D objects from their limited viewpoints as symmetrical rather than asymmetrical, and as volumetrically full rather than hollow. Furthermore, these preferences strengthen from early childhood to adulthood, leading us to hypothesize that our object completion preferences might be shaped by the characteristics of the objects we interact with in our environment. To test this idea, we randomly assigned children to freely explore novel 3D-printed objects with certain kinds of properties (e.g., objects that are all volumetrically hollow or objects that are all volumetrically full) for a minute. Then, we examined their preferences for representing a different set of objects as full volumetric versus hollow.
Preliminary results from 68 4- to 7-year-old children (Mage = 5.79 years, SD = 1.18 years, 38 females) reveal a growing difference in full volumetric preference (computed as the proportion of objects that were completed in their volumetrically full forms rather than their hollow forms) between those who were given hollow objects to explore prior to the object completion task and those who were not. Object exploration of four hollow objects transferred to weaker preference for volumetrically full forms in a new set of objects, supporting our hypothesis that the preference for full volumetric object completion are not rigid, built-in priors. Rather, they may be developed through dynamic interactions with objects in the environment across development. Critically, object characteristics such as hollowness detected during exploration can be extracted and applied to the perception of new objects.

In our ongoing study, we are also examining the object completion preferences of children after interacting with full volumetric, symmetrical and asymmetrical objects before drawing a more generalizable conclusion about how object exploration affects object completion preferences. In addition, we are systematically analyzing children’s object exploration behaviors to determine which behaviors are predictive of subsequent object completion preferences.

Authors