Through a Child's Eyes

In an article by Yee et al., called Changes in visual object recognition precede the shape bias in early noun learning, researchers explore the idea that our language development, as well as our development in object recognition, are "intertwined". They test their hypothesis on children aged 18-24 months old. To do so the researchers present various objects to the children and asked them to verbally choose an object by its name. The tasks ranged from "basic noun-caricature" recognition where somewhat familiar objects (a flower) were presented and a simple noun response of the word "flower" was required, to a slightly more difficult task of "shape bias", where novel objects were presented instead of familiar ones. The children were grouped according to their level of vocabulary. They found that children who had a more substantial vocabulary were able to more easily name objects that they were familiar with, which backs up their id...

This next article is called Parts and Relations in Young Children's Shape-Based Object Recognition. In this study, Augustine et al., propose that children ages 18-30 months use "separate information about object part shapes/relations" to recognize either familiar or novel items. They test this by seeing if the children were able to successfully group novel objects by shape and by "relation" (functional similarities). in their first experiment. The results stated that children were successful in separating the novel objects by shape by not so much by relation. In contrast, in the second experiment, where children were asked to separate familiar objects using the same paradigm as the first experiment, the children were able to separate the objects both by shape and relation. This phenomenon increased in accuracy as age increased. This backs up the idea that younger children view the world more holistically (being able to separate objects by shape) rat...

In the article Early biases and developmental changes in self-generated object views, researchers go more in-depth into just how early we begin to form a bias in the way we view objects. They examine this by studying children 12-36 months of age (with parent supervision). During the study, they place a mini video camera that is situated onto a headband on the heads of the children to monitor how they held the objects and in what positions they oriented them in. They found that the younger the children (12-18 months) were more disorganized in their viewing and held the object longer which can be identified as a type of information processing. This means they are trying to take in all the visual stimuli in front of them and make some sense of it. Wherein contrast, the older children (19-36 months) had a shorter holding time and they had a more structured form of viewing manipulation and preferred to view the objects in the predicted planar view. Raising the argument that...

In the study by James et al., titled Some views are better than others: evidence for a visual bias in object views self-generated by toddlers, researchers are exploring the differences in how children 18-24 months of age prefer to view and hold objects. They hypothesize that toddlers and adults, both tend to favor a planar view of objects which is said to be "influenced by interesting surface properties and ease of holding". They tested these biases by handing children objects that were either unencased or encased in plexiglass and placed in different orientations within the plexiglass box, one being the planar (3/4) view where the more interesting properties of the object were reported to be seen. The children were also handed the plexiglass box in "randomly chosen orientations" regardless of how the object was positioned in the box. The objects were large enough that the child could easily hold and manipulate. The researchers found that "total holding time...

Now in this article titled A Computational Account of the Development of the Generalization of shape information, the researchers find evidence that says the complete opposite of what the last article tells us about how children learn Object recognition. They claim that younger children view shapes "more metrically, or holistically" and then as they get older they start to view them more in terms of specific categories. The researchers explain their findings using a method of testing known as the "DORA model" which is a "connectionist network that learns structured (symbolic) representations of relations from unstructured (holistic) inputs". In other words, it is a system that can provide accurate simulations of how a child's learning ability changes from novel to relational with age. It does so through the use of analogical mapping to group objects that have a "semantic" or broad sense of shared properties/characteristics. Then as the pr...