The children used real world knowledge to generate and select attributes for categories. Categories (recycle, reuse, throw away) were provided as part of the task design in the Baxter Brown modeling activity. The provision of specific categories afforded categorisation opportunities that differed from those usually accessible to
172 young children in two ways. First, the children did not create categories for
themselves but constructed attributes for categories that were already allocated. This differs from typical categorisation activities for young children where they determine categories for themselves by grouping objects based on the perceived attributes such as colour, shape or size (Hanner, James, & Rohlfing 2002). Second, the allocated categories of recycle, reuse and throw away referred to events or actions, so potential attributes for the categories were not visible or perceivable from the object stimuli provided. The combined effect of these two design features was that children were required to look beyond the perceptual features of the objects in order to generate and select attributes for specific categories. Unless the children had knowledge of the recycling processes for example, there was nothing inherent in an object they had that would support determining attributes for a category it could belong to. Attributes for the assigned categories were not therefore physically perceivable and required the children to look beyond immediately apparent, physical properties to generate and select attributes.
This study found that the children used prior knowledge from various sources and initiated an active search for connections and distinctions in their experiences in order to generate and select attributes for the assigned event categories of recycle, reuse and throw away. Young children’s determination of category attributes has featured in prior data modeling studies. Lehrer and Schauble’s (2000) study of Grade 1 and 2 children (average age 7 years 1 month) examined the attributes developed to assess self-portraits, where perceptual stimuli was available to the children. They found that the children initially gave evaluative comments, and relied on adult assistance to develop more descriptive statements for attributes over several
iterations of model development. In contrast, the findings in this study are that in the absence of perceptual stimuli to support generating category attributes, the children immediately initiated descriptive statements without adult assistance. The children generated and selected complex attributes that were based on and described known actions, such as giving unwanted objects away that could be reused, and the
consequences of actions, such as being able to play again with reusable objects. Other attributes described observed or known interactions with objects that known from actions by family members, such as a child observing his or her mother place
173 cans in the recycling bin at home. Children’s descriptions of attributes made use of different kinds of informational sources including how others actions with objects and how objects are used (Gelman, 2009) in addition to their own observations and knowledge. This confirms that children learn from evidence of direct experiences and indirectly through information provided by other people (Kushnir,
Vrendenburgh, & Schneider, 2013).
The children drew from their knowledge of recycling processes to generate and select attributes using unobservable facts that went beyond first-hand
experiences. These attributes took into account whether an object could be processed and change its form by being made into something different, whether it could be used again in its original form and whether the object would be available for use again, that is, its returnability. The findings support English’s (2010) data modeling research. English (2010) found that children (mean age 6 years 8 months) provided with an event or action category, generated attributes that used unobservable facts, including melting, keeping and composting objects. The perception of objects, particularly correlations and associations between features, can affect children’s construction and use of categories (Hayes & Thompson, 2007). Perception of objects usually reveals consistent information about object structure that is relied on for feature similarity in making conceptual judgments (Goswami & Bryant, 2007). Prior data modeling studies with young children have shown that children’s reliance on resemblance has been an important part of model-based reasoning (Lehrer & Schauble, 2007a).
Real world knowledge children have, such as causal and thematic knowledge, can however affect the way children generalise properties when categorising or classifying objects, and result in generalisations that are not bound by perceptual or feature similarities (Hayes & Thompson, 2007; Hayes & Rehder, 2012). Categorising using an unseen attribute has been described as “a paramount example of inference” (Kruschke, 2005, p. 184). This study’s findings are consistent with research that children are open to and can reason about concepts that are not obvious, without depending on perceptually discernible attributes (Gelman, 2006). Perceptual resemblance in data modeling has been regarded as a bridge between form and function that helps move children “from literal similarity to analogical mapping of
174 systems of relationships” (Lehrer & Schauble, 2007b, p. 155). This study has found that children can move beyond perceptual resemblance and consistently create and apply attributes based on unseen properties. The complex and abstract attributes that the children generated contrast with the view that perceptual features of objects are a significant draw for young children when they need to arbitrate between these and relational properties such as function (Namy & Clepper, 2010). Children’s
demonstrated capacity and ability in exploring the categories meaningfully, and by inducing category properties based on known properties from prior experiences they considered to be relevant to the task at hand. The findings reveal that young
children’s capacity to draw from and meaningfully apply real world knowledge gained from experiences to more novel and complex situations of categorisation and classification is underestimated.
6.3.5 Using Data Context Knowledge to Generate, Select and Measure