Comparison and the development of knowledge

Abstract

Comparison and the Development of Knowledge Lera Boroditsky (lera@mit.edu) NE20-456, MIT, 77 Mass Ave Cambridge, MA 02139 USA Abstract This paper considers the role of comparison in the development of knowledge. Results show that comparing similar objects makes them appear more similar, while comparing dissimilar objects makes them appear less similar. The effect of comparison on similar items was especially striking since subjects judged items to be more similar after comparison even if the comparison task was to list differences between the two items. Further, this effect appears specific to comparison and does not appear to be simply due to a ifleshing outi of object representations (listing properties of two objects without comparing the objects themselves served to increase the objectsi similarity regardless of whether the objects were similar or dissimilar to start). This suggests that comparison may play a special role in partitioning bits of experience into categories, sharpening categorical boundaries, and otherwise helping us create conceptual structure above and beyond that offered by the world. Introduction Are our mental representations of things in the world simply a reflection of the structure of the world, or do we create new structures and partitions in conceptual space? Further, are our representations static, or do they change over time in systematic ways as a result of the way we process and use our knowledge? This paper suggests that some common cognitive processes (in this case, comparison) can introduce systematic biases into our representations of the world. These biases may be beneficial for separating out bits of experience into categories, sharpening categorical boundaries, and otherwise helping us create conceptual structure above and beyond that offered by the world. This paper focuses on object similarity. Similarity is a central construct in explanations of cognition. Explanations of categorization, induction, learning, and memory all rely on the construct of similarity. Things that are similar are likely to end up in the same categories, are likely to support inductive inferences for each other, will aid in the learning of other similar things, and serve as good reminders for one another in memory. But where do similarities come from? Are similarities between objects apprehended immediately and automatically, or do they develop as a function of directed processing and experience? This question has been taken up seriously in the study of categorization in the following form: Why do categories appear to contain similar things? Is it because similar things tend to end up in the same categories, or is it that putting two things in the same category makes them appear more similar? Previous research suggests that both are true. For example, Goldstone, Lippa, and Schiffrin, (2001) showed that object representations can change as a result of category-learning, with objects assigned to the same categories becoming more similar (see also Kurtz, 1998). Further, previous research by Gentner and Namy (2000) suggests that providing children with an opportunity for comparison may help them in category learning by allowing them to discover deeper relational similarities between category members (see also Kurtz & Gentner, 1998). This paper considers the role of comparison in the development of similarity. Results of four experiments suggest that comparison can play an important role in knowledge development. By making similar things appear more similar, and dissimilar things appear less similar comparison may help us partition bits of experience into categories and sharpen categorical boundaries. Four experiments explore the effects of comparison on object representation. Experiments 1 and 2 examine the effects of comparison on the perceived similarity of similar and dissimilar objects. Experiment 3 contrasts the effects of comparison with those of simple ifleshing outi or elaboration of object representations. Experiment 4 extends the findings of Experiments 1and 2 to novel objects. Experiment 1 Method Participants 132 Stanford University undergraduates participated in the study in order to fulfill a course requirement. Materials Materials consisted of a one-page questionnaire. The top of the page contained line-drawings of 4 named familiar animals (a deer, a horse, a goat, and a donkey) as shown in Figure 1a. The rest of the page contained three questions. For 73 participants, the first question asked them to describe three similarities between two of the animals (e.g., Please describe 3 similarities between the goat and the donkey. ) For the other 59 participants, the first question asked them to describe three differences between two of the animals (e.g., Please describe 3 differences between the goat and the donkey. ) Participants were given three blank lines for their responses. Which two animals were chosen for comparison was counterbalanced across subjects such that each pair of adjacent shapes was the focus of comparison equally often. Which animal was named first in the comparison was also counterbalanced across subjects.