The effects of similarity and individual differences on comparison and transfer

The Effects of Similarity and Individual Differences on Comparison and Transfer Samuel Day (day9@indiana.edu) Robert L. Goldstone (rgoldsto@indiana.edu) Dept. of Brain & Behavioral Sciences, 1001 E. 10 th St. Dept. of Brain & Behavioral Sciences, 1001 E. 10 th St. Bloomington, IN 47405 USA Bloomington, IN 47405 USA Thomas Hills (thomas.hills@unibas.ch) Institut fur Psychologie, Missionsstrasse 64A 4055 Basel, Switzerland Results such as these point to the potential power of comparison. The most common explanation for these effects is that structural alignments generated when comparing two concrete examples serve to highlight meaningful structural commonalities between them, while simultaneously taking the focus away from elements that are extraneous or irrelevant (e.g., Markman & Gentner, 2000). This, in theory, allows a more explicit representation of the structure or principle itself, making it easier to recognize when it arises in new situations. However, a great deal remains unknown about the factors that make comparison successful in transfer. Particularly, there is a surprising lack of research on how the relationship between the compared cases (such as their similarity) may influence the representations that are formed during comparison. Given that the similarities and differences between the cases are the basis for the knowledge that comparison is presumed to generate, this would seem to be a critical area for study. For instance, will transfer to new situations be best when the features of the compared cases are relatively similar to one another, or when their content is more dissimilar? There are empirical reasons to predict either of these outcomes. Evidence for “conservative generalization” (Medin & Ross, 1989) suggests that the comparison of two examples that share significant surface commonalities may lead to a representation in which many of these irrelevant features are retained. If so, one of the primary assumed benefits of comparison—a more general representation—may be lost. Comparison of dissimilar cases may therefore lead to representations with broader generalizability. On the other hand, comparisons between overtly similar cases are likely to be less cognitively demanding, and may therefore help to “boot-strap” early learning processes. Consistent with this possibility, Kotovsky and Gentner (1996) found that young children were better able to perform matches on the basis of abstract structural commonalities after performing a similar task involving more perceptually similar stimuli. A related issue is the effect of the similarity of the structures themselves. Most studies focusing on comparison and transfer have made use of cases with essentially identical relational structures. However, there are reasons to suggest that structural variation may be beneficial as well. For instance, some research has shown that comparing two “near-miss” cases (Winston, 1975), which are identical except for a crucial structural difference, may improve transfer (e.g., Gick & Paterson, 1992). This approach may be particularly effective when an individual needs to Abstract Prior research has found that while people are generally quite poor at recognizing when a new situation is structurally similar to a known case, comparison of two analogous cases greatly improves the likelihood of achieving such recognition. Our study examines the effects of varying the similarity between these compared cases, both featurally and structurally. We find that between-case similarity has a significant impact on transfer, and that these effects interact with characteristics of the learner. Introduction Our minds are filled with valuable knowledge that we are unable to use. This is particularly true of what might be the most valuable knowledge of all: general principles that can be applied across a wide range of situations. Research in analogy has repeatedly found that principles that are learned in one context often fail to be retrieved when an individual is confronted with a deeply related situation that differs in concrete or contextual ways (e.g., Gentner, Rattermann & Forbus, 1983; Gick & Holyoak, 1983; Ross, 1984). For example, in Gick and Holyoak’s classic (1980; 1983) analogy studies, individuals attempting to solve an insight problem routinely failed to recognize that the problem was analogous to one they had been taught earlier (unless given an explicit hint), and therefore failed to make use of their relevant knowledge. For both theoretical and practical reasons, researchers are keenly interested in finding ways to overcome this kind of impediment. One approach that has shown great promise is simply asking learners to compare two different examples of a principle (e.g., Gick & Holyoak, 1983; Loewenstein et al, 2003; Gentner et al, 2003; Rittle-Johnson & Star, 2007). For example, Loewenstien and colleagues (2003) conducted research with MBA students enrolled in a course on negotiation. Some of the students compared two specific cases involving a “contingency contract,” a useful but sometimes counterintuitive negotiation technique. Other students received the same two cases, but read and analyzed them separately, without any explicit comparison. The researchers found that students who had compared cases were nearly three times more likely to apply the relevant principle to a new case than those students who had analyzed the cases separately. Consistent with prior findings of poor analogical transfer in general, the students who had read but not compared cases performed no better on the transfer task than those who had received no training.