Abstract

Encoding time and allocation of attention in analogical development Nina Simms ([email protected]) Department of Psychology, 2029 Sheridan Road Evanston, IL 60202 USA Dedre Gentner ([email protected]) Department of Psychology, 2029 Sheridan Road Evanston, IL 60202 USA Abstract The aim of the current studies was to explore encoding time differences in objects and relations and to investigate whether these differences lead to differences in allocation of attention to object similarity. Using a match-to-sample paradigm with 5- to 6-year-olds and adults, we found that (1) objects were encoded faster than relations for both adults and children, and that (2) children, but not adults, preferentially allocated attention to object similarity. Ultimately, these questions are aimed at identifying the factors responsible for the development of adult-like analogical reasoning. We suggest that changes in selective attention over development may account for the pattern of results seen across these two studies. Keywords: analogical development; relational reasoning; selective attention; encoding time Introduction Reasoning by analogy is a fundamental and powerful aspect of human cognition (Gentner, Holyoak & Kokinov, 2001). Analogical reasoning is based on relational similarity. That is, two situations are analogous if they share a common relational structure (e.g., lava lamps and plate tectonics are both characterized by a system of convection); superficial commonalities like the perceptual features of the objects involved are generally irrelevant. However, reasoning on the basis of relational similarity is not trivial. Two analogous situations may share superficial commonalities that conflict with an alignment based on relational similarity. For example, to appreciate that 1:3 :: 3:9, one must understand that the relationship that holds between 1 and 3 is the same relationship that holds between 3 and 9. Based on this shared relationship, the two smaller numbers in each proportion correspond (1  3) and the two larger numbers correspond (3  9). In this case, the identity match between the two 3s must be disregarded, since this correspondence (3  3) is inconsistent with the overall relational match. Although 1:3 :: 3:9 may not seem like a particularly challenging analogy for adults, instances where relational similarity conflicts with object similarity can be very challenging for young children (Gentner, 1988; Richland, Morrison & Holyoak, 2006) In cases like these, children will often reason on the basis of object similarity rather than relational similarity. This tendency is referred to as the object bias, but over development (with age as well as experience), a relational shift occurs whereby children become increasingly adept at reasoning on the basis of relational rather than object similarity (Gentner & Rattermann, 1991). For example, Gentner and Toupin (1986) gave 6-year-old children a simple story and asked them to reenact it with new characters. They performed well when the corresponding characters were highly similar between the two stories, but performed very badly when similar characters played different roles across the two stories (the cross-mapped condition). Further studies have corroborated this finding that when relational similarity is pitted against object similarity children tend to be highly influenced by object matches and less able to attend to relational matches. For example, Richland and colleagues (2006) found the same pattern of results in a picture-matching task. The pictures depicted the same event structure, and the task was to point out correspondences based on the event patterns. When the object matches were inconsistent with the relational match, younger children were greatly impeded in choosing the correct relational match. This pattern of results, in which object similarity disrupts young children’s analogical reasoning, has been found repeatedly in a variety of analogical tasks (Gentner & Rattermann, 1991) and even across cultures (Richland, Chan, Morrison, & Au, 2010). The object bias is a robust and well-documented phenomenon, but a clear understanding of why it occurs is still lacking. Most accounts of analogical development that address the object bias implicitly or explicitly appeal to some processing difference (or differences) between objects and relations to explain the bias. Some of these differences include representational complexity, familiarity or fluency, salience, and automaticity. Improvements in relational reasoning over development are then explained by changes in this processing difference, and/or by improvements in some additional capacity that tempers the effects of these processing differences. For example, accounts of analogical development that emphasize the role of relational knowledge suggest that children are familiar with more object concepts than relational concepts, but as children gain more relational knowledge, and as this knowledge becomes more fluent, they become able to focus on relational similarity (Gentner, 1988, 2003). Other accounts appeal to the idea that object similarity is more salient than relational similarity, and that improvements in inhibitory control over development allow children to combat the