Abstract

Spatial Language and Landmark Use: Can 3-, 4-, and 5-year-olds find the Middle? Nina Simms ([email protected]) Department of Psychology, Northwestern University 2029 Sheridan Road, Evanston, IL 60208 USA Dedre Gentner ([email protected]) Department of Psychology, Northwestern University 2029 Sheridan Road, Evanston, IL 60208 USA Abstract in the second scene that played the same role as some object in the first scene. Importantly, when the object from the first scene was also in the second scene but played a different role, young children were likely to pick the identical object rather than the object in the same role in the event. The shift from a focus on perceptual properties and from objects to relations has been termed the relational shift (Gentner, 1988). Many factors have been proposed to contribute to such a shift, including maturational changes in cognitive capacity (Halford, 1993), increases in domain knowledge (Gentner & Rattermann, 1991), and the representational affordances of language (Gentner, 2003). Strikingly, children display difficulty reasoning about relations even in domains, like space, that are accessible to them and in which they are likely to have fairly extensive experience. Loewenstein and Gentner (2005) found that 3- year-olds had trouble mapping the location of an object from one three-tiered box to another, identical box. In another spatial mapping task, 3-year-olds failed to use the spatial relations in a model room to distinguish between two identical objects in the room. They were successful at locating a hidden object only when it was placed at a unique object, but chose randomly between two identical objects in different relative locations (Blades & Cooke, 1994), although having an opportunity to compare two similar models improved their performance dramatically (Loewenstein & Gentner, 2001). However, even very young children are able to use spatial relational information in certain kinds of navigation and search tasks (Bushnell et al., 1995; Newcombe et al., 1998). In particular, children can encode location relative to landmarks. There are three main types of landmark encoding strategies: beacon, vector, and relational (Figure 1). Beacon coding refers to the use of a single landmark as a general marker for location. This kind of encoding roughly corresponds to spatial relations such as at, by, or near. In Figure 1, X would be encoded as lying somewhere proximal to the landmark (e.g. somewhere inside the circle). Vector coding, in contrast, specifies a direction and distance (or vector) from a particular landmark. In this case, distal locations could be encoded relative to the landmark. In Figure 1, this is represented by the arrow from the landmark to X. Finally, relational coding marks location relative to multiple landmarks in an array, rather than to a single Evidence suggests that (a) young children have difficulty reasoning about spatial relations, and that (b) spatial language can facilitate their performance (Loewenstein & Gentner, 2005). This study investigates children’s ability to reason about a particular spatial relation, middle, which we hypothesize may be particularly challenging. We ask when children become able to encode the middle and whether this ability is related to acquisition of the words “middle” and “between.” Finally, we explored the errors children make when reasoning about middle. We gave 3-, 4-, and 5-year-old children a search task in which the hidden object was always in the middle of two landmarks (Midpoint Task), followed by a language task assessing their understanding of the spatial relational terms “middle” and “between.” Children’s accuracy on the Midpoint search task increased with age; and, more interestingly, increased with knowledge of the relevant words. Keywords: Relational reasoning; language; spatial relations; cognitive development. Introduction The ability to recognize and reason about relations is critical to the human capacity for higher order cognition. In fact, it is so powerful and so fundamental that it has been offered as a major part of the reason why “we’re so smart” (Gentner, 2003). Despite adults’ fluency with relational concepts, young children have repeatedly demonstrated difficulty with tasks that require them to focus on relations, especially in those instances where they must ignore or abstract across the identity or perceptual properties of the entities involved in the relations. For example, Christie and Gentner (2007) gave 4½- and 8½-year-olds a simple relational task that has been used in comparative studies (e.g., Thompson, Oden & Boysen, 1997). Children were shown a standard with two identical objects (e.g. two squares) and were asked to match it to one of two choices: a relational match, which had two identical new objects (e.g. two triangles) or an object match, which had two different objects, one of which was identical to the standard’s objects (e.g. a circle and a square). The 4½-year-olds significantly preferred the object match over the relational match, and even the 8½-year-olds failed to show a reliable preference for the relational match. Richland, Morrison, and Holyoak (2006) demonstrated a similar phenomenon using line drawings of event scenes. In their study they showed children two scenes, and then asked them to find an object