Integrating Multiple Strategies Efficiently to Solve an Orientation Task

Integrating Multiple Strategies Efficiently to Solve an Orientation Task Glenn Gunzelmann (glenn.gunzelmann@mesa.afmc.af.mil) Air Force Research Laboratory 6030 South Kent Street Mesa, AZ 85212 USA John R. Anderson (ja+@cmu.edu) Department of Psychology Carnegie Mellon University Pittsburgh, PA 15213 USA Murakoshi & Kawai, 2000). Our study is somewhat different than those previous studies in that it does not involve moving through a space, either by real or virtual navigation. Instead, this research looks at performance on an orientation task where participants integrate information from different static representations of a space to make a spatial judgment. In this case, a visual scene and a map of the space were shown to participants. One of the objects in the visual scene was highlighted, and the task was to identify which of the objects on the map corresponded to that highlighted object. A sample trial is shown in Figure 1. The task is described in more detail below. Abstract This research compares the general strategy described by participants doing an orientation task to two strategies described in past research on a different kind of spatial task, perspective- taking (array rotation and viewer rotation). This evaluation indicated that participants were quite flexible and efficient in their approach to the task. The strategy described in participants’ verbal reports made use of both of the perspective- taking strategies within individual trials. In addition, each alternative was applied in situations where previous research indicates that it holds an advantage over the other alternative. This research extends research on strategy use in spatial tasks by (1) showing how similar strategies can be applied to different kinds of spatial tasks and (2) illustrating how alternative strategies can be intermixed within a single task to produce efficient overall performance. Introduction Research on human performance in spatial orientation tasks has focused on the impact of misalignment on solution processes (e.g., Hintzman, O’Dell, & Arndt, 1981; Rieser, 1989; Shepard & Hurwitz, 1984). Other research has examined strategy differences in this area, showing that strategy variation can have important influences on performance (e.g., Gunzelmann & Anderson, 2004a; Huttenlocher & Presson, 1979; Just & Carpenter, 1985; Presson, 1982; Wraga, Creem, & Proffitt, 2000). This research typically uses instructional manipulations to encourage participants to use different strategies. Although this approach has been useful for uncovering differences in performance as a function of strategy use, it also leaves open the question of how strategies are selected by individuals to arrive at the solution. One motivation for this paper is to examine verbal reports of strategy use in an attempt to determine the extent to which efficiency (speed and accuracy) influences strategy selection in individuals solving spatial tasks. Some research in the area of spatial cognition has attempted to identify the strategies individuals used. In many cases, this research has explored human performance on navigation tasks, using map-drawing or other tasks to infer how participants learn and represent routes through a space (e.g., Aginsky, Harris, Rensink, & Beusmans, 1997; Figure 1: Sample trial. Participants click on the object on the map corresponding to the target. To better understand human performance on this task, verbal reports were gathered from participants after they finished. In previous research, we have used these verbal reports to infer the general strategy that the participants were using to solve the task. This strategy is described below. It provides support for a theoretical explanation for participant performance on this kind of task (see Gunzelmann & Anderson, in press). In addition, the predictions of this strategy for performance have been validated against the human data using a computational cognitive model developed in ACT-R (Gunzelmann & Anderson, 2004b). With a validated strategy for performing orientation tasks, it is possible to explore the relationship between it and strategies that have been described for other types of spatial tasks. This kind of comparison has not been performed in the past. In the next section, we briefly describe perspective- taking tasks and two strategies that have been described for doing them, array rotation and viewer rotation (Huttenlocher